T cells chronically stimulated with the same peptide tend to express exhaustion markers such as PD-1 or LAG-3. Deficiencies in the PD-1 and LAG-3 pathways have been linked to the development of autoimmune diseases. IMP761 is a LAG-3–specific humanized agonist Ab with immunosuppressive properties both in vitro and in vivo in an Ag-specific delayed-type hypersensitivity (DTH) model in the cynomolgus macaque (Macaca fascicularis). IMP761 inhibits TCR-mediated NFAT activation and Ag-induced human T cell proliferation and activation. In the DTH model, assessment of T cell infiltration and gene expression profile at the DTH biopsy site corresponds to immunosuppression of an Ag-induced T cell response. IMP761 is the first LAG-3–specific agonist product candidate, acting upstream on activated T cells, the root cause of self-Ag–specific T cell–induced autoimmune diseases.
T cell–mediated immune responses are regulated by coinhibitory signals, which are critical to maintain self-tolerance and prevent self-harm by controlling the magnitude of the response to foreign Ags. Notable examples of these inhibitory immune checkpoints are CTLA-4, PD-1, and LAG-3. In the context of cancer or infectious diseases, these molecules are often upregulated as a mechanism of immune evasion. Countering the negative regulators of T cell function via the inhibition of the immune checkpoint pathways has led to the development of several approved drugs, such as anti–CTLA-4 and anti–PD-1 mAb in immuno-oncology.
LAG-3 is expressed on the surface of activated T cells and NK cells; it binds to MHC class II molecules and regulates T cell proliferation and homeostasis of both effector T cells and T regulatory cells (1–3). Blocking PD-1 and LAG-3 pathways has been shown to have synergistic effects against tumoral immune escape (4), and many clinical studies investigating combinatory blockade of PD-1 and LAG-3 are ongoing (5). Conversely, a deficiency in inhibitory immune checkpoint signaling can lead to the development of autoimmune diseases (6) as illustrated by CTLA-4– and PD-1–deficient mice that develop autoimmune disorders (AID) (7, 8). In line with this observation, some patients with advanced cancer treated with immune checkpoint inhibitors suffer from induced autoimmunity directed at various organs (9). Achieving selective immunosuppression of activated T cells by triggering such inhibitory pathways could clinically be effective in AID.
Evidence for LAG-3 involvement in the development of AID such as colitis (10), rheumatoid arthritis (11), or diabetes (12) has been reported. Although single LAG-3– and PD-1–deficient mice display minimal immunopathologic sequelae, double LAG-3/PD-1 knockout mice develop lethal systemic autoimmunity (4). A cytotoxic LAG-3 Ab has been evaluated in a nonhuman primate (NHP) model of delayed-type hypersensitivity (DTH) (13). The depletion of LAG-3–positive T cells was linked to a reduced Th1-driven skin inflammation, and the effect lasted several months after elimination of the Ab. The results of a phase I clinical trial investigating this approach in psoriasis has been reported recently (14), and a large randomized phase II started in ulcerative colitis (NCT03893565). An alternative approach involving this pathway could be the use of an LAG-3 agonist that inhibits T cell proliferation and function without depleting LAG-3–positive T cells. This innovative strategy could potentially inhibit activated T cells without eliminating them and should preserve the ability of the immune system to defend the host against pathogens.
Choosing the right animal model is critical because they often do not mirror the human AID setting. NHP models are thought to be superior because of their genetic and physiological similarity to humans, although NHP models of AID are scarce (15). The use of a minimally invasive model of DTH to tuberculin or other foreign Ags in an NHP model to investigate cell-mediated responses has been reported (13, 16). This concept of Ag-specific DTH in vivo testing has even been extended to clinical trials to select at an early stage novel T cell immunosuppressive agents for AID (17).
We describe in this study the in vitro and in vivo immunosuppressive properties of a first-in-class humanized LAG-3 agonist Ab, termed IMP761. IMP761 inhibits TCR-mediated NFAT activation and Ag-induced T cell proliferation and activation, therefore acting upstream on activated T cells (the root cause of self-Ag–specific T cell–induced AID) and not downstream like present therapies (e.g., anti-TNF mAbs).
Materials and Methods
Agonist anti–LAG-3 generation
BALB/c mice (Charles River Laboratories) were immunized with s.c. injection of clinical-grade LAG-3Ig (IMP321) without adjuvant and received an i.v. boost of LAG-3 D1–D4 recombinant protein without Fc region. B cell hybridomas were generated by fusing mouse splenocytes to Sp2/0 myeloma cells using PEG-1500 (polyethylene glycol solution) (Roche). The hybridomas were screened for LAG-3 binding and agonist activity, and 13E2 was the selected hybridoma clone. The CDR from the V region of the 13E2 L chain and H chain were grafted into the human κ L chain and to the human IgG4 isotype with an S228P mutation to abolish Fab arm exchange (Fusion Antibodies). The resulting humanized Ab was called IMP761 and produced in CHO cells in suspension culture (Thermo Fisher Scientific).
LAG-3 binding affinity assay
Binding affinity assay was performed by Biaffin. IMP761 was covalently immobilized to a Biacore C1 sensor chip (GE Healthcare). LAG-3Ig was then passed over the captured Abs at six different concentrations (0.078–2.5 nM) in analysis buffer at 25°C with regeneration in every cycle. The binding of recombinant human LAG-3Ig protein (IMP321) to the captured Abs was analyzed on Biacore T200 (GE Healthcare), and the data were fitted using the kinetic global fit (Langmuir 1:1) model.
Primary T cell LAG-3 binding assay
Human blood from healthy volunteers came from the French National Blood Bank (Etablissement Français du Sang). All participants gave their informed consent.
PBMC were incubated with 0.5 μg/ml staphylococcal enterotoxin B (Toxin Technology) for 2 d. At the end of the incubation, PBMC were washed and incubated with different concentrations of IMP761. Cell-bound IMP761 was revealed by an FITC-conjugated goat F(ab′)2-anti-human Ig (H+L) (SouthernBiotech). PBMCs were phenotyped using CD14 allophycocyanin (MφP9), CD8 allophycocyanin-cyanine (Cy) 7 (SK1), and CD4 PerCP Cy 5.5 (L200) (all from BD Biosciences) before flow cytometry analysis.
PBMC proliferation assay
Cryopreserved PBMC were labeled with CFSE (Thermo Fisher Scientific) and incubated with 125 ng/ml CMV, Epstein–Barr, and influenza virus (CEF) peptide pool (Miltenyi Biotec) in the presence of different concentrations of IMP761. After 6 d of culture, PBMCs were stained with anti-CD4 PerCP Cy5.5 (L200), CD8 allophycocyanin Cy7 (SK1), CD14 allophycocyanin (MφP9), and CD25 PE Cy7 (M-A251) (all from BD Biosciences) before flow cytometry analysis.
LAG-3+/NFAT luciferase assay
A Jurkat T cell line expressing LAG-3 and containing the luciferase gene under the control of NFAT promoter (NFAT-luc2) (Promega) was activated using a suboptimal dose of anti-CD3 (OKT3) (eBioscience) in the presence of different concentrations of IMP761 for 24 h. Luciferase activity was then quantified using an Envision Multilabel reader (PerkinElmer) after addition of Bio-Glo reagent (Promega).
Ab-dependent cellular cytotoxicity bioassay
The Antibody-Dependent Cell-Mediated Cytotoxicity Reporter Bioassay (Promega) uses engineered Jurkat cells stably expressing the FcγRIIIa receptor and an NFAT response element driving expression of firefly luciferase as effector cells. Effector cells and CHO cells expressing high levels of LAG-3 were incubated in presence of IMP761 or IgG4 isotype control for 6 h before addition of Bio-Glo reagent and luciferase activity quantification. Raji B cells incubated with rituximab (anti-CD20) were used as positive control.
Cellular calcium concentration assay
Cellular calcium concentration was monitored using the Fluo-4 NW Calcium Assay Kit (Thermo Fisher Scientific). The dye solution was added to LAG-3–expressing Jurkat cells and incubated for 30 min at 37°C and 30 min at room temperature. Mean fluorescence intensity of cells activated with suboptimal concentration (3 or 10 ng/ml) of anti-CD3 Ab (OKT3) (eBioscience) in presence of IMP761 or IgG4 isotype control (600 ng/ml) was quantified by flow cytometry.
Animal studies were performed by Cynbiose (Ministry of Higher Education and Research of France approbation number 2016010518437560). A total of 18 male cynomolgus macaque (Macaca fascicularis) were used in this study. Animals were housed in two different enclosures based on supplier recommendations (socializable group). All animals received two bacillus Calmette–Guérin (BCG) vaccines (Biomed-Lublin) before being challenged by 40 IU of Bovituber-purified protein derivative (PPD) (Synbiotics) intradermal injection in the back. The two BCG vaccinations were injected at 2-wk intervals with the first vaccination occurring 6 or 7 wk prior to IMP761 injection. The first PPD challenge started 2 wk after the second BCG vaccination, and the second PPD challenge started 1 d after IMP761 injection. Blood samples were collected for serum isolation and measurement of circulating IMP761 concentration. At day 0, 12 animals received s.c. injection of IMP761 (six at 0.03 mg/kg, six at 0.3 mg/kg), and six animals received PBS as control. Skin biopsies were performed at the PPD injection site using 6-mm punch biopsy before and after IMP761/PBS injection, with OCT embedded and stored at −80°C for downstream immunofluorescence staining and gene expression profiling.
Interspecies cross-reactivity of IMP761
IMP761 was conjugated to PE-Cy7 fluorochrome using the Lightning Link Conjugation Kit (Novus Biologicals). Con A (Sigma-Aldrich) at 5 μg/ml was used to stimulate PBMCs incubated with different concentrations of IMP761-PE-Cy7, then stained with CD8 FITC (SK1) (BD Biosciences) and CD25 allophycocyanin (CD25-3G10) (Thermo Fisher Scientific) before flow cytometry analysis.
Measurement of IMP761 circulating concentration and antidrug Ab detection by ELISA
For IMP761 concentration measurement, Maxisorp (Nunc) plates were first coated with LAG-3Ig (IMP321) in carbonate-bicarbonate buffer and blocked with Prime Blocking Reagent (GE Healthcare Life Sciences) in PBS (Life Technologies) 0.1% Tween-20 (Sigma-Aldrich). Serum samples or IMP761 for standard were diluted in PBS Tween BSA and incubated on the plates. Anti–huIgG4-HRP (Abcam) and 3,3′,5,5′-tetramethylbenzidine substrate (BD Biosciences) were used for assay detection at 600–450 nm.
For antidrug Ab detection, plates were coated with IMP761. After protein blocking using BSA (GE Healthcare), the diluted animal serum samples were incubated. An antirhesus monkey IgG H chain-HRP (Abcam) detection Ab was used followed by tetramethylbenzidine substrate (BD Biosciences) revelation.
Impact of IMP761 on peripheral T cell subsets
Frozen cynomolgus macaque PBMC were used to monitor CD4+ and CD8+ T cell subsets longitudinally from the 0.3 mg/kg IMP761 group and the PBS control group before (day −6) and 24 h after IMP761/PBS injection (day 1). Cells were labeled with LIVE/DEAD Fixable Yellow Dead Cell Stain Kit (Thermo Fisher Scientific), then stained with anti-CD3 FITC (SP34) (BD Biosciences), anti-CD4 allophycocyanin-Cy7 (SK3) (BD Biosciences), anti-CD8 PerCP-Cy5.5 (SK1) (BD Biosciences), anti-CD25 BV421 (2A3) (BD Biosciences), anti-CD45RA PE-Cy7 (L48) (BD Biosciences), anti-CD127 allophycocyanin (MB15-18C9) (Miltenyi Biotec), and anti-CCR7 PE (G043H7) (BioLegend) before flow cytometry analysis.
Evaluation of the erythema response
Erythema diameter was measured daily on PPD injection sites at 24 h postchallenge up to erythema disappearance. An erythema was considered positive if its diameter was superior to 4 mm. To compare responses of animals to challenges, the area under the curve (AUC) of longitudinal erythema measurement was calculated using the trapezoidal approach.
Immunofluorescence staining of infiltrating cells
Slices of 7-μm thickness were performed using the OCT-embedded skin biopsies. Samples were incubated with the following primary Abs: rabbit anti-human CD3 (polyclonal) (DAKO), mouse anti-human CD4 (13B8.2) (Beckman Coulter), or mouse anti-human CD8 (B9.11) (Beckman Coulter). Then they were incubated with an Alexa Fluor 488 goat anti-rabbit or mouse IgG (H+L) Abs (Thermo Fisher Scientific). Evans blue counterstaining was carried out before mounting slices in SlowFade Gold Antifade Reagent with DAPI (Thermo Fisher Scientific) to visualize the nuclei. The s.c. inflammatory reaction for each marker was evaluated by blinded pathologists by double reading, based on the surface coverage of the tissue and according to the following grading scale: 0, no inflammation; 1, limited inflammation (<10% of the surface); 2, moderate inflammation (10% ≤ i < 30% of the surface); 3, significant inflammation (30% ≤ i < 50% of the surface); and 4, severe inflammation (i ≥ 50%).
Principal component analysis of tuberculin test readouts
The principal component analysis was performed by Cynbiose with R (V 1.7) and RStudio (V 1.1.442) software using the “prcomp” function. The figure was generated using the “ggbiplot” package, and the statistical analysis was conducted using the “vegan” package (rda and Adonis functions). Permutational multivariate ANOVA (PERMANOVA) was applied to Euclidean distances.
Gene expression profiling of skin biopsies
RNA from OCT-embedded skin biopsies was extracted using the RNeasy Fibrous Tissue Mini kit (Qiagen) after tissue homogenization on a Precellys Evolution with a mix of 2.8- and 5.0-mm ceramic beads (Bertin Instruments). RNA quality was assessed using the Agilent RNA 6000 Nano Kit, and RNA concentration was quantified on a Qubit fluorometer (Thermo Fisher Scientific). Transcriptional profiling of skin biopsies was established using the digital multiplexed nCounter NHP Immunology Kit containing 770 immune response genes (NanoString Technologies) at the Center for Translational Science and Cytometry and Biomarkers Unit of Technology and Service of the Institut Pasteur. Normalizations, quality controls, and fold change calculation were performed using nSolver software (NanoString Technologies). Housekeeping genes were selected by the GeNorm algorithm. Unsupervised analysis was performed on normalized data using principal component analysis, and differentially expressed genes were identified by two-group comparison (t test) (Qlucore Omics Explorer). Because this study is exploratory, no multiple test correction was performed. Protein–protein association and functions were investigated using the STRING database.
Except for unsupervised principal component analysis and PERMANOVA (described above), all statistical analyses were performed using Prism 7 (GraphPad Software). The value of p > 0.05 was considered not significant.
IMP761 binds to LAG-3–expressing T cells and has immunosuppressive properties
Clinical-grade LAG-3Ig fusion protein (IMP321) was used to immunize mice and generate anti–LAG-3 mAbs. Screening of LAG-3–specific mAbs in functional assays led to the identification of 13E2, an immunosuppressive murine hybridoma. A humanized version of 13E2, called IMP761, was then developed. Binding affinity assay using a Biacore system revealed that IMP761 had a very fast association rate (ka: 2.9 ± 0.2 × 107 M−1s−1), leading to a high affinity (KD: 22.8 ± 0.9 × 10−12 M) to LAG-3 (Supplemental Fig. 1). The capacity of IMP761 to bind to activated primary human T cells expressing LAG-3 was tested. As expected, the fluorescence intensity of IMP761 was higher on CD8+ than on CD4+ T cells (2, 18) (Fig. 1A left). The average EC50 for CD4+ and CD8+ activated T cells was 19 and 34 ng/ml, respectively. Fig. 1A (right) shows examples of IMP761 binding on activated CD4+ and CD8+ T cells as monitored by flow cytometry staining.
The immunosuppressive properties of IMP761 were tested in a CFSE dilution assay following antigenic stimulation. Even at low concentration, IMP761 was able to inhibit the Ag-induced proliferation of MHC class I–restricted CEF peptide pool–activated T cells (Fig. 1B, upper panel, left) with an IC50 of 13 ng/ml. Inhibition of this CD8+ T cell proliferation in response to this strong (i.e., nonself) virus Ag pool was variable among donors with a mean inhibition of 50% (Fig. 1B, upper panel, center and right). Whether a higher level of inhibition could be achieved in response to weak (i.e., self) Ags is not known. Inhibition of CD8+ T cell activation by IMP761, as monitored by CD25 expression, was also dose dependent with an IC50 of 12 ng/ml (Fig. 1B, lower panel left) and a mean inhibition of 38% for different donors (Fig. 1B, lower panel, center and right). There was a strong positive correlation between proliferation inhibition and activation inhibition (p < 0.01, r = 0.93, Supplemental Fig. 2A). Although MHC class I–restricted peptides were used for the proliferation experiments, four healthy donors displayed proliferative CD4+ T cell responses to these peptides. In all cases, there was a decrease in CD4+ T cell proliferation when the cells were incubated with IMP761, suggesting that the agonist anti–LAG-3 Ab may suppress CD4+ T cell proliferation as well (Supplemental Fig. 2B). The absence of the cytotoxic effect of IMP761 was confirmed in a robust and sensitive Ab-dependent cell-mediated cytotoxicity reporter bioassay (Supplemental Fig. 2C) (19).
IMP761 inhibits TCR-induced NFAT activation in an LAG-3–expressing cell line
Early studies reported that LAG-3 colocalizes with CD8 and CD4 upon TCR engagement and negatively regulates the CD3/TCR activation pathway (20–22). During TCR stimulation, LAG-3 cross-linking increases the calcium flux (21), and blocking LAG-3 with an antagonist Ab leads to the activation of NFAT (23). Having described the inhibitory effects of IMP761 on Ag-stimulated primary T cells, we next thought to investigate the more-direct action of IMP761 on a Jurkat T cell line expressing LAG-3 at the surface and containing the luciferase gene under the control of NFAT promoter (NFAT-luc2). The luciferase activity of the cell line is used to measure cell activation induced by a low dose of an anti-CD3 Ab. IMP761 can inhibit TCR signaling as measured by luciferase activity (Fig. 2A), and this inhibition is dose dependent with a mean IC50 of 39 ng/ml and a mean inhibition of 70% (p < 0.0001) (Fig. 2B). Our results show that IMP761 is a biologically active LAG-3 agonist Ab in vitro. Calcium cellular concentration monitoring by flow cytometry of anti-CD3–activated LAG-3–expressing Jurkat cells in the presence or not of IMP761 was also performed (Supplemental Fig. 2D). In this setting, no impact of the anti–LAG-3 agonist Ab on the Jurkat TCR-induced calcium flux was observed, suggesting that immediate LAG-3 engagement by IMP761 does not impair TCR-induced calcium flux, an early event in T cell activation.
IMP761 is well tolerated by NHP, and potentially active concentrations were achieved
Cross-species reactivity of IMP761 was confirmed by monitoring binding of IMP761 to LAG-3–expressing activated CD8+ T cells from cynomolgus macaques and humans (Fig. 3). We have previously used the tuberculin DTH approach in a baboon model to test the immunosuppressive properties of an immunosuppressive LAG-3–specific depleting mAb (13). The same methodology was used in this study for cynomolgus macaques. Eighteen animals were included in this study. They were all vaccinated twice intradermally with BCG vaccine and challenged twice intradermally with tuberculin- PPD at 40 IU (Fig. 4A). A low dose of PPD was chosen (40 IU) because it is regarded as being more physiologically relevant than a higher dose [2000 IU as used in the DTH baboon study (13)], which may induce unspecific inflammation. One day before the second tuberculin-PPD challenge, six animals were injected s.c. with 0.03 mg/kg IMP761, six animals were injected with 0.3 mg/kg IMP761, and six animals were injected with PBS (control). Skin biopsies were performed to evaluate cell infiltrate at intradermal reaction (IDR) sites before (IDR1) and after (IDR2) IMP761/PBS injection. Circulating IMP761 concentration was monitored by ELISA. Median levels of IMP761 concentration at 24 h postadministration were 165.6 ng/ml (interquartile range: 123.7–195.4 ng/ml) and 1367 ng/ml (interquartile range: 1105–2062 ng/ml) for the 0.03 and 0.3 mg/kg injected groups, respectively (Fig. 4B). These observed IMP761 concentrations are above the potentially pharmacodynamically active range (i.e., well above in vitro IMP761 IC50, as shown in Figs. 1, 2). Immunogenicity of mAbs in NHP models is a major challenge for drug development, as 93% of mAbs were found to be immunogenic in NHP (24). Although IMP761 antidrug Abs were detected 2 wk after IMP761 injection, no antidrug Abs were detected at the time of IDR2 for either of the IMP761 doses (Supplemental Fig. 3A), implying that IMP761 immunogenicity did not interfere with the outcome of the experiment. No abnormal clinical or biological adverse events were reported following IMP761 administration.
The systemic impact of IMP761 was investigated on different CD4+ and CD8+ T cell subsets using frozen PBMC from the 0.3 mg/kg IMP761 group and the PBS control group before (day −6) and 24 h after (day 1) injection (Supplemental Fig. 3B, 3C). No significant population change was monitored, showing that IMP761 does not have a systemic effect on the different circulating T cell subsets after injection. These results are in line with the fact that LAG-3–positive cells are extremely rare in the peripheral blood of the animals. Additionally, the DTH model used in this study is linked to a local skin inflammation, and a low-dose intradermal tuberculin injection is unlikely to have a detectable impact (e.g., induction of immune checkpoints like LAG-3) on the different circulating T cell subsets.
IMP761 inhibits tuberculin-induced inflammatory T cell infiltration at the DTH site
In immunized animals, tuberculin skin testing induces an erythema characterized by inflammatory T cell infiltration. Because of the waning of the tuberculin-specific response, erythema size at IDR2 was consistently lower than IDR1 for both PBS- and IMP761-injected groups (Fig. 4C, 4D, Supplemental Table I). In terms of longitudinal erythema size (IDR2 versus IDR1), there was a stronger erythema size inhibition for the animals that received IMP761 compared with the animals that received PBS, but statistical significance was not reached (p = 0.19 for the 0.3 mg/kg group, Fig. 4D).
To better compare T cell infiltration at IDR1 (before injection of IMP761/PBS) and IDR2 (after injection), markers of T cells (CD3, CD4, CD8) were assessed by immunofluorescence staining in skin biopsies of animals that received IMP761 (0.03 or 0.3 mg/kg) or PBS (Fig. 5A). The degree of T cell infiltration was estimated by blinded pathologists. To estimate the impact of IMP761 treatment on T cell infiltration, we compared the percentage of inflammatory-grade inhibition for the T cell markers (IDR2 versus IDR1) between the two animal groups injected with IMP761 at the lower and higher dose and the PBS-injected control group. There was significant inhibition of CD3+ cell infiltration in both IMP761-injected groups compared with the PBS group (Fig. 5B). Only the higher IMP761 dose was able to decrease CD8+ cell infiltration compared with PBS. CD4+ cell infiltration was not significantly affected by IMP761 (Fig. 5B). Because the inhibitory effect of LAG-3 positively correlates with the cell surface amount of LAG-3 (25), this absence of detectable effect on CD4+ cells might be related to the lower LAG-3 expression on this cell subset (2) and also to the variability of CD4+ cell infiltration in the control group. To integrate the different intradermal response readouts (erythema size and CD3+, CD4+, and CD8+ T cell infiltration), a principal component analysis was performed from the multivariate data. The unsupervised analysis shows two clearly distinct and statistically different (PERMANOVA p < 0.004) clusters, one for the group injected with PBS and another one for the group injected with IMP761 at 0.3 mg/kg (Fig. 6).
In the 0.3 mg/kg IMP761–injected group, four out of six animals showed a reduced erythema as compared with the PBS group (Fig. 4D). In the controller group, the animal with the highest IMP761 serum concentration at 24 h postinjection (2244 ng/ml, 1BJG15) showed the highest inhibition of T cell skin infiltration for two markers (−66.7% for both CD4 and CD8). Another animal with a relatively high IMP761 serum concentration (1624 ng/ml, 2BKG15) showed the highest inhibition of CD3+ T cell infiltration (−66.7%). In contrast, the only animal that did not display a CD8+ T cell infiltration decrease (2BHW15) is one of the two nonresponder animals. The difference between the responder and the nonresponder animals is illustrated by the principal component analysis (Fig. 6) in which the four responder animals (1BJF15, 1BJG15, 1BLM15, 2BKG15) form a subcluster of the 0.3 mg/kg IMP761–injected group.
IMP761 inhibits tuberculin-induced inflammatory gene signature at the DTH site
To further assess the effect of IMP761 on this Ag-specific DTH reaction, we analyzed the skin biopsies at IDR1 (before injection of IMP761/PBS) and IDR2 (after injection) using an ultrasensitive multiplexed gene expression assay. We focused on the skin biopsies of animals that received IMP761 at a high dose or PBS. Unsupervised principal component analysis revealed that IMP761-injected animals did not cluster with PBS-injected animals (Fig. 7A), demonstrating that IMP761 can change the immune gene expression profile at the tuberculin skin test site. Supervised analysis identified eight differentially expressed genes between the two groups (Fig. 7B). Gene ontology analysis revealed that among the seven genes that were downregulated after IMP761 injection, five were linked to the inflammatory response biological process (CCL20, IL-5, MIF, PLA2G2E, TICAM1). The gene encoding PD-1 (PDCD1) was also downregulated in the IMP761 group. PD-1 is expressed by activated T cells, NK cells, B cells, macrophages, and subsets of dendritic cells (26, 27) and can be considered as an immune activation marker following activation of the IFN-γ cascade. Finally, ANP32B was upregulated in the IMP761 group and is known to be important in dampening inflammation in an autoimmune encephalomyelitis mouse model (28). These results are in line with our assessment of a decreased inflammatory T cell infiltration in the IMP761 group. Taken together, these results clearly show that IMP761 at 0.3 mg/kg was able to significantly decrease a foreign Ag-specific IDR in this NHP animal model.
We aimed to develop a therapeutic approach focused on the root cause of autoimmune diseases by targeting the locally “exhausted” self-peptide–specific memory T cells expressing LAG-3. We produced IMP761, an agonistic LAG-3–specific Ab to downmodulate self-peptide–induced TCR signaling. IMP761 was able to inhibit peptide-induced T cell proliferation, activation of human primary T cells, and an Ag-specific DTH reaction in an NHP model.
The basic notion that T cells express exhaustion markers comes from chronic viral infection experiments in mice (29, 30). T cell exhaustion has been reported in human chronic viral infections, such as HIV, hepatitis B virus, and hepatitis C virus (31, 32), but also in malaria (33) and tumors (34, 35). It is not known whether this general mechanism leading to downregulation of T cells may also apply to some extent to self-Ag recognition in AID.
Expression of LAG-3 as a potential therapeutic target expressed by human T cells has mainly been reported in tumor tissues (2, 36), and testing of several blocking LAG-3–specific Abs is ongoing in immuno-oncology. As a coinhibitory receptor, LAG-3 expression identifies lymphocytes that may contribute to initiation and persistence of inflammation in patients with AID. In ulcerative colitis, the frequency of LAG-3+ cells in peripheral blood was very low (<0.5%), but in the colonic lamina propria, the frequency of LAG-3+ T cells was markedly increased compared with uninflamed controls (37). LAG-3 was expressed mainly on effector memory and CD161+ T cells, and these mucosal LAG-3+ T cells produced robust levels of IFN-γ and IL-17A (i.e., no sign of overt exhaustion) but less IL-10 when stimulated ex vivo compared with LAG-3− cells (37). This report on LAG-3+ T cells in the ulcerative colitis mucosa provides some evidence that the same mechanisms are at play at both a tumor site (i.e., chronic stimulation of the TCR by the same tumor peptide, leading to LAG-3 expression) and at an AID inflamed site (i.e., chronic stimulation of the TCR by the same self-Ag peptide). In addition, the majority of FOXP3+ T cells did not express LAG-3 (38), supporting the therapeutic potential of targeting LAG-3 for AID.
To investigate the impact of IMP761 on an Ag-induced T cell response in the skin, we used a DTH model consisting of two BCG vaccinations followed by two tuberculin challenges. This model was initially developed in the baboon, and the reaction to this strong (i.e., nonself) infectious disease Ag induced an erythema characterized by memory T cells and macrophages. It can be considered as a surrogate model for psoriasis inflammation (39). In the cynomolgus macaque, we witnessed a faster waning of the tuberculin-specific response with IDR2 consistently lower than IDR1 in all animal groups, suggesting that the targeted peptide-specific memory T cell pool declines faster in this model than in the baboon. Possibly because of the decline of DTH reaction in all animal groups, the decrease in erythema size following IMP761 treatment did not reach statistical significance, in contrast to the results obtained with the depleting anti–LAG-3 Ab in the baboon (39). Nonetheless, 50% of the animals that received IMP761 displayed complete disappearance of their erythema, whereas no animal displayed complete inhibition in the control group. Although the latter observation did not reach statistical significance, IMP761 significantly decreased CD3+ and CD8+ T cell infiltration in the DTH skin biopsy, and the results of a gene expression profiling study also supported this immunosuppressive effect.
LAG-3 binds to a restricted subset of MHC class II molecules located in membrane lipid raft microdomains (40, 41). Recently, Maruhashi et al. (42) provided new insights into the interaction between MHC class II and LAG-3 by describing how LAG-3 selectively recognizes stable complexes of peptide–MHC class II (pMHCII). Their work suggests that LAG-3 suppresses autoimmunity by inhibiting autoreactive T cells that recognize stable autoantigen–pMHCII complexes but somehow escaped thymic negative selection. Therefore, in the presence of unstable autoantigen–pMHCII complex, the LAG-3 inhibitory pathway would be neutralized, leading to autoimmunity. In this context, it remains to be investigated if IMP761 could increase LAG-3 clustering and signal transduction leading to stabilization of LAG-3/TCR complexes, possibly decreasing autoimmunity.
A PD-1 agonist Ab (CC-90006) is currently in a proof-of-concept phase I clinical study in psoriasis patients (NCT03337022). The mechanisms of PD-1 signaling in T cells are well characterized (43), but the transduction mechanisms of LAG-3 are still elusive. In addition to the KIEELE sequence in the middle of the LAG-3 intracellular region (44) and the C-terminal EP repeat binding to the LAG-3–associated protein (LAP) (45), an FXXL motif in the membrane-proximal region has been reported (25). These three motifs in the LAG-3 intracellular region have not been reported for other inhibitory coreceptors before, indicating that LAG-3 inhibits T cell activation using nonredundant inhibitory mechanisms with the other inhibitory coreceptors.
Triggering LAG-3 inhibitory immune checkpoint signaling in LAG-3–expressing self-peptide–specific memory T cells via an agonistic Ab represents a promising novel and more-targeted therapeutic approach for treating AID.
We thank Benjamin Pelletier, Emmanuelle Pradelli, Hélène Beuneu, Julie Garibal, and Françoise Gaudin for contributions to this work. We thank Joachim Confais from Cynbiose for help with the multivariate analysis. We acknowledge the Center for Translational Science and Cytometry and Biomarkers Unit of Technology and Service at Institut Pasteur for support in conducting this study.
The online version of this article contains supplemental material.
Abbreviations used in this article:
The authors are full time employees of Immitep S.A.S. F.T. is a stockholder of Immutep S.A.S. C.B. and F.T. are inventors on patents related to the technology, but their rights have been transferred to Immutep S.A.S. (as requested by French law).