Abstract

   Background/Objectives: Dendrobium nobile Lindl. (DNL), a traditional
   dietary supplement, exhibits therapeutic potential for type 2 diabetes
   mellitus (T2DM), yet its mechanisms remain unclear. Methods: T2DM was
   induced in db/db mice. DNL (10 g/kg/d) or metformin (65 mg/kg/d) was
   administered for 4 weeks. This study integrated pharmacodynamic
   evaluation and multi-omics to elucidate DNL’s anti-diabetic effects in
   db/db mice. Results: DNL intervention significantly ameliorated T2DM
   phenotypes, reducing hyperglycemia, insulin resistance, and renal
   dysfunction. Metabolomics analysis identified 39 differential
   metabolites (19 upregulated, 20 downregulated) linked to citrate cycle,
   oxidative phosphorylation, and glycerophospholipid metabolism, while
   proteomics revealed 113 differentially expressed proteins, with
   multi-omics integration highlighting DNL’s modulation of three proteins
   (Ckm, Ache, Selenbp1) and four metabolites (4-guanidinobutanoic acid,
   phosphorylcholine, homocysteine, succinic acid) across arginine/proline
   metabolism, glycerophospholipid metabolism, and sulfur metabolism.
   Pathway analysis demonstrated DNL’s restoration of dysregulated
   processes, including inflammation suppression via NF-κB and PI3K-Akt
   pathways, enhanced insulin sensitivity through glycerophospholipid
   balance, and mitigation of oxidative stress via sulfur metabolism. Key
   correlations between metabolites and proteins underscored DNL’s
   multi-target action. Conclusions: These findings systematically decode
   therapeutic mechanisms of Dendrobium nobile Lindl., emphasizing its
   role in rectifying metabolic disorders and inflammatory signaling,
   thereby providing a molecular basis for its clinical application in
   T2DM management.

   Keywords: Dendrobium nobile Lindl., metabolomics, proteomics, type 2
   diabetes mellitus, mechanism

1. Introduction

   Type 2 diabetes mellitus (T2DM) is a disease characterized by
   non-autoimmune heterogeneous progressive pancreatic islet β-cell
   insulin underproduction [[40]1], which often occurs in the presence of
   insulin resistance (IR) and the metabolic syndrome (MS) [[41]2,[42]3].
   Diabetes has grown up to be a global health burden due to its high
   morbidity, disability, and mortality. The usual treatment is the
   implementation of statins and hypoglycaemic agents. In contrast to
   these applications, which may produce significant side effects,
   Dendrobium nobile Lindl. (DNL) has been demonstrated to generate
   hypoglycaemic properties without significant side effects [[43]4].

   DNL has been used as a dietary additive for centuries [[44]5], and its
   pharmacological activities are closely related to the chemical
   components, with pharmacological activities such as anti-tumor,
   anti-aging, immune enhancement, hypoglycemic, and anti-cataract
   [[45]6]. Recent phytochemical studies identify key bioactive
   constituents in DNL, including dendrobine, polysaccharides, and
   flavonoids, which contribute to its hypoglycemic effects [[46]7,[47]8].
   DNL polysaccharides enhance insulin sensitivity via AMPK activation
   [[48]9], while dendrobine analogues can improve diabetic nephropathy by
   regulating glucose metabolism [[49]10]. These compounds collectively
   modulate metabolic pathways relevant to T2DM pathogenesis. Clinically,
   it can be used to treat patients with metabolic syndrome [[50]11].
   However, the unclear mechanisms of DNL’s anti-diabetic action limit its
   clinical translation. It is imperative to develop a method to elucidate
   the mechanism of action of DNL.

   Metabolomics can capture the panoramic changes of body metabolites
   through high-throughput technology to reveal the overall regulatory
   effect of herbs on the biological system of the organism [[51]12].
   Metabolomics contributes to the standardized assessment of the efficacy
   and mechanisms of herbs by providing data-driven, quantifiable evidence
   of metabolic changes [[52]13]. Proteomics offers a panoramic analysis
   of the expression and modification of all proteins and their dynamic
   changes in cells or tissues, a holistic feature that is highly
   compatible with the systemic mode of action of herbs [[53]14].
   Proteomics resolves potential drug targets by means of bioinformatics
   [[54]15]. The integration of proteomics and metabolomics can achieve
   the systematic analysis of herbs from phenotype to mechanism, and the
   construction of a multi-level network to reveal the multi-target
   mechanism of action of herbs [[55]16].

   In this study, we used db/db mice to construct a T2DM mouse model and
   elucidated the efficacy and mechanism of DNL through experimental
   examination, metabolomics, and proteomics analyses ([56]Figure 1). The
   ameliorative effect of DNL on T2DM provides a reference for exploring
   the indications and mechanism of action of DNL.

Figure 1.

   [57]Figure 1
   [58]Open in a new tab

   The design path of this study. Firstly, the pharmacodynamics of db/db
   mice in each group was studied, including body weight, FBG, HbA1 c
   level, and pathological observation. Secondly, based on the
   metabolomics method, the metabolites of DNL callback and their enriched
   metabolic pathways were screened. Then, based on the proteomics method,
   the proteins of DNL callback and their enriched pathways were screened.
   Finally, a combined analysis of metabolomics and proteomics was
   performed to elucidate the potential mechanism of HGD.

2. Results

2.1. DNL Significantly Attenuated the T2DM Disease Phenotype in db/db Mice

   To investigate the effects of DNL administration on T2DM, we
   established a T2DM disease model using db/db mice. The model exhibited
   pathophysiological features of T2DM, such as significantly higher body
   weight ([59]Figure 2a), fasting blood glucose ([60]Figure 2b),
   glycosylated haemoglobin ([61]Figure 2c), and insulin ([62]Figure 2d)
   in the MOD group of mice compared to the CON group. In addition, mice
   in the MOD group had significantly lower renal indices ([63]Figure 2e)
   compared with the CON group, suggesting renal impairment. Subsequently,
   we examined the serum levels of Cr, BUN, and MDA in the mice in each
   group. The results showed that Cr, BUN, and MDA levels were
   significantly higher in mice in the MOD group compared with the CON
   group ([64]Figure 2f–h). Observation of H&E staining of renal tissues
   of mice in each group revealed lesions in the renal tissues of mice in
   the MOD group ([65]Figure 2i). We treated mice with the drug for 36
   days, and salvage was obtained in all the treated groups, which
   indicates that DNL administration has a significant preventive and
   curative effect on T2DM mice.

Figure 2.

   [66]Figure 2
   [67]Open in a new tab

   Therapeutic effects of DNL on pathological indicators in db/db mice
   with T2DM. (a) Body weight dynamics: The model (MOD) group exhibited a
   progressive increase in body weight over 36 days compared to the
   control (CON) group, while DNL administration significantly attenuated
   this trend (# p < 0.01 vs. MOD). (b) Fasting blood glucose (FBG): MOD
   mice displayed sustained hyperglycemia, which was markedly reduced by
   DNL intervention (### p < 0.001 vs. MOD). (c) Glycated hemoglobin
   (HbA1c): elevated HbA1c levels in MOD mice were effectively normalized
   following DNL treatment (## p < 0.01 vs. MOD). (d) Serum insulin:
   Hyperinsulinemia in the MOD group was significantly suppressed by DNL
   (** p < 0.01 vs. MOD). (e) Kidney index (KI): DNL restored the
   decreased KI in MOD mice to near-normal levels (** p < 0.01 vs. MOD).
   (f) Creatinine (Cr): DNL reversed MOD-induced renal dysfunction, as
   evidenced by reduced Cr levels (*** p < 0.001 vs. MOD). (g) Blood urea
   nitrogen (BUN): Elevated BUN in MOD mice was ameliorated by DNL (*** p
   < 0.001 vs. MOD). (h) Malondialdehyde (MDA): Oxidative stress indicated
   by increased MDA was mitigated by DNL (*** p < 0.001 vs. MOD). (i)
   Renal histopathology: H&E staining revealed severe glomerular lesions
   in MOD mice Neutrophils (orange arrows) Renal tubular epithelial cell
   necrosis (blue arrows) Proliferation (black arrows) Fibroblasts (red
   arrows) Basement membrane thickening (brown arrows)), which were
   alleviated by DNL treatment. Scale bar: 50 μm. Abbreviations: CON,
   control group; MOD, T2DM model group; MET, metformin-treated group;
   DNL, Dendrobium nobile Lindl.-treated group. Data are expressed as mean
   ± SEM. Statistical significance: ** p < 0.01, *** p < 0.001 vs. CON; #
   p < 0.05, ## p < 0.01, ### p < 0.001 vs. MOD (two-tailed Student′s
   t-test).

2.2. Metabolic Effects of DNL on Serum Profiles of db/db Mice

   To reveal the metabolites that change after T2DM and DNL
   administration, we performed non-targeted metabolomics analysis of
   serum samples from the CON group, the MOD group, and the DNL group. As
   shown in [68]Figure 3a, the PCA plot showed that the CON group was well
   separated from the MOD group, and the DNL group had a tendency to
   deviate from the MOD group and approach the CON group. The correlation
   clustering analysis showed a clear trend of clustering among the three
   groups, with the DNL group intermediate between the CON group and the
   MOD group ([69]Figure 3b).

Figure 3.

   [70]Figure 3
   [71]Open in a new tab

   Multi-omics integration reveals DNL-mediated metabolic reprogramming in
   T2DM mice. (a) Principal component analysis (PCA) score plot:
   Unsupervised multivariate analysis demonstrated distinct clustering
   patterns among the control (CON), T2DM model (MOD), and DNL-treated
   groups. Dimensionality reduction via PCA highlighted metabolic
   trajectory restoration by DNL, with the DNL group positioned
   intermediately between CON and MOD along principal component 1 (PC1).
   (b) Hierarchical clustering heatmap: Spearman correlation analysis of
   serum metabolites across groups revealed substance-specific clustering
   patterns. The MOD group exhibited disrupted metabolic correlations
   (blue clusters), which were partially restored by DNL intervention (red
   clusters), indicating coordinated regulation of glycerophospholipid and
   amino acid metabolism. Abbreviations: CON, control group; MOD, T2DM
   model group; DNL, Dendrobium nobile Lindl.-treated group.

2.3. DNL Exhibited Effects on db/db Mice with T2DM Related Various
Metabolites

   The endogenous metabolites that change in the serum of db/db mice with
   T2DM upon DNL intervention were further investigated. The OPLS-DA
   analysis was performed to identify metabolites contributing to the
   separation of the CON group and the MOD group. The score plot showed a
   clear separation between the two groups, and cross-validation plots
   generated from 200 alignment tests showed intercepts of R2 = 0.999 and
   Q2 = 0.85, indicating good reliability and predictability of the
   respective models ([72]Figure 4a). Filtering the initial selection of
   metabolites according to the criteria of VIP > 1, p < 0.05, we screened
   to obtain 39 metabolites, of which 19 were upregulated and 20
   downregulated in the MOD group ([73]Figure 4b). The correlation
   analysis showed that differential metabolites with similar abundance
   clustered together, representing an internal relationship between
   metabolic proximity and differential metabolites ([74]Figure 4c).

Figure 4.

   [75]Figure 4

   [76]Figure 4
   [77]Open in a new tab

   Impact of DNL on metabolic profiles in db/db mice with T2DM. (a).
   OPLS-DA score plot: Supervised multivariate statistical approach
   revealing variable importance in projection (VIP) values, enabling
   preliminary identification of discriminative metabolites between the
   control (CON) and model (MOD) groups. (b). VIP–heatmap integration:
   Dual visualization combining VIP value distribution and metabolite
   abundance patterns across CON and MOD groups. (c). Correlation heatmap:
   Inter-metabolite relationship network constructed through pairwise
   correlation analysis of expression levels in CON and MOD groups. (d).
   Hierarchical clustering heatmap: Color-gradient visualization of
   metabolite patterns across CON, MOD, and DNL groups, with annotations
   highlighting significant variations in CON_VS_MOD and MOD_VS_DNL
   comparisons. (e). Stacked bar chart: Group-wise distribution and
   quantitative comparisons of differential metabolites across all
   experimental groups. (f). K-means cluster analysis: Bipartite
   stratification of upregulated/downregulated metabolites in CON/MOD
   groups, with DNL intervention recovery patterns.

   The hierarchical clustering dendrogram showed that the metabolites in
   the MOD group were significantly separated from both the CON group and
   the DNL group, while the metabolites in the CON group and the DNL group
   had a similar trend. And 24 differential metabolites were significantly
   altered in the DNL group compared to the MOD group ([78]Figure 4d). As
   shown in [79]Figure 4e, the distribution of differential metabolite
   contents within and across groups showed significant differences.
   Further, the K-means clustering analysis showed that 32 differential
   metabolites were reversed by DNL intervention (17 upregulated and 15
   downregulated), of which 24 were significantly recovered (13
   upregulated and 11 downregulated) ([80]Figure 4f and [81]Figure S1).

2.4. DNL Exerted Influences on db/db Mice with T2DM-Related Metabolic
Pathways

   Based on the KEGG database, the enrichment analysis using the OmicShare
   Tools identified potential pathways associated with T2DM, including
   citrate cycle (TCA cycle), pentose phosphate pathway, primary bile acid
   biosynthesis, oxidative phosphorylation, arginine and proline
   metabolism, glycerophospholipid metabolism, arachidonic acid
   metabolism, pyruvate metabolism, sulfur metabolism, cAMP-signaling
   pathway, etc. DNL regulates pathways such as citrate cycle (TCA cycle),
   oxidative phosphorylation, glycerophospholipid metabolism, sulfur
   metabolism, and other pathways, which are potential pathways for DNL to
   influence T2DM ([82]Figure 5a,b).

Figure 5.

   [83]Figure 5
   [84]Open in a new tab

   Modulation of T2DM-associated metabolic pathways by DNL in db/db mice.
   (a). KEGG pathway-enrichment analysis: Bar chart illustrating
   pathway-specific enrichment of differential metabolites, highlighting
   key metabolic pathways perturbed in T2DM. (b). Sankey network
   visualization: Flow relationships between enriched metabolic pathways
   and their corresponding differential metabolites, emphasizing
   pathway-metabolite connectivity.

2.5. DNL Showed Effects on Serum Proteinic Profiles in db/db Mice with T2DM

   To improve our understanding of the potential mechanisms of T2DM and
   the mechanism of action of DNL, we performed non-targeted proteomics
   analysis of serum samples from the CON group, the MOD group, and the
   DNL group. As shown in [85]Figure 6a, the PCA plot showed that the CON
   group was well separated from the MOD group, and the DNL group had a
   tendency to deviate from the MOD group and approach the CON group. The
   correlation clustering analysis showed a clear trend of clustering
   among the three groups, with the DNL group intermediate between the CON
   group and the MOD group ([86]Figure 6b).

Figure 6.

   [87]Figure 6
   [88]Open in a new tab

   Proteomic profiling reveals systemic modulation by DNL in T2DM mice.
   (a) PCA score plot: Unsupervised multivariate analysis of serum
   proteomes demonstrated distinct clustering patterns among the control
   (CON), T2DM model (MOD), and DNL-treated groups. Dimensionality
   reduction via principal component analysis (PCA) revealed progressive
   normalization of proteomic profiles in the DNL group, positioning it
   intermediately between MOD and CON along the first principal component
   (PC1). (b) Hierarchical clustering heatmap: Quantitative proteomic
   profiling identified group-specific proteomic signatures. The MOD group
   exhibited disrupted protein correlation networks (blue clusters), while
   DNL intervention partially restored physiological interaction patterns
   (red clusters), highlighting its regulatory effects on insulin
   signaling and inflammatory pathways. Abbreviations: CON, control group;
   MOD, T2DM model group; DNL, Dendrobium nobile Lindl.-treated group.

2.6. Proteins Changes of DNL on db/db Mice with T2DM

   The endogenous protein changes in the serum of db/db mice with T2DM
   upon DNL intervention were further investigated. The OPLS-DA analysis
   was performed to identify proteins contributing to the separation of
   the CON group and the MOD group. The score plot showed a clear
   separation between the two groups, and cross-validation plots generated
   from 200 alignment tests showed intercepts of R2 = 0.994 and Q2 =
   0.913, indicating good reliability and predictability of the respective
   models ([89]Figure 7a). Filtering the initial selection of proteins
   according to the criteria of FC > 1.5, p < 0.05, we screened to obtain
   113 proteins, of which 13 were upregulated and 100 were downregulated
   in the MOD group ([90]Figure 7b and [91]Figure S2). The correlation
   analysis showed that differential proteins with similar abundance
   clustered together, representing an internal relationship between
   differential proteins ([92]Figure S3).

Figure 7.

   [93]Figure 7
   [94]Open in a new tab

   Proteomic regulation by DNL in db/db mice with T2DM. (a). OPLS-DA score
   plot: Supervised multivariate analysis revealing variable importance in
   projection (VIP) scores, identifying discriminant proteins between
   control (CON) and model (MOD) groups. (b). VIP-FC lollipop plot:
   Multiparametric biomarker profiling integrating VIP values, fold change
   (FC), and statistical significance (p-values) of differential proteins
   (Ranked in the top 55): ** p < 0.01, *** p < 0.001. (c). Hierarchical
   clustering heatmap: Color-coded visualization of protein expression
   patterns across CON, MOD, and DNL groups, with annotations for
   significant differences in CON_VS_MOD and MOD_VS_DNL comparisons. (d).
   K-means cluster analysis: Bipartite classification of
   upregulated/downregulated proteins in CON/MOD groups, demonstrating
   recovery trends in the DNL intervention group.

   The hierarchical clustering dendrogram showed that the proteins in the
   MOD group were significantly separated from both the CON group and DNL
   group, while the proteins in the CON group and DNL group had a similar
   trend. And 55 differential proteins were significantly recovered in the
   DNL group compared to the MOD group ([95]Figure 7c). Further, the
   K-means clustering analysis showed that DNL recovered 76 differential
   proteins (64 upregulated and 12 downregulated) ([96]Figure 7d).

2.7. DNL Exerted Influences on db/db Mice with T2DM-Related Pathways

   Based on the KEGG database, the enrichment analyses using the OmicShare
   Tools identified potential pathways associated with T2DM, including
   oxidative phosphorylation, arginine and proline metabolism, sulfur
   metabolism, the MAPK-signaling pathway, NF-kappa B-signaling pathway,
   HIF-1-signaling pathway, mTOR-signaling pathway, PI3K-Akt-signaling
   pathway, AMPK-signaling pathway, JAK-STAT-signaling pathway, etc. DNL
   regulated arginine and proline metabolism, sulfur metabolism, the
   MAPK-signaling pathway, NF-kappa B-signaling pathway, HIF-1-signaling
   pathway, PI3K-Akt-signaling pathway, and JAK-STAT-signaling pathway,
   which served as potential pathways for DNL to influence T2DM
   ([97]Figure 8a,b).

Figure 8.

   [98]Figure 8
   [99]Open in a new tab

   Multi-dimensional pathway mapping of DNL’s therapeutic actions in T2DM
   mice. (a). The KEGG-enrichment bar chart: The enrichment results of
   differential protein pathways were displayed. (b). The sankey diagram:
   The differential proteins enriched in each pathway were shown. (c). The
   GO classification histogram: Based on the GO classification results, a
   classification histogram was generated to visually display the GO
   classification results of differential proteins.

   The classification was performed according to GO annotations:
   biological process (BP), cellular component (CC), and molecular
   function (MF). The results showed that differential proteins in the MOD
   group were significantly enriched in several biological processes,
   molecular functions, and cellular component categories compared to the
   CON group (p < 0.05). Differential proteins from the MOD group were
   involved in the regulation of metabolic processes, cellular process
   regulation, signalling, and nitrogen compound metabolism processes.
   These differential proteins were associated with the regulation of
   molecular functions, such as catalytic activity, regulators of
   molecular functions, structural molecule activity, transporter
   activity, etc. The main cellular components enriched in the MOD group
   were extracellular regions, organelles, macromolecular complexes, and
   membranes ([100]Figure 8c).

2.8. Metabolomics with Proteomics Revealed Mechanism of DNL Intervention in
db/db Mice

   Through the KEGG database, as shown in [101]Figure 9a,b, differential
   proteins and differential metabolites co-regulated six pathways,
   including oxidative phosphorylation, glycine, serine and threonine
   metabolism, cysteine and methionine metabolism, arginine and proline
   metabolism, glycerophospholipid metabolism, and sulfur metabolism.
   Among them, DNL regulated three differential proteins and four
   differential metabolites, which were involved in arginine and proline
   metabolism, glycerophospholipid metabolism, and sulfur metabolism
   ([102]Figure 9c). The results of the correlation analysis showed
   ([103]Table S4, [104]Figure 9d) that Ache had a low correlation
   (0.2–0.39) with 4-guanidinobutanoic acid, ache had moderate correlation
   (0.4–0.69) with homocysteine and succinic acid, and Ache had a high
   correlation (0.7–0.89) with phosphorylcholine; Ckm had an extremely low
   correlation (0–0.19) with 4-Guanidinobutanoic acid, Ckm had moderate
   correlation (0.4–0.69) with homocysteine, phosphorylcholine and
   succinic acid; Selenbp1 was moderately correlated with succinic acid
   and homocysteine (0.4–0.69), and Selenbp1 was highly correlated with
   4-Guanidinobutanoic acid and phosphorylcholine (0.7–0.89). In arginine
   and proline metabolism, DNL significantly modulated the levels of
   4-Guanidinobutanoic acid (p < 0.001); in glycerophospholipid
   metabolism, DNL significantly modulated the level of phosphorylcholine
   (p < 0.001); in sulfur metabolism, DNL significantly modulated the
   levels of Homocysteine (p < 0.05) and Selenbp1 (p < 0.01) ([105]Figure
   9e–g).

Figure 9.

   [106]Figure 9
   [107]Open in a new tab

   Multi-omics convergence deciphers DNL’s metabolic regulatory network in
   T2DM mice. (a) KEGG pathway intersection: Venn diagram identifies
   co-enriched pathways (q < 0.05) from metabolomic (blue) and proteomic
   (yellow) datasets, including oxidative phosphorylation and sulfur
   metabolism. (b) Cross-omics pathway mapping: Sankey diagram visualizes
   32 metabolites (left) and 55 proteins (right) converging on
   arginine/proline metabolism (node width reflects enrichment
   significance). (c) Functional convergence landscape: Circos plot
   integrates DNL-regulated pathways (outer ring) with corresponding
   metabolites (green arcs) and proteins (purple arcs), highlighting
   glycerophospholipid metabolism as the central hub. (d)
   Metabolite–protein interactome: Network heatmap reveals strong positive
   correlations (r > 0.7, red edges) between 4-guanidinobutanoic acid
   (metabolite) and Ckm (protein), alongside NF-κB pathway proteins (IκBα,
   p65) negatively correlating with homocysteine (r < −0.6, blue edges).
   Pathway-specific mechanistic insights: (e) Arginine/proline metabolism:
   DNL upregulated 4-guanidinobutanoic acid (**** p < 0.0001 vs. MOD) and
   muscle creatine kinase (Ckm), restoring ATP regeneration (solid arrows:
   enzymatic reactions; dashed arrows: regulatory interactions). (f)
   Glycerophospholipid metabolism: DNL reduced phosphorylcholine ((*** p <
   0.001) and acetylcholinesterase (Ache), mitigating membrane lipid
   peroxidation (solid lines: metabolic flux; dashed lines: inhibitory
   effects). (g) Sulfur metabolism: Intervention decreased homocysteine (*
   p < 0.05) and selenium-binding protein 1 (Selenbp1, ** p < 0.01) while
   modulating succinate-mediated inflammatory signaling (solid lines:
   direct catalysis; dashed lines: allosteric regulation). Abbreviations:
   CON, control; MOD, T2DM model; DNL, Dendrobium nobile Lindl.-treated.
   Statistical significance: * p < 0.05, ** p < 0.01, *** p < 0.001, ****
   p < 0.0001.

3. Discussion

   In this study, we constructed a T2DM model using db/db mice. Thirty-six
   days later, mice in the MOD group spontaneously developed T2DM
   symptoms, such as elevated body weight, FBG, HbA1c, and insulin levels.
   In addition, mice in the MOD group had a significantly lower kidney
   index and significantly higher serum levels of Cr, BUN, and MDA. H&E
   staining showed that the kidney tissues of mice in the MOD group were
   diseased. We used DNL to intervene in db/db mice for 36 days, and all
   the above symptoms were significantly improved, which indicated that
   DNL administration had significant preventive and therapeutic effects
   on T2DM.

   The mechanism of DNL intervention in the db/db mouse model was
   subsequently explored through the integration of proteomics and
   metabolomics. Differentially expressed proteins and metabolites were
   identified, elucidating the mechanisms of DNL intervention in T2DM.
   Serum samples from each group clustered closely in PCA score plots.
   Proteins and metabolites from the MOD group were separated from the CON
   group and DNL group, and the DNL group was closer to the CON group. The
   correlation clustering plot showed that the proteins and metabolites of
   the MOD group were separated from the CON group and DNL group, and the
   DNL group was closer to the CON group. This suggested that the
   metabolic profile of the db/db mice was altered, and that intervention
   with DNL resulted in a similar metabolic profile to that of the CON
   group of mice. By KEGG-enrichment analysis, DNL modulated three
   pathways co-enriched by metabolomics and proteomics in db/db mice and
   with altered protein and metabolite content. These pathways included
   arginine and proline metabolism, glycerophospholipid metabolism, and
   sulfur metabolism. DNL regulated differentially expressed proteins such
   as Ckm, Ache, and Selenbp1 and differentially expressed metabolites
   such as 4-Guanidinobutanoic acid, phosphorylcholine, homocysteine, and
   succinic acid.

   Arginine and proline metabolism regulate the balance between ROS, as
   well as NO production and clearance. Its disturbance accelerates T2DM
   pathology by synergistically promoting inflammatory response and
   oxidative damage, disrupting energy metabolism and insulin sensitivity
   [[108]17,[109]18]. Further, 4-Guanidinobutyric acid is a common
   arginine metabolite involved in the metabolism of arginine and proline
   (creatinine pathway), which is essential for energy metabolism
   [[110]19,[111]20]. In the present study, it was detected that the level
   of 4-guanidinobutyric acid is lower in the MOD group than in the CON
   group, and DNL-mediated upregulation of 4-guanidinobutyric acid
   suggests enhanced arginine/proline metabolism, which may contribute to
   the therapeutic effect. Ckm maintains the rapid regeneration of ATP in
   the muscle through the creatine–phosphocreatine cycle, which provides
   energy for muscle contraction and metabolism. In patients with T2DM,
   metabolic disturbances lead to a decrease in Ckm activity, which
   impairs the muscle’s ability to respond rapidly to energy demands
   [[112]21]. In the present study, the level of Ckm was found to be lower
   in the MOD group than in the CON group, and DNL significantly increased
   the level of Ckm.

   Glycerophospholipid metabolism directly affects the ratio and
   composition of cell membrane lipids. In patients with T2DM, an
   imbalance in the ratio of PC/PE in the cell membrane reduces membrane
   fluidity and impairs the efficiency of insulin receptor binding to
   insulin, leading to insulin resistance [[113]22]. Glycerophospholipid
   metabolism is also critical for maintaining pancreatic β-cell membrane
   integrity and function. Metabolic disorders may lead to cell membrane
   disruption and increased lipotoxicity, triggering β-cell apoptosis or
   dysfunction and reducing insulin secretion [[114]23]. Phosphocholine is
   an important intermediate product of phospholipid metabolism and is
   widely involved in biological processes, such as cell membrane
   structure, signal transduction, and regulation of inflammatory
   responses [[115]24]. Increased phosphocholine deteriorates insulin
   signalling and causes insulin resistance [[116]25]. Blocking the
   synthesis of lysophosphatidic acid from phosphocholine restores serum
   GLP-1 levels, insulin secretion, and glucose tolerance [[117]26]. DNL
   significantly reduced the level of phosphocholine and suppressed
   insulin resistance and inflammation in db/db mice with T2DM.

   Sulfur metabolism involves a wide range of biochemical reactions of
   elemental sulfur and is an important process in maintaining life
   activities. Sulfur metabolism is involved in the regulation of
   antioxidant defence, inflammation regulation, energy metabolism, and
   pancreatic function and is closely related to the development of T2DM
   [[118]27]. Homocysteine is an intermediate product of methionine
   metabolism, and its metabolism is dependent on folate, vitamin B6, and
   vitamin B12. When abnormal metabolism leads to homocysteine
   accumulation, it can directly interfere with the insulin-signaling
   pathway [[119]28]. It has been shown that homocysteine induces insulin
   resistance and contributes to the diabetic phenotype through protein
   cysteine–homocysteinylation of the pre-insulin receptor [[120]29].
   Homocysteine can exacerbate the chronic inflammatory state by
   activating the NF-κB pathway and exacerbating the release of
   pro-inflammatory cytokines (e.g., IL-6, TNF-α) [[121]30]. Homocysteine
   also induces the activation of pro-inflammatory macrophages (M1 type)
   and inhibits the function of anti-inflammatory macrophages (M2 type),
   which further worsens the inflammatory microenvironment and reduces
   insulin sensitivity [[122]31]. Succinic acid can act as a ligand to
   activate SUCNR1 and enhance insulin secretion. Prolonged increased
   levels of succinic acid may lead to the overactivation of the receptor,
   which in turn triggers β-cell dysfunction and affects normal insulin
   secretion [[123]32]. In macrophages, succinic acid activates
   inflammatory pathways and increases the release of pro-inflammatory
   cytokines, leading to chronic low-grade inflammation [[124]33].
   Up-regulation of SELENBP1 expression inhibits the activation of
   PI3K/AKT/mTOR signaling pathway, thereby hindering the malignant
   development of lung adenocarcinoma. [[125]34], which attenuates
   insulin-stimulated glucose uptake and promotes the development of
   insulin resistance [[126]35]. In this study, the levels of
   homocysteine, succinic acid, and Selenbp1 were detected to be
   significantly higher in the MOD group than in the CON group, and DNL
   significantly reduced the levels of cysteine, succinic acid, and
   Selenbp1, exerting anti-inflammatory, anti-insulin resistance, and
   alleviating T2DM symptoms.

   This study focuses on the overall intervention mechanism of DNL. The
   existing literature indicates that the primary active compounds in DNL
   include alkaloids (such as dendrobine), polysaccharides, and flavonoids
   [[127]6]. Dendrobine compounds can promote mitochondrial function and
   inhibit oxidative phosphorylation [[128]7]. This study observed that
   DNL significantly upregulates 4-guanidino butyric acid (a key
   intermediate in arginine/proline metabolism) and creatine kinase (CKM),
   which may be partially attributed to the alkaloids’ reprogramming
   effect on energy metabolism. This explains the restoration of ATP
   regeneration capacity after DNL intervention, alleviating energy
   imbalance in T2DM ([129]Figure 9e). Dendrobium polysaccharides have
   been shown to activate the PI3K/Akt- and AMPK-signalling pathways,
   improving glucose uptake in liver and muscle tissues [[130]8]. This
   aligns closely with the findings in this study, where DNL inhibits the
   PI3K-Akt pathway (by downregulating Selenbp1) and regulates
   glycerophospholipid metabolism ([131]Figure 9f,g). The flavonoid
   components in DNL (such as naringin and kaempferol) possess the ability
   to inhibit the NF-κB pathway and scavenge ROS [[132]36]. This directly
   supports the findings of this study: DNL blocks the NF-κB inflammatory
   cascade by downregulating homocysteine and succinate levels
   ([133]Figure 9g) and reduces oxidative stress markers such as MDA
   ([134]Figure 2h).

4. Materials and Methods

4.1. Animal Experiments

   Male db/db and db/m mice (seven-week-old, SPF-grade) were obtained from
   GemPharmatech Co., Ltd. (Nanjing, China). The experimental protocol was
   designed with strict reference to relevant ethical regulations and was
   approved by the Ethics Committee of Hainan Medical College. This
   experiment had four groups: the control (CON) group, the model (MOD)
   group, the metformin hydrochloride (MET) group, and the Dendrobium
   nobile Lindl. (DNL) group, which were fed normal diets. The mice in
   each treatment group were provided MET (65.0 mg/kg/d) and DNL (10
   g/kg/d) for 4 weeks (n = 12). The MET was purchased from Merck
   Pharmaceutical (Nantong, China) Co. The study protocols were ethically
   reviewed and approved by the Institutional Review Board of Hainan
   Medical University (Ethics Approval Number: HYLL2024-5-09, approved on
   31 May 2024).

4.2. DNL Preparation

   DNL was prepared according to ≪Pharmacopoeia of the People’s Republic
   of China (2020 Edition)≫ [[135]37]. 50 g of dried DNL herbs (Hainan
   Shengrong Biotechnology Co., Ltd., Haikou, China) were weighed, soaked
   in 400 mL of water for 1 h, decocted for 30 min, replenished with 400
   mL of water, and decocted for another 30 min to obtain 150 mL of drug
   solution. Subsequent rotary evaporation was conducted using a rotary
   evaporator (HEI-VAP CORE ML, Heidolph, Schwabach, Germany) (150 rpm; 70
   °C; 1.5 h) to obtain 30 mL of DNL herb concentrate and further made
   into freeze-dried powder.

4.3. General Status Observation and Monitoring of Indicators

   During the experimental period, mice were weighed weekly and observed
   for mental status, reactivity, activity, changes in hair colour, and
   feeding. Tail vein blood was taken every week, and FBG was measured by
   GA-3 blood glucose meter (Sanuo Biosensing (Changsha, China) Co.,
   Ltd.). Tail vein blood was taken 1 day before the end of the
   experiment, and HbA1c was measured by glycated haemoglobin kit (Sanuo
   Biosensing (Changsha, China) Co., Ltd.).

4.4. Histopathological Examination

   Kidney tissues were fixed with 4% paraformaldehyde (PFA) and then
   embedded in paraffin. Specimens were sectioned at 4 μm thickness and
   stained with haematoxylin and eosin (H&E). Photographs of random areas
   of the stained sections were taken using a light microscope (Leica
   Microsystems, Wetzlar, Germany).

4.5. Serum Biochemical Analysis

   Serum levels of creatinine (Cr), urea nitrogen (BUN), and
   malondialdehyde (MDA) in db/db mice were determined using enzyme-linked
   immunosorbent kits (Nanjing Jiancheng Institute of Bioengineering,
   Nanjing, China) according to the manufacturer’s guidelines.

4.6. Sample Preparation for Metabolomics Analysis

   The serums were centrifuged for 10 min (12,000 rpm, 4 °C). The
   supernatants were transferred to 2 mL centrifuge tubes, concentrated,
   and dried. The samples dissolved in 200 μL of 50% acetonitrile solution
   (2-chloro-1-phenylalanine (4 ppm)) were filtered with a 0.22 μm filter
   membrane and transferred to a detection bottle for LC-MS detection.

4.7. Liquid Chromatographic and Mass Spectrometry Conditions

   Metabolomics analysis was performed using the Vanquish-Liquid
   chromatograph (LC) system coupled with the Orbitrap Exploris 120 mass
   spectrometer in both positive and negative ionization modes.
   Chromatographic conditions for LC analysis of metabolomics: flow speed:
   0.3 mL/min; column temperature:40 °C; injection volume: 2 μL. Mass
   spectrometry conditions of metabolomics: spray voltage(+): 3.50 kV;
   spray voltage(−): −2.50 kV; sheath gas pressure: 40 arb; aux gas flow:
   10 arb; capillary temperature: 325 °C; MS1 resolution: 60,000 FWHM; MS1
   range: m/z 100~1000; normalized collision energy: 30%; MS/MS
   resolution: 15,000 FWHM; dynamic exclusion time: automatic. Detailed
   positive/negative ion mode gradient elution program for LC analysis was
   provided in [136]Tables S1 and S2.

4.8. Metabolomics Data Processing

   The principal component analysis (PCA) and orthogonal partial
   least-squares discriminant analysis (OPLS-DA) were used to build the
   models by the metware cloud ([137]https://cloud.metware.cn/ (accessed
   on 15–16 December 2024)). Finally, statistically significant
   metabolites were defined as those with p values < 0.05 and VIP values >
   1. Differential metabolites were subjected to a pathway analysis using
   the OmicShare Tools ([138]https://www.omicshare.com/ (accessed on 17
   December 2024)). Metabolomics-discovered compounds were then linked to
   KEGG pathways for biological interpretation of higher-level system
   function.

4.9. Proteomics Analysis of Sample Preparation

   Protein samples were incubated with 5 mM DTT at 37 °C for 1 h and then
   returned to room temperature. Then, 10 mM of iodoacetamide were added
   and incubated for 45 min at room temperature without light, and the
   sample was diluted four times with 25 mM of ammonium bicarbonate, while
   trypsin was added according to the ratio of protein to trypsin 50:1 and
   incubated overnight at 37 °C. The next day, formic acid was added to
   adjust the pH to less than 3, and then the enzyme digestion was
   terminated. The samples were desalted using a C18 desalting column,
   activated with 100% acetonitrile, equilibrated with 0.1% formic acid,
   loaded onto the column, washed with 0.1% formic acid to remove
   impurities, and finally eluted with 70% acetonitrile, and the
   flow-through solution was collected and lyophilised.

4.10. Chromatographic and Mass Spectrometry Conditions

   Proteomics analysis was performed using the RIGOL L-3000 High
   Performance Liquid Chromatography System (RIGOL TECHNOLOGIES Co., Ltd.,
   Suzhou, China) coupled with the Orbitrap Exploris^TM 480 mass
   spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). Mass
   spectrometry conditions of proteomics: compensation voltage:
   −45/−65·s^−1; ion spray voltage: 2.0 kV; ion transmission tube
   temperature: 320 °C; full scanning range of mass spectrometry: m/z
   350–1500; primary mass spectral resolution: 120,000 (m/z); automatic
   gain control: 4 × 10^5; maximum injection time: 50 ms; secondary mass
   spectral resolution: 15,000 (200 m/z); automatic gain control: 5 ×
   10^4; maximum injection time: 22 ms; and peptide fragmentation
   collision energy: 33%. The gradient elution program for LC analysis of
   proteomics was detailed and provided in [139]Table S3.

4.11. Proteomics Data Processing

   The principal component analysis (PCA) and orthogonal partial
   least-squares discriminant analysis (OPLS-DA) were used to build the
   models by the Metware Cloud ([140]https://cloud.metware.cn/). Finally,
   statistically significant proteins were defined as those with p values
   < 0.05 and FC > 1.5. Differential proteins were subjected to a pathway
   analysis using the OmicShare Tools ([141]https://www.omicshare.com/).
   Metabolomics-discovered compounds were then linked to KEGG pathways and
   GO enrichments for biological interpretation of higher-level system
   function.

4.12. Statistical Analysis

   Student’s t-test was performed using Origin version 2021 software for
   statistical comparisons. Statistical significance was determined at a
   threshold of p < 0.05, thus delineating significant differences among
   the groups being compared.

5. Conclusions

   This study systematically elucidated the anti-diabetic mechanisms of
   DNL through integrated metabolomics and proteomics in a db/db mouse
   model of T2D. DNL intervention significantly ameliorated hyperglycemia,
   insulin resistance, and renal dysfunction, restoring metabolic
   homeostasis and alleviating pathological phenotypes. Multi-omics
   analysis identified 39 differential metabolites and 113 differentially
   expressed proteins, with integrative pathway mapping highlighting DNL’s
   regulation of arginine/proline metabolism, glycerophospholipid
   metabolism, and sulfur metabolism. Mechanistically, DNL suppressed
   oxidative stress and inflammation by modulating homocysteine and
   succinic acid levels, restored insulin sensitivity via
   glycerophospholipid balance, and enhanced energy metabolism through TCA
   cycle reactivation. The correlation network between metabolites and
   proteins further validated DNL’s multi-target action, particularly its
   inhibition of NF-κB and PI3K-Akt pathways linked to insulin resistance.
   These findings not only decode DNL’s therapeutic potential but also
   provide a multi-omics framework to probe the systemic mechanisms of
   traditional herbal interventions. This work provides critical insights
   for developing DNL-based therapies and underscores the value of
   integrative omics in bridging traditional medicine with modern
   pharmacodynamics.

Supplementary Materials

   The following supporting information can be downloaded at:
   [142]https://www.mdpi.com/article/10.3390/ph18071061/s1.
   [143]pharmaceuticals-18-01061-s001.zip^ (1.6MB, zip)

Author Contributions

   Conceptualization, A.Z. and S.Q.; Methodology, Z.W., S.G., Y.C., D.X.,
   Y.W., A.Z., J.D. and S.Q.; Validation, S.Q.; Formal analysis, J.D.;
   Investigation, Z.W., X.W., S.G., Y.C., D.X., Y.W., A.Z., J.D. and S.Q.;
   Resources, X.W. and J.D.; Data curation, S.Q.; Writing—original draft,
   Z.W.; Writing—review & editing, Z.W. All authors have read and agreed
   to the published version of the manuscript.

Institutional Review Board Statement

   The study protocols were ethically reviewed and approved by the
   Institutional Review Board of Hainan Medical University (Ethics
   Approval Number: HYLL2024-5-09, 31 May 2024).

Informed Consent Statement

   Not applicable.

Data Availability Statement

   Data is contained within the article or [144]Supplementary Materials.

Conflicts of Interest

   Author Jun Dai was employed by the company Hainan Shengrong Technology
   Co., Ltd. The remaining authors declare that the research was conducted
   in the absence of any commercial or financial relationships that could
   be construed as a potential conflict of interest.

Funding Statement

   The authors are grateful for the generous support from the Program of
   Key research and development project of Hainan Province (No.
   ZDYF2024SHFZ076); Hainan Province “Nanhai New Star” Science and
   Technology Innovation Talent Platform Project by Hainan Provincial
   Department of Science and Technology (NHXXRCXM202317); Academic
   Enhancement Support Program by Hainan Medical University (XSTS2025079),
   Program of Natural Science Foundation of State (Grant No. 81973745);
   Haikou Science and technology planning project: (2022-008); Anhui
   Province Outstanding Young Teachers Training Program for Universities
   (YQYB2024055); Key Project of Natural Science Research for Universities
   in Anhui Province (2022AH051677).

Footnotes

   Disclaimer/Publisher’s Note: The statements, opinions and data
   contained in all publications are solely those of the individual
   author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI
   and/or the editor(s) disclaim responsibility for any injury to people
   or property resulting from any ideas, methods, instructions or products
   referred to in the content.

References