Early expansion of TIGIT⁺PD1⁺ effector memory CD4 T cells via agonistic effect of alefacept in new-onset type 1 diabetes

Abstract

The CD2-depleting drug alefacept (LFA3-Ig) preserved beta cell function in new-onset type 1 diabetes (T1D) patients. The most promising biomarkers of response were late expansion of exhausted CD8 T cells and rare baseline inflammatory islet-reactive CD4 T cells, neither of which can be used to measure responses to drug in the weeks after treatment. Thus, we investigated whether early changes in T cell immunophenotypes could serve as biomarkers of drug activity. We characterized T cell responses by flow cytometry and identified an exhausted-like population of CD2^low CD4 effector memory T cells coexpressing TIGIT and PD1 that expanded by 11 wk after the start of treatment. This population was not entirely spared from alefacept-mediated depletion in vivo or in vitro but recovered through homeostatic proliferation of CD2^low cells in vivo. Proliferation of TIGIT⁺PD1⁺ effector memory CD4 T cells increased with treatment, with a concomitant reduction of proinflammatory cytokine production. The persistent increase of TIGIT⁺PD1⁺ effector memory CD4 T cells was specific to alefacept treatment; 2 other T cell depleting therapies, teplizumab and anti-thymocyte globulin, induced only a transient increase in this CD4 population. Our data suggest that the expanding TIGIT⁺PD1⁺ effector memory CD4 T cell population represents a promising biomarker of early treatment effects of alefacept. The nondepleting effects on proliferation and cytokine production also suggest agonistic activity by this CD2 targeted therapy.

Graphical Abstract

• The Editors have selected this article as a highlight of the issue.

Introduction

Type 1 diabetes (T1D) is a chronic autoimmune disease in which the immune system destroys insulin-producing beta cells in the pancreas, necessitating insulin replacement. An important area of study is the use of immune-modulating agents to preserve beta cell function in T1D patients. One molecule of interest, alefacept (LFA3-Ig fusion protein), targets CD2 and preserved beta cell function in the T1DAL trial (Inducing Remission in New Onset T1DM with Alefacept (Amevive®)).¹ We sought to characterize the pharmacodynamic changes in T cell populations in participants receiving alefacept.

CD2 is expressed on all T cells but is most highly expressed by memory and effector T cells and is expressed at lower levels on naïve and regulatory T cells (Tregs).² CD2 is a costimulatory molecule that binds LFA-3 and plays a key role in immune synapse formation and intracellular signaling.^3,4 In the T1DAL trial, central memory CD4 and CD8 T cells (Tcms) and effector memory CD4 and CD8 T cells (Tems), which were predominantly CD2^high, were depleted, while predominantly CD2^low naïve T cells and Tregs were preserved.^1,5 Despite this drug effect, depletion of Tcms or Tems was not associated with a clinical response in the trial. Instead, preservation of C-peptide was associated with 2 T cell biomarkers: (1) higher levels of an exhausted TIGIT⁺KLRG1⁺ subset of CD8 T cells 78 wk after initiation of therapy that was maintained out to 2 yr later^6,7 and (2) lower levels of rare inflammatory islet-reactive CD4 T cells at baseline.⁸ Changes in the CD8 TIGIT⁺KLRG1⁺ T cells were also detected within 12 wk in a distinct T1D trial with teplizumab (anti-CD3) and the frequency of these cells correlated with clinical response.⁹ Early biomarkers, like the CD8 TIGIT⁺KLRG1⁺ T cells with teplizumab, can be helpful in designing trials to determine optimal biologic dosing for immunotherapies. With the similar goal of developing a biomarker for immune modulators targeting the CD2 pathway, we investigated changes in immune markers after alefacept treatment. We previously reported that alefacept therapy increased the proportion of CD4 memory T cells expressing the inhibitory receptor PD1 after the first 12-wk treatment cycle.⁵ Given the role of PD1 and other immune checkpoint molecules in limiting T cell activity, we further characterize the effects of alefacept treatment on CD4 T cell phenotype and function with the intent of identifying early biomarkers of treatment. Our results demonstrate that alefacept therapy results in an expansion of exhausted-like TIGIT⁺PD1⁺ CD4 Tems with reduced inflammatory cytokine production. We propose that these effects are linked to the agonistic activity of alefacept on CD2^low CD4 T cells and thus could be a more robust marker for immune modulation.

Materials and methods

Human subjects and sample handling

Participants with T1D (ages 12–35 yr) enrolled in the randomized placebo-controlled T1DAL trial have been described previously.¹ Participants in the active drug arm received 2 courses of alefacept (12-weekly subcutaneous injections) separated by a 12-wk drug-free window. Participants in the placebo arm were dosed with saline on the same schedule. Additional T1D individuals were selected from the Benaroya Research Institute (BRI) Diabetes Control Registry Program. Written informed consent was obtained from all participants according to institutional review board–approved protocols at clinical sites (T1DAL) and by the BRI institutional review board (protocol number IRB07109). Publicly accessible data were used from 2 additional Immune Tolerance Network (ITN) clinical trials, AbATE (Autoimmunity-Blocking Antibody for Tolerance in Recently Diagnosed Type 1 Diabetes)¹⁰ and START (Effect of Antithymocyte Globulin on Preserving Beta-cell Function in New-onset Type 1 Diabetes Mellitus).¹¹ Written informed consent was obtained from all participants in these trials according to institutional review board–approved protocols at clinical sites.

Peripheral blood mononuclear cells (PBMCs) were isolated from whole blood within 36 h of blood draw using a Ficoll gradient and the ITN standard operating procedure (https://www.immunetolerance.org/for-researchers/lab-protocols). All assays were performed on samples in a blinded manner and all assay batches included participants from all cohorts along with an internal control.

Reagents

LFA3TIP (preclinical alefacept), CD2-targeted LFA3-Ig fusion protein, was obtained from Biogen Idec.

Cell culture with alefacept

Alefacept was added to the culture medium (RPMI 1640 medium with 5% heat-inactivated human AB serum, 1% penicillin/streptomycin, 5 mM L-glutamine, 25 mM HEPES) at 50 µg/mL. PBMCs were cultured at 0.5 × 10⁶/mL in a round bottom 96-well plate for 4, 24, and 72 h at 37 °C and 5% CO₂ prior to live cell counting and flow cytometry analyses.

Flow cytometry staining

Ex vivo cell staining from clinical trial samples was performed as described previously.^6,9,12 In short, PBMCs were thawed and stained with a surface and intracellular panel (Table SI) prior to acquisition on a BD LSRFortessa analyzer using Diva software (BD). Data were analyzed using FlowJo version 9.4 (TreeStar) and are available at the ITN TrialShare website (https://www.itntrialshare.org/). Counts of cell populations were calculated from gating in FlowJo, and complete blood counts were measured at ICON Central Labs.⁵

Cell cultures were harvested at the timepoints indicated and were washed with phosphate-buffered saline. Cells were resuspended at 1 × 10⁶/mL and stained with live/dead fixable viability dye Zombie NIR (Cat. 423106; BioLegend) for 15 min in phosphate-buffered saline at room temperature. After washing, cells were stained with an antibody cocktail (Table SI) for 30 min at 4 °C. Cells were washed and resuspended in fluorescence-activated cell sorting buffer prior to acquisition on an LSRII (BD) using Diva software, and postacquisition analysis using FlowJo 9.4.

Alefacept binding was performed on isolated CD4 T cells to avoid the functional consequences of antibody dependent cellular cytotoxicity. No touch Miltenyi enrichment columns were used to isolate CD4 T cells (CD4⁺ T Cell Isolation Kit, human; Cat. 130-091-155; Miltenyi Biotec). Cells were incubated with alefacept for 4 h and then stained with a surface cocktail, including a secondary goat anti-mouse IgG antibody (Table SI).

For cytokine staining, PBMCs were stimulated for 4 h using PMA (500 ng/mL) and ionomycin (500 ng/mL), and cytokine secretion was blocked using brefeldin A and monensin (both at 5 mg/mL), added the last 3 h of stimulation.

TCR profiling

TIGIT⁺PD1⁺ CD4 Tems from T1DAL participants were index sorted (gated as in Fig. 1) from PBMCs from 3 placebo-treated and 3 alefacept-treated T1DAL participants at baseline and week 35 on a BD Aria Fusion sorter. Complementary DNA was synthesized using the SMART-Seq v4 ultra-low input RNA kit (Takara Bio), and libraries were prepared for sequencing using a Nextera XT DNA library preparation kit (Illumina). Libraries were sequenced on a Illumina HiSeq2500 sequencer. T cell receptor (TCR) sequences were extracted from the sequencing data¹³ using the MiXCR¹⁴ Java package (version 3.0.5).

Figure 1.

TIGIT⁺PD1⁺ CD4 Tems selectively expand after treatment with alefacept. PBMCs from alefacept-treated (n = 28) and placebo-treated (n = 12) participants were collected periodically from baseline to week 104 and markers on cell subsets were analyzed in a univariate manner by flow cytometry. (A) Volcano plot showing the frequencies of cell subsets (memory T cell, Tcm, Tem) and markers (PD1, TIGIT, EOMES, etc.) with significant changes from baseline at week 35 after alefacept therapy compared with placebo. Values were not corrected for multiple tests due to the limited features and exploratory nature of the analyses. The x-axis represents fold change of alefacept treatment compared with placebo and the y-axis represents log base 10 P value. Cell types are noted by symbols; markers with significant changes were annotated. (B) Longitudinal analysis of KLRG1⁺ as a percentage of CD3⁺CD4 ⁺ Tems (CD45RA^–CCR7^–, gating in Fig. S1A). (C) Representative gating of TIGIT and PD1 expression on CD4 Tems at baseline and week 35 in placebo- and alefacept-treated participants (also in Fig. S1B). (D) Longitudinal analysis of TIGIT and PD1 subsets of CD4 Tems. Boxes along x-axes depict the timing of treatment with alefacept. Analysis of variance: *P < 0.05, **P < 0.01, ***P < 0.001.

Statistical analysis

Calculations were completed in Microsoft Excel and graphs and statistics were produced in GraphPad Prism version 10 (GraphPad Software).

Volcano plots showing the log₂ fold change of the frequency of T and natural killer (NK) cell subsets between alefacept and placebo arms, normalized to baseline levels, were generated in R (version 3.6.2; R Foundation for Statistical Computing). Statistics were performed by Mann-Whitney U test. P values <0.05 were considered significant.

Experimental data were presented as mean ± SE or as individual data points with lines at mean ± SE. For all longitudinal comparisons, statistics were computed with a mixed-effects model with Tukey correction for multiple comparisons, except for the comparison between drugs, which used the Dunnett’s correction for multiple comparisons, and the comparison between responders and nonresponders, which used the Sidak correction for multiple comparisons. All comparisons for the cell culture with alefacept were computed with a 2-way analysis of variance with Sidak correction for multiple comparisons. Comparisons required an adjusted P value of <0.05 to be considered statistically significant. Shannon entropy was calculated using an online calculator for TCR alpha and beta chains separately (https://www.endmemo.com/bio/shannonentropy.php). For Shannon entropy, the 2-stage linear step-up procedure of Benjamini, Krieger, and Yekutieli was used to control for the false discovery rate in the multiple comparisons.¹⁵

Results

Alefacept treatment proportionally increases the TIGIT⁺PD1⁺ CD4 Tem population

To understand early changes in inhibitory and activation receptor expression from baseline to wk 35, we first analyzed selected features of CD4 and CD8 memory T cells, NK cells, and Tregs by flow cytometry. Comparing fold change from baseline in 18 parameters, we found decreased frequencies of the majority of parameters in memory CD4 T cells and CD56 on NK cells (Fig. 1A). This is consistent with depletion of cells that express high levels of CD2,^1,16 as shown in longitudinal analysis of KLRG1⁺ CD4 Tems in placebo- and alefacept-treated participants (Fig. 1B). In contrast, the proportion of CD4 Tcms and Tems expressing PD1 and/or TIGIT increased in alefacept-treated participants, with the most significant increase of these parameters noted on the Tem subset (Fig. 1A).

To better understand the relationship between the kinetics of treatment and TIGIT and PD1 expression on CD4 T cells, we quantified TIGIT and PD1 subsets after alefacept treatment (Fig. 1C). We measured frequencies of TIGIT⁺PD1⁺, TIGIT⁺PD1^–, TIGIT^–PD1⁺, and TIGIT^–PD1^– CD4 Tems longitudinally. Both TIGIT⁺ (PD1⁺ and PD1^–) populations increased in treated participants, with a near doubling of TIGIT⁺PD1⁺ as a proportion of Tems by week 24 after the start of treatment that persisted for over a year (P$\le$ 0.015) (Fig. 1D). There were no statistical differences in the expansion of this population between pediatric (age 12–18 yr) and adult (age 19–34 yr) participants (P > 0.08), and no statistical differences consistent across time points based on sex (P > 0.18). In contrast, the TIGIT^–PD1^– population decreased, consistent with the total CD4 Tem population (P < 0.01) (Fig. 1D).

TIGIT⁺PD1⁺ CD4 Tems are depleted by alefacept and undergo homeostatic proliferation

To address whether TIGIT⁺ and PD1⁺ CD4 Tem were depleted by alefacept during the T1DAL trial, we calculated the absolute numbers of each TIGIT and PD1 CD4 Tem subset. All were reduced by week 11 after the start of treatment (P < 0.05) (Fig. 2A). However, in comparison with baseline, TIGIT⁺PD1⁺ CD4 Tems were reduced to a lesser degree and recovered in frequency by week 78, while TIGIT^–PD1^– CD4 Tems did not (Fig. 2B). We then compared the TIGIT⁺PD1⁺ CD4 Tem frequency with percent changes from baseline in CD4 counts obtained as real-time safety parameter during the T1DAL trial. At week 11, the frequency TIGIT⁺PD1⁺ of CD4 Tems inversely correlated with the % change in CD4 counts after treatment (R² = 0.30, P = 0.005) (Fig. 2C). In contrast, the frequency TIGIT^–PD1^– of CD4 Tems positively correlated with the % change in CD4 counts (R² = 0.23, P = 0.017) (Fig. 2C). Thus, the modest alefacept-mediated reduction of TIGIT⁺PD1⁺ CD4 Tems resulted in a relative expansion of the subset within the CD4 T cell pool.

Figure 2.

TIGIT⁺PD1⁺ CD4 Tems recover rapidly from depletion due to increased proliferation after alefacept treatment. (A, B) Absolute counts of CD4 Tem subsets were calculated from the frequency data in Fig. 1 and corresponding complete blood counts (CBCs) (A) for all subsets based on TIGIT and PD1 expression and (B) normalized to baseline for TIGIT⁺PD1⁺ and TIGIT^–PD1^– CD4 Tems (Fig. S1A and B, gating). Alefacept-treated T1DAL participants, n = 28. P values are for the comparison of baseline with the indicated visit. (A) P values for the TIGIT⁺PD1⁺, TIGIT^–PD$1_{,}^{+}$ and TIGIT⁺PD1^– are indicated in the same color as the symbols. Placebo (n = 12) for TIGIT⁺PD1⁺ indicated in gray bar representing 95% confidence interval. (C) Absolute counts of CD4 T cells were measured in CBC analysis as part of the T1DAL clinical trial at weeks 0 and 11. The % change from baseline to wk 11 was calculated for each alefacept-treated subject (n = 28) and plotted against frequency TIGIT⁺PD1⁺ (left) or TIGIT^–PD1^– (right) of CD4 Tems at week 11. Simple linear regression computed for correlations. (D) % Ki67⁺ of indicated CD4 Tem subsets at baseline. (E) % change in Ki67 (Fig. S1C, gating) from week 0 in participants from the T1DAL trial, with treatment group indicated (T1DAL trial, n = 5 treated, 5 placebo). (F) TIGIT⁺PD1⁺ CD4 Tems were sorted and TCR CDR3 regions sequenced using single-cell RNA sequencing. Shannon entropy was calculated and normalized to 1 for each participant and time point for TCR alpha and beta separately, with TCR beta depicted on the graph (T1DAL trial, n = 3 treated, 3 placebo). Analysis of variance: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. For Shannon entropy, the 2-stage linear step-up procedure of Benjamini, Krieger, and Yekutieli was used to control for the false discovery rate in the multiple comparisons.¹⁵

Given the rapid recovery of TIGIT⁺PD1⁺ CD4 Tem frequency, we analyzed their inherent proliferative capacity at steady state based on Ki67 expression. In individuals with T1D, TIGIT⁺PD1⁺ CD4 Tems had a higher frequency Ki67⁺ than other subsets of CD4 Tems (P < 0.01) (Fig. 2D). The Ki67⁺ frequency of TIGIT⁺PD1⁺ CD4 Tems further increased after treatment, specifically in alefacept-treated and not in placebo-treated participants (P < 0.05) (Fig. 2E). Elevated proliferation can result from clonal (antigen-specific) or homeostatic expansion. To differentiate between these possibilities, we sequenced TCR from alefacept- and placebo-treated T1DAL participants to measure changes in TCR clones. The TCR repertoire diversity of TIGIT⁺PD1⁺ CD4 Tems was high at baseline, and it was not reduced following alefacept treatment, suggesting that the expansion was polyclonal (P > 0.68 for TCR beta and P > 0.24 for TCR alpha) (Fig. 2F). These data further support the nondepletional mechanism of alefacept in promoting proliferation of TIGIT⁺PD1⁺ CD4 Tems.

TIGIT⁺PD1⁺ CD4 Tems are not resistant to alefacept-mediated depletion

The lower degree of reduction of TIGIT⁺PD1⁺ and CD4 Tems (Fig. 2B) raised the question of whether this Tem subset was resistant to depletion by alefacept. First, we found that alefacept selectively bound CD2^high cells and did not interfere with CD2 staining (Fig. 3A). We next analyzed CD2 levels on CD4 Tems longitudinally in T1DAL participants to demonstrate that alefacept reduced CD2^high CD4 Tems, consistent with our understanding of its mechanism of action (Fig. 3B). This treatment response persisted to at least wk 52 (P < 0.0001) (Fig. 3B), 4 mo after the completion of treatment.

Figure 3.

TIGIT⁺PD1⁺ CD4 Tems that express high levels of CD2 are depleted by alefacept. (A) Schematic (left) and representative plots of anti-CD2 staining and alefacept binding in media (middle) and 50 µg/mL alefacept-treated isolated CD4 T cells (right) after 4 h culture. (B) PBMCs from alefacept-treated (n = 33) and placebo-treated (n = 16) participants in the T1DAL trial were stained for flow cytometry (Table SI), gated for CD2 levels on naïve T cells and Tems (Fig. S1A, gating), and on CD4 T cells (left) and frequency of CD2 high within CD4 Tems over time (right). (C) PBMCs from T1D individuals not enrolled in T1DAL (n = 10) were analyzed for CD4 Tem subsets and CD2 expression (Table SI, gating in Fig. S2). Representative CD2 staining on indicated CD4 Tem subsets (left) and frequency of CD2 high and low CD4 Tems expressing PD1, TIGIT, or KLRG1 (right). (D) Schematic of PBMCs from T1D individuals cultured with media or 50 µg/mL alefacept for 4 or 72 h. (E) Loss of CD2^high cells on CD4 Tems treated with 50 µg/mL alefacept or etanercept (Fc binding control) after 72 h culture. Loss calculated as % CD2^high from treatment/media control. (F) % of CD4 Tems that are CD2^high (top) or CD2^lo (bottom) and TIGIT⁺PD1⁺ (left) or KLRG1⁺ (right) after 72 h culture. Participants from (A) T1DAL, n = 16 to 33; (B–F) BRI registry, n = 10. **P < 0.01, ****P < 0.0001, analysis of variance.

To further study how alefacept affected different T cell populations, we studied alefacept in vitro using PBMC samples from untreated T1D individuals not enrolled in T1DAL. CD2^low CD4 Tems, in contrast to CD2^high CD4 Tems, were more likely to be TIGIT⁺ and KLRG1⁻ (Fig. 3C). When culturing PBMCs from individuals with T1D (Fig. 3D), CD2^high CD4 Tems were specifically depleted by alefacept in vitro, not by the control fusion protein etanercept that also contains the Fc region of IgG1 and targets tumor necrosis factor α (TNFα) (P < 0.05) (Fig. 3E). CD2^high CD4 Tem TIGIT⁺PD1⁺ and KLRG1⁺ populations were depleted in vitro, while CD2^low CD4 Tems of either subset were not (P < 0.01) (Fig. 3F), indicating that CD2 density is the primary driver of alefacept-mediated depletion across distinct T cell subtypes in vitro. This is consistent with previous research indicating that T cell activation is driven by multimeric, not monomeric, interactions of LFA-3 and CD2, requiring occupation of 10⁴ to 10⁵ CD2 sites.¹⁷ Therefore, the relative expansion of TIGIT⁺PD1⁺ CD4 Tems with alefacept therapy is not explained by resistance to depletion.

TIGIT⁺PD1⁺ CD4 Tems have reduced proinflammatory cytokine production

To further characterize the function of the expanded TIGIT⁺PD1⁺ Tem population, we measured cytokine production in TIGIT⁺PD1⁺, TIGIT⁺PD1^–, TIGIT^–PD1⁺, and TIGIT^–PD1^– subsets as compared with proinflammatory KLRG1⁺ Tems. While KLRG1⁺ and KLRG1^–TIGIT^–PD1^– CD4 Tems produced interleukin (IL)-2, TNF-α, and interferon γ (IFN-γ) in response to PMA/ionomycin stimulation, TIGIT⁺ and PD1⁺ subsets produced very low levels of these cytokines (P < 0.05) (Fig. 4A). Further, at week 35 after treatment, the frequencies of TIGIT⁺PD1⁺ CD4 Tems expressing TNF-α and IFN-γ decreased in alefacept-treated but not in placebo-treated participants (P < 0.01) (Fig. 4B), suggesting that treatment rendered TIGIT⁺PD1⁺ CD4 Tems hyporesponsive.

Figure 4.

TIGIT⁺PD1⁺ CD4 Tems have reduced proinflammatory cytokine production after alefacept treatment. (A) % IL-2⁺, TNF-α⁺, or IFN-γ⁺ of indicated subsets of CD4 Tems in T1D individuals (n = 3). Stimulation conditions described as in Fig. 3 (Fig. S3, gating). (B) The % change from week 0 to week 35 in expression of IL-2, TNF-α, or IFN-γ or (C) from week 0 to weeks 24 and 35 in expression of (left) IL-10 or (right) IL-21 (Fig. S3, gating) on TIGIT⁺PD1⁺ CD4 Tems in participants from the T1DAL trial, with treatment group indicated (n = 5 treated, 5 placebo). Analysis of variance: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

The changes in IL-2, TNF-α, and IFN-γ represent a decrease in inflammatory cytokines, so we next addressed other cytokines, specifically IL-10, which is associated with T cell tolerance,^18,19 and IL-21, which is associated with loss of immune tolerance.²⁰ In alefacept-treated participants, there was an increase in IL-10 within the TIGIT⁺PD1⁺ subset by week 35 (P < 0.01) (Fig. 4C) and an increase in IL-21 by week 24 (P < 0.05) (Fig. 4C). Thus, the TIGIT⁺PD1⁺ CD4 Tem expansion is associated with decreases in certain proinflammatory cytokines concomitant with increases in other cytokines.

Alefacept treatment uniquely leads to sustained expansion of TIGIT⁺PD1⁺ CD4 Tems

Increases in TIGIT⁺PD1⁺ CD4 Tem numbers along with changes in proliferation and cytokine production likely demonstrate a nondepletional effect of the drug. We sought to further explore the nature of the selective expansion of TIGIT⁺PD1⁺ CD4 Tems, and in particular whether the effect was specific to alefacept by comparing frequencies of TIGIT⁺PD1⁺ CD4 Tems with alefacept treatment to 2 other T cell targeting agents, ATG¹¹ and teplizumab.¹⁰ Teplizumab is a monoclonal antibody targeting CD3,²¹ while ATG is a preparation of polyclonal antibodies.²² Both ATG and teplizumab induced increased TIGIT⁺PD1⁺ Tems shortly after treatment, but the population returned to baseline levels within 12 wk (Fig. 5A). In contrast, the increase produced by alefacept was more gradual and was maintained for at least a year (Fig. 5A). Comparing responders and nonresponders across the 3 studies demonstrates that levels of TIGIT⁺PD1⁺ CD4 Tems were elevated in responders (Fig. 5B). Thus, the TIGIT⁺PD1⁺ CD4 Tem expansion is correlated with response and has distinct kinetics in treatment with alefacept and other therapies. These findings suggest another mechanism of action in addition to the hypothesized depletional effect of alefacept.

Figure 5.

Distinct kinetics of the TIGIT⁺PD1⁺ Tem population with alefacept as compared with other T cell therapies. (A) Frequency TIGIT⁺PD1⁺ of CD4 Tems measured in alefacept-treated participants in T1DAL, ATG-treated participants from the ITN START trial, teplizumab-treated participants from the ITN AbATE trial, and in placebo participants from all 3 trials (Fig. S1A and B, gating). Data for placebo are depicted as 95% confidence interval around the mean. T1DAL: n = 28; START: n = 38; AbATE: n = 34; placebo: n = 47. Analysis of variance: **P < 0.01, ***P < 0.001, ****P < 0.0001. Main effects were significantly different between treatments (P < 0.01). P values on plot represent main row effect for the time points present in T1DAL. (B) Fold change in % TIGIT⁺PD1⁺ of CD4 Tems calculated for responders and nonresponders in T1DAL, AbATE, and START. As defined as in T1DAL, responders were those subjects with no loss of C-peptide, and nonresponders were those subjects with loss of 50% or more C-peptide.⁵ Responders: n = 17; nonresponders: n = 43. Analysis of variance: *P < 0.05, **P < 0.01. P values on the plot represent comparison between responders and nonresponders for that time point.

Discussion

Alefacept was selected as an agent in the T1DAL clinical trial to antagonize T cell activation and selectively deplete CD2^high cells, in particular memory cells. This follow-up study demonstrates that while the drug depleted Tems, a subset of TIGIT⁺PD1⁺ CD4 Tems also expanded after treatment. This subset was not resistant to depletion but instead underwent homeostatic proliferation, leading to a proportional increase in the frequency of this subset in the CD4 pool. In addition, we found that this subset is predominantly CD2^low, produced low levels of inflammatory cytokines, and had evidence of increased proliferation. All of these phenotypes suggest that alefacept has functional effects outside of its capacity for depletion. While TIGIT⁺PD1⁺ Tems expanded shortly after treatment with other depleting agents, such as anti-CD3 or ATG, the trajectory of this population is qualitatively and quantitatively distinct with alefacept treatment, characterized by a much more gradual and stable increase. The early expansion of TIGIT⁺PD1⁺ CD4 Tems correlates with the depletion of total CD4 T cells by alefacept. Thus, TIGIT⁺PD1⁺ CD4 Tems may represent an early biomarker of treatment effect that demonstrates distinct effects of different drugs. This merits further study to determine if expansion of this subpopulation is protective in T1D, and whether changes in this population can be used to understand drug mechanism of action, including whether this population can be used as a biomarker to identify optimal drug dosage.

An important remaining question is the role of this TIGIT⁺PD1⁺ CD4 Tem population after alefacept treatment. TIGIT and PD1 are both expressed by exhausted T cells, and blockade of either molecule can reverse exhaustion.^23,24 The low cytokine production is also consistent with an exhausted state.^25,26 However, the proliferative phenotype (Ki67⁺) of these cells demonstrates that they are not terminally exhausted.²⁵ Therefore, the TIGIT⁺PD1⁺ Tems may represent an early exhausted population induced by alefacept. Their proliferative phenotype may also indicate that TIGIT⁺PD1⁺ CD4 T cells develop from highly proliferative precursors. Follow-up studies will address this possibility. Alefacept-induced TIGIT⁺PD1⁺ Tems may be similar to an exhausted TIGIT⁺PD1⁺ CD4 population that also expresses LAG3 and is associated with HIV persistence,²⁷ in contrast to a TIGIT⁺PD1⁺ population that lacks CXCR5 and is associated with rheumatoid arthritis disease activity.²⁸ TIGIT and PD1 expression has been associated with an exhaustion phenotype specifically in autoreactive memory T cells and Tregs from individuals with chronic autoimmune diseases.²⁹ TIGIT-expressing Tregs have been associated with stable suppressive function.³⁰ While the TIGIT⁺PD1⁺ Tems in our study were gated to exclude Tregs, this shared phenotype along with the low cytokine expression suggests features in common with Tregs. TIGIT has also been shown to play a role in the differentiation state of PD1⁺ cells, marking germinal center T follicular helper (Tfh) cells as compared with CD127^high memory cells.^31,32 TIGIT⁺ Tfh cells have been associated with autoimmunity in studies of other diseases.^33,34 In a mouse model, TIGIT agonism inhibited Tfh cells, thereby promoting protection against experimental autoimmune encephalomyelitis and lupus.³⁵ It is possible that alefacept has an agonistic effect on TIGIT⁺PD1⁺ cells that limits Tfh cell differentiation and instead promotes an exhausted-like phenotype.

Exhausted T cells have been associated with response to therapy in new-onset T1D trials of teplizumab^9,12,36 and ATG,³⁷ and in a T1D prevention trial of teplizumab.³⁸ Individuals with T1D who maintain C-peptide for at least 5 yr can be distinguished from those who lose C-peptide by a more exhausted phenotype in their islet-specific CD8 T cells.³⁹ The majority of studies of exhausted T cells have focused on CD8 T cells, though there are an increasing number of studies of exhausted CD4 T cells in cancer and autoimmune diseases,^40–42 including T1D.^12,37 Further study of the epigenetic state and gene expression of TIGIT⁺PD1⁺ CD4 Tems will be important for defining their function. It will also be important for future studies to address the possibility that this Tem subset does not interfere with the autoimmune process directly, but rather is a biomarker of other protective immunological changes. In addition, if the TIGIT⁺PD1⁺ Tems are early exhausted cells, analyzing their differentiation into terminally exhausted cells will be important to understand the full implications of the alefacept-induced expansion.

CD2 expression levels affect T cell function through differential integration within the TCR-costimulation interaction complex.⁴³ We show that CD2^high cells are depleted, and some CD2^low cells are modulated with therapy. The changes in cytokine, proliferation, and phenotype specifically after treatment with alefacept suggest an agonistic impact of alefacept specifically on TIGIT⁺PD1⁺ Tems, even though this population expanded in a polyclonal manner. Prior study of alefacept treatment for psoriasis has identified partial agonism based on modulation of gene expression in total PBMCs.⁴⁴ Consistent with our work, the psoriasis study found increased expression of genes associated with proliferation and downregulation of TCR signaling.⁴⁴ These effects may be mediated by alefacept binding to CD2 and mediating changes in cell signaling on cells with inhibitory receptor expression (i.e., TIGIT, PD1) that may already be dampening TCR signals. Agonism of inhibitory receptors is an active area of study for the treatment of autoimmune disease.^45,46 The distinct changes in TIGIT⁺PD1⁺ Tems with alefacept, teplizumab, and ATG treatment (Fig. 5) demonstrate that this agonistic effect is likely specific to alefacept. This specificity of increased TIGIT⁺PD1⁺ Tems to alefacept could be explained by specificity to CD2 itself, by the differential timing of treatment between the 3 studies (T1DAL, AbATE, and START), or by differences in signaling induced by antibody (teplizumab, ATG) versus soluble receptor (alefacept). Follow-up studies with anti-CD2 antibody⁴⁷⁻⁴⁹ will be important in distinguishing between the specificity to CD2 and the drug delivery system.

An intriguing possibility for application of this finding is in pharmacodynamics. Our prior analysis of exhausted CD8 T cells found an increase in exhausted cells at week 78, which correlates with response to alefacept.⁶ However, this is a time point at which the change in C-peptide is already apparent. A recent study found that islet-reactive CD4 T cells at baseline inversely correlate with response to alefacept,⁸ suggesting that CD4 T cells may represent an earlier biomarker than CD8 T cells. The expansion of TIGIT⁺PD1⁺ Tems was detectable in 85% of participants by week 11 and statistically significant by week 24; the TIGIT⁺PD1⁺ Tem population doubled as a percentage of Tems by week 35. The magnitude of depletion of TIGIT⁺PD1⁺ Tems was half that of TIGIT^–PD1^– Tems and correlated with total CD4 depletion (Fig. 2B, C). Tracking increases in this population, therefore, could provide an early biomarker of activity of alefacept or other anti-CD2–targeting drugs, and may be more robust than measuring the reduction of circulating cells that may be confounded by nondeletional effects of treatment. For example, a dose-finding study could rely on changes in TIGIT⁺PD1⁺ Tems by week 11 or 24 after start of treatment in lieu of waiting for C-peptide results at week 52. TIGIT⁺PD1⁺ Tems could also be used to define “treat-to-target” therapy,⁵⁰ with dosage and/or duration of CD2-targeted drugs determined based on the expansion of this population. In contrast, a decreasing population like total CD4 T cells could also be used as a biomarker of depletion but has only a limited correlation to the nondepletional changes observed and may also result from cells homing to another tissue site rather than depletion. Biomarkers of population expansion like TIGIT⁺PD1⁺ Tems could provide a means of screening for responders early after treatment initiation and selecting more intensive or alternate intervention for those participants anticipated to be nonresponders.

There are several factors that are important to consider in interpreting the results of this study. First, all the functional analysis was in vitro on a small number of individuals with T1D and may not reflect the fate of the cells in vivo. Second, our ex vivo analysis provides a detailed picture of immunity immediately prior to and up to 2 years after treatment with alefacept but does not provide insight on longer-term effects. Third, when comparing the expansion of TIGIT⁺PD1⁺ CD4 Tems between responders and nonresponders with therapy, we had to compile analysis across therapies in order to have sufficient sample size to analyze. Studies in a larger cohort will be important to identify the role of this population in response to CD2-targeted treatment specifically. Finally, ongoing studies of CD2 as a target of immune-modulating therapies will focus on antibodies or new versions of LFA3-Ig because alefacept has been discontinued.⁵¹ Therefore, it will be important to conduct further studies to determine if the same mechanism applies to these other CD2-targeted therapeutics.^48,52

In conclusion, we have demonstrated that while CD4 Tems are depleted by alefacept, a proliferative subpopulation coexpressing TIGIT and PD1 with low proinflammatory cytokine production selectively expands after treatment with alefacept. Our data suggest that this expansion results from agonistic effects of alefacept on CD4 Tems expressing the inhibitory receptors TIGIT and PD1 and low levels of CD2.

Supplementary material

Supplementary material is available at The Journal of Immunology online.

Funding

Research reported in this publication was performed as a project of the Immune Tolerance Network and supported by the National Institute of Diabetes and Digestive and Kidney Diseases and the National Institute of Allergy and Infectious Diseases of the National Institutes of Health under Award Numbers NO1-AI-15416 and UM1AI109565. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Conflicts of interest

The authors have no conflicts of interest to disclose.

Data availability

The data underlying this article are available at ITN TrialShare (https://www.itntrialshare.org/T1DALJI.url).