Abstract

   Heart failure (HF) is a chronic condition that progressively worsens
   and continues to be a major financial burden and public health concern.
   The “gut-heart” axis provides an innovative perspective and therapeutic
   strategy for preventing and treating heart failure. Shenfu injection
   (SFI) is a Traditional Chinese Medicine-based treatment demonstrating
   potential as a therapeutic strategy for heart failure. However, the
   precise therapeutic mechanisms of SFI in heart failure are not
   completely characterized. In this study, HF models were established
   utilizing subcutaneous multipoint injection of isoproterenol (ISO) at a
   dosage of 5 mg kg^−1·d^−1 for 7 days. Serum levels of inflammatory
   biomarkers were quantified using protein microarrays. Rat feces were
   analyzed using untargeted metabolomics research and 16S rRNA
   sequencing. The link between gut microbiota and metabolites was
   examined using a MetOrigin and Spearman correlation analysis. Our
   results show that Shenfu injection effectively enhances cardiac
   function in rats with ISO-induced heart failure by potentially
   modulating pro-/anti-inflammatory imbalance and reducing serum and
   urine Trimethylamine-N-oxide (TMAO) levels. Moreover, SFI significantly
   increases the abundance of Bacteroidota at the phylum level, thereby
   improving disrupted gut microbiota composition. Additionally, SFI
   supplementation enriches specific genera known for their capacity to
   produce short-chain fatty acids. SFI was found to be associated with
   three key metabolic pathways, as revealed by fecal metabonomics
   analysis, including the pentose phosphate pathway, pyrimidine
   metabolism, and purine metabolism. Metabolite tracing analysis revealed
   that Taurine and hypotaurine metabolism was found to be specific to the
   microbial community. The biosynthesis of Pyrimidine metabolism, Purine
   metabolism, beta-alanine metabolism, Naphthalene degradation,
   Pantothenate, and CoA biosynthesis were identified as co-metabolic
   pathways between microbes and host. The Spearman correlation analysis
   was also significantly correlated to differentially expressed
   metabolites regulated by SFI and the gut microbiota. These results
   suggest that SFI improves ISO-induced heart failure by modulating
   co-metabolism and regulating the TMAO-inflammation axis.

   Keywords: gut microbiota, 16S rDNA, metabolomics, inflammation, Shenfu
   injection

Graphical Abstract

   [38]graphic file with name FPHAR_fphar-2024-1412300_wc_abs.jpg

1 Introduction

   Heart failure (HF) is a complex and potentially fatal condition caused
   by a variety of cardiovascular disorders. It is marked by high costs,
   substantial morbidity and mortality, and therefore presents a
   significant worldwide health concern ([39]Groenewegen et al., 2020).
   Over 64 million people worldwide are affected by chronic heart failure
   (CHF), and despite ongoing advancements in diagnosis, treatment, and
   management, re-admission and mortality rates have not decreased
   considerably ([40]Savarese et al., 2023). Therefore, identifying
   potential therapeutic targets for heart failure is crucial for
   preventing disease progression.

   The expanding focus on the potential role of the gastrointestinal
   system in the development of HF has sparked increasing interest
   ([41]Tang et al., 2019). The gut-heart axis, which describes the
   interaction between the gut and the heart, presents a unique viewpoint
   and potential therapeutic avenue for managing heart failure ([42]Madan
   and Mehra, 2020; [43]Bui et al., 2023). Tang et al. pioneered the “gut
   hypothesis of heart failure” which postulated that reduced cardiac
   output in heart failure could compromise intestinal perfusion,
   resulting in mucosal ischemia and subsequent damage to the intestinal
   mucosa. Intestinal barrier dysfunction can cause increased
   permeability, leading to malnutrition, bacterial translocation, and
   higher levels of endotoxins in the bloodstream ([44]Tang et al., 2019).
   This can trigger inflammation linked to heart failure, with factors
   like epithelial dysfunction, gut barrier compromise, microbiota
   imbalance, and abnormal gut metabolites playing a role in the
   development and worsening of the condition ([45]Tang et al., 2019;
   [46]Cui et al., 2023). Increased systemic inflammation is reported to
   be linked to gut dysbiosis, especially in HF patients ([47]Madan and
   Mehra, 2020).

   Trimethylamine-N-oxide (TMAO) is a metabolite derived from the gut
   microbiota, originating from phosphatidylcholine, choline, betaine, and
   L-carnitine. These nutrients are abundant in seafood, dairy products,
   egg yolks, muscle meat, and organ meats. The gut microbiota’s
   trimethylamine (TMA) lyase hydrolyzes these nutrients to form the TMAO
   precursor TMA, which is further oxidized by hepatic flavin
   monooxygenase to form TMAO ([48]Zhang et al., 2021). It has been
   suggested that dietary choline and TMAO may contribute to
   cardiovascular disease (CVD) ([49]Canyelles et al., 2023). The increase
   in blood TMAO levels has been linked to increased inflammatory genes
   and cytokines, resulting in increased oxidative stress ([50]Saaoud et
   al., 2023). The pathophysiology of numerous inflammatory conditions is
   considerably affected by the activation of inflammatory pathways and
   the production of inflammatory cytokines by TMAO. The development of
   heart failure has been linked, in part, to inflammation, thus,
   alleviating inflammation is crucial for enhancing the clinical
   manifestations and prognosis of HF. Consequently, targeting the
   TMAO-inflammation axis could be a unique therapeutic strategy for
   treating heart failure.

   Based on recent research, Traditional Chinese Medicine (TCM) can
   prevent the development of cardiovascular diseases by regulating the
   gut microbiota and reducing inflammation ([51]Huang et al., 2020). TCM
   is a valuable tool for improving cardiac function, mitigating medical
   conditions, and improving the overall wellbeing of patients. In
   treating post-acute myocardial infarction heart failure, recent
   pharmacological evaluations—such as those conducted by [52]Wu et al.
   (2022)—have demonstrated the reliability and effectiveness of SFI,
   highlighting its potential to improve heart function and reduce
   associated symptoms ([53]Wu et al., 2022). Researchers have shown that
   TCM is effective in modulating the dysbiosis of gut microbiota,
   stimulating the proliferation of beneficial microorganisms, suppressing
   the proliferation of detrimental microorganisms, harmonizing the
   abundance of commensal and pathogenic bacteria, and sustaining a
   favorable gut milieu ([54]Huang et al., 2020; [55]Jia et al., 2020).
   Shenfu injection (SFI) is one of the representative prescriptions of
   the warming Yang method. It originates from the classic formula Shenfu
   Decoction and continues to be applied extensively in the treatment of
   heart failure with remarkable curative effects ([56]Luo et al., 2021).
   Currently, SFI is a formulation created with Panax ginseng C.A. Mey
   [Araliaceae; Ginseng radix et rhizoma rubra] and Aconitum carmichaelii
   Debx [Ranunculaceae; Aconiti lateralis radix praeparata] ([57]Chen et
   al., 2017) utilizing cutting-edge technologies. Previous
   pharmacological experiments have confirmed that Shenfu Injection can
   enhance myocardial contractility, increase cardiac output, and inhibit
   cardiomyocyte apoptosis caused by myocardial ischemia/reperfusion
   injury, among other cardiovascular effects ([58]Wu H. et al., 2019;
   [59]Wang et al., 2021). In addition, a study by Zhu et al. demonstrated
   that this intervention can mitigate myocardial injury through the
   prevention of mitochondrial apoptosis and facilitating vasodilation by
   augmenting eNOS activity via the PI3K/Akt signaling pathway ([60]Zhu et
   al., 2020). Nevertheless, the precise underlying mechanism remains
   incompletely understood. Therefore, this research endeavors to delve
   into the mechanisms underlying the enhancement of cardiac function in
   chronic heart failure through the administration of SFI.

2 Material and method

2.1 Shenfu injection

   Shenfu injection (batch number:221111AK05) was purchased from Huarun
   Sanjiu Pharmaceutical Co., Ltd. (Ya’an, Sichuan, China). This injection
   is a solution extracted from Panax ginseng C.A. Mey and Aconitum
   carmichaelii Debeaux, as described in [61]Table 1. The two Chinese
   herbs—Panax ginseng C.A. Mey and Aconitum carmichaelii Debeaux—in crude
   form were soaked and concentrated into solutions of 1 mg/mL and
   2 mg/mL, respectively, and then mixed to form the SFl injection
   ([62]Huang et al., 2024). All voucher specimens are deposited in the
   herbarium center of Huarun Sanjiu Pharmaceutical Co., Ltd. The plant
   was identified was conducted by a botanist at Huarun Sanjiu
   Pharmaceutical Co., Ltd.

TABLE 1.

   Information of raw herbs in SFl.
   Latin name Plant part English name Chinese name Ratio (%) Origin of
   place Month of harvest
   Panax ginseng C.A Mey Root Red Ginseng Root Hongshen 33.3 Jilin, China
   September
   Aconitum carmichaelii Debeaux Tuber Aconite Fuzi 66.6 Sichuan, China
   August
   [63]Open in a new tab

2.2 Reagents and chemicals

   Metoprolol tartrate was bought from AstraZeneca Pharma (NO2302069)
   (Jiangsu, China). Isoproterenol (C[11]H[17]NO[3]▪HCL) was purchased
   from Aladdin (Shanghai Aladdin Biochemical Technology Co., Ltd.,
   Shanghai, China). Protein chip detection: A protein chip assay kit
   (QAR-INF-1) was purchased from Raybiotech, Inc. (Norcross, GA, United
   States of America). The TMAO kit was provided by Xiamen Luncanshuo
   Biotech Co., Ltd. The NT-proBNP and LPS kit was provided by Wuhan
   Huamei Biological Engineering Co., Ltd.

2.3 Animals and ethics statement

   This study received approval from the Ethical Committee of Hunan
   University of Chinese Medicine (Hunan, China) under Approval Number
   LL202309230002. Thirty-two male Sprague-Dawley (SD) rats, aged 6 weeks
   that weighed 200–220 g were procured from Hunan SJA Laboratory Animal
   Co. Ltd. (Hunan, China) and maintained under standard husbandry
   conditions.

2.4 Quality control of the shenfu injection

   Ultra-High Performance Liquid Chromatography Quadrupole-Orbitrap Mass
   Spectrometry (UHPLC-QE-MS) was performed to control the quality of the
   Shenfu injection. The analysis was performed using a Vanquish UHPLC
   system (Thermo Fisher) with a Waters UPLC BEH C18 column ([64]Xiao et
   al., 2020). Samples (5 μL) were eluted at 0.5 mL/min with a mobile
   phase of 0.1% formic acid in water (A) and acetonitrile (B). A Q
   Exactive Focus mass spectrometer with Xcalibur software was used for MS
   and MS/MS data acquisition (IDA mode, m/z range: 100–1,500). The top
   three ions per cycle were selected for MS/MS.

2.5 Establishment of animal model and drug administration

   The animals in the study were divided into six groups at random (n =
   8), namely, the control group (CON group), model group (MOD group),
   positive drug group (metoprolol, the beta-adrenergic antagonist, MET
   group, 10 mg/kg) ([65]Lujan and DiCarlo, 2020), the Shenfu injection
   low dose group (3.0 mL/kg, SFI_L group), the Shenfu injection middle
   dose group (6.0 mL/kg, SFI_M group), and the Shenfu injection high dose
   group (12.0 mL/kg, SFI_H group). According to prior research, rats were
   administered subcutaneous multipoint injections of isoproterenol at a
   dosage of 5 mg·kg-1·day-1 for 7 days to induce the heart failure model
   ([66]Huang et al., 2022). After adaptive feeding continued for 2 weeks,
   the CON and MOD groups were administered distilled water by gavage and
   had normal saline injected intraperitoneally, while the SFI group was
   intraperitoneally injected with SFI and administered purified water by
   gavage, the MET group was injected with metoprolol intragastrically and
   intraperitoneally with sterile water for 15 days ([67]Figure 1A).

FIGURE 1.

   [68]FIGURE 1
   [69]Open in a new tab

   (A) Schematic illustration of the animal experimental design. (B) The
   UHPLC-QE-MS chromatogram illustration of three batches of SFI samples
   detected in positive ion mode (B–D) and negative ion mode (E–G).

2.6 Sampling and bioassays

   At the end of the experimental period, the rats were euthanized using
   urethane anesthesia (1.0 g/kg, i.p.). The blood samples were obtained
   from the abdominal aorta and allowed to clot at room temperature (25°C
   ± 2°C). Subsequently, centrifugation was performed to obtain serum.
   Myocardial and colonic tissues were fixed in a 4% paraformaldehyde
   solution for 48 h. Hematoxylin and eosin (H&E) staining was performed
   on tissue sections fixed in paraffin.

   The scoring of cardiac pathological changes was based on the criteria
   established in the study by [70]Rezkalla et al. (1990). Three
   high-power fields were observed in each slice to calculate the
   percentage of inflammatory infiltration and necrosis area compared to
   the total area. A scoring system was established as follows: 0 points
   for no lesion, one point for a lesion area of less than 25%, two points
   for a lesion area of 25%–49%, three points for a lesion area of
   50%–75%, and four points for a lesion area greater than 75% ([71]Niu et
   al., 2017; [72]Xiao et al., 2019). After being extracted from the
   colon, fecal samples were kept in liquid nitrogen at −80°C. Six rats
   were randomly chosen from each group for 16S rRNA sequencing and
   microbiome analysis.

2.7 Echocardiography

   For the calculation of the Left Ventricular Ejection Fraction (LVEF)and
   Left Ventricular Fractional Shortening (LVFS), the computer software
   (VINNO6, Suzhou VINNO Technology Co., Ltd., Suzhou, China) was used
   ([73]Li et al., 2023).

2.8 UHPLC-OE-MS fecal metabolic profiling

   Ultra-High Performance Liquid Chromatography-Ion Trap-Orbitrap Mass
   Spectrometry (UHPLC-OE-MS) was peformed using HPLC system connected to
   a Waters ACQUITY UPLC BEH Amide column coupled to Orbitrap Exploris 120
   mass spectrometer (Orbitrap MS, Thermo). Conditions for UHPLC-MS
   analysis can be found in [74]Supplementary Material ([75]Supplementary
   Material 1). A Quality control (QC) mixture was created by combining
   extracts from all samples to assess variability. [76]Supplementary
   Figure S1A illustrates the remarkable stability and reproducibility of
   the instruments utilized in this investigation. The distinct clustering
   in the QC samples demonstrates consistent and reliable performance.

   Results were processed as specified in previous publications ([77]Li et
   al., 2021). In this study, 15,533 in two (positive and negative) ion
   modes were detected, and 8,888 metabolites were found after relative
   standard deviation de-nosing. Metabolite identification was carried out
   using the R program and BiotreeDB (V3.0) ([78]Chen et al., 2023).
   Biotree company in Shanghai, China, provided support for the LC-MS
   detection.

   The Metabolomics dataset was analyzed using SIMCA14.1 software from
   Umetrics, Sweden. Peak numbers, sample identifiers, and normalized peak
   areas were imported for analysis. Principle Component Analysis (PCA)
   and Orthogonal Projections to Latent Structures Discriminate Analysis
   (OPLS-DA) were used, followed by a seven-fold cross-validation to
   determine R ^2 and Q^2 values. Metabolites with Variable Importance in
   the Projection (VIP) > 1.0 and p < 0.05 were considered significantly
   changed. Identified markers were then analyzed using MetaboAnalyst to
   pinpoint the metabolic pathways they influenced. The origin and
   function analysis of metabolites was performed using MetOrigin ([79]Liu
   et al., 2023). This methodology aligns with the approach described in a
   previous publication ([80]Zhao et al., 2023).

2.9 16S rRNA gene sequencing

   Tiangen Fecal Genomic DNA was used to obtain total genomic DNA
   following the protocols provided by the manufacturer ([81]Liang et al.,
   2022). After total DNA extraction from the samples, specific primers
   containing barcodes were created using the complete primer sequences.
   The following steps included PCR amplification, purification,
   quantification, and normalization to generate a sequencing library. The
   libraries underwent quality control before being sequenced using the
   PacBio Sequel II system. The output data was converted to CCS files
   using SMRT Link analysis software, which were then used to identify
   samples based on their barcode sequences and converted to FASTQ format
   data. UCHIME v4.2 software was used to detect and eliminate chimera
   sequences. The resulting effective CCS sequences were used for further
   analysis ([82]Edgar, 2013). Based on the species annotation information
   of 16S rDNA sequences in the Silva database
   ([83]https://www.arb-silva.de/), the operational taxonomic units (OTUs)
   were classified into various taxonomic levels, including phylum, class,
   order, family, and genus ([84]Ma et al., 2022). Alpha diversity was
   assessed using Chao, ACE, Simpson, and Shannon indices. Chao and ACE
   indices estimate species richness, while Simpson and Shannon indices
   gauge diversity. The Biotree company in Shanghai, China, provided
   support for the 16S rRNA gene sequencing.

2.10 Protein chip detection

   The protein chip assay kit QAR-INF-1 was utilized to measure the levels
   of inflammatory factors in four distinct groups. The assay was carried
   out following the manufacturer’s protocol.

2.11 Statistical analysis

   Results were presented as mean ± SD. Statistical analyses were
   performed using IBM SPSS Statistics 25. The Kolmogorov-Smirnov test was
   employed to evaluate the normal distribution of the data. Parametric
   tests were applied to normally distributed data, whereas non-parametric
   tests were used for data that did not exhibit normal distribution.
   One-way ANOVA or Dunnett’s T3 test was used based on variance
   homogeneity. Differences with p < 0.01 were substantially significant,
   while p < 0.05 was significant. Correlations were calculated using
   Spearman’s rank correlation (presented as Spearman rho). Prism GraphPad
   software was used for figure creation.

3 Results

3.1 Qualitative detection of SFI under LC-MS conditions

   UHPLC-QE-MS analysis was employed for the quality control assessment of
   SFI. [85]Figure 1B presents representative base peak intensity
   chromatograms of three different batches of SFI samples. These
   chromatograms demonstrated effective separation and detection of major
   SFI components in both positive and negative ion modes, using optimized
   chromatographic and MS conditions ([86]Figures 1B–G). The chromatograms
   clearly indicated stable and consistent composition across the three
   SFI batches. The core compounds were selected for Total ion
   chromatogram (TIC) icon peak labeling, the details on the compounds are
   illustrated in [87]Table 2.

TABLE 2.

   The core compounds of SFI.
   No Name MZ RT Formula Type MS2
   1 (20S,23E)-3beta,12beta,20,25-tetrahydroxydammarane-23-ene
   20-O-beta-D-glucopyranoside|ginsenoside Rh13 653.43 367.382
   C[36]H[62]O[10] NEG 653.427309; 668.887099; 336.878879; 537.416933;
   581.417049
   2 Ginsenoside F1 683.44 337.19 C[36]H[62]O[9] NEG 683.442099;
   637.424548; 92.675844; 101.024143; 475.379881
   3 Chikusetsu saponin IVa 793.44 425.282 C[42]H[66]O[14] NEG 793.432508;
   71.013628; 59.013837; 101.023884; 113.023864
   4 Ginsenoside Ro 955.49 328.119 C[48]H[76]O[19] NEG 955.487604;
   955.515614; 793.434039; 569.389402; 75.008669
   5 Ginsenoside Re 945.54 364.179 C[48]H[82]O[18] NEG 945.538413;
   945.48327; 71.013666; 101.024208; 89.024148
   6 6-Gingerol 293.18 408.432 C[17]H[26]O[4] NEG 221.152945; 236.105298;
   220.145786; 293.175077; 71.013682
   7 Ginsenoside Rg2 783.49 322.2315 C[42]H[72]O[13] NEG 783.480972;
   783.50177; 475.380023; 92.678382; 59.013959
   8 Ginsenoside Rg3 783.48 483.441 C[42]H[72]O[13] NEG 783.480964;
   783.501762; 101.024211; 71.013629; 113.023869
   9 Ginsenoside Rg3 (R-FORM) 829.49 487.319 C[42]H[72]O[13] NEG
   783.481909; 783.502707; 101.024255; 71.013675; 113.024865
   10 Ginsenoside-Rg1 845.49 439.588 C[42]H[72]O[14] NEG 799.484899;
   799.506337; 637.427691; 101.024146; 92.675001
   1 Bullatine B 438.28 94.0406 C[24]H[39]NO[6] POS 438.282997;
   420.277593; 439.154571; 388.245091; 406.256222
   2 Ginsenoside Rf 823.48 296.459 C[42]H[72]O[14] POS 823.475297;
   823.520123; 365.102344; 91.497063; 245.062527
   3 Ginsenoside RG1 823.48 217.128 C[42]H[72]O[14] POS 823.475313;
   823.520139; 643.41145; 203.051766; 91.497901
   4 Benzoylaconine 604.31 211.611 C[32]H[45]NO[10] POS 604.317144;
   105.03308; 109.100833; 95.084826; 123.11597
   5 Ginsenoside Rd 969.54 364.299 C[48]H[82]O[18] POS 969.533842;
   789.468488; 789.510567; 969.476581; 970.537505
   6 Ginsenoside Rb1 1,091.60 314.367 C[54]H[92]O[23] POS 85.028256;
   145.049096; 325.11008; 127.038605; 163.060486
   7 14-Deoxy-11,12-didehydroandrographolide 315.20 536.445 C[20]H[28]O[4]
   POS 133.100608; 315.195615; 189.090703; 157.10117; 105.070083
   8 Benzoylhypaconine 574.30 226.736 C[31]H[43]NO[9] POS 574.302648;
   542.27759; 105.033157; 63.810472; 140.107385
   9 Aconine 500.28 46.2104 C[25]H[41]NO[9] POS 500.284; 450.24443;
   55.586708; 468.255394; 418.218269
   10 Benzoylmesaconine 590.30 189.104 C[31]H[43]NO1[0] POS 590.293131;
   105.033194; 540.259757; 65.588159; 558.264448
   [88]Open in a new tab

   RT, retention time; NEG, negative ion mode; POS, positive ion mode; MS,
   mass spectrum.

3.2 Pharmacodynamic evaluation

3.2.1 SFI improved cardiac function in rats with ISO-induced HF

   The Representative echocardiographic images are shown in [89]Figure 2A.
   As shown in [90]Figures 2B, C, the LVEF and LVFS levels in the MOD
   group were significantly decreased compared to the CON group, but the
   LVFS levels in the SFI_L group, SFI_M group, SFI_H group, and MET group
   were increased compared with the MOD group ([91]Figure 2B). However,
   only the SFI_M and MET groups had markedly increased LVEF levels
   ([92]Figure 2C). As shown in [93]Figure 2D, there was a notable rise in
   the NT-proBNP levels in the MOD group. However, treatment with SFI and
   MET led to a significant reduction in NT-proBNP levels in the rats,
   with the most pronounced effect observed in the SFI_M group.

FIGURE 2.

   [94]FIGURE 2
   [95]Open in a new tab

   (A) Mode echocardiography illustration for the six groups. (B) Left
   ventricular fractional shortening. (C) Left ventricular ejection
   fraction. (D) Serum level of NT-proBNP, as measured by ELISA. (E)
   Pathological score of H&E staining (n = 3). (F) Representative H&E
   staining images of myocardial tissue (×200). Notes: Data were reported
   as mean ± SD, n = 8. ^## p < 0.01, compared with the CON group; ^** p <
   0.01, compared with the MOD group.

   The H&E staining in the MOD group showed myocardial dissolution and
   inflammation infiltrating the tissue, while myocardial structures were
   vastly improved after SFI and MET treatment, and the most significant
   effect was observed in the SFI_M group ([96]Figures 2E, F). The
   improvement observed after SFI treatment was not in a dose-dependent
   manner.

3.2.2 SFI attenuates gut function in rats with ISO-induced HF

   The H&E-stained colonic segment of the MOD group displayed compromised
   intestinal mucosa integrity, inflammatory cell infiltration into the
   colon tissue, and lymphoid hyperplasia ([97]Figure 3A), indicating
   potential damage to the intestinal barrier (leaky gut) in the MOD
   group. Following SFI and MET administration, the colon tissue
   structures were primarily improved, and the SFI_M treatment attenuated
   gut function better than SFI_L and SFI_H treatment. As shown in
   [98]Figure 3B, the serum Lipopolysaccharide (LPS) level in the MOD
   group was considerably higher than in the CON group, and following
   SFI_M administration, there was a reduction in the level of LPS.

FIGURE 3.

   [99]FIGURE 3
   [100]Open in a new tab

   (A) Representative H&E staining images of colon tissue (×200). (B)
   Serum level of LPS, as measured by ELISA. Notes: Data were reported as
   mean ± SD, n = 8. ^## p < 0.01, compared with the CON group; ^** p <
   0.01, compared with the MOD group.

   These findings indicate that SFI potentially preserves both cardiac and
   gut function in a manner that is not strictly dose-dependent, with the
   most significant impact observed at intermediate doses. Consequently, a
   dose of 6 mL/kg was selected for subsequent experiments, including
   cytokine microarray analysis, 16S rRNA sequencing, and metabolomics
   studies.

3.2.3 SFI improves inflammation and TMAO levels in rats with ISO-induced HF

   Cytokine microarray was employed to examine the anti-inflammatory
   influence of SFI on rats with ISO-induced HF ([101]Figure 4A).
   [102]Figures 4B–D illustrate that the three highly expressed
   pro-inflammatory cytokines (TNF-α, IL-1β, and IL-2), in the serum of HF
   rats were significantly downregulated in the presence of SFI. However,
   in the HF rat model ([103]Figures 4E–G), the anti-inflammatory
   cytokines IL-4, IL-10, and IL-13 were elevated, reflecting the
   pro-/anti-inflammation imbalance. It has been shown that the levels of
   TMAO were related to inflammation ([104]Wang Q. et al., 2022). As
   illustrated in [105]Figures 4H, I, TMAO levels in serum and urine in HF
   were substantially elevated but decreased after treatment with SFI and
   MET.

FIGURE 4.

   [106]FIGURE 4
   [107]Open in a new tab

   (A) The heatmap of inflammatory cytokines (n = 3); Cytokine microarray
   result of (B) TNF-α, (C) IL-1β, (D) IL-2, (E) IL-10, (F) IL-4 and (G)
   IL-13. (H) Serum level of TMAO, as measured by ELISA. (I) Urine level
   of TMAO, as measured by ELISA. (J) Spearman correlations between
   inflammatory cytokines and TMAOs.

3.2.4 Correlation analysis between TMAO and inflammation

   Red denotes a positive Spearman correlation, while blue denotes a
   negative Spearman correlation. As shown in [108]Figure 4J positive
   correlation was observed between inflammatory factors IL-1β and IL-2
   and serum TMAO and urine TMAO, as well as between urine TMAO and LPS.

3.3 16s rRNA sequencing results

3.3.1 OTU analysis

   A 97% similarity threshold was used to identify 3,870 operational
   taxonomic units (representing gamma diversity). The CON, MOD, SFI_M,
   and MET groups had 888, 845, 1,188, and 949 OTUs, respectively. This
   revealed that the number of operational taxonomic units (OTUs) in HF
   rats was decreased, however, the number of OTUs became elevated
   following SFI processing ([109]Figure 5A).

FIGURE 5.

   [110]FIGURE 5
   [111]Open in a new tab

   (A) Venn diagrams of the four groups. (B) The Shannon curve of the four
   groups. (C) PCoA analysis of the four groups; (D) NMDS analysis of the
   four groups (n = 6).

3.3.2 Alpha and beta diversity analysis

   The Shannon curve was generated based on OTU numbers to assess the
   quality of gut microbiota sequencing ([112]Figure 5B). The results
   indicated a consistent and minimal fluctuation in the curve, suggesting
   an adequate amount of sequencing data. Alpha diversity analysis was
   utilized to estimate the microbial diversity within individual samples.
   The Chao, Shannon, Simpson, and ACE indices in [113]Supplementary
   Figure S1B analysis showed that the CON and MOD groups differed from
   one another. Following SFI administration, there was a trend toward
   recovery in microbial richness, although this trend was not
   statistically significant. Nevertheless, weighted unifrac PCoA of beta
   diversity indicated a distinct divergence of the four groups. The four
   groups are distinctly segregated, as evidenced by ANOSIM analysis
   (ANOSIM: R = 0.3, p = 0.001). As shown in [114]Figure 5C, the CON and
   MOD groups could be separated, and the SFI_M group was close to the CON
   group, revealing that SFI_M had a therapeutic influence. The NMDS
   analysis corroborated the PCoA results (model stress = 0.14 < 0.2)
   ([115]Figure 5D).

3.3.3 Composition of gut microbiota and their differential analysis

   The barplot in [116]Figure 6A illustrates gut microbiota composition at
   the phylum and genus levels. Firmicutes, Bacteroidota, Proteobacteria,
   Actinobacteria, and Campylobacterota were the most abundant bacteria
   discovered at the phylum level. Firmicutes emerged as the predominant
   phylum, with relative abundances of 65.1%, 78.5%, 74.5%, and 83.2% in
   the CON, MOD, SFI_M, and MET groups. Furthermore, compared to the CON
   group, the MOD group showed a considerable decrease in bacteria with
   relative abundances of 24.3% and 11.2%, respectively. Additionally,
   treatment with SFI greatly enriched Bacteroidota (19.6% vs. 11.2%)
   compared with the MOD group.

FIGURE 6.

   [117]FIGURE 6
   [118]Open in a new tab

   (A) The histogram of species distribution at the phylum and (B) genus
   levels in four groups; (C) Differential analysis of gut microbiota at
   phylum and genus level (n = 6).

   At the genus level, there were 29 genera with an abundance of over 1%
   ([119]Figure 6B). Ruminococcus, unclassified_f_Muribaculaceae, and
   unclassified_f_ Erysipelotrichaceae were less abundant in the MOD group
   than in the CON group. In contrast, in the MOD group, Romboutsia,
   Clostridium_sensu_stricto_1, Allobaculum, and Limosilactobacillus were
   more productive when compared to the CON group. The levels of the
   microorganisms mentioned above may recover to varied degrees following
   SFI therapy ([120]Figure 6C).

   To ascertain the specific bacteria linked to SFI, we used linear
   discriminant analysis (LDA) and effect size (LEfSe) to determine the
   predominant taxon ([121]Figure 7A). Our analysis revealed 56 OTUs as
   key discriminants, with Bacteroidota showing significant
   overrepresentation (LDA scores >4.5) in the CON group. The microbiota
   of Prevotellaceae_NK3B31_group was identified to be the most prevalent
   in the CON group as well (LDA scores >3.6) ([122]Figure 7B). After
   additional LDA analysis, 15 OTUs were enriched in the CON group, 3 OTUs
   were enhanced in the MOD group, 24 OTUs were enriched in the SFI_M
   group, and 6 OTUs in the MET group. From our findings, it is evident
   that the microbiota in the gut of the rats can be improved through SFI
   intervention.

FIGURE 7.

   [123]FIGURE 7
   [124]Open in a new tab

   (A) Cladogram illustrating the phylogenetic distribution of microbiota
   correlated with the CON, MOD, SFI_M, and MET groups. (B) The variations
   in the abundance of microbiota among the CON, MOD, SFI_M, and MET
   groups (n = 6).

3.4 Study of the fecal metabolomics

3.4.1 Metabolic profile analysis

   The metabolic changes in the four experimental groups were visualized
   in the current study using PCA. [125]Figures 8A, B illustrate that the
   CON and MOD groups can be separated, and the SFI_M groups were
   scattered between them, indicating that the drug intervention regulated
   the metabolic disturbances induced by HF.

FIGURE 8.

   [126]FIGURE 8
   [127]Open in a new tab

   PCA score plots among four groups (R^2X = 0.54): (A) 2D scatter and (B)
   3D scatter. (C) OPLS-DA score plots of CON and MOD group (R^2X = 0.753,
   R^2Y = 1, Q^2 = 0.893). (D) The permutation test (n = 200) for the
   OPLS-DA model of CON and MOD group. (E) OPLS-DA score plots of MOD and
   SFI_M group (R^2X = 0.413, R^2Y = 0.997, Q^2 = 0.643). (F) The
   permutation test (n = 200) for the OPLS-DA model of MOD and SFI_M
   group. (G) OPLS-DA score plots of MOD and MET group (R^2X = 0.486, R^2Y
   = 0.995, Q^2 = 0.755). (H) The permutation test (n = 200) for the
   OPLS-DA model of MOD and MET group (n = 6).

3.4.2 Differential metabolites analysis

   OPLS-DA analysis shows that the four groups were separated
   ([128]Figures 8C, E, G). The OPLS-DA model was also statistically
   validated through 200 permutation tests, as illustrated in [129]Figures
   8D, F, H. The model was reliable. Forty-eight metabolites with VIP >1.0
   and p < 0.05 were chosen as the metabolites linked to the pathological
   alteration of HF. All the metabolites showed a callback trend, and 28
   metabolites were significantly changed ([130]Table 3).

TABLE 3.

   Differential metabolites of four groups.
   No RT(s) Var ID Formula KEGG Ion form MZ Trend
   M/C S/M M/M
   1 66.8 2,4-Dimethylphenol C[8]H[10]O [131]C14582 NEG 121.066 ↓^## ↑^**
   ↑
   2 66.8 3-(2-Hydroxyphenyl)propanoic acid C[9]H[10]O[3] [132]C01198 NEG
   165.056 ↓^## ↑^** ↑
   3 66.8 3-(3-Hydroxyphenyl)propanoic acid C[9]H[10]O[3] [133]C11457 NEG
   165.056 ↓^## ↑^** ↑
   4 242.1 Edetic acid edetic acid C[10]H[16]N[2]O[8] [134]C00284 NEG
   291.084 ↓^## ↑^* ↑^*
   5 59.7 Indolelactic acid C[11]H[11]NO[3] [135]C02043 NEG 204.067 ↓^##
   ↑^* ↑^*
   6 249.8 2-methylcitrate C[7]H[10]O[7] [136]C02225 NEG 187.025 ↓^## ↑^*
   ↑
   7 132.2 Caffeic acid C[9]H[8]O[4] [137]C01197 NEG 179.039 ↓^## ↑^* ↓
   8 20.5 PC(15:0/P-16:0) C[39]H[78]NO[7]P [138]C00157 POS 704.566 ↑^## ↓
   ↓
   9 120 Sedoheptulose C[7]H[14]O[7] [139]C02076 NEG 209.067 ↓^## ↑^** ↑
   10 111.8 D-Glucosaminic acid C[6]H[13]NO[6] [140]C03752 POS 196.083
   ↑^## ↓^* ↑^*
   11 182.8 2-Hydroxyhexanedioic acid C[6]H[10]O[5] [141]C02360 NEG
   143.035 ↓^## ↑^* ↑
   12 146.9 Pseudo uridine C[9]H[1]2N[2]O[6] [142]C02067 NEG 243.063 ↑^##
   ↓ ↑
   13 159.1 Glycocholic acid C[26]H[43]NO[6] [143]C01921 NEG 464.302 ↓^##
   ↑ ↑
   14 182.8 D-Ribose C[5]H[10]O[5] [144]C00121 NEG 149.046 ↓^## ↑^** ↑
   15 28.5 Gentisic acid C[7]H[6]O[4] [145]C00628 NEG 153.020 ↓^## ↑^** ↑
   16 68.1 2-Hydroxyethanesulfonic acid C[2]H[6]O[4]S [146]C05123 NEG
   124.992 ↓^# ↑^* ↓
   17 22.6 4-Methyl-5-thiazoleethanol C[6]H[9]NOS [147]C04294 POS 144.048
   ↓^# ↑^** ↑^*
   18 20.4 Arachidonic acid (AA) C[20]H[32]O[2] [148]C00219 NEG 303.234
   ↓^## ↑ ↑
   19 168.4 Urobilin C[33]H[42]N[4]O[6] [149]C05794 POS 591.3187 ↓^# ↑^* ↑
   20 185.5 Tyrosine C[9]H[11]NO[3] [150]C00082 POS 182.081 ↓^## ↑ ↑
   21 135.6 Glycochenodeoxycholic acid C[26]H[43]NO[5] [151]C05466 NEG
   448.307 ↓^## ↑ ↑
   22 21 Pyrocatechol C[6]H[6]O[2] [152]C00090 NEG 109.030 ↓^# ↑ ↑
   23 139.9 Cholic acid C[24]H[40]O[5] [153]C00695 NEG 407.282 ↓^# ↑ ↑^*
   24 182.8 Fructose C[6]H[12]O[6] [154]C02336 NEG 179.057 ↓^## ↑ ↓
   25 240.1 Nicotinamide riboside (NR) C[11]H[15]N[2]O[5] [155]C03150 POS
   255.097 ↓^## ↑ ↓
   26 163.1 Phenylalanine C[9]H[11]NO[2] [156]C00079 POS 166.086 ↓^# ↑ ↑
   27 19.2 Phenacetin C[10]H[13]NO[2] [157]C07591 POS 180.102 ↓^## ↑^** ↑
   28 294.6 S-Adenosylmethionine C[15]H[22]N[6]O[5]S [158]C00019 POS
   399.145 ↓^# ↑ ↓
   29 229.9 5-Aminopentanoic acid C5H11NO2 [159]C00431 POS 118.086 ↓^#
   ↑^** ↓
   30 134.3 Nicotinate C6H5NO2 [160]C00253 NEG 122.0250 ↑^# ↓ ↑^*
   31 177.6 1-Aminopropan-2-ol C3H9NO [161]C05771 POS 76.076 ↓^# ↑^** ↓
   32 127.4 Xanthine C5H4N4O2 [162]C00385 NEG 151.026 ↑^## ↓ ↑^*
   33 87.5 Hypoxanthine C[5]H[4]N[4]O [163]C00262 POS 137.046 ↑^## ↓^* ↓
   34 141 Mevalonic acid C[6]H[12]O[4] [164]C00418 NEG 147.067 ↓^# ↑ ↓
   35 257.5 Histidine C[6]H[9]N[3]O[2] [165]C00135 POS 156.077 ↓^# ↑ ↓
   36 76.7 1-Methylguanine C[6]H[7]N[5]O [166]C04152 POS 166.072 ↑^## ↓^*
   ↑
   37 186.8 Methylimidazoleacetic acid C[6]H[8]N[2]O[2] [167]C05828 NEG
   139.052 ↓^## ↑ ↑^*
   38 93.3 3-Hydroxyisovaleric acid C[5]H[10]O[3] [168]C20827 NEG 117.056
   ↓^# ↑ ↓
   39 190.3 Proline C[5]H[9]NO[2] [169]C16435 POS 116.070 ↓^# ↑ ↑
   40 215.5 4-Guanidinobutyric acid C[5]H[11]N[3]O[2] [170]C01035 POS
   146.092 ↓^# ↑^** ↑
   41 26.4 Euscaphic acid C[30]H[48]O[5] [171]C17890 POS 489.358 ↑^# ↓^**
   ↓^*
   42 223.6 4-Aminobutyric acid (GABA) C[4]H[9]NO[2] [172]C00334 POS
   104.071 ↓^## ↑ ↓
   43 35.8 Uracil C[4]H[4]N[2]O[2] [173]C00106 NEG 111.020 ↑^# ↓^* ↓
   44 23.4 L-tyrosine-methyl-ester C[10]H[13]NO[3] [174]C03404 POS 196.097
   ↓^# ↑^* ↑
   45 235.4 Chloroquine C[18]H[26]C[l]N[3] [175]C07625 POS 320.185 ↑^#
   ↓^** ↓
   46 101.4 Deoxyinosine C[10]H[12]N[4]O[4] [176]C05512 POS 253.093 ↑^#
   ↓^* ↑
   47 31.9 Thymine C[5]H[6]N[2]O[2] [177]C00178 NEG 125.036 ↑^# ↓^** ↑
   [178]Open in a new tab

   RT, retention time; NEG, negative ion mode; POS, positive ion mode; MS,
   mass spectrum.

   ^# p < 0.05. ^## p < 0.01 compared with the CON group; *p < 0.05, **p <
   0.01 compared with the MOD group. M/C: MOD group compared with the CON
   group. S/M: SFI_M group compared with the MOD group. M/M: MET group
   compared with the SFI_M group.

3.4.3 Key metabolic pathway analysis for different metabolites

   For pathway enrichment analysis, the 28 differential metabolites were
   imported into MetaboAnalyst 6.0 to investigate further the mechanisms
   underlying the SFI effects on HF. [179]Figure 9A presents the three
   metabolic pathways that were influenced (p < 0.05): (1) Pentose
   phosphate pathway; (2) Pyrimidine metabolism; and (3) Purine
   metabolism. [180]Supplementary Table S1 contains details on all the
   impacted pathways.

FIGURE 9.

   [181]FIGURE 9
   [182]Open in a new tab

   (A) MetPA analysis of differential metabolites pathway. (B) Venn
   diagram of differential metabolites. (C) Histogram of differential
   metabolites. (D) Venn diagram of enrichment analysis of differential
   metabolites. (E) Histogram of enrichment analysis of differential
   metabolites.

3.4.4 MetOrigin tracing analysis of different metabolites

   Metabolite tracing analysis identified 28 differential metabolites
   associated with the Shenfu injection: seven bacterial-host
   co-metabolites, 11 bacterial metabolites, and one host-specific
   metabolite (including 10 drug-related, 21 food-related, two
   environment-related, and one unknown) ([183]Figures 9B, C). Metabolite
   pathway enrichment analysis (MPEA) showed that 7, 2, and seven relevant
   metabolic pathways matched with host, bacterial, and co-metabolic
   pathway databases, respectively ([184]Figures 9D, E). Based on
   origin-based functional analysis, Taurine and hypotaurine metabolism
   was found to be specific to the microbial community. The biosynthesis
   of Pyrimidine metabolism, Purine metabolism, beta-alanine metabolism,
   Naphthalene degradation, Pantothenate, and CoA biosynthesis were
   identified as co-metabolic pathways between microbes and hosts. To
   better illustrate the co-metabolic relationships between microbiota and
   hosts, a Bio-Sankey network based on MetOrigin analysis further
   visualized the statistical correlations and biological relationships
   between microbiota and metabolites ([185]Figures 10A, B).
   [186]Supplementary Tables S2, S3 contains details of microbiota and
   hosts.

FIGURE 10.

   [187]FIGURE 10
   [188]Open in a new tab

   (A) Sankdy diagram of metorigin analysis of Pyimidine metabolism and
   (B) Purine metabolism, (C) Network diagram of correlation analysis
   (Spearman’s R-value >0.5, p < 0.05).

3.4.5 Correlation analysis between metabolomics and 16S rRNA sequencing
results

   As shown in [189]Figure 10C, correlation analysis revealed a close
   relationship between intestinal microflora and different metabolites
   ([190]Supplementary Table S4). A correlation was considered significant
   only when the absolute value of Spearman’s correlation coefficient r
   was greater than 0.5 ([191]Figure 10). When the total value of R was
   higher than 0.6, the results were considered to show a strong linear
   correlation between the two variables. For example, Lactobacillus was
   strongly correlated with D-Glucosaminic acid and positively correlated
   with EDTA. Ligilactobacillus was substantially inversely correlated to
   Hypoxanthine, Deoxyinosine, Chloroquine, Uracil, and 1-methylguanine,
   but positively correlated to 2-methyl citrate, Phenacetin, Caffeic
   acid, Gentisic acid, Indolelactic acid, 2,4-Dimethylphenol,
   3-(2-Hydroxyphenyl)propanoic acid, 3-(3-Hydroxyphenyl)propanoic acid,
   and 4-Guanidinobutyric acid. Clostridium_sensu_stricto_1 was
   substantially inversely correlated to Hypoxanthine,1-Methylguanine,
   D-Glucosaminic acid, but positively correlated to 2-methylcitrate,
   L-tyrosine-methyl-ester, Phenacetin,2-Hydroxyethanesulfonic acid,
   Caffeic acid, Gentisic acid, 2,4-Dimethylphenol,
   3-(2-Hydroxyphenyl)propanoic acid, 3-(3-Hydroxyphenyl) propanoic acid,
   and 4-Guanidinobutyric acid. These results suggest that the SFI could
   alter the microbe-metabolic axis.

4 Discussion

   The prevalence of heart failure is on the rise globally due to the
   aging population ([192]McDonagh et al., 2021). Emerging research
   indicates that this trend is significantly influenced by the gut-heart
   axis ([193]Madan and Mehra, 2020). The symbiotic cooperation between
   the gut microbiota and the host is vital to maintaining overall health.
   The intricate nature of TCM poses challenges in elucidating its
   therapeutic mechanisms; however, using multi-omics strategies presents
   novel opportunities for a comprehensive understanding of the
   therapeutic mechanism. In this investigation, 16S rRNA sequencing and
   non-targeted metabolomics were employed to examine the underlying
   pathways of SFI treatment. The most important findings from this study
   are as follows ([194]Figure 11): (1) SFI effectively enhanced cardiac
   function in rats with ISO-induced heart failure, (2) SFI effectively
   modulated inflammatory imbalance and reduced serum and urine TMAO
   levels, (3) Treatment with SFI altered the gut microbiota and the
   composition of intestinal metabolites, enriched SCFA-producing bacteria
   such as Ruminococcus and Erysipelotrichaceae at the genus level and
   influenced the metabolites particularly those associated with the
   pentose phosphate pathway, pyrimidine metabolism, and purine
   metabolism, and (4) A close relationship was established between
   intestinal microflora and different metabolites, indicating that the
   SFI could alter the microbe-metabolic co-metabolism axis.

FIGURE 11.

   [195]FIGURE 11
   [196]Open in a new tab

   Shenfu injection improves cardiac function by modulating co-metabolism
   and the TMAO-inflammation axis. (This figure was drawn using Figdraw).

   SFI has been extensively utilized in China for more than 3 decades in
   the treatment of cardiovascular diseases. Its primary constituents
   consist of ginsenosides and aconitum alkaloids. The content of
   ginsenosides is 676–742 μg/mL, and the content of aconite alkaloids is
   3–7 μg/mL ([197]Song et al., 2015). Aconite has certain toxicity, but
   the combination of aconite and ginseng exhibits a “detoxifying and
   enhancing efficacy” effect ([198]Xu et al., 2024). Ginsenosides can
   promote the metabolism of toxic components like aconitine, prolonging
   the elimination half-life of aconitine, benzoylmesaconine, and
   hypaconitine, thereby significantly increasing the exposure of
   effective components in the body. An example of ginseng’s
   detoxification mechanisms towards aconitine involves ginsenoside Rg1’s
   regulation of ion channels pathways ([199]Xu et al., 2022). This
   regulation accelerates aconitine metabolism and enhances the absorption
   of benzoylaconine ([200]Xu et al., 2020).To explore the effects and
   toxicity of the Shenfu Injection at varying concentrations on the
   organism, we administered three different concentrations of Shenfu
   Injection. Our investigation entailed assessing both cardiac function,
   through techniques like echocardiography, Nt-proBNP level measurements,
   and histological analyses of cardiac tissues using H&E staining, as
   well as gut function, involving examination of intestinal tissues via
   H&E staining and measurement of LPS levels. However, dose dependency
   was not obvious in this study. The moderate dose (6 mL/kg) of the
   Shenfu Injection exhibited the most significant improvement in cardiac
   and gut function.

   The metabolite TMAO, produced in a process dependent on gut microbiota,
   is linked to the pathogenic causes of HF, and it may operate as a
   predictive marker for identifying patients at risk of disease
   advancement ([201]Zhang et al., 2021). Results from a systematic review
   indicate that TMAO is related to HF severity, brain natriuretic peptide
   (BNP), estimated glomerular filtration rate (eGFR), and adverse
   outcomes such as hospitalizations and fatalities ([202]Anderson et al.,
   2022). Research conducted in different population groups indicates that
   higher plasma TMAO levels are correlated positively to inflammation,
   impaired endothelial function, type 2 diabetes, central obesity, and
   high blood pressure ([203]Thomas and Fernandez, 2021). Research has
   indicated that TMAO further induces vascular inflammation by activating
   the NLRP3 inflammasome ([204]Chen et al., 2017). Overexpressing SIRT3
   alleviated the activation of the endothelial NLRP3 inflammasome induced
   by TMAO ([205]Wu J. et al., 2019). In our study, we observed increased
   levels of TMAO in the serum and urine of rats with heart failure.
   However, significant downregulation of TMAO levels in both serum and
   urine, along with pro-inflammatory cytokines TNF-α, IL-2, and IL-1β,
   was observed in SFI-treated HF rats, aligning with previous research
   findings ([206]Yang et al., 2019). Spearman correlation analysis
   demonstrated positive associations between inflammatory factors IL-1β
   and IL-2 with serum and urine TMAO, highlighting the pivotal role of
   the TMAO-inflammation axis targeted by SFI.

   The 16S rRNA sequencing results demonstrated a reduction in the
   quantity of OTUs and Alpha diversity changes (including Chao, Shannon,
   Simpson, and ACE), indicating the dysfunction of gut microbiota in HF
   rats. A previous study has reported similar findings ([207]Li et al.,
   2021). At the phylum level, Firmicutes in the intestine primarily
   specialize in the hydrolysis of carbohydrates and proteins.
   Bacteroides, on the other hand, predominantly target steroids,
   polysaccharides, and bile acids, which aid in the absorption of
   polysaccharides and contribute to protein synthesis within the body. In
   our research, when compared to the CON group, the Firmicutes were found
   to increase and the Bacteroidota reduced substantially in the MOD
   group. Shenfu Injection significantly increased the levels of
   Bacteroidota. A research study discovered that infants with heart
   failure had significantly altered intestinal microbiota compared to
   those without heart disease, with an increased number of Firmicutes at
   the phylum level ([208]Zhang et al., 2023). At the genus level,
   Ruminococcus, Erysipelotrichaceae, and unclassified_f_ Muribaculaceae,
   which produce short-chain fatty acids (SCFAs) from fermenting dietary
   polysaccharides, are observed to be decreased ([209]Shaidullov et al.,
   2021; [210]Yamaguchi et al., 2021), which is consistent with our
   results. It is widely recognized that SCFAs play a crucial role in
   regulating inflammation, a process intricately linked to the
   pathophysiology of CHF. SCFAs are also integral in preserving the
   integrity of the gut barrier, modulating immune responses, and serving
   as an energy source for colonocytes. Ruminococcus is a genus within the
   Ruminococcaceae family. Species within the Ruminococcus genus are
   mainly known for designing cellulose and resistant starches, thereby
   contributing to the fermentation process that produces SCFAs
   ([211]Michels et al., 2022). These metabolic activities are vital for
   maintaining a healthy gut environment and the host’s overall metabolic
   health ([212]Zhu et al., 2019). The bacterial levels listed above could
   be restored to varied degrees following SFI treatment. In addition,
   results from a study by LEfse showed that the SCFA producers,
   Ruminococcaceae and Lachnospiraceae, were significantly overrepresented
   in the SFI_M group, these findings reveal that SFI is instrumental to
   SCFA-producing bacteria.

   Gut microbiota can affect an organism by modulating the host metabolic
   products or through its metabolic products. In a metabolomics study by
   [213]Li et al. (2023), metabolic imbalances such as disturbances in the
   metabolism of amino acids, lipids, and glucose, were established to be
   associated with heart failure ([214]Li et al., 2023). LC-MS analysis
   was conducted to investigate the influence of gut microbiota on the
   host metabolism of the ISO-induced HF rat model. In our research
   involving metabolites, 48 differential metabolites were detected in the
   HF rat model, and all the metabolites showed a callback trend, with 28
   significantly changed metabolites. These findings indicate that the SFI
   has a regulating effect on the metabolic disturbances induced by HF.
   Untargeted metabolomics have become a powerful tool for evaluating the
   efficacy of TCM. As a cutting-edge technological system in biology,
   metabolomics embraces a holistic perspective akin to TCM, enabling
   in-depth exploration of TCM’s complex conditions and multifactorial
   nature. Tian et al. used untargeted metabolomics to investigate the
   mechanism of action of QFPDD in treating coronavirus-induced pneumonia
   in mice ([215]Tian et al., 2022). UPLC-Q-TOF/MS is also used for TCM
   syndrome differentiation ([216]Ye et al., 2021).

   The results from our KEGG pathway analysis revealed that the regulation
   of metabolites by SFI focuses on the pentose phosphate pathway,
   pyrimidine metabolism, and purine metabolism. Glycometabolism,
   recognized for its critical function in energy production, encompasses
   three primary metabolic pathways: oxidative phosphorylation,
   glycolysis, and the pentose phosphate pathway. It is vital to maintain
   cardiovascular health to ensure the efficient progression of
   glycometabolism ([217]Peng et al., 2023). D-ribose is a pentose sugar
   present in every living cell and is a critical component of several
   crucial biomolecules such as RNA, nucleotides, and riboflavin. It is
   synthesized from glucose via the pentose phosphate pathway (PPP) within
   the cell. Through the non-oxidative phase of the PPP, d-ribose plays an
   essential role in producing adenosine triphosphate (ATP). As a result,
   it is utilized as a therapeutic agent to improve cardiac function in
   patients with HF. Research shows that Supplemental d-ribose reduces the
   symptoms of HFpEF and increases EF ([218]Pierce et al., 2022). Our
   study indicates that SFI can increase D ribose and Sedoheptulose
   levels. Altogether, these data provide the basis for glycometabolism as
   an essential target of SFI. Growing evidence suggests that the levels
   of intermediates from purine degradation reflect the energy status of
   myocardial cells. Under normal conditions, an increase in the heart’s
   energy demands a need for increased levels of purine nucleotides and
   their by-products. In contrast, a considerable decrease in total purine
   production suggests that myocardial cells are conserving energy, hence
   maintaining the myocardium’s energy balance. In pathological states,
   such as myocardial ischemia, ATP breaks down into xanthine and
   accumulates in the tissue. The produced xanthine is converted into uric
   acid by xanthine oxidase, which generates a significant amount of
   superoxide anions, causing cellular damage. Under low oxygen
   environments, the byproducts of the breakdown of adenosine and inosine
   serve as more efficient energy sources than extracellular glucose,
   thereby slowing the build-up of nicotinamide adenine dinucleotide
   (NADH) and offering a degree of protection to the cells ([219]Wang G.
   et al., 2022). In the present study, hypoxanthine and deoxyinosine
   levels were significantly increased in the HF rat model, indicating the
   presence of an energy imbalance in the HF rat model, which is
   consistent with research findings ([220]Berry and Hare, 2004).

   A significant crosstalk exists between the gut microbiota and the host
   through substrate co-metabolism and metabolic exchange. Therefore, it
   may be possible to infer host and microbiome co-metabolism through
   untargeted metabolomics, as [221]Zhao et al. (2022) showed that
   interactions between fecal metabolomes and gut microbiomes can reveal
   functional markers in cerebral ischemic stroke ([222]Zhao et al.,
   2022). Chen employed fecal metabonomics in conjunction with 16S rRNA
   gene sequencing to examine alterations in gut microbiota in rats
   exhibiting kidney-yang deficiency syndrome and the impact of the
   You-gui pill intervention. Tong’s study found that Gushudan could
   protect rats with kidney-yang deficient disease by modulating the
   gut-kidney axis ([223]Tong et al., 2022). MetOrigin is a bioinformatics
   tool designed to pinpoint the specific bacterial species involved in
   various metabolic reactions and elucidate their roles in these
   processes ([224]Fan et al., 2023). In our study, metabolite tracing
   analysis showed that Taurine and hypotaurine metabolism was found to be
   specific to the microbial community. The biosynthesis of Pyrimidine
   metabolism, Purine metabolism, beta-alanine metabolism, Naphthalene
   degradation, Pantothenate, and CoA biosynthesis were identified as
   co-metabolic pathways between microbes and hosts. The Spearman
   correlation analysis in our study also showed a significant correlation
   between differentially expressed metabolites regulated by SFI and the
   gut microbiota, indicating that SFI could modulate the co-metabolism of
   ISO-induced HF.

   Our research may give some insights into the relationship between gut
   microbiota and heart failure. However, the validation of this relevant
   hypothesis based on animal model testing faces several challenges.
   Specifically, we need to determine whether the improvement in dysbiosis
   and metabolome abnormality is merely a consequence of heart failure
   amelioration or if it plays a critical role in heart failure
   progression. Additionally, distinguishing changes in gut microbiota
   induced by heart failure from those induced by specific interventions
   like ISO is essential. To address these challenges, future studies
   should consider conducting fecal transplantation trials and expanding
   both sample sizes and clinical data collection. These steps will
   enhance our understanding and validation of the relationship between
   gut microbiota and heart failure.

5 Conclusion

   This study utilized a comprehensive methodology, incorporating various
   techniques such as echocardiography, protein chip detection,
   histopathology, 16S rDNA sequencing, and metabolomics, to investigate
   the effectiveness of SFI in treating heart failure. The comprehensive
   findings from our study demonstrated significant benefits of SFI
   treatment in rats with isoproterenol-induced heart failure. SFI
   effectively modulated inflammatory imbalance, reduced serum and urine
   TMAO levels, and improved gut microbiota composition. Specifically, SFI
   substantially increased the abundance of Bacteroidota at the phylum
   level and enriched SCFA-producing bacteria such as Ruminococcus and
   Erysipelotrichaceae at the genus level. Furthermore, SFI influenced the
   composition of intestinal metabolites, particularly those associated
   with the pentose phosphate pathway, pyrimidine metabolism, and purine
   metabolism. Our findings highlight SFI’s ability to address dysbiosis
   within the gut microbiota-host co-metabolism.

Acknowledgments