Stress-trained microalgae robots with probiotics backpack and intestinal brake for inflammatory bowel disease management

Abstract

Inflammatory bowel disease (IBD) involves elevated intestinal reactive oxygen species (ROS) and microbial imbalance. A key challenge is that current delivery systems cannot adequately protect active agents, such as antioxidants and probiotics, through the harsh gastric environment nor precisely deliver them to inflamed intestinal sites. Here, we present the EcN@PDA@HP biohybrid robot, composed of stress-trained microalgae, probiotics, and an intestinal braking system, enabling rapid gastric passage and precise retention in inflamed regions. Stress-trained Haematococcus pluvialis (HP) developed a thickened cell wall for gastric resilience, maintained flagella-driven motility, and enhanced astaxanthin (AST) production. Polydopamine (PDA)-coated Escherichia coli Nissle 1917 (EcN) anchors onto HP via host-guest interactions. In vivo, PDA preserves EcN activity and enables specific adhesion to inflamed sites. By combining antioxidizing AST and bacteriotherapy, EcN@PDA@HP effectively alleviates male murine IBD models via ROS scavenging and microbiota restoration, offering a promising strategy for diverse IBD conditions.

IBD involves intestinal ROS overproduction and microbial imbalance, with current delivery systems lacking effective protection and targeting. Here, the authors show the EcN@PDA@HP biohybrid robot alleviates murine IBD via astaxanthin and probiotic

Introduction

The onset and progression of inflammatory bowel disease (IBD), encompassing ulcerative colitis (UC) and Crohn’s disease (CD), present significant medical challenges^1,2. Patients with IBD often suffer from reduced quality of life and face an elevated risk of developing colon cancer^3,4. The pathogenesis and progression of IBD are frequently associated with overproduction of reactive oxygen species (ROS) and dysbiosis in microbiota^5,6. Excessive ROS levels induce lipid peroxidation and DNA mutations, resulting in damage in intestinal endothelial cells, further impairing protein functions, altering epithelial permeability, and disrupting the intestinal barrier—factors that contribute to the initiation or exacerbation of IBD^7,8. The biopsies of IBD patients reveal significantly elevated mucosal ROS concentrations, ranging from 10 to 100 times higher at the inflamed site than normal levels⁹. Antioxidants play a well-established role in scavenging ROS in the intestines, rendering them a valuable therapeutic strategy for IBD management^7,10. However, conventional antioxidant formulations face limitations, including poor oral stability, nonspecific distribution, and short intestinal retention time, which collectively hinder their ROS-scavenging efficacy and result in inconsistent therapeutic outcomes for IBD.

Recent advances in micro-and nanorobotics have highlighted actively propelled robots as effective carriers for biomedical delivery, overcoming physiological barriers in diverse applications^11,12. Among these, green algae-based microrobots have garnered particular attention, due to their inherent motility advantages, including prolonged autonomous self-propulsion, high-speed propulsion, robust directional motion, autofluorescence for imaging, efficient cargo delivery, and scalability in production^13,14. These attributes distinguish algae-based microrobots from other types, such as magnetically, chemically, or bacteria-driven robots^15,16. By modifying the algae surface with specific components, researchers have successfully integrated advanced cargo delivery capabilities, demonstrated in applications like IBD treatment. For instance, algae robots loaded with cell membrane nanoparticles have shown significant efficacy in clearing inflammatory cytokines, along with high biocompatibility and extensive distribution throughout the gastrointestinal (GI) tract¹⁵. Despite these advancements, the acidic environment of the stomach can impair the functionality of both the cargo and the microalgae robot, necessitating intricate protective measures^15–17. Moreover, the robots’ high-speed propulsion, while advantageous, presents a challenge for targeted accumulation in the intestinal tract^14,15. Therefore, developing a microalgae-based delivery system capable of both withstanding gastric acid and intelligently braking at inflammatory sites in the GI tract would enable prolonged and targeted cargo release, enhancing therapeutic efficacy.

Many metabolites biosynthesized by microorganisms exhibit immunomodulatory and antioxidant properties, making them valuable in diverse applications such as food and pharmaceutical additives^18–20. Notably, Haematococcus Pluvialis (HP) responds to stressful condition^21–23 by thickening its cell wall and biosynthesizing astaxanthin (AST, 3,3′-dihydroxy-β, β′-carotene-4,4′-dione), a powerful antioxidant. AST offers various ROS scavenging enzyme activities, such as catalase (CAT), superoxide dismutase (SOD), and ascorbate peroxidase (APX)²⁴. The thicker cell wall of stress-conditioned HP provides resilience, protecting it from damage in the acidic stomach microenvironment while preserving motility in the intestine. Thus, HP was selected as the bio-robot for efficient oral delivery in this study, functioning as a robust cargo carrier for IBD treatment.

The pathogenesis and progression of IBD are intricately linked to the dysbiosis of colonic microflora^25–27. In a healthy gut, the microbiota synthesizes essential short-chain fatty acids and vitamins, and safeguarding against pathogen colonization^28,29. However, in IBD, disruptions in the intestinal flora contribute to chronic inflammation, increase the production of harmful substances, and impair host metabolism^5,28. Oral probiotics (Lactobacillus rhamnosus GG, Bifidobacterium longum, and Escherichia coli Nissle 1917) have shown promise in restoring gut microbiome balance for treating GI disorders^30–33. However, their efficacy is fundamentally constrained by gastric acid vulnerability, causing >90% viability loss at pH 1–2 within 2 h³⁴, compounded by bile salt-mediated colonization failure in the small intestine and transient colonic retention limiting sustained microbiome modulation³⁵. These intrinsic biological barriers necessitate innovative delivery platforms for maintaining probiotic functionality and improving colonization in the right place.

Herein, we developed a stress-trained algae-based microrobot equipped with an intelligent braking system for efficient GI delivery of biosynthesized astaxanthin and probiotics, aimed at managing IBD by mitigating excessive ROS and restoring intestinal flora homeostasis (Fig. 1). Haematococcus Pluvialis (HP) was pre-conditioned under stress (nitrogen starvation, continuous light, and mild acidity) to induce astaxanthin (AST) production and thicken its cell wall, thereby enhancing resistance to gastric acidic, maintaining self-motility, and allowing rapid passage through the harsh gastric environment. Probiotic EcN was then coated with polydopamine (PDA) to construct EcN@PDA for enhanced intestinal adhesion property^36,37. Host and guest molecules (β-cyclodextrin (β-CD) on HP and adamantane (ADA) on EcN), were respectively introduced to enable the formation of HP-EcN hybrid microrobots (EcN@PDA@HP) through supramolecular complexation. Thanks to HP’s motility and acid-resistant properties acquired through stress-conditioning, these hybrid microrobots traverse the stomach rapidly without compromising the activity of AST and EcN. Upon reaching the intestine, the PDA coating on EcN prolongs the retention time of EcN@PDA@HP robots in inflamed areas, supporting sustained AST release and effective EcN colonization. As a result, EcN@PDA@HP robots, equipped with an intelligent braking system, significantly inhibited colitis induced by dextran sodium sulfate (DSS) and Salmonella Typhimurium (STm), including effects like macrophage reprogramming, reduction of epithelial apoptosis, intestinal barrier restoration, inflammation reduction, ROS scavenging, and microbiota modulation. Overall, our findings demonstrate that EcN@PDA@HP alleviates inflammation and ameliorates colonic disease pathology, advancing the clinical potential of probiotic-loaded microalgae robots. This approach shows promise as an oral formulation with biosynthetic and intelligent motility for IBD combination therapy.

Fig. 1. Schematic illustration on the construction of EcN@PDA@HP robots and their mechanism for IBD treatment.

A Schematic illustration on the fabrication process of EcN@PDA@HP robots. Stress training enables HP to biosynthesize the antioxidant AST and thicken its cell wall for acid resistance, resulting in a robust microalgal robot for active delivery. Probiotic EcN is then coated with PDA for gastric acid protection effect and adhesive property in inflamed colon. Finally, by modifying host and guest molecules on the surface of microalgae and probiotics, respectively, the hybrid EcN@PDA@HP robots are constructed through host-guest interaction. B Mechanism for IBD treatment: Upon oral administration, EcN@PDA@HP robots quickly and safely traverse the stomach, leveraging their natural motility and acid resistance to remain active despite exposure to gastric acid and digestive enzymes. The PDA coating on EcN provides a brake for the EcN@PDA@HP robots to adhere effectively at intestinal lesion sites. Once there, the robots release AST, which works synergistically with the anchored EcN to treat IBD by scavenging ROS and modulating gut microbiota.

Results

Preparation and characterization of EcN@PDA@HP robots

Polydopamine (PDA) was selected to coat the surface of EcN, enhancing the adhesion ability of the EcN@PDA@HP robots. Previous studies have shown that in situ polymerized PDA adheres to bacterial surfaces via covalent or hydrogen bonds³⁶. Both transmission electron microscopy (TEM) and scanning electron microscopy (SEM) images showed bare EcN with a typical rod shape and clear edges (Fig. 2A and Supplementary Fig. 1A). After PDA coating, a rough layer was observed (Fig. 2B and Supplementary Fig. 1B), and the hydrated size increased from 1560 to 1893 nm (Fig. 2C). Additionally, the zeta potential of EcN@PDA shifted from −30.26 mV for EcN to −26 mV (Fig. 2D). These alterations in size and surface potential confirm the successful PDA coatings of probiotics.

Fig. 2. Preparation and characterization of Haematococcus pluvialis–EcN hybrid microrobots (denoted as “EcN@PDA@HP”).

Typical TEM images of EcN (A) and EcN@PDA (B). Scale bar, 1 μm. C Zeta potential of EcN and EcN@PDA (n = 3 independent experiments). Significance was determined by t test. D Size distribution of EcN and EcN@PDA (n = 3 independent experiments). E GFP-labeled E. coli decoration with ADA-PEG-NHS, and then stained with CD-Cy5 for 5 min. Bare EcN@PDA without ADA surface modification served as the control group. The scale bar, 2 μm. The experiment was repeated twice independently with similar results and the representative data is shown. F HP was decorated with β-CD-PEG-NHS, and then stained with ADA-FITC for 5 min. Bare HP without β-CD surface modification served as the control group. The scale bar, 20 μm. The experiment was repeated twice independently with similar results and the representative data is shown. G Pseudocolor SEM imaging of EcN@PDA and HP with different formulations (Bare HP), (CD (−), ADA (−)), (CD (−), ADA (+)), (CD (+), ADA (−)), (CD (+), ADA (+)). The scale bar, 10 μm. The experiment was repeated twice independently with similar results and the representative data is shown. H Fluorescent images of EcN@PDA and HP with different formulations (CD (−), ADA (−)), (CD (−), ADA (+)), (CD (+), ADA (−)), (CD (+), ADA (+)). Autofluorescence of natural algae chloroplast in Cy5 channel; Fluorescent EcN in GFP channel. The scale bar, 20 μm. I, J Proportions of the HP carrying GFP labeled EcN in different groups (n = 3 independent experiments). Significance was determined by one-way ANOVA. K, L The number of EcN attached in HP surface from different groups (n = 3 independent experiments). Significance was determined by one-way ANOVA.

To verify the successful modification of adamantane (ADA) and β-cyclodextrin (β-CD) groups on bacteria and Haematococcus pluvialis (HP), respectively, we conducted the following experiments. EcN@PDA was functionalized with ADA by reacting ADA-PEG-NHS (adamantane-poly (ethylene glycol)-N-hydroxysuccinimide) ester with primary amine groups on the EcN@PDA surface. The availability and stability of ADA on EcN@PDA were evaluated using Cyanine 5 (Cy5)-modified β-CD. After treating ADA-EcN@PDA with β-CD-Cy5 for 5 min, bright red fluorescence was observed on the bacterial surface due to host-guest interactions between β-CD and ADA (Fig. 2E), which persisted even after 24 h in fresh media (Supplementary Fig. 3A). Haematococcus pluvialis (HP), pre-trained under nitrogen starvation, continuous light and weak acid for 6 days to enhance AST production (Supplementary Fig. 2) and thicken cell walls (Supplementary Fig. 4), were modified with β-CD-PEG-NHS. The successful β-CD modification was confirmed using fluorescein isothiocyanate (FITC)-labeled ADA. Incubation with ADA-FITC for 5 min resulted in green fluorescence on the HP surface (Fig. 2F), which also remained stable for 24 h (Supplementary Fig. 3B). In contrast, unmodified HP showed negligible fluorescence.

Subsequently, ADA-modified EcN@PDA were conjugated onto the β-CD modified HP surface through host-guest interaction, forming “EcN@PDA@HP” robots. SEM images confirmed the attachment of EcN@PDA on the surface of HP (Fig. 2G). The binding required the specific modifications of both components via host and guest molecules, respectively, as unmodified counterparts showed no significant attachment, and this conjugate could be stable for at least 24 h (Supplementary Fig. 5). To track the conjugation process, GFP-labeled EcN@PDA was used (whereas HP has intrinsic red fluorescence). After 2 h of incubation, green fluorescence signals were observed on HP surfaces, confirming successful attachment (Fig. 2H). In contrast, HP without β-CD modification or EcN@PDA without ADA modification showed minimal attachment. Flow cytometry (FCM) analysis revealed a binding efficiency of ~75.3% between GFP-labeled EcN@PDA and HP (Fig. 2I, J). To systematically investigate the long-term stability of EcN@PDA@HP, we conducted time-dependent binding rate and viability analyses under physiologically relevant conditions. The engineered EcN@PDA@HP system retained approximately 59% binding rate between EcN@PDA and HP after incubation for 2 h in simulated gastric fluid (SGF, pH 3.5) (Supplementary Fig. 6C, D). Similarly, the sustained binding stability was observed in simulated intestinal fluid (SIF, pH 6.8), and the combination rate of EcN@PDA and HP (in the form of EcN@PDA@HP) was as high as 65.8% after incubation for 24 h (Supplementary Fig. 6A, B). The EcN@PDA@HP system exhibited good protective capacity, with HP’s viability declining to 57.2% following 2 h exposure to SGF (Supplementary Fig. 7C, D). Notably, the viability was enhanced in the intestinal conditions, demonstrating 65.7% HP survival (34.3% mortality) even after 24 h incubation (Supplementary Fig. 7A, B). Furthermore, quantitative analysis using the spread plate method demonstrated that each HP cell could carry over 300 EcN cells on its surface (Fig. 2K, L), underscoring the robust host-guest interaction in forming the EcN@PDA@HP robots.

The bacterial viability of EcN@PDA@HP robots was investigated. After 24 h of culture, the OD600 values of EcN@PDA@HP robots were similar to those of untreated EcN and EcN@PDA, indicating that the modifications did not compromise bacterial viability (Fig. 3A). Next, the resistance of EcN@PDA@HP robots to GI environmental challenges was assessed. Bare EcN, EcN@PDA, and EcN@PDA@HP were incubated with SGF (pH 3.5) containing pepsin. As shown in Fig. 3B and Supplementary Fig. 8, both EcN@PDA and EcN@PDA@HP demonstrated significantly higher viability compared to bare EcN after 0.5, 1, and 2 h, attributed to the protective effect of the PDA coating against acidic conditions. Similarly, simulated intestinal fluid (SIF, pH 6.8) containing trypsin was used to evaluate bacterial survival in the intestinal environment. As shown in Fig. 3C and Supplementary Fig. 9, the survival rates of EcN@PDA and EcN@PDA@HP in SIF were comparable to those of untreated EcN, indicating that the PDA coating did not impair bacterial viability in the intestine. Collectively, these results demonstrate that EcN@PDA@HP robots are resistant to adverse physiological conditions, ensuring the safe oral delivery of both EcN and HP.

Fig. 3. Environmental tolerance, motility, and in vitro antioxidant evaluation of EcN@PDA@HP robots.

A Growth curves of EcN, EcN@PDA, EcN@PDA@HP cultured in LB medium at 37 °C (n = 3 independent experiments). B Survivals of EcN, EcN@PDA, and EcN@PDA@HP after exposure to SGF (pH = 3.5) (n = 3 independent samples). C Survival rates of EcN, EcN@PDA, and EcN@PDA after exposure to SIF (pH = 6.8) (n = 3 independent samples). Moving trajectories of OHP (D), HP (E), and EcN@PDA@HP (F) for 10 s after incubation in SGF (pH = 3.5) for different durations (0, 15, 30, 60, and 120 min). Moving trajectories of OHP (G), HP (H), and EcN@PDA@HP (I) for 10 s after incubation in SIF (pH = 6.8) for different durations (0, 2, 4, and 8 h). J EcN@PDA@HP (SGF) was obtained by incubating EcN@PDA@HP in SGF for 2 h, and the moving trajectories of EcN@PDA@HP (SGF) were recorded for 10 s after incubation in SIF (pH = 6.8) for different durations (0, 2, 4, and 8 h). K The average velocities of OHP, HP, and EcN@PDA@HP after incubation in SGF (pH = 3.5) for different durations (0, 15, 30, 60, and 120 min) (n = 20 independent samples). L The average velocities of OHP, HP, EcN@PDA@HP, and EcN@PDA@HP (SGF) after incubation in SIF (pH = 6.8) for different durations (0, 2, 4, and 8 h) (n = 20 independent samples). H₂O₂ (M), ·O₂⁻(N), and ·OH (O) scavenging performances of EcN@PDA@HP in response to various treatments (n = 3 independent experiments). P Schematic diagram of ROS scavenging by EcN@PDA@HP in NCM 460 cells via a transwell model. Q, R FCM analysis of cellular ·O₂⁻ levels after different treatments. (n = 3 independent experiments). S, T FCM analysis of cellular H₂O₂ levels after different treatments. (n = 3 independent experiments). Data are presented in the form of mean values  ±  SD. Statistical analysis of (B, C, K, L) were conducted using Two-Way ANOVA. All the other statistical analyses were conducted using one-way ANOVA.

Motion behavior and ROS scavenging of EcN@PDA@HP robots

We first investigated the motion characteristics of trained Haematococcus pluvialis (HP) in SGF, and compared them to those of untrained Haematococcus pluvialis (original HP (OHP)). As shown in Supplementary Movie 1, OHP exhibited significant decline in activity after incubation in SGF for 2 h, making it difficult to maintain self-motility. In contrast, pre-trained HP with (Supplementary Movie 2) and EcN@PDA@HP (Supplementary Movie 3) exhibited better resistance to SGF and retained inherent motility. Regarding movement trajectory, both EcN@PDA@HP (Fig. 3F and Supplementary Movie 4) and HP (Fig. 3E and Supplementary Movie 5) maintained good motility after incubation in SGF for different durations. In comparison, the control group (OHP) exhibited a rapid decline in motility, particularly noticeable after 30 min of incubation in SGF (Fig. 3D and Supplementary Movie 6), with a marked reduction in the distance traveled. Interestingly, while the average velocity of trained HP (79.35 μm s⁻¹) was slightly lower than that of free OHP (94.78 μm s⁻¹), HP maintained a high-speed level (61.91 μm s⁻¹) after 120 min incubation in SGF, similar to that of EcN@PDA@HP (60.70 μm s⁻¹, Fig. 3K). These results indicate that gastric acid environment and surface attachment of EcN had negligible impact on the motility of HP. This finding is further supported by the movement distances of microalgae robots recorded over 10 s at various time points post-SGF incubation (Supplementary Fig. 10A). Subsequently, we further evaluated the motion performance of microalgae robots in SIF. Figure 3G–I) and Supplementary Movies 7–9 show that OHP, trained HP, and EcN@PDA@HP all sustained motility in SIF for extended periods. To stimulate the GI transit via the oral route, EcN@PDA@HP was first incubated in SGF for 2 h (designated as EcN@PDA@HP (SGF)), and its movement trajectory in SIF was recorded over 10 s at different time points. As shown in Fig. 3J and Supplementary Movie 10, EcN@PDA@HP (SGF) demonstrated robust motility in SIF, which might be due to the thickened HP cell wall induced by pre-training under harsh conditions, allowing it to overcome the stomach acid barrier. The average speed and cumulative distance traveled by the various microalgae robots in SIF over 10 s, as shown in Fig. 3L and Supplementary Fig. 10B, respectively, further corroborate these above-discussed results.

In response to a harsh environment, HP was previously reported to produce antioxidant astaxanthin (AST) that could be used to reduce oxidative stress and alleviate inflammation for potential IBD treatment^38,39. The rich flora in the intestines produces a variety of polysaccharide-degrading enzymes, including cellulase and pectinase that may degrade HP cell walls^40–42, therefore, cellulase and pectinase were employed to lyse the cell wall of HP to obtain protoplasts, which was conducive to release biosynthesized astaxanthin, in order to further evaluate the antioxidant activity in vitro. The biosynthesized AST was quantified (Supplementary Fig. 11). The clearance of H₂O₂, •O₂⁻, and •OH was subsequently measured to determine the antioxidative performance of EcN@PDA@HP, whereas vitamin C (VC) and chemosynthetic AST were used for comparison. As shown in Fig. 3M–O), EcN@PDA@HP exhibited the strongest ROS-scavenging capability towards H₂O₂, •O₂⁻, and •OH, which was further confirmed in NCM 460 cells via a transwell model. We performed cytotoxicity assays using NCM460 cells to evaluate the biocompatibility of AST, HP, and EcN@PDA@HP as well as their potential protective effects on these cells. As determined by CCK-8 assay (Supplementary Fig. 12A), AST, HP, and EcN@PDA@HP did not exhibit cytotoxicity even after 24 h incubation, indicating the excellent in vitro biocompatibility. In addition, apoptosis analysis via FCM revealed that all of the treatment groups induced negligible apoptotic cell death in NCM460 cells, further corroborating their favorable biocompatibility profile (Supplementary Fig. 12B). NCM460 cells were seeded in the basal chamber of the transwell plate, and different antioxidants and protoplasts were added in the apical chamber (Fig. 3P). NCM460 cells were treated with H₂O₂ to induce excessive ROS production, serving as a positive control in the experiment. When pre-trained HP or EcN@PDA@HP was added to cells incubated with H₂O₂, the intracellular ·O₂⁻ (Fig. 3Q, R) and H₂O₂ (Fig. 3S, T) concentrations decreased significantly. Collectively, EcN@PDA@HP robots retain excellent motility in SGF and release biosynthesized AST in response to enzymatic lysis for efficient ROS scavenging, likely providing an oral option for IBD treatment.

Evaluation of the mucoadhesive (“braking”) ability of EcN@PDA@HP

Next, we assessed the ability of EcN@PDA@HP to rapidly pass through the stomach in vivo. As shown in Fig. 4A, B, the entire GI tract tissues were collected at 1 h after oral administration of various formulations to mice. The fluorescence of OHP was mainly concentrated in the stomach, while the fluorescence of the intestine was not obvious. This showed OHP activity was reduced, losing its own powerful motility characteristics to quickly pass through the stomach. However, both trained HP and EcN@PDA@HP quickly avoided the erosion of gastric acid and then accumulated in the intestine, due to their adaptation to harsh gastric acid and their own mobility. Therefore, trained EcN@PDA@HP robots protected them from exposing to stomach acid and maintained motility for quick passing through the stomach, contributing to the preservation of AST antioxidant activity to treat IBD.

Fig. 4. In vivo evaluation of the passage through the stomach and mucoadhesive (“braking”) capability of EcN@PDA@HP in the intestine.

A Fluorescent pictures of the entire GI tract at 1 h after oral administration. B Quantitative analysis of the fluorescence intensity of mice stomachs and intestines in each group at 1 h after oral administration (n = 3 independent samples). C Fluorescence images of ex vivo intestines after incubation with HP, EcN@PDA + HP, and EcN@PDA@HP, respectively. D Quantitative analysis on the fluorescence intensity of ex vivo intestines after incubation with HP, EcN@PDA + HP, and EcN@PDA@HP, respectively (n = 3 independent samples). E Fluorescence images of mice treated with various formulations at different time points. F Quantitative analysis on the in vivo fluorescence intensity of treated mice at different time points (n = 3 mice). G, H Fluorescent images of the entire GI tract at 24 h after oral administration. Cy5 channel: HP; FITC channel: GFP labeled EcN. Data are presented in the form of mean values  ±  SD. Statistical analysis of (F) was conducted using two-way ANOVA. All the other statistical analyses were conducted using one-way ANOVA.

To assess the “braking” function of the PDA layer in the intestines HP, EcN@PDA physically blended with HP (EcN@PDA + HP), and EcN@PDA@HP were incubated with freshly harvested mouse intestines in PBS for 1 h, followed by imaging via IVIS after being washed with PBS. The fluorescence of chlorophyll in microalgae enables IVIS imaging, showing their in vivo distribution. As shown in Fig. 4C, D, the fluorescence intensity of the mice intestines in the EcN@PDA@HP treated group was much higher than those of HP and EcN@PDA + HP treated groups, suggesting that PDA layer could improve the adhesion of HP microalgae robot onto the intestines during the oral delivery of EcN@PDA@HP. Furthermore, the intestinal biodistribution of EcN@PDA@HP was further investigated in vivo. Figure 4E, F showed that the retention time of the fluorescence in mice treated with EcN@PDA@HP was much prolonged in comparison to those of mice treated with HP and EcN@PDA + HP, respectively. After oral administration for 24 h, the fluorescence in EcN@PDA@HP treated mice was still very strong, attributed to the excellent adhesion of EcN@PDA in the inflammatory intestines. The adhesion ability of EcN@PDA in the intestine was also verified in the frozen tissue sections (Supplementary Fig. 13). To further investigate the enhanced retention of HP mediated by EcN@PDA in the intestines, we quantitatively analyzed AST levels in colonic tissues. As shown in Supplementary Fig. 14, the EcN@PDA@HP treated group of mice exhibited the highest level of colonic AST, attributable to prolonged intestinal retention facilitated by EcN@PDA. Furthermore, bio-SEM analysis qualitatively demonstrated the colonic retention of different formulations. Supplementary Fig. 15 reveals that EcN@PDA@HP exhibited robust retention in the colon with intact EcN colonization, whereas negligible retention was observed for the HP or EcN@PDA + HP physical mixture groups. In addition, there was negligible fluorescence in the other organs, suggesting a likely good safety profile of oral EcN@PDA@HP (Supplementary Fig. 16). Finally, the entire GI tracts of mice were collected at the endpoint of the in vivo experiment. HP was imaged under Cy5 channel, and GFP-labeled EcN was selected to allow observation under FITC channel. As depicted in Fig. 4G, owing to the robust motility of HP and intestinal peristalsis, only a minimal quantity of HP remained in the entire GI tract from mice treated HP and EcN@PDA + HP, respectively. In contrast, EcN@PDA@HP remained at a high level of HP fluorescence intensity in the intestine. In addition, the green fluorescence EcN essentially overlapped with the red fluorescence of HP (Cy5), indicating a sustained structure stability and hand-in-hand delivery of EcN@PDA@HP via oral administration (Fig. 4H). In contrast, there was negligible overlap area between the red (HP) and green (EcN) fluorescence in the mice treated with a physical mixture of HP and EcN@PDA. These results show that EcN@PDA@HP went through the stomach quickly via HP’s motility, and efficiently accumulated in the inflammatory intestine with a braking system effect of PDA coating.

EcN@PDA@HP effectively treated DSS-induced acute colitis

Considering the efficient oral delivery of EcN@PDA@HP to the inflammatory intestine, the therapeutic efficacy of EcN@PDA@HP was evaluated on DSS-induced acute colitis in mice. The mice were administered with drinking water containing 3% of DSS for 5 days to induce colitis (Fig. 5A). After treatment, EcN@PDA@HP significantly inhibited the body weight loss in colitis mice (Fig. 5B) and downregulated the disease activity index (DAI, Fig. 5C), in comparison to those of mice treated with EcN@PDA, HP, and EcN@PDA + HP, respectively. Colitis mice receiving treatments of EcN@PDA, HP, and EcN@PDA + HP suffered from hematochezia, whereas EcN@PDA@HP treated mice did not show obvious hematochezia during the same period (Supplementary Fig. 17A). The colons of all treated mice were collected, photographed and measured on day 10. As shown in Fig. 5D, E, the colon length of EcN@PDA@HP treated group was significantly longer than that of other groups. Further histopathological characteristics of all treated mice were evaluated by H&E staining. DSS-induced mice model exhibited characteristic inflammatory pathology, including mucosal epithelium cell shedding (yellow arrows), distorted crypt architecture, and marked depletion of goblet cells (red arrows), and dense infiltration of polymorphonuclear leukocytes within the lamina propria (blue arrows). In contrast, mice treated with EcN@PDA@HP demonstrated preserved mucosal integrity with well-organized crypt structures, significantly attenuated inflammatory cell infiltration, and restoration of mucin-producing goblet cell populations. These observations collectively indicated substantial mitigation of colonic inflammation and epithelial barrier recovery in the treatment group (Fig. 5F).

Fig. 5. Validation of the therapeutic efficacy of EcN@PDA@HP in acute colitis.

A Schematic illustration on the treatment protocol of DSS-induced acute colitis mice. B Curve chart variations of murine body weight of the six groups, normalized to the percentage of day 0 body weight (n = 5 mice). C Curve chart analysis on the DAI scores of treated mice (n = 5 mice). D Photographs of murine colons after oral treatment with different formulations. E Histogram analysis on the colon length of treated mice (n = 5 mice). F Histological images of H&E-stained sections (scale bar, 200 μm) (black arrows: intact crypt structure, yellow arrows: mucosal epithelium cell shedding, blue arrows: inflammatory cell infiltration, and red arrows: loss of crypt and goblet cells). G CD86 staining (M1-type macrophages) and CD206 staining (M2-type macrophages) on the colons of treated mice. The scale bar, 200 μm. H, I The level of MDA and the activity of SOD in the collected serum of treated mice (n = 5 mice). J DHE staining on the colons of treated mice. The scale bar, 200 μm. K–M Levels of TNF-α, IL-6, and IL-1β in colonic tissues isolated from different groups (n = 5 mice). N Immunofluorescence images of Occludin and ZO-1-stained colon tissues. The scale bar, 200 μm. All values were presented as means ± SD. All statistical analyses were conducted by using one-way ANOVA.

Modulation of macrophage polarization can ameliorate symptoms associated with colitis¹². Therefore, whether EcN@PDA@HP could modulate immune homeostasis and exert anti-inflammatory effects by influencing macrophage polarization was investigated. Immunofluorescence staining was performed to detect M1 macrophages expressing CD86 and M2 macrophages expressing CD206, respectively (Fig. 5G and Supplementary Fig. 17F, G). Abundant CD86⁺ macrophages (M1 phenotype) were observed in the colonic epithelium of colitis mice, while fewer CD206⁺ macrophages (M2 phenotype) infiltrated the impaired colonic epithelium. In contrast, EcN@PDA@HP treated mice showed a marked decrease in CD86⁺ macrophages in the inflamed colon. Single-cell suspensions obtained from the colon tissues were further assessed for quantification on M2 and M1 macrophage populations using flow cytometry. As depicted in Supplementary Fig. 17D, E, EcN@PDA@HP treatment significantly increased the proportion of M2-type macrophages (F4/80⁺CD11b⁺CD206⁺ cells) by approximately 4-fold, as well as decreased M1-macrophages (F4/80⁺CD11b⁺CD86⁺ cells), when compared to that of the untreated colitis mice, indicating that EcN@PDA@HP promoted polarization of macrophage phenotype from M1 to M2. Malondialdehyde (MDA) and superoxide dismutase (SOD) are crucial indicators for assessing lipid oxidative damage^12,43. As shown in Fig. 5H, I, the level of MDA in the colon tissue of the colitis mice was notably high, while the activity of SOD was severely inhibited, suggesting that the antioxidant ability of the colon tissue in colitis mice was reduced and the oxidative damage was severe. In comparison to EcN@PDA, HP and EcN@PDA + HP treated groups, the MDA content of EcN@PDA@HP treated group was significantly lower, while the SOD activity was significantly enhanced. To evaluate oxidative stress and antioxidant response in colitis, we first employed dihydroethidium (DHE) fluorescence probe to detect colonic ROS levels. As shown in Fig. 5J and Supplementary Fig. 17B, EcN@PDA@HP-treated mice exhibited significantly attenuated red fluorescence intensity compared to other groups, demonstrating potent ROS scavenging capability of EcN@PDA@HP. Mechanistically, immunofluorescence co-staining of Nrf2 and HO-1 revealed that this antioxidative effect was mediated by activation of the Nrf2/HO-1 signaling axis. EcN@PDA@HP treatment markedly upregulated nuclear translocation of Nrf2 (Supplementary Fig. 17H, I) and subsequent HO-1expression (Supplementary Fig. 17J, K). We further demonstrated apoptosis rate in the colon tissues using deoxyribonucleotide terminal transferase (TdT) dUTP nick end labeling (TUNEL) staining. As depicted in Supplementary Fig. 17L, M, the green fluorescence intensity of TUNEL-stained colon section was obviously decreased in the EcN@PDA@HP-treated group, indicating a notable reduction in apoptosis rate, in comparison to other treatment groups. Finally, EcN@PDA@HP treatment obviously reduced the level of local pro-inflammatory factors, such as tumor necrosis factor-α (TNF-α), interleukin-6 (IL-6) and interleukin-1β (IL-1β) (Fig. 5K–M). Based on these findings, we further explored the effects of EcN@PDA@HP on inflammatory colonic epithelial cells. Tight junction-associated proteins, such as ZO-1 and Occludin, play pivotal roles in preserving the structure and functionality of the intestinal epithelium as well as maintaining intestinal barrier integrity⁴³. Immunofluorescence analysis demonstrated that EcN@PDA@HP effectively rescued colitis-induced depletion of these junctional proteins, thereby restoring epithelial barrier integrity in acute colitis models (Fig. 5N and Supplementary Fig. 17C). These results indicate that oral administration of EcN@PDA@HP significantly ameliorated IBD symptoms in DSS-induced acute colitis models in mice.

The pathogenesis of colitis is intricately linked to the dysbiosis of gut commensal microbiota. To evaluate the regulatory effects of EcN@PDA@HP on intestinal microbiota, we analyzed the composition of the gut flora using 16S rRNA gene amplicon sequencing of fecal samples. Nonmetric multidimensional scaling (NMDS) plots revealed a distinct separation between the microbiota profiles of DSS-induced colitis mice and healthy controls. Notably, the intestinal microbiota profile of EcN@PDA@HP-treated mice closely resembled that of normal, healthy mice (Fig. 6A). The hierarchical clustering tree revealed similar findings, with EcN@PDA@HP exhibiting the closest proximity to the healthy mice group at the Amplicon Sequence Variants (ASV) level (Fig. 6B). This suggested a smaller disparity in community composition between these two groups. Quantification of microbial dysbiosis index (MDI) demonstrated obvious elevation in DSS-induced colitis mice models, indicative of severe ecological imbalance, whereas EcN@PDA@HP intervention achieved the most pronounced MDI reduction among all treatment cohorts (Fig. 6C). Taxonomic profiling at the phylum level uncovered striking compositional parallels between EcN@PDA@HP-treated and healthy groups (Fig. 6D), with particular enhancement of Bacteroidota–a phylum associated with anti-inflammatory polysaccharide metabolism (Fig. 6E). Concurrently, EcN@PDA@HP effectively suppressed the expansion of Proteobacteria, a recognized microbial hallmark of intestinal inflammation and IBD progression¹²(Fig. 6F). Genus-level heat map analysis further characterized the microbiota-modulating effects of EcN@PDA@HP (Fig. 6G). Treatment with EcN@PDA@HP significantly increased the relative abundance of norank_f__Muribaculaceae (Fig. 6H), a commensal genus linked to mucosal immunity and metabolic homeostasis, and markedly reduced colonization by inflammation-associated taxa including Escherichia-Shigella and Enterococcus–both recognized contributors to intestinal barrier dysfunction and pro-inflammatory signaling (Fig. 6I, J). This targeted modulation of microbial ecology positions EcN@PDA@HP as a therapeutic candidate for restoring colonic microbial balance.

Fig. 6. EcN@PDA@HP recovered the intestinal flora homeostasis in mice with DSS-induced acute colitis.

A NMDS score plot illustrated the gut microbiome β-diversity. B Hierarchical clustering tree of ASV level among the various groups. C Microbial dysbiosis index among the different formulations (n = 5 independent samples). D Heat map (columns) of relative abundance of taxon (rows) at the phylum level for each group. The relative abundance of gut microbiota is expressed as relative percentages. E, F The relative phylum abundance of Bacteroidota and Proteobacteria (n = 5 independent samples). G The histogram of genus abundance among groups. H–J The genus abundance of norank_f__Muribaculaceae, Escherichia-Shigella, and Enterococcus after different administrations (n = 5 independent samples). In the box plots, the middle lines indicate the median (central line). The boxes show the 10th–90th percentile with the median, and whiskers show the minimum–maximum. All values were presented as mean ± SD. Significance was assessed using one-way ANOVA.

Preventative efficacy of EcN@PDA@HP against mouse colitis

Preventing the onset and progression of IBD holds significant importance in reducing medical and economic burdens, avoiding severe complications, and decreasing the risk of associated cancers. To evaluate the preventive potential of EcN@PDA@HP against DSS-induced colitis, we performed longitudinal gut microbiota analysis following 48-h pretreatment. Comparative assessment of α-diversity indices (observed species, Chao, and Shannon) revealed comparable microbial diversity and abundance between EcN@PDA@HP-pretreated mice and the healthy control group (Supplementary Fig. 18A–C), indicating preserved baseline microbiome homeostasis. Core microbiome analysis via Venn diagrams further demonstrated 75.93% taxonomic overlap between groups (Supplementary Fig. 18D), suggesting prophylactic maintenance of commensal flora composition. Notably, despite comparable diversity metrics, EcN@PDA@HP pretreatment significantly elevated the Gut Microbiome Health Index (GMHI) (Supplementary Fig. 18E) and selectively enriched the relative abundance of Bifidobacterium (EcN@PDA@HP vs Normal group, P = 0.0221), a keystone genus associated with anti-inflammatory activity (Supplementary Fig. 18F). We next examined the preventative efficacy of EcN@PDA@HP in a DSS-induced acute colitis model, with the schematic procedure shown in Supplementary Fig. 19A. The typical symptoms of colitis, body weight loss, elevated DAI, hematochezia, and reduced colon length, were evident in DSS-treated mice (Supplementary Fig. 19B–F). In contrast, mice treated with EcN@PDA@HP showed the most significant recovery, including restored body weight, reduced DAI, and preserved colon length, highlighting the treatment’s preventative efficacy. Histological analysis with H&E staining further confirmed severe colonic structural damage and inflammatory infiltration in the DSS model group, whereas EcN@PDA@HP-treated mice displayed intact tissue structure with no evident inflammation (Supplementary Fig. 19G). Additionally, the EcN@PDA@HP group demonstrated a significant reduction in malondialdehyde (MDA) levels and enhanced superoxide dismutase (SOD) activity compared to other treatment groups (Supplementary Fig. 19H, I). Moreover, cytokine analysis revealed that myeloperoxidase, TNF-α, IL-6, and IL-1β levels in the colonic tissue lysates of EcN@PDA@HP-treated mice remained within normal ranges, even under DSS-induced conditions (Supplementary Fig. 19J–L). These findings underscore the potential of EcN@PDA@HP as an effective strategy for colitis prevention.

Treatment of STm-induced colitis mouse model by EcN@PDA@HP

Given the inherent ability of EcN to suppress S. typhimurium (STm), we evaluated the therapeutic potential of EcN@PDA@HP in a murine model of STm-induced colitis. In addition to chemical-induced colitis, bacteria such as Salmonella typhimurium can colonize the gut, invade intestinal tissues, and cause enterocolitis. Given the high infection rate of Salmonella typhimurium (STm), we assessed the therapeutic potential of EcN@PDA@HP in a murine model of STm-induced colitis⁴⁴. An overview of the experimental procedure is presented in Supplementary Fig. 20A. Mice were pre-treated with streptomycin 1 day prior to oral administration of 1 × 10⁷ CFUs of STm to induce the colitis model. The body weight in STm-infected group decreased dramatically (Supplementary Fig. 20B), while the groups treated with EcN@PDA, HP, and EcN@PDA + HP showed various responses. As excepted, the body weight of the Normal group increased and remained stable, while the EcN@PDA@HP treatment group showed minimal change in body weight. Additionally, EcN@PDA@HP treatment led to the most pronounced recovery, maintaining normal colon length (Supplementary Fig. 20C, D). The STm number in GI tract is a critical indicator for intestinal infection, which is the direct cause of enteritis. As shown in Supplementary Fig. 20F, G, the number of STm in Model group increased dramatically in comparison to that of the healthy group and mice administrated with EcN@PDA@HP dramatically reduced colonization of STm in comparison with the other formulations. H&E staining results demonstrated that EcN@PDA@HP exhibited minimal epithelial damage, inflammatory cell infiltration, and architectural alteration in colonic tissue (Supplementary Fig. 20E). Subsequently, we examined the impact of EcN@PDA@HP on colonic epithelial cells and the compromised integrity of colonic epithelial barrier. According to the immunofluorescence staining results (Supplementary Fig. 20H–J), STm treatment markedly reduced the expression of ZO-1 and Occludin in colon tissue. In comparison to EcN@PDA, HP, and EcN@PDA + HP treatments, EcN@PDA@HP treatment effectively upregulated the expression of ZO-1 and Occludin. Additionally, to further validate the potential of astaxanthin in HP for ameliorating oxidative stress associated with intestinal inflammation, we assessed the levels of MDA and SOD in the blood of mice. EcN@PDA@HP group showed a significant decrease in MDA content, along with a significant increase in SOD activity (Supplementary Fig. 20K, L). Finally, the levels of TNF-α, IL-6, and IL-1β in the colon tissue of mice treated with STm were remarkably upregulated, while EcN@PDA@HP treatment obviously decreased the levels of these proinflammatory cytokines (Supplementary Fig. 20M–O). Taken together, EcN@PDA@HP robots effectively alleviated STm-induced colitis by modulating the gut bacterial composition and scavenging ROS.

Efficacy of EcN@PDA@HP in the treatment of DSS induced chronic colitis

Given the prevalence of chronic colitis in clinical settings, this study also evaluated the therapeutic efficacy of EcN@PDA@HP in chronic colitis, aiming to enhance its clinical reference value⁹. An overview of the experimental procedure is presented in Fig. 7A. The body weights and DAI scores of chronic colitis mice were evaluated after treatment with various therapeutic formulations. EcN@PDA@HP demonstrated significant amelioration in weight loss (Fig. 7B) and a reduction in DAI scores in mice with chronic colitis (Fig. 7E), highlighting the potential therapeutic efficacy of this intervention. After undergoing various treatments, mice in the Model, EcN@PDA, HP, and EcN@PDA + HP groups exhibited varying degrees of hematochezia. In contrast, EcN@PDA@HP treated mice showed no obvious hematochezia at the study endpoint (Fig. 7C). The colons of treated mice were collected, photographed and measured on day 29. As shown in Fig. 7D, G, the colon length of EcN@PDA@HP treated group was longer than that of other groups, indicating that EcN@PDA@HP inhibited typical colon shortening in chronic colitis model. Histopathological assessment via H&E staining delineated the therapeutic efficacy of EcN@PDA@HP in mitigating colonic inflammation. The DSS-induced model group manifested hallmarks of pathological features, including architectural crypt distortion, substantial loss of mucin-secreting goblet cells, and prominent infiltration of polymorphonuclear leukocytes within the mucosal layer. Conversely, EcN@PDA@HP intervention preserved crypt-villus structural integrity, attenuated inflammatory cell recruitment, and facilitated goblet cell repopulation, collectively demonstrating restoration of mucosal homeostasis and suppression of inflammatory progression (Fig. 7F). Furthermore, the MDA level in the EcN@PDA@HP group was significantly lower, and the SOD level was significantly higher, indicating a reduction in oxidative damage in chronic inflammatory colon (Fig. 7H, I). We further assessed apoptosis in colon tissues by employing TUNEL staining. As shown in Fig. 7M, N, the TUNEL fluorescence intensity (green) was significantly diminished in the EcN@PDA@HP greated group, indicating a marked reduction in apoptosis in the colonic region associated with colitis. Quantitative analysis of colonic inflammatory mediators demonstrated significant downregulation of proinflammatory cytokines (TNF-α, IL-6, IL-1β) in the EcN@PDA@HP-treated group compared to other controls (Fig. 7J–L). Besides, immunofluorescence evaluation of intestinal barrier components showed marked upregulation in the expression and distribution patterns of tight junction proteins ZO-1 and Occludin, key molecular determinants of epithelial integrity (Fig. 7O–Q). Taken together, these findings suggest that EcN@PDA@HP offers a promising therapeutic approach for preventing and treating DSS-induced chronic and acute colitis, as well as STm-induced colitis.

Fig. 7. Efficacy of EcN@PDA@HP in the treatment of chronic colitis.

A Schematic illustration on the treatment profile of DSS-induced chronic colitis mice. B Curve chart variations of murine body weight across time among the six groups, normalized to the percentage of day 0 body weight (n = 5 mice). C Representative photographs of mice anus showing diarrhea and stool bleeding in different groups. D Photographs of murine colons collected from treated mice. E Curve chart analysis of DAI scores among treated mice (n = 5 mice). F Histological images of H&E-stained sections (black arrows: intact crypt structure, yellow arrows: mucosal epithelium cell shedding, blue arrows: inflammatory cell infiltration, and red arrows: loss of crypt and goblet cells) (The scale bar, 200 μm). G Histogram analysis of colon length among different formulations (n = 5 mice). H, I The level of MDA and the activity of SOD in the collected serum (n = 3 mice). J–L Levels of TNF-α, IL-6, and IL-1β in colonic tissues isolated from different groups (n = 5 mice). M, N Fluorescence photos of Apo-BrdU expression in colon tissues and quantitative analysis of Tunel in colon tissues (n = 3 mice). The scale bar, 200 μm. O–Q Immunofluorescence images of Occludin and ZO-1 in colon tissues and quantitative analysis of Occludin and ZO-1 in colon tissues (n = 3 mice). The scale bar, 200 μm. All values were presented as means ± SD. All statistical analyses were conducted using one-way ANOVA.

Furthermore, the in vivo safety profile of EcN@PDA@HP in mice with chronic colitis was evaluated after oral administration. The results of physicochemical serum parameters in treated mice showed no significant differences to the normal mice in enzyme activities, including lactate dehydrogenase, creatine kinase, creatine kinase, blood urea nitrogen, aspartate aminotransferase, and alanine aminotransferase (Supplementary Fig. 21A). Histological evaluation of major organs (heart, liver, kidneys, lungs, and spleen) confirmed that no obvious cell damage or pathological abnormalities were observed in EcN@PDA@HP treated mice ((Supplementary Fig. 21B). Long-term safety assessment through a 30-day oral administration study in healthy mice demonstrated the favorable biosafety profile of EcN@PDA@HP. Histopathological evaluation via H&E staining revealed little pathological alterations in major organs, including the heart, liver, spleen, lungs, kidneys, and colon, confirming systemic biocompatibility (Supplementary Fig. 22).

Discussion

Patients with IBD experience significantly reduced quality of life and an elevated risk of colon cancer. Rapid clearance of ROS and restoration of gut microbiota balance are essential for effective IBD treatment. However, traditional bacterial or small molecule delivery systems lack autonomous mobility, rendering them vulnerable to gastric acid degradation and non-specific adsorption by intestinal contents and mucus, which limits their therapeutic impact in managing intestinal inflammation^6,30,39,45. In response, we have developed an oral, robust, biohybrid robotic platform equipped with an intelligent brake system, combining stress-trained Haematococcus pluvialis (HP) with probiotics to enable rapid passage through the harsh stomach and enhance targeted adhesion in the intestines, ROS scavenging, and microbiota modulation for improved IBD management. A key finding of our research is that stress training HP in adverse conditions leads to the biosynthesis of antioxidative astaxanthin (AST) and a thickened, acid-resistant cell wall, while maintaining strong motility. This allows both HP and its therapeutic cargo to remain active and traverse the harsh gastric acid barrier rapidly. In vitro motility assays show that trained HP and hybrid microalgae robots (EcN@PDA@HP) retain mobility, while untrained HP loses activity in simulated gastric fluid. Additionally, in vivo biodistribution studies confirm that both trained HP and the hybrid robots efficiently cross the gastric acid barrier post-oral administration. The hybrid robots also exhibit potent ROS-scavenging activity, attributed to the biosynthesized AST, which protects colonic epithelial cells from oxidative stress-induced damage.

Compared to previous studies that used live algae as intestinal drug carriers, the strong motility of algae alone often leads to shorter retention times at sites of intestinal lesions^15,16,20. In contrast, our algae-based microrobots achieve superior intestinal retention due to the inflammatory adhesion properties of EcN@PDA. This design makes these microalgal robots an ideal choice for targeted delivery of therapeutic agents in the inflammatory intestines. In vitro and in vivo adhesion studies confirm that host-guest conjugation between EcN@PDA and HP significantly enhances the residence time of hybrid robots in inflamed intestines. In murine models of DSS-induced acute colitis (both preventative and delayed treatment), chronic colitis, and STm-induced colitis, the biohybrid robots exhibited prolonged retention at colitis sites. This enables sustained release of AST and EcN, leading to effective ROS scavenging, modulation of microbiota homeostasis, reduction of intestinal inflammation, macrophage reprogramming, and restoration of the intestinal barrier. The microbiota-restoring capability of EcN@PDA@HP highlights its potential as a universal therapeutic strategy for inflammatory bowel disorders. The dual-action mechanism combining ROS scavenging and microbiota modulation may transcend etiology-specific limitations, positioning this strategy as a broadly applicable anti-inflammatory approach across diverse intestinal pathologies. Over a 30-day period of daily oral administration in colitis and healthy mice, no adverse effects on blood biochemistry or major organs were observed, confirming the excellent biosafety profile of the formulation. This combination of biosynthetic and intelligent motility, as well as braking functionalities in microrobots presents a versatile approach for IBD treatment, positioning these microrobots as a potential oral formulation for clinical translation.

The notable performance of our oral hybrid microalgae robot system not only makes it effective in alleviating IBD symptoms but also highlights its potential as a treatment for other GI diseases, such as bacterial infections, irritable bowel syndrome, and colon cancer, due to its excellent biocompatibility, intelligent motility, and targeted residence. Looking forward, the application of biohybrid microrobots based on green algae for human treatment offers exciting prospects but also presents challenges. First, it is crucial to minimize contamination risks, rigorously monitor the stability of algae and probiotics-based products, and ensure batch-to-batch consistency during manufacture, storage and transportation. Additionally, exploring alternative methods to enhance EcN-microalgae integration, such as covalent conjugation or electrostatic adsorption, could potentially facilitate preparation process. To standardize hybrid microalgae robot production, quality control measures should include monitoring the algae’s stress-training process, AST content, and probiotic quantity. Beyond batch-to-batch variability, process economics and scalability are key translational challenges: current laboratory-scale production features high labor and equipment costs with limited output, insufficient to meet commercial demands. To address this, it is necessary to optimize low-cost renewable culture media, design high-efficiency photobioreactors, and integrate automated control systems to cut manual input and boost yield. Meanwhile, regulatory hurdles also persist: novel biohybrid robots lack clear regulatory guidelines alongside the need for comprehensive long-term safety evaluations. Therefore, future research should focus on developing real-time integration monitoring systems, modifying microrobot surfaces with biocompatible materials to enhance adaptability and storage stability, collaborating with regulators to establish evaluation standards, and partnering with industry to optimize large-scale manufacturing for potential clinical translation.

Overall, our research introduces oral hybrid microalgae robots (EcN@PDA@HP) with unique mobility and braking capabilities specifically tailored to treat IBD. This multifaceted approach not only demonstrates its potential for clinical translation but also paves the way for further research and development in the field.

Methods

Animals

Male BALB/c mice (22–25 g) were obtained from SPF (Beijing) Biotechnology Co., Ltd (Beijing, China). Mice were housed under specific pathogen-free (SPF) conditions, with a 12 h light/dark cycle (lights on at 7:00 a.m.), a controlled temperature of 25 °C, and a relative humidity of 50–60%. All animals had ad libitum access to autoclaved drinking water and a standard rodent chow (catalog no.: SPF-F01-002) and this chow contained approximately 18% crude protein, 5% crude fat, and 5% crude fiber. Terminal anesthesia was induced with intraperitoneal pentobarbita. Animals were then euthanized via exsanguination followed by CO₂ inhalation.

Bacterial strains and culture conditions

Escherichia coli Nissle 1917 (EcN) (Catalog number: B71907) was provided by Mingzhou Biotechnology Co., Ltd (Ningbo, China). EcN was cultured aerobically at 37 °C with constant shaking at 200 rpm in Lysogeny Broth (LB) medium containing 10 g/L tryptone, 5 g/L yeast extract, and 10 g/L NaCl, purchased from Solarbio (Beijing, China).

Materials

Haematococcus pluvialis and BBM (Bold Basal Medium) were purchased from Shanghai Guangyu Biological Technology Co., Ltd. β-CD-PEG2000-NHS, ADA-PEG2000-NHS, ADA-FITC, and β-CD-Cy5 were purchased from Xi’an Ruixi Biological Technology Co., Ltd (China). Dopamine was purchased from Aladdin (Shanghai, China). Dihydroethidium (DHE) probe was provided by US EVERBRIGHT Biotechnology Co., Ltd (Suzhou, China). ROS Green^TM H₂O₂ Probe was purchased from MKBio Co., Ltd. (Shanghai, China). SYTOX Green Dead nucleic acid stain was purchased from Tianjin Alphabio Co., Ltd (China). Interleukin (IL)-6 kit and tumor necrosis factor-α (TNF-α) kit for mouse were obtained from 4 A Biotech (Suzhou, China). Interleukin (IL)-1β kit for mouse was purchased from Elabscience Biotechnology Co., Ltd (Wuhan, China). The following primary antibodies were commercially sourced: anti-CD86 from Boster Biological Technology (Wuhan, China), anti-CD206, anti-HO-1, anti-ZO-1, and anti-Occludin from Protein tech (Wuhan, China), Anti-Nrf2 from Abcam (UK). Dextran sodium sulfate (DSS; molecular weight, 36–50 kDa) was purchased from MP Biomedicals (USA). NCM 460 cells was purchased from ATCC (USA).

Preparation of EcN@PDA and ADA group decoration of bacteria

EcN@PDA was prepared by shaking the bacteria in Tris-HCl buffer containing dopamine^37,45. Briefly, EcN and EcN (GFP labeled) were cultured in LB medium containing kanamycin for 8 h and then washed with PBS. Afterward, 2 × 10⁹ CFU of EcN were added into 2 ml of Tris-HCl buffer containing 1 mg/ml of dopamine and shaken for 30 min at room temperature. EcN@PDA were collected by centrifugation (4000 × g), purified by washing with PBS for 3 times and then suspended in PBS. ADA-PEG2000-NHS was then added into the suspension to a final concentration of 100 μM. After incubation for 2 h, the bacteria were centrifuged at a speed of 4000 × g for 3 min and then resuspended in PBS. To demonstrate the successful modification of EcN@PDA by ADA group, β-CD-Cy5 was added, and the EcN@PDA were incubated for 5 min at room temperature. After following washes to remove unbound β-CD-Cy5, fluorescence imaging experiments were performed on by confocal laser scanning microscopy (CLSM, Leica TCS SP8, German).

Biosynthesis of HP and β-CD group decoration of HP

Without training original HP (denoted“OHP”) were grown photo autotrophically in liquid BBM (Bold Basal Medium) at 30 μmol m⁻² s⁻¹ (side, white light) for 10 days²². The OHP biosynthesized astaxanthin is produced under high light intensity, nitrogen deficiency and weak acid conditions^21–23. Briefly, OHP were treated under nitrogen starvation, continuous light (150 μmol m⁻² s⁻¹) and weak acid (pH 4) for 6 days. HP were collected by centrifugation (800 × g), purified by washing with PBS for 3 times and then suspended in PBS. β-CD-PEG2000-NHS was then added into the suspension to a final concentration of 100 μM. After incubation for 2 h, the HP were centrifuged at a speed of 800 × g for 3 min and then resuspended in PBS. Similarly, ADA-FITC was added and the HP were incubated for 5 min to prove the successful modification β-CD group on HP and fluorescence image was performed by CLSM.

Construction of EcN@PDA@HP

To carry EcN on the surface of HP, 1 × 10⁸ CFU ADA modified EcN@PDA were added to PBS with CD modified HP and cocultured for different time points (0.5, 1, 2 h)⁴⁶. After co-incubation at room temperature, the PBS was removed and HP were rinsed with PBS three times slightly and centrifuged at 800 × g for 3 min to wash the unattached bacteria.

Characterization of EcN@PDA@HP

To perform SEM characterization, EcN@PDA@HP were first fixed with 2.5% glutaraldehyde overnight at 4 °C and then washed in ultrapure water. After overnight drying, EcN@PDA@HP were coated with palladium for SEM characterization using an acceleration voltage of 3 kV (ThermoFisher SEM instrument). Bare algae and other control group were treated and examined using the same methodology. Subsequently, we calculated the mean fluorescent intensity (MFI) of EcN@PDA (GFP) and HP (Cy5) using FCM. To evaluate the stability and viability of EcN@PDA@HP under GI conditions, samples were incubated in SGF and SIF at predetermined time points. Test the stability of the EcN@PDA@HP robot directly using FCM. EcN@PDA@HP were stained with a SYTOX Green Dead nucleic acid Kit for testing viability, followed by FCM using 488 nm excitation. To further quantify the number of bacteria on HP, the amount of HP in EcN@PDA@HP was quantified by cell counter (RWD). Following dilution, 100 μL of EcN@PDA@HP was spread on solid agar plates, and the bacterial count was determined by counting the colonies after 24 h of incubation in microbiological incubators.

Growth curves and in vitro resistance analysis of EcN@PDA@HP

The EcN@PDA@HP were incubated in a bed temperature incubator at 37 °C with gentle shaking. The OD values of the culture media in 96-well plates were recorded at 600 nm for 2 h intervals by a microplate reader. Bare bacteria served as a control. EcN, EcN@PDA, and EcN@PDA@HP (1 × 10⁸ CFU) were resuspended into the 1 mL of SGF (pH 3.5 with pepsin). At specific time points, 100 μL of each sample was taken, diluted with PBS, and spread on LB solid plates. Following the same procedure, the same amounts of EcN, EcN@PDA, and EcN@PDA@HP (1 × 10⁸ CFU) were resuspended into 1 mL of SGF (pH 6.8 with trypsin). One hundred microliters of each sample was taken at a predetermined time point, centrifuged to obtain precipitation, and washed with PBS. The obtained bacterial suspension was coated on solid LB plates after a serial dilution. The colonies were counted after incubation at 37 °C overnight.

Motion analysis of EcN@PDA@HP

The motion of HP, OHP, and EcN@PDA@HP motors was analyzed in SGF and SIF. The Dmi8 (Leica) was used to record the motion behavior of motors for 10 s after incubation in SGF or SIF at a specific time points. The motion trajectories of the HP, OHP, and EcN@PDA@HP were recorded. Subsequently, the tracking image sequences and the speed of the nanomotors were analyzed by ImageJ plugin manual tracking according to the previous reports.

ROS scavenging capacity

The ROS scavenging capacity of EcN@PDA@HP was assessed using H₂O₂, •O₂⁻, and •OH as representative ROS. The H₂O₂ scavenging capacity was measured with an H₂O₂ assay kit (Elabscience, China). The 405 nm absorbance was measured. The ability to scavenge •O₂⁻ was evaluated by superoxide anion scavenging capacity assay kit (Solarbio, China). In addition, the •OH scavenging ability of EcN@PDA@HP was determined by hydroxyl free radical scavenging capacity assay kit (Elabscience, China). The full wavenumber scanning curve of the solution was detected, and the absorbance of the solution at 510 nm was recorded. Intracellular ROS levels of NCM 460 cells were analyzed with the DHE and H₂O₂ probe. Cells were incubated with H₂O₂ (300 μM) and samples for 24 h. The untreated cells served as the control. Next, the cells were incubated with probe, and then washed with PBS. At last, ROS was measured by FCM and the flow data were analyzed by FlowJo software.

Evaluation the mucoadhesive (“braking”) ability of EcN@PDA@HP in vitro and in vivo

For in vitro adhesion analysis, freshly separated mouse intestines were washed and sliced⁴⁷. HP, EcN@PDA + HP (physical mixture of EcN@PDA and HP), and EcN@PDA@HP were incubated with mouse intestines at 37 °C for 1 h, then the intestines were cleaned with PBS 3 times, and IVIS (PerkinElmer, USA) was used for imaging. Next, to evaluate the ability of EcN@PDA@HP to overcome the gastric acid barrier in mice, male BALB/c mice (8 weeks, n = 3) were orally administered with OHP, HP, and EcN@PDA@HP and sacrificed after 1 h. The stomach, small intestine, and large intestine were collected and imaged.

To further evaluate the in vivo mucosal adhesion effect of HP, EcN@PDA + HP (EcN, expressing GFP) and EcN@PDA@HP (EcN, expressing GFP) were orally administered to male BALB/c mice, and the IVIS was used to track the distribution of EcN@PDA@HP at each predetermined time point. Mice were sacrificed, and major tissues (whole GI tract, five organs) were collected and imaged. Mice were imaged for GI tract using the Cy5 channel (HP) and the FITC channel (GFP), respectively. Afterward, frozen sections of colon tissue were made to observe the adhesive ability of GFP-labeled-EcN using CLSM.

Validation the ability of EcN@PDA@HP in the treatment of acute colitis

Male BALB/c mice aged 8 weeks were fed with 3% DSS for 5 days. Then the mice were randomly divided into different groups and administered with EcN@PDA, HP, EcN@PDA + HP, EcN@PDA@HP, (EcN, 10⁸ CFU/day), or PBS by oral gavage for 5 days. Daily changes in body weight were evaluated over a 10-day experimental period. Feces were collected for intestinal flora analysis on the final day, then mice were executed, and the blood and the entire colons were collected. Then, colon tissues were used for H&E staining, cytokine detecting, immunofluorescence staining, and ROS staining. The fecal samples from each mouse were gathered to extract total genomic DNA using the QIAamp-DNA Stool Mini Kit. The bacterial 16S rRNA gene (V3–V4 region) was amplified using the primer pairs 338F (5’-ACTCCTACGGGAGGCAGCAG-3’) and 806R (5’-GGACTACHVGGGTWTCTAAT-3’) in the PCR thermocycler. Microbiota composition was determined according to the standard protocols.

Validation the ability of EcN@PDA@HP in the preventative treatment of UC

Eight-week-old male BALB/c mice were housed in groups of five mice per cage and acclimatized for 1 week before inclusion in the study. To evaluate the potential preconditioning effects of EcN@PDA@HP on gut microbiota homeostasis, fecal samples were collected from two cohorts: (1) healthy mice receiving EcN@PDA@HP via oral gavage once daily for 2 days, and (2) age-matched untreated controls. The fecal samples from each mouse were gathered to extract total genomic DNA using the QIAamp-DNA Stool Mini Kit. In the preventative therapy of IBD, healthy mice were randomly divided into six groups (n = 5): (1) Normal group, (2) Model group, (3) EcN@PDA, (4) HP, (5) EcN@PDA + HP, and (6) EcN@PDA@HP. The mice received DSS on the second day, and drinking water with DSS was replaced with normal drinking water on the seventh day. The mice received drug treatment from day 0 to 7. Body weight, visible stool consistency, and fecal bleeding were evaluated were assessed daily overt the 10-day experimental period. The colon was collected and then gently rinsed with saline. A partial section of the colon was taken for H&E staining. For an inflammation level evaluation, the amounts of proinflammatory cytokines were quantified by a commercial ELISA kit.

Treatment of an STm-induced colitis mouse model

Male BALB/c mice aged 8 weeks were randomly divided into 6 groups (n = 5). Model and treatment groups were pretreated with 100 μL of streptomycin (200 mg/mL) 1 day before infection by oral administration of 1 × 10⁷ CFUs of STm. Since day 1, the mice were daily oral gavaged with PBS, EcN@PDA, HP, EcN@PDA + HP, and EcN@PDA@HP for 6 days. At day 7, the mice were euthanized for sample collection and analysis. Body weight was recorded daily and colon length was measured on day 7. Fecal samples were collected from each group, homogenized with PBS, and plate counting was conducted for STm. Colon tissues were used for H&E staining, cytokines detecting, immunofluorescence staining.

Validation the ability of EcN@PDA@HP in the treatment of chronic colitis

Chronic colitis was induced by repeated administration of 2% DSS solution and normal water, and treatment was given during normal drinking water for three cycles. Every day, the weight of the mice was recorded. On the final day of the experiment, the mice were humanely euthanized and the whole colon was extracted. The length of the colon was assessed, and it was delicately rinsed with PBS. Subsequently, a section of the intestine was utilized for immunofluorescence staining and histological assessment. The remaining colon tissues from the mice were used to quantify the levels of inflammatory cytokines. The physicochemical parameters of the serum and H&E staining of major internal organs were employed to assess the safety profile of EcN@PDA@HP.

Ethical statement

All animal experiments were performed according to the standards set forth by the Guide for the Care and Use of Laboratory Animals. The animal experiments were approved by the Animal Ethics Committee of the Chengdu University of Traditional Chinese Medicine (Permit NO. 2024011).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

The project was conceptually designed by R.W., C.G., and R.L. The majority of the experiments were conducted by R.L., assisted by J.L. and L.D. Data analysis and interpretation was done by R.L., J.L., L.D., Q.Z. and T.D. The manuscript was prepared by R.L., C.G., and R.W. All authors discussed the results and implications and commented on the manuscript.

Peer review

Peer review information

Nature Communications thanks Ryuichi Okamoto and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The authors declare that all data supporting the findings of this study are available within the article and the Supplementary Information/Source Data file. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-66692-x.