Integrin αvβ8–mediated TGF-β1 activation regulates the sustained response in immune thrombocytopenia after TPO-RA withdrawal

Key Points

• •Integrin αvβ8–mediated TGF-β1 activation sustains ITP remission after TPO-RA withdrawal by enhancing immune tolerance.
• •D-mannose plus TPO enhance αvβ8 expression and TGF-β1 activation, offering a novel strategy to prolong ITP remission.

Visual Abstract

Abstract

Only 30% to 50% of patients with immune thrombocytopenia (ITP) exhibit a sustained response upon thrombopoietin receptor agonists (TPO-RA) withdrawal, underscoring the necessity for mechanistic elucidation. We enrolled 49 patients treated with TPO-RA for 4 months and performed a follow-up study for 3 months, classifying them into sustained responders (n = 21), and nonsustained responders (n = 28). Compared with total transforming growth factor β1 (TGF-β1) levels, activated TGF-β1 levels (3854 ± 4380 vs 943 ± 1500 pg/mL; P < .001) were significantly elevated in sustained responders, with integrin αvβ8 regulating TGF-β1 activation and restoring immune tolerance. We established a passive ITP model using platelet factor 4–TGF-β1 conditional knockout (CKO) mice, which exhibited a shorter duration of sustained response than wild-type (WT) mice. CKO mice demonstrated a reduced regulatory T-cell (Treg) population, an increased M1-to-M2 macrophage ratio, and more severe megakaryocyte destruction after anti-CD41 injection. Exogenous administration of αvβ8 (250 ng/kg) effectively activated TGF-β1 and prolonged remission after TPO discontinuation in WT mice. Additionally, CD4⁺ T cells were transfected with lentiviral small interfering RNA or short hairpin RNA to modulate integrin β8 expression and these were injected into severe combined immunodeficiency mice undergoing an active model of ITP. Results showed that β8 overexpression increased Tregs and reduced megakaryocyte damage. Mechanistically, TPO-RA modulated αvβ8-mediated TGF-β1 activation through the activator protein 1family and Smad family member 2 signaling pathways. Furthermore, D-mannose combined with TPO prolonged the response in ITP mice by upregulating αvβ8 and activating TGF-β1. Overall, the integrin αvβ8–mediated activation of TGF-β1 pathway represents a promising therapeutic target for ITP, with substantial potential for clinical application.

Approximately 80% of patients with immune thrombocytopenia (ITP) treated with thrombopoietin receptor agonists (TPO-RA) have a significant platelet increase, but only 30% to 50% have durable responses after TPO-RA withdrawal. Mei and colleagues used a combination of patient samples and animal models to reveal that the immune tolerance needed for persistent response, represented by an increased number of regulatory T cells, is sustained by integrin αvβ8–mediated transforming growth factor β1 (TGF-β1) activation and clinically is enhanced by initial therapy with TPO-RA plus d-mannose. Their work offers a potential novel therapeutic approach to prolonging remissions in ITP.

Introduction

The pathogenesis of immune thrombocytopenia (ITP) involves T-cell hyperactivation, reduced regulatory T cells (Treg), and antibody- or cytotoxic T lymphocyte–mediated platelet destruction.¹ Thrombopoietin receptor agonists (TPO-RA) are a cornerstone of second-line therapy for ITP, with efficacy rates ranging from 60% to 90%.^2,3 Despite their rapid onset of effect and high efficacy, most patients experience a swift decline in platelet counts within 2 weeks after drug withdrawal. Retrospective studies indicate that about 30% to 50% of patients off TPO treatment maintain safe or normal platelet counts thereafter, a sustained response.^4,5 However, the high relapse rate after TPO-RA discontinuation increases patient dependency on the medication, and exacerbates the financial burden on health care systems.

Previous studies have shown that patients responding to TPO-RA therapy exhibit enhanced Treg and regulatory B-cell function, likely reflecting the immunoregulatory effects of TPO-RA.6, 7, 8, 9 In murine ITP models, TPO-RA not only increased platelet counts but also reduced antiplatelet antibody titers, supporting their role in immune modulation.¹⁰ These effects may be mediated by elevated platelet-derived transforming growth factor β1 (TGF-β1), which promotes Treg differentiation.^7,11,12 Responders also show higher TGF-β1 levels and increased Treg counts, although the underlying mechanisms remain unclear.^6,7 Recently, Wang et al reported that TPO-RA enable platelet-derived TGF-β1 to reprogram myeloid-derived suppressor cells via the TGF-β/Smad pathway, suggesting that platelets may contribute to immune homeostasis and regulate their own survival.^13,14

TGF-β1 is typically secreted by immune cells, platelets, and megakaryocytes, binds to latency-associated peptide but remains inactive under physiological conditions until its activation.¹⁵ This activation is facilitated by integrins αvβ6 and αvβ8, after which TGF-β1 plays a crucial role in immune regulation, particularly in influencing T cells and monocytes.^15,16 Integrin αvβ6 is exclusively expressed in epithelial cells; mice lacking αvβ6 exhibit a mild inflammatory phenotype.¹⁷ In contrast, loss of integrin αvβ8 on dendritic cells in mice leads to autoimmune disease.^11,18,19 A previous study has shown that supraphysiological levels of D-mannose promote Treg differentiation by enhancing TGF-β activation, which is mediated through the upregulation of integrin αvβ8.²⁰ Based on these findings, we hypothesized that integrin αvβ8–mediated activation of TGF-β1 plays a critical role in the immunoregulatory effects of TPO-RA in patients with ITP. Combining TPO-RA with D-mannose may help sustain long-term therapeutic responses.

This study used TGF-β1 conditional knockout (KO; CKO) mice to specifically target platelets and megakaryocytes in a passive ITP model. Additionally, CD61KO mice were immunized to generate antibodies, and spleen cells were transferred into severe combined immunodeficiency (SCID) mice to establish an active ITP model. Our goal was to thoroughly investigate the role of integrin αvβ8–mediated activation of TGF-β1 in maintaining treatment efficacy after TPO-RA withdrawal in ITP.

Methods

Participants

This study enrolled patients from August 2018 to June 2023 at Wuhan Union Hospital, with ethics approval from the ethics committee of Wuhan Union Hospital. All participants provided informed consent and met the diagnostic criteria for ITP.²¹ Inclusion criteria and data collection details are provided in the supplemental Methods 2, available on the Blood website.

Patients were treated with TPO-RA for 4 months, followed by a 3-month follow-up, and all samples were collected at the time of drug discontinuation. Then, patients were categorized into 2 groups based on treatment response and maintenance after drug cessation: sustained response (ITP-sus; platelet count of ≥50 × 10⁹/L at week 28 [or at the last follow-up, if earlier] and in ≥50% of assessments); and nonsustained response (ITP-nonsus; platelet count of <50 × 10⁹/L at week 28 or in >50% of assessments, or if additional ITP medication was required).

Establishment of passive and active ITP models

In this study, we used the CRISPR-associated protein 9 system to generate CKO mice with platelet- and megakaryocyte-specific deletion of TGF-β1 by targeting the platelet factor 4 (PF4) promoter. Mice with PF4 Cre⁺ Tgfb1^fl/fl (Cre⁺) showed specific loss of TGF-β1 in platelets and megakaryocytes, whereas littermate Cre⁻ mice served as controls (almost equivalent to wild-type mice). Mice were randomly assigned to 5 groups: control, ITP, TPO, TPO+αvβ8^low, and TPO+αvβ8^high. The control group received saline, whereas the other groups received intraperitoneal injections of the CD41 antibody at 5 μg per 20 g on days 0 and 2, 7.5 μg per 20 g on days 4 and 6, and 10 μg per 20 g every 2 days from day 8 to day 28. Concurrently, TPO (1500 U/kg) was administered subcutaneously from days 1 to 14 in all groups except the control group. The TPO+αvβ8^low and TPO⁺αvβ8^high groups also received IV injections of recombinant integrin αvβ8 (25 μg per 20 g or 250 μg per 20 g, respectively) on days 1, 7, and 14. Blood samples were collected every other day from day 0 to day 28 (Figure 3A).

Figure 3.

TPO combined with integrin β8 prolongs sustained platelet response and improves health in ITP mice. (A) Schematic diagram of the establishment of the ITP passive model and the treatment time points. (B-C) Levels of activated TGF-β1 in the peripheral blood and bone marrow of Cre⁻ (nearly WT) and Cre⁺ (CKO) mice. (D) Dynamic monitoring of platelet levels in saline control and ITP groups (MWReg30) in Cre⁻and Cre⁺ mice. (E) Dynamic monitoring of platelet levels in the TPO treatment group (MWReg30+TPO) in Cre⁻ and Cre⁺ mice. (F) Dynamic monitoring of platelet levels in the TPO combined with low-dose recombinant integrin αvβ8 treatment group (MWReg30+TPO+αvβ8^low) and TPO combined with high-dose recombinant integrin αvβ8 treatment group (MWReg30+TPO+αvβ8^high) in Cre⁻ and Cre⁺ mice. (G) Expression of peripheral blood Treg on day 14 among the groups. (H) Schematic diagram of the establishment of the ITP active model and interventions in each group (saline; TPO, TPO + CD4⁺ T-cell treatment; TPO+β8^siT, TPO + CD4⁺ T-cell overexpressing integrin β8; TPO+β8^shT, TPO + CD4⁺ T-cell knockdown of integrin β8). (I) Dynamic monitoring of platelet levels in each intervention group of the active ITP model. (J) Survival curves of mice in each intervention group of the active ITP model. (K) Expression of peripheral blood Treg on day 14 in each intervention group of the active ITP model. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001. APC, allophycocyanin; BM, bone marrow; PB, peripheral blood; PE, phycoerythrin; PLT, platelet.

For the active ITP model, methods were adapted from Chow et al.²² Platelets (10⁸) were isolated from wild-type (WT; C57BL/6) mice and injected into CD61-KO mice via the tail vein once a week for 4 weeks. Six- to 8-week-old SCID mice (C57BL/6) were randomly assigned to 5 groups: saline (control), TPO monotherapy (TPO), TPO with CD4⁺ T cells overexpressing integrin β8 (TPO+β8^siT), TPO with CD4⁺ T cells with β8 knockdown (TPO+β8^shT), and TPO with D-mannose (TPO+D-man). After 180 cGy X-ray irradiation, SCID mice were injected with 2 × 10⁴ spleen cells isolated from CD61-KO mice. Mice in the TPO+β8^siT and TPO+β8^shT groups also received 2 × 10⁴ CD4⁺ T cells transduced with either β8 or anti-β8 virus, whereas the other groups received an equal number of WT CD4⁺ T cells. Recombinant TPO was administered subcutaneously from days 1 to 14. Blood samples were collected on days 0, 7, 14, 21, and 28.

Flow cytometry analysis of lymphocyte subsets and megakaryocyte

Peripheral blood mononuclear cells (PBMC) were isolated from patients at TPO-RA cessation and incubated with flow cytometry antibodies (1 × 10⁶ cells, 30 minutes, room temperature). Samples were centrifuged and analyzed using a flow cytometer.

Mouse samples were collected after anesthesia, lymphocyte subsets were analyzed by incubating 1 × 10⁶ cells per mL in RPMI 1640 with PMA (phorbol-12-myristate-13-acetate; 50 ng/mL), ionomycin (1 μg/mL), and brefeldin A (3 μg/mL) for 6 hours before antibody staining. Cells were stained for surface markers, fixed, and stained intracellular with antibodies specific to T helper (TH) cells, Treg, and macrophages. Bone marrow cells were isolated from femurs and tibias, stained with antibodies, fixed, permeabilized, and analyzed by flow cytometry. Detail antibodies and methods were shown in supplemental Materials 1.

Cytometric bead array for cytokine detection

A standard curve was generated using serial dilutions of standard proteins. Each flow tube was loaded with 100 μL of either standard or test sample and 25 μL of capture beads. After 1 hour of incubation, 25 μL of phycoerythrin-conjugated detection antibodies was added. After a 2-hour incubation, samples were analyzed via flow cytometry.

Real-time qPCR analysis

RNA was extracted from cells or tissues using TRIzol, followed by chloroform extraction. RNA was reverse transcribed into complementary DNA (cDNA) using the Vazyme RNA reverse transcription kit. Quantitative polymerase chain reaction (qPCR) was performed to quantify gene expression levels, and primer sequences were provided in the supplemental Materials 3.

Western blotting

Cells or tissues were lysed in protein extraction buffer and subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Proteins were transferred onto polyvinylidene fluoride membranes, blocked, and incubated with primary and secondary antibodies. Protein bands were visualized using chemiluminescence, and densitometry analysis was performed using ImageJ software.

Human T-cell functional assay

CD4⁺ T cells from patients with ITP and healthy controls were isolated and cultured with RPMI 1640 medium, 10% fetal bovine serum, interleukin-2 (5 ng/mL), and anti-human CD3/CD28 antibodies (1.5 μg/mL), along with respective drug treatments, for 72 hours. Cells were stained with fluorescein isothiocyanate–Annexin V and propidium iodide for apoptosis analysis.

Treg (CD4⁺CD25⁺) and effector T cells (Teff; CD4⁺CD25⁻) were isolated using magnetic beads. Teff were labeled with CFSE (carboxyfluorescein succinimidyl ester; 5 μmol/L) by incubating for 15 minutes, followed by a 5-minute incubation with fetal bovine serum at 4°C. Treg (5 × 10⁴ cells per well) and Teff (2 × 10⁵ cells per well) were cocultured at a 1:4 ratio in U-bottom 96-well plates for 6 days, after which flow cytometry was used to analyze T-cell functionality.

Mouse T-cell sorting and transfection

CD4⁺ T cells were isolated from WT C57 mouse spleens using magnetic beads and resuspended at 5 × 10⁵ cells per mL. Cells were cultured in X-VIVO medium supplemented with interleukin-2 (5 ng/mL) and anti-mouse CD3/CD28 antibodies (1.5 μg/mL) for 48 hours. After replacing the medium, 100 μL of lentiviral transfection reagent and 100 μL of lentiviral particles were added. Cells were further incubated for 12 to 24 hours before experiments.

Giemsa staining, hematoxylin and eosin staining and immunofluorescence

Bone marrow cells were fixed with 4% paraformaldehyde and stained with 1× Giemsa solution for 45 minutes to assess megakaryocyte number and developmental stages. Whole femurs were fixed in 4% paraformaldehyde, then decalcified, dehydrated, and embedded in paraffin. Tissue sections, 4 μm thick, were stained with hematoxylin and eosin for 1 minute to detect megakaryocytes. For immunofluorescence, after dewaxing, antigen retrieval was performed using heat, acid, or alkaline methods. Nonspecific binding was blocked with Fc-blocking reagent, followed by incubation with primary and secondary antibodies, then the sections were visualized using a fluorescence microscope.

Single-cell sequencing

PBMC were isolated from patients with ITP. Single cells were separated using microfluidic technology and fixed in chip wells. RNA was extracted from the cells, reverse transcribed into cDNA, and amplified to generate a cDNA library. Sequencing was performed using 10x Genomics technology, and the raw data were analyzed with bioinformatics tools.

Statistical analysis

Continuous variables with a normal distribution were presented as mean ± standard deviation, whereas nonnormally distributed data were presented as median and interquartile range. Categorical variables were summarized as counts and percentages, with comparisons made using the χ² test. For comparisons among ≥3 groups, the Kruskal-Wallis test was used, followed by pairwise comparisons with significance adjustments. And statistical significance was set at a 2-sided P value <.05. All statistical analyses were performed using SPSS software (Chicago, IL), graphical representations were generated using GraphPad Prism (La Jolla, CA) and Adobe Illustrator (San Jose, CA).

Results

Patient characteristics

Between August 2018 and June 2023, 224 patients were screened at Wuhan Union Hospital. The following reasons led to exclusions: 35 patients had secondary thrombocytopenia, 42 patients failed to achieve complete response (platelet count of ≥100 × 10⁹/L and absence of bleeding) during treatment, 29 patients lacked available samples, 45 declined bone marrow evaluation, and 7 patients did not undergo platelet monitoring during follow-up. As a result, 49 patients were included and categorized into 2 groups: ITP-sus (n = 21), and ITP-nonsus (n = 28). The median age was 45 years, with a higher proportion of females. Eltrombopag and hetrombopag were the most commonly prescribed TPO-RA. No statistically significant differences were observed in sex, age, or comorbid conditions between the groups. All patients had received at least 1 other form of ITP therapy, but a significantly higher proportion of patients in the ITP-nonsus group had undergone ≥5 previous treatments compared with the ITP-sus group (Table 1).

Table 1.

Demographics and clinical characteristics at baseline

Baseline characteristic	ITP-sus (n = 21)	ITP-nonsus (n = 28)
Age, median (IQR), y	45 (18-67)	45 (19-73)
Sex, female, n (%)	15 (71.4)	16 (57.1)
Baseline PLT, median (IQR), ×109/L	20 (5-41)	9 (5-32)
ITP phase, n (%)
Persistent	5 (23.8)	8 (28.5)
Chronic	16 (76.1)	20 (71.4)
Comorbidity, n (%)	10 (47.6)	13 (46.4)
Hypertension	4 (40.0)	3 (23.0)
Diabetes mellitus	1 (10.0)	6 (46.1)
Hepatitis	2 (20.0)	5 (38.4)
Impaired renal function	5 (50.0)	6 (46.1)
Previous infections	4 (40.0)	8 (61.5)
Surgery	2 (20.0)	5 (38.4)
No. of previous treatments, n (%)	21 (100.0)	28 (100.0)
0	0 (0.0)	0 (0.0)
1-2	8 (38.1)	6 (21.4)
3-4	11 (52.3)	8 (28.5)
≥5	2 (9.5)	14 (50.0)
TPO-RA type, n (%)
Eltrombopag	7 (33.3)	10 (35.7)
Hetrombopag	9 (42.8)	12 (42.8)
Avatrombopag	4 (19.0)	5 (17.8)
Romiplostim	1 (4.7)	1 (3.5)

IQR, interquartile range; PLT, platelet.

Biomarkers associated with sustained response after drug withdrawal

Compared with patients with ITP-nonsus, patients with ITP-sus had higher platelet counts, both at baseline and upon drug discontinuation. No notable difference in plasma levels of total TGF-β1 was observed between the groups. Conversely, activated TGF-β1 levels were markedly increased in patients with ITP-sus (ITP-sus vs ITP-nonsus, 3854 ± 4380 pg/mL vs 943 ± 1500 pg/mL; P < .001). Additionally, Treg cells were elevated in patients with ITP-sus. These findings suggest that higher activated TGF-β1 levels may promote Treg differentiation (Table 2; Figure 1B-F).

Table 2.

Clinical characteristics of patients at the end of TPO-RA treatment

Variable	ITP-sus (n = 21)	ITP-nonsus (n = 28)	P value
PLT at drug withdrawal, ×109/L	88 (51-178)	63 (39-156)	.001
Total TGF-β1, pg/mL	7002 ± 763	5882 ± 464	.197
Activated TGF-β1, pg/mL	3436 ± 647	1013 ± 338	.001
MCP-1, pg/mL	43.97 ± 2.69	57.39 ± 7.73	.154
IP-10, pg/mL	271.40 ± 50.49	275.01 ± 45.81	.958
IL-10, pg/mL	51.37 ± 10.23	61.44 ± 10.33	.502
IL-12, pg/mL	381.92 ± 169.77	368.90 ± 136.19	.221
IL-21, pg/mL	507.05 ± 222.33	419.18 ± 141.18	.241
IL-17A, pg/mL	16.36 ± 5.07	18.37 ± 5.72	.801
IL-5, pg/mL	20.93 ± 6.62	14.98 ± 2.47	.354
IL-2, pg/mL	29.80 ± 1.25	30.43 ± 1.34	.738
IFN-γ, pg/mL	12.71 ± 1.18	11.87 ± 1.06	.604
IL-1α, pg/mL	24.17 ± 2.04	25.38 ± 1.71	.649
IL-1β, pg/mL	50.96 ± 3.81	55.89 ± 4.56	.433
IL-6, pg/mL	72.53 ± 16.69	67.27 ± 12.95	.945
IL-4, pg/mL	15.10 ± 0.48	16.66 ± 1.29	.326
TNF-α, pg/mL	34.77 ± 1.05	36.09 ± 1.65	.538
CD3+ T cells, %	57.35 ± 2.06	52.48 ± 1.52	.526
CD3+CD4+ T cells, %	36.91 ± 2.35	35.87 ± 2.71	.897
CD3+CD8+ T cells, %	19.23 ± 1.70	20.39 ± 3.01	.344
CD4+/CD8+ T cells, %	2.07 ± 0.22	2.91 ± 0.73	.285
Th1 (CD4+CXCR3+) cells, %	20.12 ± 1.37	23.45 ± 2.07	.220
Th2 (CD4+CCR4+) cells, %	9.88 ± 0.84	7.67 ± 0.55	.034
Th17 (CD4+CCR6+) cells, %	19.00 ± 2.04	21.71 ± 2.27	.396
Treg (CD25+CD127−) cells, %	6.55 ± 0.38	4.07 ± 0.26	.000
NK (CD3−CD56+) cells, %	28.58 ± 2.58	23.89 ± 2.78	.223
NKT (CD3+CD56+) cells, %	9.10 ± 0.79	9.30 ± 1.13	.998
γδT (CD3++) cells, %	5.27 ± 0.67	5.28 ± 0.71	.993
B (CD19+) cells, %	23.21 ± 1.51	19.85 ± 1.05	.066
Breg (CD24+CD38+) cells, %	2.20 ± 0.36	2.61 ± 0.28	.363
Monocytes (CD14+) cells, %	11.95 ± 1.22	12.23 ± 1.15	.871
TPO, pg/mL	120.34 ± 47.12	128.10 ± 28.63	.883
Total MK number	89.2 ± 86.0	127.8 ± 131.4	.455
Immature MK, %	3.5 (4.3)	7.0 (6.3)	.794
Granular MK, %	66.1 (80.7)	100.7 (79.3)	.932
Platelet-producing MK, %	9.9 (12.1)	17.3 (10.1)	.458
Naked MK, %	2.4 (2.9)	2.2 (4.0)	.216
MAIPA+	11 (31.4)	19 (39.5)	.312

PLT was shown as median (IQR), the other variables were shown as mean ± SD or number (percentage). A P value <.05 indicates statistical significance.

Breg, regulatory B cells; IFN-γ, interferon gamma; IL-1α/1β/2/4/5/6/10/12/17A/21, interleukin-1α/1β/2/4/5/6/10/12/17A/21; IP-10, interferon–inducible protein 10; MAIPA, monoclonal antibody-specific immobilization of platelet antigen; MCP-1, monocyte chemoattractant protein 1; MK, megakaryocyte; NK, natural killer cells; NKT, natural killer t cells; NR, no response; TNF-α, tumor necrosis factor α.

Figure 1.

Biomarkers associated with sustained response after drug withdrawal. (A) Dynamic changes in platelet counts between patients with ITP-sus and patients with ITP-nonsus. (B) Total TGF-β1 and (C) activated TGF-β1 levels in patients with ITP-sus and patients with ITP-nonsus. (D) Relative ITGB8 gene expression, (E) Treg cell levels, and (F) Th2 cell levels in patients with ITP-sus and patients with ITP-nonsus. (G) Protein levels of integrin αv, integrin β8, and p-Smad2 in patients with ITP-sus and those with ITP-nonsus. (H) Activated TGF-β1 levels in the bone marrow of 7 patients with ITP. (I) Megakaryocyte ploidy distribution in the bone marrow of 7 patients with ITP. (J) Protein expression in nucleated bone marrow cells from 7 patients with ITP. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001. AKT, protein kinase B; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; ns, no significance; P1, patient 1; p-AKT, phosphorylated AKT; PLT, platelets.

Integrin αvβ8–activated TGF-β1 inhibits T-cell activation

Gene (ITGB8) and protein analyses revealed a significant reduction in integrin β8 expression in patients with ITP-nonsus. The phosphorylation level of Smad2 (phosphorylated Smad2 [p-Smad2]), commonly used as a surrogate marker to assess TGF-β1 activation, followed a similar pattern to integrin β8 expression and Treg cell levels (Figure 1G).

CD4⁺ T cells were cocultured with recombinant TGF-β1, recombinant integrin αvβ8, or anti-β8 antibody, with no observed effect on cell apoptosis. qPCR analysis showed significant upregulation of Foxp3 gene expression with TGF-β1 and integrin αvβ8. Flow cytometry demonstrated that recombinant TGF-β1 and integrin αvβ8 promoted Treg differentiation, an effect reversed by the anti-β8 antibody (Figure 2A-C).

Figure 2.

Integrin αvβ8 activated TGF-β1 inhibits T-cell activation. (A-B) Differentiation of Treg from CD4⁺ T cells of healthy volunteers and patients with ITP after culture with PBS, TGF-β1, recombinant integrin αvβ8, and anti-β8 antibody blocking agents. (C) Relative mRNA expression levels of FOXP3 in CD4⁺ T cells from healthy volunteers and patients with ITP after culture with PBS, TGF-β1, recombinant integrin αvβ8, and anti-β8 blocking agents. (D) CFSE assay assessing T-cell proliferation in PBMC from patients with ITP and healthy volunteers after 96 hours of treatment with TGF-β1, recombinant integrin αvβ8, and anti-β8 antibody blocking agents. (E) Bar graph displaying the division index of T cells. (F) Western blot analysis of protein expression in PBMC from patients with ITP and healthy volunteers after 96 hours of treatment with TGF-β1, recombinant integrin αvβ8, and anti-β8 antibody blocking agents. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HC, healthy controls; PBS, phosphate-buffered saline; PE-CY7, phycoerythrin-Cy7 conjugate.

CFSE assays indicated that T-cell proliferation was significantly inhibited by recombinant TGF-β1 and integrin αvβ8, with this effect reversed by anti-β8 antibody treatment (Figure 2D-E). p-Smad2 signaling proteins were upregulated in the recombinant TGF-β1 and integrin αvβ8 groups, whereas they were downregulated in the anti-β8 antibody group (Figure 2F).

Activated TGF-β1 regulates megakaryocyte differentiation

Activated TGF-β1 levels were also assessed in bone marrow samples from 7 patients with ITP, ranked from highest to lowest (patient 1 to patient 7). A positive correlation was observed between polyploid megakaryocytes (≥16 N) and activated TGF-β1 levels (Figure 1H). Additionally, western blot analysis demonstrated a decrease in phosphorylated STAT5, phosphorylated protein kinase B, and p-SMAD2 signaling from patient 1 to patient 7, indicating that lower activated TGF-β1 levels are associated with reduced phosphorylation of proteins that promote platelet production (Figure 1I-J).

TPO combined with integrin β8 prolongs sustained platelet response after drug withdrawal in ITP mice

We generated PF4-TGF-β1 CKO mice and established a passive ITP model, in which total and activated TGF-β1 levels were nearly undetectable in the bone marrow and ∼50% lower in the peripheral blood at baseline in Cre⁺ mice. After intervention, total and activated TGF-β1 levels were elevated in the TPO+integrin αvβ8^high group in both Cre⁻ and Cre⁺ mice (Figure 3A-C).

At baseline, platelet counts were higher in Cre⁺ mice than Cre⁻ mice (875 × 10⁹/L ± 102 × 10⁹/L vs 733 × 10⁹/L ± 98 × 10⁹/L). After anti-CD41 injection, platelet levels dropped rapidly and remained low throughout the treatment and follow-up period in Cre⁺ mice compared to Cre⁻ mice. In the TPO-monotreatment group, platelets returned to baseline by day 8 and stabilized at above 300 × 10⁹/L by day 28 in Cre⁻ mice, whereas they returned to baseline by day 10 but decreased to their lowest levels by day 21 in Cre⁺ mice. Notably, TPO+integrin αvβ8^high treatment restored platelet count to baseline by day 5 and maintained levels of >400 × 10⁹/L on day 28 in Cre⁻ mice (Figure 3D-F).

In the active ITP model, baseline platelet count did not significantly differ between the groups. After injection of spleen cells from immunized CD61 KO mice, platelet levels remained <100 × 10⁹/L in TPO monotherapy and TPO+β8^shT therapy groups. In contrast, the TPO+β8^siT group maintained significantly higher platelet levels during and after treatment, reaching 587 × 10⁹/L ± 63 × 10⁹/L by day 28 (Figure 3I-J). The levels of platelet autoantibodies are provided in supplemental Figure 3.

TPO combined with integrin β8 improves health in ITP mice

In the passive ITP model, Treg were significantly lower in Cre⁺ mice than in Cre⁻mice. TPO treatment led to a gradual increase in Treg, whereas TPO+integrin αvβ8^high treatment significantly enhanced Treg proportions (Figure 3G). Additionally, both TPO and TPO+integrin αvβ8^high treatments promoted macrophage polarization from M1 to M2. However, Th1, Th2, and Th17 cell populations showed no differences between the groups (supplemental Figure 2).

Similarly, Treg were significantly elevated in the TPO+β8^siT group and markedly suppressed in the TPO+β8^shT group (P < .001) in the active ITP model.

TPO combined with integrin β8 promotes the differentiation of high-ploidy megakaryocytes in ITP mice

Both hematoxylin and eosin staining and immunofluorescence analysis revealed that Cre⁺ mice had smaller megakaryocyte diameters. After CD41 antibody injection, both Cre⁻ and Cre⁺ mice exhibited megakaryocyte destruction. TPO+integrin αvβ8^high treatment significantly enhanced megakaryocyte diameter, the number of platelet-producing megakaryocytes, and the proportion of polyploid megakaryocytes in Cre⁻ mice but not in Cre⁺ mice (Figure 4A-C).

Figure 4.

TPO combined with integrin β8 promote the differentiation of high-ploidy MK in ITP mice. (A) Hematoxylin and eosin (H&E) staining (original magnification ×20) and immunofluorescence (original magnification ×40) were used to evaluate MK in Cre⁻ and Cre⁺ mice across different groups. (B-C) Measurement of MK number and diameter in Cre⁻ and Cre⁺ mice across groups. (D) Ploidy analysis of MK in Cre⁻ and Cre⁺ mice across groups using flow cytometry. (E) Giemsa staining to assess the proportion of MK stages in Cre⁻ and Cre⁺ mice across groups. (F) H&E staining (×20) to evaluate MK in the ITP active model across groups. (G) Giemsa staining to determine the percentage of MK at various stages in the ITP active model. (H-I) Flow cytometric analysis of MK polyploidy in the ITP active model across groups. ∗P < .05; ∗∗P < .01. gra MK, granular megakaryocyte; MK, megakaryocytes; nak-mk, naked megakaryocyte; PI, propidium iodide; pla-mk, platelet-forming megakaryocyte; pro-mk, promegakaryocyte.

Compared with the passive model, the active model exhibited more severe megakaryocyte damage and clearance (Figure 4F). In the saline-treated group, megakaryocyte numbers were largely destroyed, with extensive erythrocyte aggregation indicating severe hemorrhage. The TPO group showed a few intact megakaryocytes, albeit smaller in diameter. The TPO+β8^siT group did not significantly increase megakaryocyte numbers compared with the TPO group, but there was a slight increase in megakaryocyte diameter. Notably, the TPO+β8^shT group displayed extensive megakaryocyte destruction, with almost no intact megakaryocytes, suggesting accelerated destruction due to β8 subunit deficiency (Figure 4F). Additionally, the proportion of polyploid megakaryocytes and the number of platelet-producing megakaryocytes increased in the TPO+β8^siT group but decreased in the TPO+β8^shT group (Figure 4G-I).

TPO-RA regulate integrin β8 through the AP-1 family

Single-cell sequencing was performed on samples from 4 patients with ITP-sus and 4 patients with ITP-nonsus. Cluster analysis of T cells showed that the ITP-sus group had high expression levels of FOS-JUN-CD4⁺ T cells, FOS-JUN-natural killer cells, and FOXP3⁺ Treg, whereas the ITP-nonsus groups had elevated levels of CD4⁺ and CD8⁺ cytotoxic T lymphocyte (Figure 5A). Gene analysis revealed that activator protein 1 (AP-1) family genes, including FOS, FOSB, JUN, JUNB, and JUND, as well as CD69 were significantly elevated in the ITP-sus group, whereas markedly reduced in ITP-nonsus group. This suggests that treatment with TPO-RA leads to the restoration of gene regulation in the sustained-remission group. Pathway enrichment analysis of the differential genes revealed significant associations with AP-1 signaling pathways, Smad signaling pathways, immune modulation, and integrin binding (Figure 5B-C).

Figure 5.

TPO-RA regulate integrin β8 through the AP-1 family. (A) Percentage of T-cell subpopulations in patients with ITP-sus and patients with ITP-nonsus. (B) Pathway enrichment analysis. (C) Differential gene expression in single-cell sequencing. (D) qPCR analysis of differential gene expression in CD4⁺ T cells from patients with ITP-sus and patients with ITP-nonsus after treatment with different concentrations of Eltro. (E) Differential protein expression in peripheral blood samples from patients with ITP-sus and patients with ITP-nonsus. (F) Differential protein expression in CD4⁺ T cells from patients with ITP-sus and patients with ITP-nonsus after treatment with different concentrations of Eltro. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001. Eltro, eltrombopag; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GZMH_MK, megakaryocyte; NK, natural killer cells; P-STAT5, phosphorylated STAT5; UBB, ubiquitin B; UBC, ubiquitin C.

Subsequent in vitro validation was conducted using CD4⁺ T cells from patients with ITP treated with increasing concentrations of eltrombopag. In patients with ITP-sus, gene and protein expression of JUN, JUND, FOS, and FOSB increased in a dose-dependent manner, a pattern not observed in patients with ITP-nonsus. Integrin β8 expression and p-Smad2 showed similar trends. Consistent with previous findings, these results highlight AP-1 family members as upstream regulators of ITGB8 transcription (Figure 5D-F).

TPO combined with D-mannose improves the sustained response

PBMC were cultured and treated with eltrombopag (2000 ng/mL) or a combination of eltrombopag and D-mannose (25mM). The combination treatment significantly enhanced Treg differentiation and increased integrin β8 and p-Smad2 levels, whereas markedly inhibiting CD4⁺ T-cell proliferation (Figure 6A-D).

Figure 6.

TPO combined with D-man enhances sustained response in ITP mice. (A) Flow cytometric analysis of Treg proportions from HC and patients with ITP treated with PBS, Eltro, or Eltro + D-man. (B) CFSE assay to assess T-cell proliferation from HC and patients with ITP after treatment with PBS, Eltro, or Eltro + D-man. (C) qPCR analysis of ITGB8 gene expression in CD4⁺ T cells from HC and patients with ITP treated with PBS, Eltro, or Eltro + D-man. (D) Western blot analysis of integrin β8 protein and phosphorylation levels of Smad2 signaling pathway in CD4⁺ T cells treated with PBS, Eltro, or Eltro + D-man. (E) Platelet dynamics during and after treatment with TPO, D-man, or TPO + D-man in Cre⁻ and Cre⁺ mice. (F) Levels of active TGF-β1 in the peripheral blood and bone marrow of Cre⁻ and Cre⁺ mice after 2 weeks of treatment with TPO, D-man, or TPO + D-man. (G) Treg levels in Cre⁻ and Cre⁺ mice after 2 weeks of treatment with TPO, D-man, or TPO + D-man. (H-I) MK analysis in the bone marrow of Cre⁻ and Cre⁺ mice after 2 weeks of treatment with TPO or TPO + D-man (original magnification, ×20 for H&E and ×40 for immunofluorescence). (J) Platelet dynamic during and after treatment in ITP active models with TPO, D-man, or TPO + D-man. (K) Expression of integrin β8 and Smad2 phosphorylation in the spleens of ITP passive and active models treated with TPO or TPO + D-man. (L-M) Treg levels in ITP active model after 2 weeks of treatment with TPO or TPO + D-man. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001. D-man, D-mannose; Eltro, eltrombopag; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HC, healthy controls; P-STATS3, phosphorylated STAT3; PBS, phosphate-buffered saline; PLT, platelet.

In ITP mice, TPO combined with D-mannose significantly elevated activated TGF-β1 levels in both the peripheral blood and bone marrow, with a more pronounced effect in Cre⁻ mice compared with Cre⁺ mice (P < .001; Figure 6F). Platelet counts were higher and had a longer sustained response time in the TPO+D-man group than with TPO monotherapy, particularly in Cre⁻ mice (Figure 6E). In the active ITP model, this combined treatment also resulted in higher platelet counts and better maintenance after treatment. Additionally, increased integrin β8 expression and Smad2 phosphorylation were observed in spleen cells of the TPO+D-man group (Figure 6I-J).

Flow cytometry revealed a significant increase in Treg in the TPO+D-man group in both the passive and active models (Figure 6G,K-L). However, there were no significant differences in megakaryocytes maturation between the groups.

Discussion

With the emergence of TPO-RA, their role in ITP management has greatly expanded. However, challenges remain in reducing drug dependence and achieving lasting remission after discontinuation. In this study, we found that activated TGF-β1 levels were significantly elevated in those who maintained a sustained response after stopping TPO-RA. Further research demonstrated that TPO-RA upregulate the integrin β8 subunit via the AP-1 pathway, leading to TGF-β1 activation in patients with sustained responses. Activated TGF-β1 increased Treg, promoted M1-to-M2 macrophage polarization, enhanced immunoregulation, and facilitated the differentiation of high-ploidy megakaryocytes. Furthermore, combining D-mannose with TPO-RA upregulates integrin αvβ8 and activates TGF-β1, prolonging the sustained response after TPO discontinuation in ITP mice.

Studies have shown that treatment with TPO-RA leads to a significant enrichment of Treg in the spleen, highlighting the immunomodulatory potential of TPO-RA.⁹ Research also reveals pronounced activation of tumor necrosis factor α signaling via the NF-κB pathway in nonsustained responders, suggesting distinct immune regulatory effects.⁵ Notably, TPO-RA responders exhibit elevated TGF-β1 levels.⁶ Recently, Wang et al reported that TPO-RA promote platelet-derived TGF-β1–mediated reprogramming of myeloid-derived suppressor cells via the TGF-β/Smad pathway, suggesting that platelets may help maintain immune homeostasis and regulate their own survival.¹⁴ TGF-β1, a multifunctional cytokine primarily secreted by platelets and megakaryocytes in its latent form, requires integrin αvβ8 for activation. Once activated, it promotes Treg expansion and modulates adaptive immunity, underscoring the growing recognition of immunoregulatory functions of TPO-RA.

TGF-β1 plays a critical role in immunosuppression, promoting immune escape through Treg differentiation.^16,23 Neutralizing TGF-β1's immunoregulatory effects is a promising anticancer strategy. Furthermore, in melanoma and breast cancer models, high expression of integrin αvβ8 activates TGF-β1, impairing CD8⁺ cytotoxic T-cell function.^19,23 Conversely, the absence of integrin β8 or use of anti-β8 antibodies disrupts TGF-β1 signaling, restores T-cell cytotoxicity, and controls tumor growth.¹⁹ However, in a graft-versus-host disease model, blocking β8-exacerbated graft-versus-host disease by inhibiting Treg-mediated immune suppression.²⁴

In autoimmune diseases characterized by heightened cytotoxic responses and reduced Treg, moderate integrin αvβ8 expression can activate TGF-β1, enhancing Treg differentiation and suppressing cytotoxic T-cell activity, offering a potential therapeutic approach.²⁰ Travis et al reported that integrin αvβ8 deficiency in leukocytes led to severe inflammatory bowel disease and age-related autoimmune disorders in mice, characterized by reduced Treg levels and elevated circulating immunoglobulin A (IgA), IgG, and IgE.²⁵ Of interest, Kelly et al found that integrin αvβ8–mediated TGF-β1 activation promoted immunomodulation in human monocytes and macrophages, increasing M2 macrophage populations whereas reducing M1 macrophage levels.¹⁸ Furthermore, studies on thrombocytopenia after allogeneic transplantation reported an increase in M1 macrophages and a decrease in M2 macrophages, a shift resulting in impaired megakaryocyte maturation.²⁶ Our approach to ITP therapy using integrin αvβ8 showed that combining TPO with integrin αvβ8 significantly increased Treg and shifted macrophages from proinflammatory M1 to anti-inflammatory M2 phenotype, restoring immune tolerance.

The role of TGF-β1 in megakaryocyte regulation remains debated. Early studies suggested that TGF-β1 inhibits the differentiation of CD34⁺ stem cells into megakaryocytes.²⁷ However, Badalucco et al later found that human megakaryocytes continuously release TGF-β1 in vitro, regulating proplatelet formation and megakaryocyte maturation.²⁸ Recent research showed that mice lacking TGF-β1 in megakaryocytes and platelets exhibited a 30% increase in megakaryocytes and a 15% increase in platelets.²⁹ Consistently, we also observed increased baseline platelet and megakaryocyte numbers in TGF-β1 CKO mice. However, after establishing an ITP model, CKO mice showed more severe platelet and megakaryocyte damage due to the loss of TGF-β1–mediated immune protection, suggesting that TGF-β1 regulates the differentiation of megakaryocytes.

AP-1 is a procancer transcription factor involved in cell proliferation, migration, and invasion.^30,31 TGF-β1 activates JUN via the TGF-β–JNK axis by phosphorylating serine residues, and JUN family proteins interact directly with Smad proteins to regulate downstream gene expression.^32,33 The phosphoinositide 3-kinase–JNK–AP-1 signaling pathway synergistically stimulates TGF-β1 expression, which is crucial for Treg differentiation and functional activation.34, 35, 36 In systemic lupus erythematosus, AP-1 expression is markedly reduced in the peripheral blood, likely due to immune cell dysregulation.^37,38 In lung fibroblasts, nuclear extract analysis showed that AP-1 complexes containing JunD and JunB bind to the cis-regulatory element‌ site in the ITGB8 core promoter, essential for ITGB8 promoter function.³⁰ This suggests that AP-1 regulates TGF-β1 activation through the transcriptional control of integrin β8, influencing downstream signaling.

D-mannose enhances integrin αvβ8 expression and reactive oxygen species production in T cells, thereby promoting Treg generation.²⁰ Our in vitro experiments demonstrated that combining TPO with D-mannose significantly increased Treg numbers. In vivo, this combination markedly elevated activated TGF-β1 levels and enhanced p-Smad2 phosphorylation in ITP mice. Increased integrin β8 expression was also observed in spleen cells of mice treated with this combination. These findings suggest that combining TPO with D-mannose enhances long-term maintenance after drug withdrawal, offering significant clinical value in improving patient outcomes.

This study has several limitations. First, we were unable to perform longer-term follow-up and to dynamically assess activated TGF-β1 levels at multiple time points. Second, we did not thoroughly investigate the regulatory relationship between activated TGF-β1 and the increase in high-ploidy, platelet-producing megakaryocytes, although this phenomenon was observed. Third, although single-cell sequencing data confirmed increased TGF-β1 activation and AP-1 family expression in sustained responders after drug discontinuation, the complex regulatory mechanisms of AP-1 proteins were not fully explored. Last, although D-mannose showed efficacy in ITP mice, clinical trials are necessary to validate its combination with TPO-RA in patients with ITP. These areas warrant further investigation.

In conclusion, this study demonstrates that plasma-activated TGF-β1, rather than total TGF-β1 levels, is significantly elevated in sustained responders after TPO-RA discontinuation in patients with ITP. Integrin αvβ8 regulated TGF-β1 activation, promoting Treg expansion, M1-to-M2 macrophage polarization, and the differentiation of high-ploidy megakaryocytes. TPO-RA upregulated integrin β8 expression through AP-1 family molecules. In ITP mice, combining TPO with D-mannose enhanced integrin αvβ8–mediated TGF-β1 activation, extending the sustained response after drug withdrawal. Overall, TGF-β1 emerges as a promising therapeutic target for ITP with significant potential for clinical translation.

Conflict-of-interest disclosure: The authors declare no competing financial interests.