Methylphenidate hydrochloride extended-release tablets (Batch No.: HJ20225002) were purchased from Xi’an Janssen Pharmaceutical Ltd (Xi’an, China). Primary antibodies against glial fibrillary acidic protein (GFAP, 60190-1-Ig), dopamine deceptor D1 (DRD1, 17934-1-Ap), and BDNF (66292-1-Ig) were obtained from Proteintech (Wuhan, China). RNA was extracted using the EZB-RN10 kit from EZBioscience (Shanghai, China). Reverse transcription and quantitative PCR were performed respectively using the reverse transcription (kitRR092A) and the QuantityNova SYBR Green PCR kit (RR420A), both of which were purchased from Takara Bio (Beijing, China).

LMNXD comprises 15 herbal components: raw Os Draconis (15 g), raw Concha Ostreae (15 g), Magnetitum (15 g), Rehmanniae Radix Praeparata (9 g), Plastrum Testudinis (9 g), Rhizoma Acori Tatarinowii (9 g), Radix Polygalae Praeparata (9 g), Fructus Alpiniae Oxyphyllae (9 g), Radix Ophiopogonis (9 g), Radix Asparagi (9 g), Radix Angelicae Sinensis (9 g), Poria cum Radix Pini (9 g), Fructus Schisandrae (6 g), Triticum aestivum (15 g), Rhizoma Dioscoreae (15 g). All herbs were purchased from Longhua Hospital, Shanghai University of Traditional Chinese Medicine, and processed into granular form.

The chemical constituents of LMNXD were analyzed by ultra-high performance liquid chromatography-Orbitrap mass spectrometry (UHPLC-Orbitrap-MS). To prepare the LMNXD extract, 300 μL of supernatant was combined with 1,000 μL of an extraction solution (methanol:acetonitrile:water = 2:2:1, v/v/v) containing isotope-labeled internal standards. This mixture was vortexed for 30 s, sonicated in an ice bath for 5 min, incubated at −20 °C for 1 h, and centrifuged at 12,000 rpm for 15 min at 4 °C. The resulting supernatant was passed through a 0.22 μm membrane filter to yield the final filtrate. Chromatographic separation was carried out on a Vanquish UHPLC system (Thermo Fisher Scientific) fitted with a Phenomenex Kinetex C18 column (2.1 mm × 100 mm, 2.6 μm). The mobile phase comprised 0.01% acetic acid in water (solvent A) and a 1:1 (v/v) mixture of isopropanol and acetonitrile (solvent B). Mass spectrometric detection employed an Orbitrap Exploris 120 MS operating in both positive and negative ion modes, with the following key parameters: sheath gas and auxiliary gas flow rates of 50 and 15 Arb, respectively; a capillary temperature of 320 °C; full MS resolution of 60,000; MS/MS resolution of 15,000; and spray voltages of ±3.8/3.4 kV. Data were processed using ProteoWizard and an in-house R software package integrated with the Biotree TCM/BT-HERB database.

Thirty 4-week-old male SHR/NCrl rats, representing an age equivalent to 9-year-old human males (https://www.taletn.com/rats/age/), and six age-matched male WKY/NCrl rats, were sourced from Beijing Vital River Laboratory Animal Technology Co., Ltd. (SCXK (Jing) 2021-0006). All animals were housed under standard conditions at 23 °C ± 2 °C, with relative humidity maintained between 50% and 60% and a 12-h light/dark cycle. Food and water were available ad libitum. All experimental procedures strictly followed the guidelines of the Chinese Institute of Laboratory Animal Sciences and internationally recognized principles for the care and use of laboratory animals. The experimental protocol was approved by the Institutional Animal Care and Use Committee of SHRM (Approval No.: LL-2025-299).

SHR/NCrl rats were randomly divided into five groups: the model group (SHR), the LMNXD low-dose group (LMNXD-LD), the LMNXD medium-dose group (LMNXD-MD), the LMNXD high-dose group (LMNXD-HD), and the MPH group (4.22 mg/kg/d). WKY/NCrl rats served as the blank control group.The clinically effective pediatric dose of LMNXD is 162 g/d. The dose administered to the animals was based on the body surface area of a 9-year-old (body mass 26 kg) child (0.991 m², Stevenson’s formula = 0.0061×height (cm)+0.0128×body weight (kg)-0.1529) and 4-week-old (body mass 60 g) SHR pups body surface area (0.013947 m², Meeh-Rubner formula = 9.1 × body weight (g)^2⁄3/10,000) were converted in parallel (Li et al., 2024). This conversion yielded a clinically equivalent dose of 38 g/kg/d for SHR/NCrl rats. The LMNXD-MD group received this clinically equivalent dose (38 g/kg/d), the low dose was set to half the clinically effective dose (19 g/kg/d), and the high dose was twice the clinically effective dose (76 g/kg/d). The TCM granules were dissolved in distilled water before administration. Intragastric administration was conducted at 1 mL/100 g body weight, twice daily (8:00–9:00 a.m. and 14:00–15:00 p.m.), over four consecutive weeks. The SHR and WKY groups received an equivalent volume of normal saline via intragastric administration.

After behavioral testing, the rats were fasted for 12 h and anesthetized with Zoletil^® 50 (50 mg/kg). The animals were then decapitated using a guillotine. The PFC, striatum, and hippocampus were rapidly dissected on ice with glass needles after removing the skin and skull. Any residual blood or hair was carefully rinsed away with cold phosphate-buffered saline (PBS). The dissected tissues were immediately frozen in liquid nitrogen and stored at −80 °C.

For immunohistochemistry, three rats were randomly selected from each group. PFC tissue was fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned at a thickness of 3–4 μm. The sections underwent deparaffinization, antigen retrieval, endogenous peroxidase quenching, and blocking with 3% BSA. They were then incubated overnight at 4 °C with a primary antibody DRD1 (1:100). Following PBST washes, the sections were incubated for 2 hours at room temperature with an HRP-labeled goat anti-rabbit IgG secondary antibody and developed with DAB, which produced a brown-yellow color indicative of positive staining. Nuclei were counterstained with hematoxylin for 3 minutes before the sections were dehydrated and mounted. Finally, the sections were scanned under a microscope, and images from three randomly chosen fields were captured for DRD1 expression analysis using ImageJ software (Mizote et al., 2024).

Hippocampal tissues from three randomly selected rats per group were processed for immunofluorescence staining. Sections were dewaxed, underwent antigen retrieval, and blocked with FBS. They were then incubated overnight at 4 °C with GFAP antibody (1: 100). After washing, the sections were treated with a fluorophore-conjugated secondary antibody and counterstained with DAPI to label nuclei. Fluorescence microscopy was used to scan the sections, and three random high-power fields per sample were selected for evaluation. The mean value from these three fields served as the final result for each sample. GFAP expression was quantified using ImageJ software.

Striatum tissues from six randomly selected rats per group were analyzed by Western blot. The tissues were homogenized and lysed before centrifugation to collect the supernatant, whose protein concentration was measured with a BCA protein quantification kit (P0012, Beyotime, China). Proteins were separated using 10% SDS-PAGE, loading 40 μg per lane. After electrophoresis and transfer, the membranes were blocked for 30 min with a rapid protein-free blocking solution (BL1032B). They were then incubated overnight at 4 °C with the appropriate primary antibodies. Following washes, the membranes were incubated with corresponding secondary antibodies for 1 hour at room temperature. Protein bands were visualized via ECL detection, and the grayscale values of target bands were quantified using ImageJ software.

The structures and SMILES identifiers for the 70 compounds identified in the LMNXD mass spectrometry analysis were retrieved from the PubChem database and ChemDraw. These compounds then underwent ADME screening via the SwissADME platform. Their structures were submitted to the SwissTargetPrediction and STITCH databases to predict potential targets, which were subsequently compiled and deduplicated to yield the final target set. Disease targets related to ADHD were collected by querying the OMIM, GeneCards, DrugBank, and PharmGKB databases with the keyword “Attention-deficit/hyperactivity disorder.” The intersection of LMNXD and ADHD targets was defined as the potential therapeutic target genes of LMNXD for ADHD. A protein-protein interaction (PPI) network for these overlapping targets was built using the STRING database. The resulting network file was imported into Cytoscape for visualization. Finally, a “component-target-disease-pathway” network was constructed in Cytoscape.

Hippocampal tissues were randomly selected from four rats in the WKY group, SHR group, and LMNXD-M group for transcriptomic analysis. Total RNA was extracted using TRIzol reagent. The integrity and quantity of RNA were assessed using the Agilent 2100 bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Purity and concentration were simultaneously determined using the NanoDrop ND-1000 (NanoDrop, DE, USA). Total RNA was used as the starting material for mRNA enrichment and fragmentation, followed by cDNA synthesis. The cDNA underwent end repair, adapter ligation, size selection, and amplification to construct the libraries, which were then quality-controlled using Qubit, Agilent 2100, and qRT-PCR. After passing quality control, these libraries were sequenced on the Illumina Novaseq 6000 platform, generating 150-bp paired-end reads. Raw data were converted into FASTQ files, and clean reads were obtained after filtering out low-quality sequences. HISAT2 was used to align the reads to the reference genome. Novel transcripts were predicted with StringTie, and featureCounts quantified reads mapped to each gene. Principal Component Analysis (PCA) was conducted on the gene expression values (FPKM) of all samples to evaluate biological replicates. Differential expression analysis was performed using DESeq2 to identify significantly differentially expressed genes (DEGs), with thresholds established at |log2(Fold Change)| ≥ 1 and an adjusted P-value (Padj) ≤ 0.05.

To investigate the mechanisms underlying LMNXD treatment for ADHD, we conducted an integrated and annotated analysis of network pharmacology targets and transcriptomic DEGs using DAVID (https://david.ncifcrf.gov/). This included functional annotation through Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. The GO and KEGG enrichment results were then visualized with bioinformatics tools.

Total RNA was extracted from 40 mg of hippocampal tissue with Trizol reagent. cDNA was synthesized from the extracted RNA using the Invitrogen First Strand cDNA Synthesis Kit (RR092A). Primers targeting ATPase Plasma Membrane Ca²⁺Transporting 4 (ATP2B4), Glutamate Receptor Ionotropic NMDA 3A (GRIN3A), Oxytocin Receptor (OXTR), Collagen Type VI Alpha 2Chain (COL6A2), and Integrin Subunit Alpha 1 (ITGA1) were designed via NCBI Primer-BLAST based on GenBank sequences, with their specific sequences provided in Table 1. Amplification of cDNA was then carried out with a SYBR Green PCR kit under the following cycling protocol: 95 °C for 30 s, then 40 cycles of 95 °C for 5 s and 60 °C for 10 s. The detailed composition of the PCR reaction mixture is specified in Table 2. Relative mRNA expression levels for the target genes were calculated according to the 2^−ΔΔCT method. The RT-qPCR analysis included four biological replicates per experimental group.

Data analysis was performed with SPSS 25.0 software, and measurement data are presented as the mean ± standard deviation. For the data of each index, we first performed the normality test and the homogeneity of variance test. Intergroup comparisons employed one-way analysis of variance (ANOVA), while multiple comparisons used the LSD and Dunnet methods. Datasets involving treatments and time points were analyzed with a two-factor repeated measures analysis of variance. A P value <0.05 was considered statistically significant. Graphs were generated using GraphPad Prism 8.0 software.

A comprehensive analysis using UHPLC-OE-MS qualitatively and quantitatively determined the chemical constituents of LMNXD, leading to the identification of 70 distinct chemical structures (Supplementary Table 1). These constituents comprised seven flavonoids, such as Liquiritigenin, six phenylpropanoids including N-p-trans-Coumaroyltyramine, and six phenolic acids like 4-Hydroxybenzaldehyde. The analysis also identified five terpenoids, for example, Nootkatone, four lignans such as Schisandrin B, three alkaloids including 3-Pyridinemethanol, and 38 other components (Figures 2A,B).

The body weights of the rats in each group are shown in Table 3, and LMNXD had no significant effect on the weight gain of the rats.

The OFT assessed the effects of LMNXD on rat locomotor activity by measuring the total distance moved, average speed, and resting time. Before treatment, the SHR, LMNXD-LD + SHR, LMNXD-MD + SHR, LMNXD-HD + SHR, and MPH + SHR groups exhibited significantly greater total distance and average speed (P < 0.0001, Figures 3A,B) and significantly less resting time (P < 0.0001, Figure 3C) than the WKY group.

Following 4 weeks of treatment, the SHR group continued to display significantly elevated total distance and average speed (P < 0.0001, Figures 3A,B) and significantly reduced resting time (P < 0.0001, Figure 3C) relative to the WKY group. By contrast, the LMNXD-MD + SHR, LMNXD-HD + SHR, and MPH + SHR groups showed significant reductions in total distance and average speed (P < 0.0001, Figures 3A,B) and a significant increase in resting time (Figure 3C) compared to the SHR group. Furthermore, relative to their pre-treatment baselines, the LMNXD-LD + SHR, LMNXD-MD + SHR, LMNXD-HD + SHR, and MPH + SHR groups demonstrated significantly decreased total distance and average speed (Figures 3A,B) and significantly increased resting time (Figure 3C) after the 4-week treatment period. These findings indicate that LMNXD effectively ameliorates hyperactive and impulsive behaviors in SHR/NCrl rats, with definitive efficacy at medium and high doses. Representative movement trajectories for each group after administration are presented in Figure 3D.

The MWM test assessed the effect of LMNXD on spatial learning and memory in SHRs/NCrl rats. From day 3 onward, the escape latency was significantly shorter in the LMNXD-MD + SHR, LMNXD-HD + SHR, MPH + SHR, and WKY groups than in the SHR group (Figure 3E). On day 4, the LMNXD-LD + SHR group also displayed a significantly shorter escape latency than the SHR group (P < 0.05, Figure 3E).

During the spatial probe trial on day 6, all four treatment groups showed an increased number of platform crossings relative to the SHR group, but significant differences emerged only for the MPH + SHR and LMNXD-MD + SHR groups (P < 0.01, Figure 3F). The SHR group spent significantly less total distance and time in the target quadrant than the WKY group (P < 0.01, Figures 3G,H). In contrast, the total swimming distance and time in the target quadrant were significantly greater in the LMNXD-LD + SHR, LMNXD-MD + SHR, and MPH + SHR groups than in the SHR group (Figure 3G). The LMNXD-HD + SHR group covered a significantly longer distance in the target quadrant compared to the SHR group (P < 0.05, Figure 3H), although the increase in time spent there did not reach statistical significance. These findings demonstrate that LMNXD effectively enhances spatial memory in SHR/NCrl rats, with the medium dose yielding the most substantial therapeutic effect. Representative swimming paths for each group are presented in Figure 3I.

GFAP serves as a classical marker of astrocyte activation. This activation promotes a pro-inflammatory state, which is significantly associated with the disease status and severity of ADHD. Immunofluorescence analysis evaluated the effect of LMNXD on the GFAP-positive area percentage in the rat hippocampus. The SHR group exhibited a significantly increased GFAP-positive area compared to the WKY group (P < 0.01) (Figures 4A,D). In contrast, the LMNXD-MD + SHR, LMNXD-HD + SHR, and MPH + SHR groups all showed a significant reduction relative to the SHR group (P < 0.01). Although the LMNXD-LD + SHR group also displayed a decrease, this difference was not statistically significant.

Distributed primarily in the PFC and striatum, DRD1 mediates excitatory G-protein signaling and regulates attention, impulse inhibition, and reward learning, making it a key node in ADHD-associated dopamine dysfunction. Immunohistochemistry assessed the impact of LMNXD on the DRD1-positive area percentage in the rat PFC. The DRD1-positive area was significantly lower in the SHR group than in the WKY group (P < 0.01) (Figures 4B,E). Compared to the SHR group, significant increases were observed in the LMNXD-MD + SHR (P < 0.01) and LMNXD-HD + SHR (P < 0.05) groups. The LMNXD-LD + SHR and MPH + SHR groups also showed increases, but these differences were not statistically significant.

As the most abundant neurotrophic factor in the central nervous system, BDNF regulates axonal branching, dendritic growth, and synaptic plasticity. Western blot analysis of striatal tissue revealed significantly lower BDNF protein expression in SHR rats compared to the WKY group (P < 0.05) (Figures 4C,F). BDNF protein expression was significantly increased in the LMNXD-MD + SHR, LMNXD-HD + SHR, and MPH + SHR groups relative to the SHR group (Figures 4C,F).

Seventy compounds identified through mass spectrometry were screened according to Lipinski’s Rule of Five, selecting those with at least two “Yes” outcomes and high gastrointestinal absorption, which yielded 45 candidate bioactive constituents (Supplementary Table 2). These constituents were submitted to the SwissTargetPrediction and STITCH databases for target prediction, yielding 722 unique targets after merging and deduplication. From the OMIM, GeneCards, DrugBank, and PharmGkb databases, 4528 ADHD-related targets were collected. A subsequent Venn diagram identified 425 potential targets implicated in LMNXD’s action against ADHD (Figure 5A). A PPI network constructed from these 425 overlapping targets contained 360 nodes and 1,287 edges (Figure 5B), where node size and color intensity reflected degree centrality. The outermost ring highlighted targets with a degree ≥10, among which PIK3R1, HSP90AA1, SRC, STAT3, PIK3CA, PIK3CB, AKT1, and PIK3CD emerged as potential core targets. GO and KEGG enrichment analyses performed via DAVID revealed that biological processes (BP) were mainly associated with xenobiotic stimulus response, membrane potential regulation, peptide hormone response, and muscle system processes. Key molecular functions (MF) encompassed protein tyrosine kinase activity, neurotransmitter receptor activity, transmembrane receptor protein tyrosine kinase activity, and postsynaptic neurotransmitter receptor activity. The principal cellular components (CC) included the Schaffer collateral–CA1 synapse, T-tubule, potassium channel complex, and mitotic spindle (Figure 5C). KEGG analysis highlighted significant enrichment in the PI3K–Akt, calcium, and cAMP signaling pathways (Figure 5D). An integrated “LMNXD–Constituents–Core Targets–Pathways–Disease” network, built using Cytoscape (Figure 5E), illustrated how individual constituents can modulate multiple targets and how multiple targets converge on common pathways. In summary, these network pharmacology results offer valuable mechanistic insight into LMNXD’s potential role in ADHD treatment.

To investigate the biological mechanisms underlying LMNXD’s therapeutic effects, we conducted RNA sequencing on hippocampal tissues. The DEGs across the experimental groups are displayed in Figures 6A,B. Relative to the WKYgroup, the SHR group displayed 390 upregulated and 571 downregulated genes. Treatment with LMNXD resulted in 59 upregulated and 210 downregulated genes compared to the SHR group. We identified a total of 214 overlapping DEGs common to all three groups (Figures 6C,D). These 214 overlapping DEGs were subsequently subjected to GO, KEGG enrichment pathways, and PPI analyses. GO analysis indicated significant enrichment in processes and components including extracellular matrix organization, skin development, extracellular matrix, collagen-containing extracellular matrix, extracellular matrix structural constituent, extracellular matrix structural constituent conferring tensile strength. Figure 6E presents the ten most significantly enriched terms for BP, CC, and MF. KEGG pathway analysis further revealed significant enrichment in the cAMP, PI3K-Akt, Serotonergic synapse and Cholinergic synapses (Figure 5F). We performed PPI analysis using the STRING database and visualized the interaction network with Cytoscape software, yielding a network of 190 core nodes connected by 229 edges (Figure 6G). The DEGs were ranked according to the degree value of their corresponding nodes, which reflects their number of connections. This analysis identified Col1a1, Col3a1, Col6a2, Fbn1, and Dcn as the top five hub genes.

The PI3K-Akt and cAMP signaling pathways were significantly enriched in both network pharmacology and transcriptomics analyses, suggesting that they might be the key pathways for LMNXD in the treatment of ADHD. In addition, we screened out the genes enriched in these two pathways, with significant differences and related to the pathogenesis of ADHD from the transcriptome results, and conducted qPCR experiments for verification. The results indicated that the relative mRNA expression levels of ATP2B4, OXTR, GRIN3A, ITGA1, and COL6A2 were lower in the SHR group than in the WKY group. Conversely, the LMNXD-MD + SHR group exhibited significantly increased expression levels of these genes relative to the SHR group (Figures 7A–E).

This study has several limitations. First, the limited sample size and considerable dispersion in some data may have reduced statistical power, potentially preventing certain intergroup differences from reaching significance. Second, the validation of DEGs was relatively narrow; subsequent work should confirm these findings at the protein level using methods such as Western blot analysis. Third, although this study provides preliminary evidence for the potential involvement of the cAMP and PI3K-Akt signaling pathways, the complete upstream and downstream molecular mechanisms still require systematic investigation. Finally, while this research examined the overall therapeutic effect of the compound formulation, the specific active constituents responsible for its efficacy remain to be identified.