Acute graft-vs.-host disease arises when donor immune cells transplanted into a patient recognize the recipient’s tissues as foreign and attack them. The gastrointestinal form is particularly dangerous: it is the leading cause of non-relapse mortality after allogeneic hematopoietic stem cell transplantation, the curative cellular therapy for leukemias, lymphomas, and disorders such as severe aplastic anemia. When the gut barrier breaks down, bacterial products flood into the bloodstream, driving sepsis and uncontrolled inflammation, while severe diarrhea causes fluid, electrolyte, and protein losses. Despite prophylaxis regimens such as calcineurin inhibitors combined with methotrexate or post-transplant cyclophosphamide, aGVHD still affects roughly 40 to 60 percent of transplant recipients, and treatment options beyond corticosteroids remain limited for those who do not respond.

The research team, based at the Army Medical University in Chongqing and collaborators, first turned to network pharmacology, a computational strategy suited to multi-component herbal formulas. Screening databases for compounds meeting criteria for oral bioavailability and drug-likeness, they identified 183 candidate molecules from GSZD and predicted 705 target genes. They then compiled 816 aGVHD-related genes from seven public databases. Intersecting the two lists yielded 109 genes shared between the formula’s predicted targets and aGVHD biology.

To capture genes specifically tied to gut damage rather than generic immune processes, the researchers analyzed two public transcriptomic datasets from gut tissues of patients with GI-aGVHD, comprising 70 samples in total. After correcting batch effects between datasets using ComBat and ComBat-seq algorithms, they applied differential expression analysis with DESeq2, identifying 2,378 differentially expressed genes, and weighted gene co-expression network analysis (WGCNA), which grouped genes into modules correlated with disease status. The intersection of these approaches produced 161 genes closely associated with aGVHD gut damage, 11 of which were also predicted targets of GSZD compounds.

Protein-protein interaction network analysis of the 109 shared genes ranked IL6, TNF, IL1B, AKT1, STAT3, PIK3CD, MTOR, NFKB1, TLR4, and BCL2 as the most connected nodes, and enrichment analyses highlighted the PI3K-Akt signaling pathway among the top disease-relevant pathways. Because AKT1 was highly connected and prior studies have implicated excessive PI3K-Akt activation in alloreactive T cell responses, the team selected AKT1 and FGF2, an upstream activator of the pathway found among the gut-damage genes, for exploratory molecular docking. Docking predicted favorable binding energies for 31 AKT1-compound pairs and two FGF2-compound pairs, though the authors stress these are computational estimates that do not demonstrate actual binding or functional inhibition.

The decisive test came in mice. The researchers used a classical fully mismatched model, transplanting bone marrow and splenocytes from C57BL/6 donor mice into lethally irradiated BALB/c recipients. GSZD, prepared by traditional decoction and quality-controlled by targeted UPLC-MS/MS that quantified nine marker compounds, was given orally from day 1 to day 7 after transplantation at a dose scaled from a clinical adult dose. Mice receiving GSZD survived significantly longer than untreated aGVHD mice, lost less body weight, and showed milder clinical scores through day 30.

The protective effects extended to the gut itself. Histological examination on day 14 revealed that aGVHD mice suffered severe intestinal injury, with shortened villi and disordered architecture in the small intestine and crypt damage in the colon, while GSZD-treated mice retained much better tissue structure. A fluorescent dextran assay showed that GSZD reduced the leakiness of the gut barrier, and levels of the tight junction proteins Occludin, Claudin-1, and ZO-1, which normally seal the intestinal lining, were partially restored. Blood tests also showed that the cytokine storm characteristic of aGVHD, with elevated IFN-gamma, TNF-alpha, IL-6, and IL-17A and depressed IL-4, was partially normalized by the formula.

To probe the mechanism, the team measured FGF2 in colon tissue and found it elevated in aGVHD mice but reduced by GSZD, alongside lower phosphorylation of PI3K and Akt. In a clever causal test, they delivered recombinant FGF2 directly into the colon of some mice. FGF2 alone did not significantly worsen disease, possibly because aGVHD was already severe enough to leave little room for further damage. But in GSZD-treated mice, supplemental FGF2 partially reversed the formula’s protective effects on survival, weight, histology, and barrier function, accompanied by reactivation of PI3K/Akt signaling. This suggests the FGF2/PI3K/Akt axis contributes meaningfully to the formula’s benefit, though the authors note FGF2’s effects may be context-dependent, since it promotes tissue repair in some inflammatory gut conditions.

The study has important limitations. It was conducted exclusively in male mice with a single donor-recipient strain combination and a single dose, and control mice received saline rather than a vehicle matched to the decoction’s physical properties. The transcriptomic datasets contained far more aGVHD samples than controls, and the compound screening criteria excluded some abundant constituents, including the most plentiful marker compound, timosaponin BII. The authors also caution that suppressing Th1 and Th17 responses, while beneficial against aGVHD, could theoretically weaken the graft-vs.-tumor effect that helps prevent relapse, an open question requiring dedicated experiments. No human trials have yet tested GSZD in transplant patients, so these findings remain preclinical. Still, the study represents the first systematic evaluation of this formula in GI-aGVHD and provides a computational-plus-experimental template for identifying how complex herbal mixtures might complement modern transplant medicine.

Subject of Research: Preclinical evaluation of Guizhi-Shaoyao-Zhimu Decoction against gastrointestinal acute graft-vs.-host disease via FGF2/PI3K/Akt signaling

Article Title: Guizhi-Shaoyao-Zhimu decoction alleviates gastrointestinal acute graft-vs.-host disease and suppresses FGF2/PI3K/Akt signaling: insights from network pharmacology, WGCNA, molecular docking, and animal experiment

Article References: Wan, X., Li, N., Yang, X., Yang, W., Liu, S., & Zhang, X. (2026). Guizhi-Shaoyao-Zhimu decoction alleviates gastrointestinal acute graft-vs.-host disease and suppresses FGF2/PI3K/Akt signaling: insights from network pharmacology, WGCNA, molecular docking, and animal experiment. Frontiers in Medicine, 13, Article 1908959. https://doi.org/10.3389/fmed.2026.1908959

Image Credits: AI Generated

DOI: 10.3389/fmed.2026.1908959

Keywords: acute graft-vs.-host disease, bone marrow transplantation, network pharmacology, WGCNA, molecular docking, FGF2, PI3K/Akt signaling, traditional Chinese medicine, gut barrier, mouse model, transplant immunology, herbal formula