Crystal engineering optimizes emodin-tetramethylpyrazine combination: From cocrystal design to in vivo anti-colitis efficacy assessment

Abstract

Emodin (EMO) shows therapeutic promise for ulcerative colitis (UC), yet its clinical utility is hampered by low bioavailability. To rationally overcome this limitation, this study employed cocrystal engineering, strategically selecting tetramethylpyrazine (TMP)—a natural compound from traditional Chinese medicine—as the cocrystal coformer (CCF). The selection of TMP was guided by a systematic CCF screening strategy, incorporating extensive literature analysis of natural compound CCF candidates, computational chemistry methods to predict favorable hydrogen-bonding interactions and interaction sites with EMO, and machine learning assessment of cocrystallization propensity. Utilizing this rational design approach, we successfully synthesized and characterized a novel EMO-TMP cocrystal through comprehensive solid-state characterization techniques. The resulting cocrystal significantly enhanced the aqueous solubility of EMO while preserving its intrinsic bioactivity. Pharmacokinetic studies confirmed that the cocrystal formulation markedly improved the oral bioavailability of EMO. In a dextran sulfate sodium (DSS)-induced ulcerative colitis (UC) model, the EMO-TMP cocrystal demonstrated superior efficacy compared to EMO alone, effectively alleviating colitis symptoms and associated pathological markers. This enhanced in vivo efficacy is attributed to the significantly improved systemic exposure achieved through the rationally designed cocrystal. Our findings establish the EMO-TMP cocrystal as a highly promising strategy to surmount the physicochemical barriers of EMO, unlocking its full clinical potential for UC treatment. Critically, this work not only validates TMP as an efficient and safe CCF specifically suited for active pharmaceutical ingredients (APIs) rich in hydrogen-bond donors, but also exemplifies the value of leveraging formulation principles and compatible components inherent in traditional Chinese medicine through advanced crystal engineering approaches.

Graphical abstract

1. Introduction

Natural products are pivotal in drug discovery (Chopra and Dhingra, 2021), yet their clinical translation is often hindered by poor aqueous solubility and low bioavailability (Jia et al., 2024; Pradhan et al., 2021; Pi et al., 2019; Kim et al., 2018; Lestari et al., 2023), particularly in key phytochemical classes like flavonoids, anthraquinones, and alkaloids (Peng et al., 2023). Cocrystal engineering has proven to be a powerful, rational design-based strategy that optimizes the key physicochemical properties and pharmacokinetic characteristics of the active pharmaceutical ingredient (API) by carefully selecting cocrystal formers (CCFs), without altering the covalent structure of the API (Roshni and Karthick, 2025; Guo et al., 2021; Shaikh et al., 2018).

Ulcerative colitis (UC) is typically characterized by ulceration in the lumen or lining of the colon, affecting the proximal colon and rectum (Sutaria and Mok, 2023; Le Berre et al., 2023). Emodin (EMO, Fig. 1a), a bioactive anthraquinone compound, exhibits a wide range of pharmacological activities, including potent anti-tumor, hepatoprotective, anti-inflammatory, antioxidant, cardioprotective, and antimicrobial effects (Dong et al., 2016; Zheng et al., 2021; Wang et al., 2023c; Li et al., 2020). EMO represents a promising therapeutic agent for UC (Luo et al., 2018; Luo et al., 2022), demonstrating dual mechanisms of anti-inflammatory action and mucosal repair (Pu et al., 2021), However, emodin, as a planar and highly conjugated anthraquinone derivative, has poor water solubility, resulting in low oral bioavailability, which significantly limits its clinical translation (Xu et al., 2023; Wang et al., 2020). Rational cocrystal design targeting a suitable CCF thus aims to unlock EMO's therapeutic potential against UC and associated cardiovascular risks (Robertha et al., 2018).

Fig. 1.

Structural analysis. (a)The structures of EMO and TMP. (b)The results of MEPS calculations on the cocrystal of EMO and TMP. (c) Crystal structure of the EMO-TMP cocrystal. (d) π-π stacking of EMO-EMO dimer. (e) Energy frameworks of EMO-TMP cocrystal.

The successful formation of a cocrystal is by no means accidental. It relies on a deep understanding and precise design of intermolecular interactions (especially hydrogen bonding), spatial complementarity, and thermodynamic stability. Traditional experimental trial-and-error methods are labor-intensive and resource-wasting. In recent years, computational chemistry and machine learning technologies have continuously provided strong support for the prediction of cocrystal formation. Based on the rational design principles, we screened a natural compound CCF candidate library and identified tetramethylpyrazine (TMP, Fig. 1a) as a CCF. Primarily, the molecular structure of TMP provides efficient and complementary hydrogen bond acceptor sites (especially the nitrogen atom on its pyrazine ring) (Xie et al., 2024), which lays a solid molecular design foundation for constructing stable and predictable cocrystal structures with APIs containing specific functional groups (such as phenolic hydroxyl and carboxyl) like EMO. Importantly, TMP has been successfully applied in previous studies to construct cocrystals with various APIs (Li et al., 2021; Baraldi et al., 2024; Wang et al., 2022), fully demonstrating its powerful and reliable design performance as a CCF. The cocrystals formed with TMP typically exhibit improved physicochemical properties, validating the effectiveness of its design. Additionally, TMP offers excellent druggability design assurance, as it is a widely used cardiovascular drug (such as TMP injection), with a clear safety profile and favorable pharmacokinetic properties, providing critical safety design assurance for the final druggability of the cocrystal. A deeper design consideration for selecting TMP lies in its significant pharmacological synergy with EMO in in treating cardiovascular complications caused by UC. This therapeutic synergy by design is expected to achieve a “1 + 1 > 2” therapeutic effect, going beyond the mere improvement of physicochemical properties. The long-standing history and well-established efficacy of the Rheum-chuanxiong combination in traditional Chinese medicine (TCM) (Song et al., 2025b) provide unique inspiration and intrinsic logical support for selecting the EMO-TMP combination. This heuristic design based on traditional wisdom offers a valuable starting point and validation for modern cocrystal engineering. The potential for cocrystal formation was then predicted using computational chemistry methods and machine learning models.

This study focuses on the active design of the EMO-TMP cocrystal to systematically address EMO's limitations. Beyond this specific pair, it aims to validate TMP's broader potential as a versatile CCF. TMP's demonstrated efficient hydrogen bonding complementarity, excellent druggability, and proven cocrystal formation ability make it a promising candidate for a general CCF platform, applicable to many APIs containing specific functional groups. This study will provide detailed reports on the rational design, preparation, structural characterization, physicochemical property evaluation (such as solubility, dissolution, stability, etc.), and preliminary in vitro pharmacological evaluation of the EMO-TMP cocrystal. It will emphasize the performance improvements brought about by the cocrystal design and evaluate its potential as a novel therapeutic agent for UC. At the same time, this study will provide important theoretical and experimental support for expanding the application of TMP as a general CCF.

2. Materials and methods

2.1. Materials

Emodin (purity >98 %) and Tetramethylpyrazine (purity >98 %) were purchased from Hubei Wande Chemical Co., Ltd. Analytical grade solvents employed for crystallization experiments were sourced exclusively from Sinopharm Chemical Reagent Beijing Co., Ltd. (Beijing, China).

2.2. Cocrystal syntheses

Emodin (540 mg) and tetramethylpyrazine (136 mg) were subjected to liquid-assisted grinding with methanol (2 mL) at 20 °C ± 5 °C for 10 min until complete solvent evaporation. Subsequently, sequential grinding was performed with two additional 2 mL aliquots of methanol (total 4 mL) over 30 min until solvent removal. The resultant solid was air-dried under a fume hood for 30 min prior to further characterization. Weigh precisely 57.6 mg of emodin and 13.6 mg of tetramethylpyrazine, and add them to a vial. Add 6 mL of isopropanol solution, heat at 50 °C for 10 min, and wait until the sample completely dissolves. After cooling, filter the solution and leave it to stand at room temperature for 7 days to allow the solution to evaporate naturally, resulting in yellow, transparent, sheet-like crystals.

2.3. Theoretical calculation

Geometrical optimizations of all hydrogen atoms and single-energy calculations of cocrystal were calculated at the level of B3LYP (GD3BJ) /6-31G* and B3LYP (GD3BJ) /6-311G (2d, 2p) using Gaussian package (Frisch et al., 2016), respectively. The interaction energies between API and CCF were calculated at the same computational level, employing the counterpoise correction method. The Multiwfn 3.8 program was used for all wave function analyses (Lu, 2024; Lu and Chen, 2012). Hirshfeld surface and energy frameworks were calculated using CrystalExplorer 25.03 software (Spackman et al., 2021).

2.4. Characterization

2.4.1. Single crystal X-ray diffraction (SCXRD) analysis

Single crystal X-ray diffraction data was acquired on a Rigaku XtaLAB Synergy four-circle diffractometer equipped with Cu-Kα radiation (λ = 1.54178 Å; Rigaku, The Woodlands, TX, USA). All intensity measurements were conducted at 293 K with empirical absorption correction applied using the CrystalClear program package (Rigaku). Structural determination was carried out through direct methods, followed by refinement procedures using SHELXL (Sheldrick, 2015) and Olex2(Dolomanov et al., 2009) soft-ware packages. The final structural models were optimized by full-matrix least-squares refinement. Non‑hydrogen atoms were subjected to anisotropic displacement parameter refinement, while hydrogen atoms bonded to carbon, nitrogen, and oxygen atoms were geometrically positioned using standard riding models. The crystallographic data for EMO-TMP cocrystal has been deposited in the Cambridge Crystallographic Data Centre as supplementary publications under the CCDC numbers 2,390,219. This information can be obtained free of charge from the Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

2.4.2. Powder X-ray diffraction (PXRD) analysis

Experiments were carried out in a Rigaku D/max-2550 diffractometer (Rigaku Corpo-ration, Tokyo, Japan) operating with Cu-Kα radiation (λ = 1.54178 Å) at 40 kV and 150 mA. The powder samples were continuously scanned over a range from a starting angle of 3° to a final angle of 40°, with a scan rate of 8°/min. Simulated PXRD patterns were calculated using Mercury software (version 2025.2.2, Cambridge Crystallographic Data Center, UK) with a step size of 0.02° and a full width at half maximum of 0.15°.

2.4.3. Attenuated total reflectance fourier transform infrared spectroscopy (ATR-FTIR)

IR spectra were measured by a Spectrum 400 Fourier transform infrared spectrometer (PerkinElmer, Waltham, MA, USA). with a spectral scanning range of 4000–550 cm⁻¹, a resolution of 4.000 cm⁻¹, and a scan number of 16.

2.4.4. Raman spectroscopy analysis

Raman spectra were acquired on a grating-type DXR Raman microscope (Thermo Fisher, USA). The laser wavelength was 780 nm with a power of 24 mW, the grating had 400 lines/mm, and the aperture was a 50 μm slit. The exposure time was 1.00 s, with 16 exposures and 16 background exposures.

2.4.5. Solid-state nuclear magnetic resonance (ssNMR) analysis

Solid-state nuclear magnetic resonance (ssNMR) analysis was performed on a Bruker Avance 600WB spectrometer equipped with a 5 mm cryogenic probe. ¹³C experiments were conducted at an operational frequency of 150.9 MHz using a 3.2 mm rotor. Cross polarization (CP) was employed with a sampling interval of 11.000 μs, and 1000 scans (ns) were accumulated for all samples.

2.4.6. Differential scanning calorimetry (DSC) analysis

The DSC thermograms were performed on a DSC 1 (Mettler Toledo, Greifensee, Switzerland) at a constant heating rate of 5 °C/min, 10 °C/min and 15 °C/min across a temperature range of 30–300 °C under air atmosphere. Samples were placed in an aluminum crucible and sealed with a lid featuring a hole. The instrument was calibrated using indium and zinc certified reference materials prior to the start of the experiment. Data analysis was performed using STARe Evaluation software version 16.0.

2.4.7. Thermogravimetric analysis (TGA)

A TGA/DSC1 instrument (Mettler Toledo, Switzerland) was used, with a N₂ atmosphere (50 mL/min). Sample with masses between 5 and 8 mg were used in aluminum oxide crucibles, and the temperature was scanned from 30 to 500 °C with a heating rate of 5 °C/min.

2.5. Dynamic vapor sorption (DVS) experiment

The hygroscopicity of EMO-TMP was studied using a dynamic vapor sorption instrument (DVS Adventure, Surface Measurement Systems, UK). The experiment was conducted at 25 °C, with relative humidity (RH) increasing stepwise by 10 % RH, from 0 % to 90 %, and then decreasing back to 0 %. The switching conditions for humidity steps were set as follows: a change in sample weight of less than 0.02 % within 10 min; if this condition was not met, the sample was maintained at the current humidity level for no longer than 120 min.

2.6. Particle size distribution experiment

The particle size distribution was determined using a laser diffraction particle size analyzer. Water was selected as the dispersing medium for the sample to form a stable suspension. Experimental data were recorded under a flow rate of 60 %, with the fluid refractive index at 1.333, assuming a spherical particle shape and a particle refractive index of 1.59. The refractive index of the measured particles is 1.59 under these conditions.

2.7. Solid phase stability evaluation

The EMO-TMP cocrystal was placed in an open clean glass dish and exposed to high temperature (set at 60 °C ± 2 °C), high humidity (relative humidity 90 % ± 5 %), and light (light intensity 4500 lx ± 500 lx) conditions. Samples of 50 mg were taken on 0, 5, and 10 days. The solid-state properties at different time points were analyzed using powder X-ray diffraction.

2.8. Solubility experiments

The solubility studies were carried out by the shake-flask method. Saturated solutions were obtained by stirring an excess of EMO and EMO-TMP in 5 mL of buffer at pH 1.2, pH 4.5, and pH 6.8 and water at 37 °C. After 72 h, the sample solutions were centrifuged, and the supernatant were collected, then directly measured using high-performance liquid chromatography (HPLC). In the in vitro powder dissolution experiment, we selected two buffer media with pH values of 6.8 and 7.0 for the dissolution tests. The experiment employed an RC12AD dissolution meter (Tianjin TIANDA TIANFA Pharmaceutical Testing Instrument Manufacturer, Tianjin, China), equipped with an automatic sampling system RZQ-12D, to study the dissolution of EMO and EMO-TMP. We weighed 60 mg of EMO and an equivalent amount of EMO-TMP, and placed them into dissolution tanks containing 900 mL of dissolution medium. The entire experimental process was controlled at a temperature of 37 °C and a propeller speed of 100 rpm/min. At 10, 30, 45, 60, 90, 120, 240, 480, and 720 min, 1 mL of sample was taken by the automatic sampling system. The obtained samples were centrifuged at 13,400 rpm for 10 min, and the supernatant was extracted for analysis using HPLC. The experiments were repeated three times. A modified method based on the Chinese Pharmacopoeia (Commission, C.P, 2020) was utilized for the analysis. The separation was achieved using an Agilent 1200 HPLC system under the optimized conditions. Chromatographic analysis was achieved using a mobile phase consisting of 0.1 % phosphoric acid in water and methanol (15:85), delivered at a flow rate of 1 mL/min. An Odyssil-C18 column (4.6 × 250 mm, 5 μm) was employed, with the column temperature maintained at 30 °C and the detection wavelength set at 214 nm.

2.9. In vivo pharmacokinetic study

Twelve of male Sprague Dawley rats (250 ± 20 g) were purchased by SiPeiFu (Beijing) Biotechnology Co., Ltd. All animal procedures complied strictly with NIH Guidelines for Laboratory Ani-mal Care and Use, with protocols approved by the Animal Ethics Committee of the Institute of Materials Medical, CAMS, and PUMC (Approval No.: 00008390). Following a one-week acclimation period under standard conditions with free access to food and water, rats were randomly divided into two groups (n = 6): one group received EMO (100 mg/kg), and the other received EMO-TMP cocrystal at a dose equimolar to 100 mg/kg of EMO. The dosage of EMO was determined as an intermediate level with reference to previous pharmacokinetic studies using 80 mg/kg (Wang et al., 2023a) and 150 mg/kg (Zhang et al., 2024). Serial blood sampling (400 μL) via the retro-orbital plexus occurred at 0, 0.083, 0.25, 0.5, 0.75, 1, 2, 4, 6, 8, 10, 12, and 24 h post-dose. Samples stood for 30 min prior to centrifugation (4000 rpm, 10 min, 4 °C), with plasma stored at −80 °C until analysis. For sample preparation: 100 μL thawed plasma was mixed with 20 μL internal standard (chrysophanol, 500 ng/mL), 300 μL acetonitrile, and 20 μL methanol in 1.5 mL Eppendorf tubes. After 3 min vortexing and centrifugation (13,000 rpm, 10 min), 100 μL supernatant was combined with 100 μL deionized water, vortexed (1 min), and recentrifuged (13,000 rpm, 3 min). The final supernatant (3 μL) underwent HPLC-MS/MS analysis. Chromatographic separation employed a Shimadzu HPLC system coupled to an API 6500+ triple quadrupole mass spectrometer (AB Sciex) using a Synergi™ C18 column (50 × 2.0 mm, 4 μm) at 40 °C. The mobile phase consisted of 0.1 % aqueous formic acid (A) and acetonitrile (B) (Di et al., 2015) at 0.3 mL/min with gradient: 0–0.5 min, 20 % B; 0.5–1.5 min, 20–75 % B; 1.5–4.0 min, 75 % B;4.0–4.5 min, 75–20 % B; 4.5–6.0 min, 20 % B. Chrysophanol was used as the internal standard (IS). Negative-ion MRM monitored m/z 269.0/225.1 (EM) and 253.0/224.9 (IS). Optimized parameters included: declustering potential (DP) −142 V (EM) / −120 V (IS), collision energy (CE) −37 V, curtain gas (CUR) 40 psi, Ion Source Gas 1 (GS1)/ Ion Source Gas 2 (GS2) 50 psi, ion source temperature (TEM) 550 °C, internal standard voltage −4500 V, and collision gas (CAD) 19. Plasma EMO concentrations were quantified using DAS 2.0 for pharmacokinetic profiling. Data are presented as mean ± SD.

2.10. Pharmacological evaluation

2.10.1. Animals and treatment

Male C576/J mice (18–22 g) were purchased from SiPeiFu (Beijing) Biotechnology Co., Ltd. All animal procedures were performed in accordance with the principles of the NIH Guide for the Care and Use of Laboratory Animals, and the protocols were approved by the Institutional Animal Care and Use Committee of the Institute of Materials Medical, CAMS, and PUMC (Permit number: 00008239). The animals were allowed to acclimatize for 3 days in the facility, maintained at a temperature of 22 ± 3 °C and 45 ± 10 % humidity, following a regular 12 h/12 h light/dark cycle with access to food and water.

Sixty mice were randomly divided into the following six groups: control group, model group, mesalazine group, emodin group (EMO), tetramethylpyrazine group (TMP), and EMO-TMP cocrystal group. Except for the control group, the mice were allowed to freely drink 3 % DSS (MP Biomedicals, Co., Ltd., USA) solution (W/V) for 7 days to induce ulcerative colitis model using a previously described method with minor adjustments (Yang and Yang, 2024). At the same time, mesalazine (100 mg/kg), EMO (76 mg/kg), TMP (34 mg/kg) and EMO-TMP (100 mg/kg) were dissolved in 0.5 % sodium carboxymethylcellulose (CMC-Na) and given by oral administration for 7 days, while the control group and model group were given 0.5 % CMC-Na. The dosage of mesalazine referred to previous literatures (Zhang et al., 2025; Ren et al., 2023). In order to compare the efficacy of EMO-TMP and mesalazine, the dosage of EMO-TMP was also 100 mg/kg. The doses of EMO (76 mg/kg) and TMP (34 mg/kg) were set based on molecular weight calculations and our laboratory's preliminary experiments, which demonstrated good efficacy without significant toxicity in the DSS-induced colitis model.

2.10.2. Disease activity index evaluation

During the experiment, the body weight, feces shape and rectal bleeding were observed daily to determine the occurrence of UC. The disease activity index (DAI) score was used to evaluate the overall disease severity (Kwon et al., 2021). DAI = (Weight loss score + Stool consistency score + Hematochezia score) / 3. Normal stools are dry and granular, semi-formed stools are paste-like but do not stick to the anus, and liquid stools are watery and stick to the anus. Occult blood in stools was detected using the Sirlami method (Yang et al., 2023b). The weight loss score is divided into 5 levels: 0 points: No weight loss, 1 point: Weight loss 1 %–5 %, 2 points: Weight loss 6 %–10 %, 3 points: Weight loss 11 %–15 %, 4 points: Weight loss >15 %. Stool consistency: Normal stool is scored as 0, semi-formed stool is scored as 2, and liquid stool is scored as 4. Hematochezia: Negative for occult blood is scored as 0, positive for occult blood is scored as 2, and visible blood in stool is scored as 4.

2.10.3. Colon length and inflammatory factors detection

At the end of experiment, colon tissue was gathered to measure colon length. Blood was collected from the abdominal vein and centrifuged at 5000 rpm for 10 min to obtain the serum. The levels of inflammatory factors (TNF-α and IL-1β) in serum were detected according to the instructions of the ELISA kits (Jianglai Biotechnology Co. Ltd.).

2.10.4. Histopathological analysis

Hematoxylin and eosin (H&E) staining was used for the histopathological observation of colonic tissues. Colonic tissues were fixed in 4 % paraformaldehyde and then embedded in paraffin. The same position of colonic tissues was cut into 5 mm thick sections and stained with H&E. The slices were evaluated under a light microscope for determination of pathological changes.

2.11. Statistical analysis

Statistical analysis was performed using GraphPad Prism 9.0 (GraphPad Software, San Diego, CA, USA). Data were presented as the mean ± SEM. Statistical comparisons were performed using student's t-test or one-way ANOVA. Statistical significance was defined as P < 0.05.

3. Results and discussions

This study focuses on the interactions between EMO and TMP molecules and their potential for forming cocrystals. Through molecular structure analysis of EMO and TMP, we found that the EMO molecule contains multiple hydrogen bond donors (three phenolic -OH groups) and acceptors (carbonyl oxygen atoms). In contrast, the TMP molecule has a symmetric, electron-deficient pyrazine ring structure with two nitrogen atoms positioned close to each other, making it an effective hydrogen bond acceptor. These structural features provide favorable conditions for the formation of a stable cocrystal structure between EMO and TMP molecules via hydrogen bonding.

3.1. Cocrystal formation prediction

Through molecular electrostatic potential surface (MEPS) (Wang et al., 2023b; Lu and Chen, 2012; Lu, 2024) analysis, we quantified the pairing energy differences at the interaction sites of EMO and TMP. The results showed that when the stoichiometric ratio of the two molecules is 1:1, the cocrystal formation energy difference ΔE is −1.62 kJ/mol, indicating that the thermodynamic driving force for cocrystal formation at this ratio is weak. The MEPS map (Fig. 1b) further revealed that the O₂ site of EMO exhibits the global maximum electrostatic potential (+57.94 kcal/mol), while the N_1T site of TMP shows the global minimum (−33.34 kcal/mol), suggesting that the two may form a directional hydrogen bond via O₂-H…N_1T. However, our earlier data-driven XGBoost-based machine learning approach demonstrated the promising potential for these two components to cocrystallize (Yang et al., 2023a).

3.2. Structural analysis

3.2.1. SXRD analysis

We successfully obtained EMO-TMP cocrystal single crystals via solution crystallization. Single crystal X-ray diffraction (SXRD) analysis showed that the EMO-TMP cocrystal crystallizes in the space group $P\overline{1}$, with each unit cell containing two chemical formula units (Z = 2). The crystal of the EMO-TMP cocrystal has a complex hydrogen bonding network. Two EMO molecules form a trimer structure with one TMP molecule via O₂-H₂…N_1T hydrogen bonds, generating a $D_2^2\left(6\right)$ motif. Additionally, adjacent EMO molecules form O₃-H₃…O₄ hydrogen bonds through 8-hydroxy and 9‑carbonyl groups, constructing a $R_2^2\left(16\right)$ dimer motif. Hydrogen bonds within the EMO molecule (O₁-H₁…O₄; O₃-H₃…O₄) also exist, and the EMO molecules are linked to the TMP molecules, maintaining stable three-dimensional molecular arrangements through hydrogen bonding and van der Waals forces. Additionally, the crystallographic information of the EMO-TMP cocrystal structure is listed in Table 1.

Table 1.

Crystal cell parameters and structure refinement of the EMO-TMP cocrystal.

Parameters	EMO-TMP cocrystal
Empirical formula	C15H10O5·0.5(C8H12N2)
Crystal size (mm)	0.05 × 0.20 × 0.20
Description	plate
Crystal system	Triclinic
Space group	$P\overline{1}$
a (Å)	5.898(2)
b (Å)	8.777(1)
c (Å)	15.366(1)
α (°)	82.073(3)
β (°)	88.481(3)
γ (°)	82.702(3)
Z	2
Volume (Å3)	781.41(4)
density (g·cm−3)	1.438
Goodness-of-fit on F2	1.072
R1 (I > 2σ(I))	0.0531
wR2 (I > 3σ(I))	0.1608
Reflections collected	2485
independent reflections	3133
Completeness	0.993
CCDC deposition no.	2,390,219

To assess the importance of each interaction, we calculated the interaction energy between each pair of molecules in the crystal. The calculation results showed that the prominent interactions in the one-dimensional chain structure of EMO and TMP molecules are stabilized by conventional hydrogen bonds, with an interaction energy of −13.70 kcal/mol. In contrast, the EMO-EMO interactions (marked in orange) exhibit non-classical hydrogen bonding with significantly weaker energy (−4.34 kcal/mol). This difference is further confirmed by geometric parameters: the O-H⋯O bond distance in the EMO dimer is 2.904 Å, and the bond angle is not ideal (∠O-H⋯O = 117.18°), whereas the EMO-TMP heterodimer exhibits a shorter contact distance (O-H⋯N = 2.782 Å) and a nearly linear geometry (∠O-H⋯N = 158.21°). Although the face-to-face distance between the two EMO dimers (Fig. 1d) is 5.308 Å, exceeding 4 Å, and the contact angle is 55.342°, resulting in minimal overlap of π orbitals, their interaction energy is second only to the contribution of the hydrogen bonding between EMO and TMP molecules in stabilizing the crystal. Further calculations using CrystalExplorer 25.03 software, including curvature and shape index analysis (Fig. 1d), showed that in this atypical π-π stacking, the molecular structure might undergo local optimization or be influenced by other factors, but still exhibits typical π-π stacking characteristics. In summary, EMO and TMP molecules form a cocrystal structure through a complex network of hydrogen bonds and π-π stacking interactions.

3.2.2. Cocrystal energy framework

To further understand the stability of the EMO-TMP cocrystal structure, we used CrystalExplorer 25.03 software (Spackman et al., 2021) to calculate the cocrystal energy framework, as shown in Fig. 1e. The tube size is 10.00, with a cutoff value of 10.00 kJ/mol. Under these conditions, there is no polarization energy, so it is not displayed in the figure. In the framework, red represents coulomb energy, green represents dispersion energy, red-brown represents exchange energy, and yellow cylinders represent repulsive interactions, which weaken the stability of the crystal structure and reveal the sources of repulsive forces that contribute to the instability of the crystal. Blue represents total energy. The radius of the cylinders is proportional to the magnitude of the interaction energy. The calculations show that Coulomb energy is the dominant energy form in the interaction between EMO and TMP, while dispersion energy is the dominant energy form in the EMO dimer. The spatial distribution of the yellow cylinders reveals the local instability regions present within the crystal.

Systematic multi-dimensional solid-state characterization was performed to confirm the crystalline purity and physicochemical properties of bulk EMO-TMP cocrystal powder, essential for pharmaceutical developability assessment. PXRD serves as a powerful technique for cocrystal screening, leveraging the unique diffraction pattern exhibited by each solid-phase component—including the API and CCF—to rapidly identify the formation of a new chemical entity (Pantwalawalkar et al., 2025). Comparative analysis of PXRD patterns (Fig. 2a) revealed distinct phase transformation in the EMO-TMP cocrystal. The cocrystal diffractogram displayed emergent characteristic peaks concurrent with disappearance of key diffraction signals from the individual API and CCF, confirming formation of a novel crystalline phase. Furthermore, close correspondence between experimental and simulated PXRD patterns—derived from single-crystal structure data—in peak positions (2θ) and relative intensities (I/I₀) validated high phase purity of the synthesized cocrystal material.

Fig. 2.

Characterization of the powder sample. (a) PXRD patterns of EMO, TMP, prepared EMO-TMP, and simulated EMO-TMP from SCXRD data. (b) IR spectra of EMO, TMP, and EMO-TMP. (c) Raman spectra of EMO, TMP, and EMO-TMP. (d) ssNMR spectra of EMO, TMP, and EMO-TMP. (e) DSC thermograms of EMO, TMP, and EMO-TMP. (f) TG curve of EMO-TMP. (g) DVS analysis of EMO-TMP. (h) Particle size distribution of EMO-TMP. (i) PXRD patterns of EMO-TMP under accelerated stability testing. (j) Activation energy for the loss of TMP evaluated using the DSC Kissinger method. (k) Colorimetric analysis of EMO, EMO-TMP, and EMO-PIP.

3.3. FTIR spectroscopy analysis

FTIR spectroscopy has been widely used in the study of pharmaceutical solids and polymorphism (Chakraborty et al., 2014), with infrared bands related to the bond strength of different groups in the molecules. IR peaks are typically assigned by comparing the spectra of multicomponent crystals with those of single-component crystals, and the formation of cocrystals is often accompanied by frequency shifts (Iwata et al., 2017). Significant spectral differences were observed in the IR spectrum (Fig. 2b) of the EMO-TMP cocrystal compared to its individual components, particularly in the functional group regions: ν_O-H (3600–2800 cm⁻¹), ν_N-H (3400–3100 cm⁻¹), and fingerprint region (1300–400 cm⁻¹). These alterations, primarily attributable to hydrogen bond formation, confirm the generation of a new solid phase and provide additional evidence for successful cocrystallization.

3.4. Raman spectroscopy analysis

Raman spectroscopy is a scattering technique where changes in molecular polarizability induce Raman signals. Vibrational spectra form unique fingerprint patterns based on characteristic stretching and bending vibrations of chemical bonds within molecules. This serves as an auxiliary method to confirm cocrystal formation by identifying alterations in specific functional groups within complexes (Yaghoobi et al., 2021). By comparing EMO-TMP with EMO and TMP, we observed that the characteristic Raman peaks of both EMO and TMP exhibit varying degrees of attenuation and frequency shifts (Fig. 2c). The C-H stretching vibration signal of methyl groups (–CH₃) from TMP (3000–2800 cm⁻¹) became undetectable in EMO-TMP. The Raman characteristic peak for the out-of-phase stretching vibration of the quinone double bond in EMO with adjacent hydroxyl groups shifted from 466.14 cm⁻¹ to 465.69 cm⁻¹. The Raman characteristic peak for half of the central ring stretching vibration in EMO shifted from 940.84 cm⁻¹ to 938.13 cm⁻¹. The three Raman characteristic peaks in the 1556–1610 cm⁻¹ range for in-plane vibrations of the ring showed changes in both shape and number, while the strong stretch of the C8 hydroxy oxygen shifted from 1284.32 cm⁻¹ to 1277.10 cm⁻¹ (Simeral and Hafner, 2022).

3.5. ssNMR analysis

ssNMR is highly sensitive to the molecular crystal packing effects and weak intramolecular non-covalent bonds. It plays an important role in identifying compounds, determining molecular structures (Xu et al., 2020). Based on the experimental ssNMR spectra (Fig. 2d), a significant chemical shift change was observed for the C3 peak of EMO at 166.05 ppm upon EMO-TMP cocrystal formation, which is attributed to hydrogen bonding interaction between the hydroxyl group attached to C3 and the N atom of TMP after cocrystallization. Additionally, the carbon atoms of the pyrazine ring in TMP-originally resonating as a doublet at 149.39 and 148.71 ppm-manifest as a singlet upon cocrystallization. This transformation demonstrates that cocrystal formation reconfigures the electronic density distribution of the pyrazine ring through intermolecular hydrogen bonding.

3.6. Thermodynamic analysis

DSC thermograms (Fig. 2e) of the EMO-TMP cocrystal exhibited two endothermic events between 30 and 300 °C. The second sharp endotherm corresponds to the melting point of emodin. Consequently, the preceding endothermic peak at 180 °C is attributed to TMP loss during thermal decomposition.

TGA curve (Fig. 2f) revealed a distinct mass loss step between 80 and 200 °C for the EMO-TMP cocrystal, corresponding to the first endothermic event in the DSC thermogram. Based on the TG analysis in Fig. 2f, TMP is lost at this point. The activation energy for the loss of TMP, measured using the DSC Kissinger method, is 100.74 kJ/mol, which is higher than the sublimation enthalpy of ligustrazine (94.6 kJ/mol) provided by NIST. Therefore, it is inferred that: During the heating process, when the absorbed heat is sufficient to overcome the lattice binding and the energy has reached the sublimation enthalpy of the component, the component will volatilize. Therefore，the DSC will show a curve similar to that of a solvate complex. This distinctive thermal behavior likely stems from both the intrinsically low melting point of TMP and the specific energy landscape of API-CCF intermolecular interactions. A similar phenomenon has also been reported in previous studies (Yuan et al., 2023).

3.7. DVS analysis

Notably, DSC revealed a blunt endothermic peak at 110 °C in pure EMO, indicating a dehydration event and its inherent hygroscopicity. Given that moisture may affect the stability and dissolution performance of the cocrystal, we systematically evaluated the hygroscopicity of EMO-TMP using dynamic vapor sorption (DVS). Additionally, the particle size distribution was analyzed, as it may modulate both moisture affinity and solubility. Hygroscopic behavior can lead to alterations in various physicochemical properties of drugs, such as stability, powder flowability, and solubility (Thakur and Thakuria, 2020). Dynamic Vapor Sorption (DVS) can be used to analyze the change in drug weight as a function of humidity (An et al., 2024). DVS analysis (Fig. 2g) revealed that under 90 % relative humidity, the moisture absorption rate of EMO-TMP remains below 0.06 %, indicating that the introduction of TMP effectively reduces the interaction between EMO and water molecules. For comparison, the DVS profile of the physical mixture was also evaluated under identical conditions, with detailed results provided in Fig. S1 (see Supporting Information).

3.8. Particle size distribution

In the crystallization process of pharmaceutical compounds, particle morphology has a significant impact on the development of drug products with desired critical quality attributes (CQAs). The particle size distribution (PSD) of the API directly influences downstream processing operations and the drug's performance both in vitro and in vivo (Liu et al., 2022; Rosenbaum et al., 2019). The particle size distribution of EMO-TMP was determined using wet laser diffraction (Fig. 2h). The results show that the particle sizes of EMO-TMP are primarily concentrated between 8.27 and 22.66 μm, with a D₅₀ of 17.07 μm and a span of 0.84. The narrow span indicates a uniform particle size distribution, suggesting that the solid powder of EMO-TMP is consistent, which results in more consistent hygroscopic behavior.

3.9. Solid phase stability

Solid phase stability studies are crucial for ensuring the quality, efficacy, and safety of active pharmaceutical ingredients (APIs) and products (Gonzalez-Gonzalez et al., 2022; Song et al., 2025a). Cocrystal technology is an effective strategy to improve the stability of APIs. A thorough understanding of the occurrence of stability issues and their determining factors is necessary to prevent or minimize the formation of undesirable products (Liu et al., 2022b). Stability studies of the EMO-TMP cocrystal (1:0.5) were conducted under high temperature, high humidity, and light conditions (Fig. 2i). The physical mixture was also subjected to the same stability investigation, and the corresponding results are provided in Fig. S2 of the Supporting Information. The characteristic diffraction peak positions of the PXRD spectra remained consistent under these conditions. However, the relative peak intensity ratios in the PXRD spectra showed slight changes after 5 and 10 days of stability testing.

TMP has a high sublimation rate at ambient temperature, and this instability in its physical state has a negative impact on its clinical application (Shi et al., 2024). The activation energy (100.74 kJ/mol) obtained using the DSC Kissinger method (Fig. 2j) and the sublimation enthalpy of TMP (94.6 kJ/mol) provided by NIST (National Institute of Standards and Technology, n.d) were used to calculate the following values via the Arrhenius equation:

$$k=A\bullet \mathit{exp}\left(\frac{-E_a}{\mathit{RT}}\right)$$

$$\frac{k_{\mathit{TMP}}}{k_{\mathit{TMP}\_\mathit{CC}}}=\frac{A_{\mathit{TMP}}\bullet \mathit{exp}\left(\frac{-E_{a,\mathit{TMP}}}{\mathit{RT}}\right)}{A_{\mathit{TMP}\_\mathit{CC}}\bullet \mathit{exp}\left(\frac{-E_{a,\mathit{TMP}\_\mathit{CC}}}{\mathit{RT}}\right)}$$

Assuming that the pre-exponential factor is similar.

$$\frac{k_{\mathit{TMP}}}{k_{\mathit{TMP}\_\mathit{CC}}}=e^{\frac{E_{a,\mathit{TMP}\_\mathit{CC}}-E_{a,\mathit{TMP}}}{\mathit{RT}}}$$

Under 25 °C conditions: The formation of the cocrystal increases the stability of TMP by approximately 11.9 times. Under 40 °C conditions: The formation of the cocrystal increases the stability of TMP by approximately 10.6 times. These results indicate that the formation of the cocrystal significantly enhances the thermal stability of TMP, with this enhancement slightly varying at different temperatures. This provides important scientific evidence for the development of cocrystal formulations, optimization of storage conditions, and prediction of shelf life.

3.10. The color analysis of EMO cocrystals

In the study of cocrystals of EMO, an interesting phenomenon was observed: the introduction of different CCFs results in cocrystals exhibiting different colors (Fig. 2k). Distinct intermolecular interactions induce redistribution of electron density and modulation of band gaps, resulting in chromatic transition from yellow to deep red in emodin (EMO) cocrystals (Li et al., 2018). Through theoretical calculations, we found that the color differences observed in EMO and its cocrystals are positively correlated with the extent of proton delocalization at the hydroxyl group on the C3 position. The study uses AIM theory, estimating the hydrogen bond strength in EMO, EMO-TMP, and the previously reported EMO-PIP cocrystal (Li et al., 2025) through the BCP (3,−1) electron density of the corresponding hydrogen bonds to characterize the delocalization extent, with stronger interactions indicating higher delocalization. It can be seen that the weaker the interaction, the lower the delocalization of the hydroxyl proton at C3, and the lighter the color, which is yellow. As the interaction strength increases, the delocalization also increases, and the color deepens to reddish-brown (in the EMO-PIP cocrystal). The intermediate-strength EMO exhibits an orange color. It can be speculated that when the hydroxyl proton at C3 forms a salt, the color should be the darkest. This distinct color provides a rapid quality control marker for cocrystal formation.

3.11. Solubility analysis

A significant factor limiting the clinical application of emodin (EMO) stems from its poor solubility. Therefore, we first investigated the ability of the EMO-TMP cocrystal to enhance the in vitro solubility of EMO. EMO material exhibited negligible solubility in purified water and three other media. The EMO-TMP cocrystal demonstrated significantly increased equilibrium solubility compared to pure emodin in both water and pH 6.8 phosphate buffer (Fig. 3a). Notably, the cocrystal displayed pH-dependent solubility behavior, with solubility markedly decreasing at lower pH values. Fig. 3b shows the apparent dissolution results of EMO-TMP cocrystal under pH 6.8 and pH 7.0 conditions; EMO is not included due to undetectability. The cumulative dissolution of EMO-TMP reached approximately 333.31 μg in pH 6.8 medium and 370 μg in pH 7.0 medium. In parallel, the dissolution behaviors of EMO, EMO-TMP, and their physical mixture (PM) were investigated in a medium containing 0.2 % SDS. The results indicated that the PM only provided a slight enhancement in the dissolution rate of EMO during the initial 20 min, and this effect was much weaker than that of EMO-TMP. Moreover, the PM failed to improve the final dissolution extent of EMO, while the EMO-TMP cocrystal markedly improved both the dissolution rate and the solubility of EMO. Detailed results are provided in Fig. S3 (Supporting Information).

Fig. 3.

In vitro and in vivo evaluation of the EMO-TMP Cocrystal. (a)The results of equilibrium solubilities of the pure EMO, the EMO-TMP. (b) Powder dissolution profiles of EMO-TMP in pH 6.8 and 7.0 media. (c)Plasma concentration −  time profiles of EMO and EMO from the cocrystals after oral administration. (d)The body weight changes of DSS-induced ulcerative colitis mice. (e)The DAI scores of DSS-induced ulcerative colitis mice. (f)The colon picture of DSS-induced ulcerative colitis mice. (g)The colon length of DSS-induced ulcerative colitis mice. (h) Colonic tissues of DSS-induced ulcerative colitis mice. (i)The concentration of TNF-α in the serum of DSS-induced ulcerative colitis mice. (j)The concentration of IL-1β in the serum of DSS-induced ulcerative colitis mice. (k)The concentration of TNF-α in the colon of DSS-induced ulcerative colitis mice. (l)The concentration of IL-1β in the colon of DSS-induced ulcerative colitis mice. Data were shown as mean ± SEM, Scale bar = 100 μm. ^#P < 0.05, ^##P < 0.01, ^###P < 0.001, compared with Control group; ^⁎P < 0.05, ^⁎⁎P < 0.01, ^⁎⁎⁎P < 0.001, compared with Model group, n = 5–7.

3.12. In vivo pharmacokinetics

In vitro experiments demonstrated that the EMO-TMP cocrystal enhanced the solubility of EMO to some extent. Based on the comprehensive evaluation of both the physical mixture and the cocrystal, the physical mixture exhibited poor physical stability and demonstrated no significant advantage in dissolution performance over pure EMO. Therefore, considering these clear pharmaceutical limitations, we proceeded with the animal study using only the EMO-TMP cocrystal. To further investigate the enhancement effect of the EMO-TMP cocrystal on the bioavailability of EMO, we conducted pharmacokinetic studies. Using high-performance liquid chromatography-mass spectrometry technology to determine the plasma concentration of EMO, describe the plasma concentration-time curves of EMO-TMP cocrystal, and pure EMO after oral administration. In addition, the calculated pharmacokinetic parameters are summarized in Table 2. The pharmacokinetic results (Fig. 3c) showed that the formation of cocrystal caused significant changes in the EMO pharmacokinetic behavior. The C_max of the EMO-TMP group was 427.91 ± 134.44 μg/L, whereas that of the pure EMO group was 273.58 ± 159.62 μg/L, indicating that the cocrystal significantly increased the peak concentration of EMO. In addition, a significant decrease in CLz suggested that the EMO-TMP cocrystal slowed the elimination rate of EMO in vivo compared to pure EMO. Most importantly, the AUC(0–t) provided clear and consistent evidence: Unlike the extrapolation-based AUC(0–∞), AUC(0–t) is entirely derived from measured plasma concentration data and is recognized as a more reliable metric of drug exposure. Our data showed that the AUC(0–t) of EMO-TMP was 2.16 times that of pure EMO, and the difference was statistically significant (p < 0.05). This result is fully consistent with the increase in C_max and the decrease in CLz, collectively providing strong evidence for the enhanced bioavailability conferred by the cocrystal. It should be specifically noted that the extrapolation-based AUC(0–∞) calculation method was overly sensitive to the abnormally low elimination rate constant in individual mice, leading to an abnormally high calculated value.

Table 2.

Pharmacokinetic parameters after oral administration of EMO, and EMO-TMP cocrystals in rats (Data were shown as mean ± SEM. ^#P < 0.05, compared with EMO, n = 6).

Parameter	EMO	EMO-TMP
AUC(0-t) (μg·L−1·h)	2368.22 ± 947.10	5117.72 ± 2403.85#
AUC(0-∞) (μg·L−1·h)	33,450.00 ± 63,426.55	27,198.53 ± 23,199.35
MRT(0-t) (h)	10.35 ± 1.94	10.58 ± 1.82
MRT(0-∞) (h)	5.90 ± 5.41	4.73 ± 3.83
t1/2z (h)	4.13 ± 1.97	3.70 ± 2.19
Tmax (h)	11 ± 1.67	8.5 ± 3.89
CLz (L·h−1·kg−1)	0.004 ± 0.003	0.002 ± 0.002
Cmax (μg/L)	273.58 ± 159.62	427.91 ± 134.44

3.13. Pharmacological evaluation

The 2.16-fold increase in EMO bioavailability from the cocrystal (AUC(0-t): 5117.72 ± 2403.85 vs 2368.22 ± 947.10 μg·L⁻¹·h) directly translated to enhanced systemic exposure. To confirm whether this pharmacokinetic advantage conferred superior therapeutic outcomes in a disease-relevant model, we evaluated the anti-colitis efficacy of EMO, TMP, and EMO-TMP cocrystal in ameliorating DSS-induced ulcerative colitis. Body weight changes, DAI scores and colon length were measured to evaluated the effects of EMO-TMP on DSS-induced ulcerative colitis. Compared with the control group, the model group exhibited significant weight loss (Fig. 3d), increased DAI scores (Fig. 3e), reduced colon length (Fig. 3f, g), indicating that these mice developed severe colitis. Treatment with EMO, TMP, and EMO-TMP remarkedly improved this phenomenon induced by DSS, playing protective effect. Consistent with this conclusion, colonic tissues of model group showed substantial histopathological lesions (Fig. 3h). The administration of EMO, TMP, and EMO-TMP ameliorated the histopathological changes induced by DSS, including inflammatory cell infiltration, crypt damage, and the disruption of the mucosal epithelium.

To assess the inflammatory response, the levels of pro-inflammatory cytokines TNF-α and IL-1β were measured in serum and colon. Compared with the control group, the levels of TNF-α and IL-1β in serum and colon both significantly increased in the model group, whereas treatment with EMO, TMP, and EMO-TMP dramatically reduced the levels of TNF-α and IL-1β. EMO-TMP displayed stronger anti-inflammatory effect (Fig. 3i-3l). These results suggested that EMO-TMP effectively alleviated DSS-induced ulcerative colitis.

4. Conclusion

This study demonstrates that a rationally designed EMO-TMP cocrystal effectively overcomes the low bioavailability of emodin (EMO), a promising but clinically limited therapeutic agent for ulcerative colitis (UC). Tetramethylpyrazine (TMP), a natural compound from Traditional Chinese Medicine (TCM), was strategically selected as the cocrystal former (CCF) through a multi-step screening approach integrating literature analysis, computational chemistry (docking/simulation), and machine learning-based cocrystal prediction. The resulting EMO-TMP cocrystal significantly enhanced EMO's aqueous solubility while preserving its intrinsic bioactivity. Pharmacokinetic studies confirmed improved oral bioavailability, and in vivo evaluation in a DSS-induced colitis model demonstrated superior therapeutic efficacy compared to EMO alone.

The work highlights two key advances: First, TMP is established as an effective CCF for APIs rich in hydrogen bond donors, validated through rational design and experimentation. Second, it presents a cocrystal engineering strategy that synergistically combines TCM formulation principles with modern pharmaceutical techniques, offering a viable approach to enhance the clinical potential of bioactive natural products with poor physicochemical properties.

This research not only provides a methodological reference for TCM-inspired drug development but also expands the application of computational and data-driven methods in cocrystal design, paving the way for the efficient translation of natural products into clinically viable therapeutics.

CRediT authorship contribution statement

Meiru Liu: Writing – original draft, Methodology, Investigation, Data curation. Yinru Jiang: Writing – original draft, Methodology, Investigation. Penghui Yuan: Writing – review & editing, Methodology, Investigation. Shuang Li: Methodology, Investigation, Formal analysis. Baoxi Zhang: Methodology, Formal analysis. Xia Zhou: Methodology, Formal analysis. Bin Su: Investigation, Formal analysis. Yifei Xie: Methodology, Formal analysis. Dezhi Yang: Writing – review & editing, Software, Funding acquisition, Formal analysis. Linglei Kong: Writing – review & editing, Validation, Supervision. Li Zhang: Validation, Supervision. Yang Lv: Writing – review & editing, Supervision, Project administration, Funding acquisition, Conceptualization. Guanhua Du: Supervision, Project administration.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Data availability

Data will be made available on request.