English

• Glioblastoma (GBM) is one of the most aggressive brain tumors for it causes heavy inflammation of its surrounding cells and leads to peritumoral brain edema (PTBE) [1]. GBM can create an immunosuppressive microenvironment and escape from T-cell recognition [2]. This immune-counter ability of GBM is believed to be associated with the accumulation of tryptophan in GBM cells [3], while the impacts of tryptophan or kynurenine on glioma cells and the other cerebral cells are different [4,5]. The tryptophan metabolic enzyme, indoleamine 2,3-dioxygenase (IDO), is expressed in over 90% of GBA cells but IDO is usually expressed in liver cells [6]. However, tryptophan in the human body is mainly derived from protein which is also digested by intestinal microbes, therefore, the abnormal metabolism of tryptophan in GBM induces over-generation of tryptophan in the intestine [7]. Since the significance of intestinal bacteria to human diseases has shifted the way of metabolism research, the concept of gut-brain axis (GBA) is much more concerned in lots of studies of cerebral physiology [8]. Tryptophan metabolism has become a central issue in models of GBA, in particular, the kynurenine pathway is critical for it is responsible for about 95% of tryptophan metabolism [9,10]. In detail, IDO catalyzes the conversion of tryptophan towards kynurenine, which is the very first reaction on the whole pathway [11]. Kynurenine is catabolized into neuroactive inflammatory mediators such as quinolinic acid [12]. As a result, IDO level or kynurenine accumulation both indicate the inflammation of the tumor microenvironment [13]. Based on the above facts, it is necessary to study the influence of tryptophan metabolism on the inflammatory responses of GBM cells from a GBA perspective.

  Researchers have proposed great examples via clinical tests and animal models [14–16]. However, establishing an in vitro disease model is inevitable in the future of pathological fields to avoid ethical problems [17]. As an in vitro model of GBA for metabolism study, it requires that this system has a similar arrangement of simulated gut/brain modules. Recent studies on the interaction between gut microbiota and cells as well as their associated impacts are beneficial for understanding some disease mechanisms [18–20]. The primary challenge of cells and bacteria co-culture model lies in their distinct growth cycles and culture conditions. Organoid-based Transwell method is a common option to realize the coexistence of intestinal cells and bacteria but it cannot recreate a GBA-like structure limited by the fixed device [21–23]. Based on the flexible design and operation, 3D bioprinting is free to create any desired structure while the morphogenesis of cells is restricted when encapsulated in biomaterials and it focuses less on metabolism recreating [24,25]. So far, the organ-on-chip (OOC) technique is a more potential solution for an OOC is easy to contain multiple tissues/organs by altering the chip design [26,27]. Some matured OOCs have been used for mimicking human intestine epithelium and studying the microbes in the intestine but these situations require long-term morphogenesis of Caco2 (derived from colon carcinoma and usually used as intestinal epithelial cells in simulation) [28–30]. The villi formation of Caco2 takes about 5-7 days under continuous flow in an aerobic device or within a Transwell [31]. This feature sometimes neglects the microbes and focuses less on metabolism analysis. Therefore, the use of alginate hydrogel microparticles (AHMPs) may alternatively meet these requirements to establish a GBA model. Because AHMPs are capable of carrying microbes and are more convenient to manipulate in a microchip. Several cases have shown that AHMPs can be used for recreating multicomponent systems such as tumor spheroids or mesenchymal-epithelial transition without affecting the metabolism [32–34]. The combination of AHMPs and OOCs is worthy of exploring to facilitate the co-culture of different cells to form the functions of GBA, and it is capable of including E. coli within the system. Furthermore, the combination strategy is suitable for the analysis of the morphology of various cells and the determination of metabolites separately.

  Therefore, we developed a multi-component microchip (MCM) that supported the co-culture of the intestinal epithelium and GBM cells to monitor the inflammatory responses of GBM cells, and the tryptophan influence on cell behavior was analyzed using this method. Caco2, U87 and E. coli are modeled as intestinal epithelium, GBM cells and intestine-colonized microbes, respectively. The edema of U87 cells could be observed after incubation, with corresponding alterations including depolarization, loss of cell area, and the decreasing F-actin level. Also, the increasing IDO levels detected by enzyme linked immunosorbent assay (ELISA) suggested the microenvironment emerged by pro-inflammatory factors. At the same time, key molecules from the tryptophan pathway were analyzed via LC-MS. Levels of tryptophan and kynurenine were increasing after the incubation, indicating E. coli was over-generating tryptophan and kynurenine molecules were in excess in the GBA microenvironment. Meanwhile, indole-3-propionic acid (IPA) is also accumulating. Thus, persistent accumulation of tryptophan indeed causes an inflammatory response of U87. Here, we presented the fabrication of this GBA-like MCM, the establishment of continuous co-culturing, the monitoring of cell morphogenesis, the analysis of tryptophan metabolism and how to discover novel results by using our method.

  The MCM has two chambers for cell seeding (Figs. S1a and b in Supporting information), which are connected by microchannels of 40 µm widths. Microchannels are responsible for barrier AHMPs (100-150 µm) without blocking the medium flow. Caco2 and U87 were located in each chamber on the microchip and E. coli was seeded in the same chamber of Caco2, forming a simplified structure of GBA. It is interesting to note that E. coli was loaded in the alginate hydrogel microparticles to make the operation more convenient in the MCM and avoid excessive reproduction. The fabrication of bacteria-laden AHMPs (E-AHMPs) were described in Figs. S1c and d (Supporting information). E. coli in AHMPs can digest protein derived from FBS in the medium to generate excess tryptophan near the microenvironment of the Caco2 layer in Ch1 (Fig. 1a). Tryptophan molecules were then transported via the continuous flow to the microenvironments of U87 in Ch2. Accumulating amounts of tryptophan were then catabolized by IDO in U87 to generate more kynurenines and indole-3-propionic acid (IPA) molecules. As a result, inflammatory responses of U87 mainly on cell morphological alterations like severe edema appeared. However, Caco2 cells were relatively insensitive to tryptophan for there was no significant difference was found neither at the molecular level nor in cell morphology.

  Figure 1

  

  Figure 1.  Illustration of the influence of tryptophan metabolism among cells and the manipulation for GBA simulation on MCM. (a) Three types of changes were noted in three colors: blue for chemical reactions; green for morphological responses of cells; and red for the direction of tryptophan transportation. (b) Seeding protocol: step 1, Caco2 and U87 were seeded in Ch1 and Ch2, respectively; step 2, delivery of E-AHMPs (E. coli) into Ch1 after cells adhesion; step 3, continuous flow by the direction of 3a to 3b while the other four inlets were blocked.

  DownLoad: Full-Size Img PowerPoint

  The GBA-like MCM was established by three steps described in Fig. 1b: Cell seeding, E-AHMPs injection and continuous culture setting. In detail, Caco2 cells were seeded in Ch1 and cultured statically for at least 24 h to promote their adhesion and basic differentiation. Then the continuous flow started after E-AHMPs were seeded (Fig. S2 in Supporting information). E-AHMPs were formed by flow-focusing of DP1 (E. coli in saline), DP2 (alginate with EDTA-Ca) and perpendicular CP1 (oil containing surfactant), which was demonstrated in Fig. S3 (Supporting information). Surfactants in CP1 can stabilize the oil-water surface of AHMPs. Hydrogen ions in CP2 are responsible for solidifying because that hydrogen ions induce the release of calcium ions in EDTA-Ca and free calcium ions can cross-link alginate. It was feasible to control the sizes of E-AHMPs and make their diameters much larger than the widths of microchannels linking Ch1 and Ch2. So, E-AHMPs were fix-located with Caco2 cells in Ch1 when continuous flow started. MCM was evaluated by the viability of cells (U87 and Caco2) and E. coli. In the MCM model, Caco2 cells had three stages of morphogenesis after cell seeding over time: island-shaped chunks, lamellar cell populations and epithelium-like layer (above the row in Fig. S4 in Supporting information). As for U87, glioblastomas appeared as time (lower row in Fig. S4).

  In the following step of cells and E. coli co-culturing, the time when continuous culture started was considered as t₀. As the viable/dead cells characterized by Calcien-AM/PI shown in Fig. 2a, Caco2 cells maintained the structure of epithelium-like monolayer at t₀ + 36 h and green fluorescence there was still uniformly distributed. However, U87 cells became morphologically spindle from astroglial at t₀+36 h. Cell viability (%) was defined as the ratio of (I_G - I_R)/I_R. Results in Fig. 2b and c presented good viability of both Caco2 and U87 and the average viability at t₀ + 36 h were about 85% (no significant difference shown compared to t₀ + 36 h and t₀ + 24 h), although, the viability decreased subtly. The viability of E. coli was characterized by SYTO9/PI. Similarly, corresponding fluorescence images were obtained and values were calculated by ImageJ. Results shown in Fig. 2d described the growth of E. coli colonies in an AHMP since green fluorescence intensity increased apparently. There was no dead E. coli from the signal of PI (red image). These results all demonstrated the adaptability of cells in MCM within 12 h and proved the compatibility of alginate hydrogel for bacteria incubation.

  Figure 2

  

  Figure 2.  Sustainability of cells and E-AHMPs in MCM. (a) Images of viable/dead cells characterized by Calcien-AM/PI at t₀ + 24 h and t₀ + 36 h. (b, c) Histograms of cell viability at t₀ + 24 h and t₀ + 36 h. Significance was calculated by unpaired t-test: P = 0.3629 for Caco2 and P = 0.6541 for U87, both non-significant results. (d) Images of viable E. coli in AHMPs characterized by SYTO9 and no PI signal were captured during the process (E. coli concentration of 1 × 10⁴ CFU/mL).

  DownLoad: Full-Size Img PowerPoint

  Some inflammatory responses, most notably the edema of U87 cells could be observed after incubation [35]. Phase contrast images in Fig. S5 (Supporting information) illustrated the differences in cell morphology before and after 12 h. In control groups (Caco2 + U87), epithelium-like monolayer with villi of Caco2 cells formed after 12 h and the tight junction of the Caco2 layer was enhanced as the vacuoles of the monolayer became less, meanwhile, the morphology of U87 stayed relatively stable. In the co-culture groups (Caco2 + U87 + E. coli), the epithelium-like structure of Caco2 remained relatively unchanged while the spindle-shaped U87 cells became more circular. An edematous U87 cell is typically swelled, depolarized, and adhesion-reduced. As a result, parameters such as polarization, cell area, and circularity are employed to characterize the extent of inflammation in these cells [35]. These phase contrast images were further analyzed via ImageJ to demonstrate morphological alterations. For statistical exhibiting these morphological alterations of cells, each U87 cell in the view was marked with its circumscribed ellipse, with the semimajor axis denoted as b and the semiminor axis denoted as a (Fig. 3a). Therefore, the polarization rate of the cell was able to be defined as the ratio of b to a, using to describe the stretch tendency of a cell; the roundness of the cell was defined as the ratio of cell area to the area of its smallest circumscribed circle (with radius equal to b), and used to describe the circular extent of the cell. As Fig. 3b described, the cell polarization rate of U87 stayed nearly constant in the control groups (Caco2 + U87 group, without E. coli), while the parameters in co-culture groups (Caco2 + U87 + E. coli) decreased dramatically. This indicated that U87 cells tended to become de-spindle when co-culture with germs. Results in Fig. 3c showed that areas of cells were relatively stable in both conditions, indicating that adhesion of U87 cells remained normal. However, statistics in Fig. 3d illustrated that cell circularity in control groups (Caco2 + U87) unchanged but increased significantly in the co-culture groups. Combining these above parameters, it could be determined that U87 cells maintained their adhesion abilities, showing a strong tendency to depolarize, which was consistent with the edematous appearances.

  Figure 3

  

  Figure 3.  Parameters describing the morphological alterations of U87 cells before and after the co-culture (control referred to Caco2 + U87, co-culture referred to Caco2 + U87 + E. coli). (a) Graph of U87 cells after being transformed into black and white 2D graphics in ImageJ software, where the yellow border in the right panel indicates the external ellipse of the cell, and a and b are the half-length and half-short axes of the external ellipse, respectively. (b-d) Polarization rate (P = 0.3233, P < 0.0001), cell area (P = 0.9763, P = 0.3328), and roundness (P = 0.5017, P = 0.0018) of U87 cells in the control groups versus the co-culture groups, respectively. (e) Fluorescence views of Caco2 and U87 with or without E-AHMPs. Nucleus stained with DAPI and F-actin labeled with ATTR. (f) Quantification of average F-actin level of Caco2/U87 cells. Significance was calculated by a two-way t-test (P = 0.0103 for U87 and P = 0.5041 for Caco2).

  DownLoad: Full-Size Img PowerPoint

  To further explore the edema mechanism, F-actins, as the primary components of the cytoskeleton, were characterized using an ATRR fluorescence probe. Abnormal cytoskeletal structures of edematous cells usually imply a loss of F-actin. Fluorescence images were obtained by combining the results of DAPI-visualized-nucleus and ATRR-visualized F-actins. Illustrated in Fig. 3e, cytoskeletal structures of Caco2 remained similar, however, levels of F-actin decreased according to the fluorescence view of U87 and the morphological shift from spindle to circular was also observed. Cell numbers were obtained by counting DAPI-labeled nucleus, the average F-actin level per cell could be calculated by the ratio of fluorescence intensity of F-actins to the cell number. Thus, statistical results in Fig. 3f confirmed that F-actin levels of U87 declined significantly when co-cultured with E. coli, nevertheless, the F-actin level of Caco2 nearly maintained at an average level. The decline of F-actin in U87 explained the morphological alterations mentioned in the previous discussion.

  With the morphological alterations of U87 cells, reactive oxygen species (ROS) also changed. ROS in each condition is characterized by DCFH-DA, as they are usually tested as the typical indicators in the inflammatory cell microenvironment. The group without Caco2 was considered to eliminate the potential pro-inflammatory effect of Caco2 on U87. The nucleus of U87 was labeled by Hoechst 33342. Illustrated in Fig. S6a (Supporting information), ROS intensity in the co-culture group (Caco2 + U87 + E. coli) was higher than that in the no E. coli group. Otherwise, U87 cells maintained the spindle structure in groups of no Caco2 and no E. coli, whereas a depolarization was clearly observed in a group of co-culture, corresponding to the F-actin results. Shown in Fig. S6b (Supporting information), the ROS level of each cell was obtained by dividing the fluorescence intensity in each view field by the cell number, and the results from the co-culture group were still the highest among the three conditions, which verified that there were severer inflammatory responses within the U87 microenvironment evenly. Notably, ROS levels in the co-culture group were higher than that in no Caco2 group, confirming that Caco2 cells were not inducing the inflammation, while Caco2 cells were playing a positive regulatory role in creating inflammatory microenvironment of U87 cells when E. coli existed.

  At the same time, the key molecules from the tryptophan pathway were analyzed via LC-MS. In general, kynurenine and IPA are both catabolites of tryptophan but generated in different in vivo locations and they kept the same metabolic manner in the co-culture of MCM (Fig. 4a). Kynurenine is believed to be pro-inflammatory while IPA is more considered to be anti-proinflammatory [36]. Levels of tryptophan, kynurenine and IPA in the medium collected from the MCM were shown in Fig. S7, Tables S1 and S2 (Supporting information). Levels of tryptophan and kynurenine were increasing after the incubation, indicating E. coli was over-generating tryptophan and kynurenine molecules were in excess in the GBA microenvironment. IPA is also accumulated, which is an anti-inflammatory factor [37]. The concentration values of each compound at t₀ + 24 h (c₀) and t₀ + 36 h (c_t) were first obtained and their difference value (∆c = c_t - c₀) was eventually used to describe the metabolic dynamics in the system. From Figs. 4b-d, when E. coli existed in the MCM, ∆c values of all three compounds were positive in both groups (Caco2 + U87 + E. coli and U87 + E. coli). However, all ∆c values were negative when E. coli were not co-cultured, namely, Caco2 and U87 were incubated without E. coli. Compared to the situation when U87 and E. coli were incubated without Caco2 cells, ∆c values of these compounds were higher when Caco2, U87 and E. coli were together incubated. Conclusions could be drawn based on the above analysis: (1) E. coli can stimulate the accumulation of tryptophan, kynurenine and IPA in the MCM; (2) Under the condition where E. coli were not present, these three compounds would be consumed over time.

  Figure 4

  

  Figure 4.  Tryptophan metabolism results analyzed by LC-MS. (a) Two metabolic pathways of excess tryptophan produced by E. coli in AHMPs. (b) ∆c(Trp) values from left to right are 26.62, 17.32 and 20.23 µmol/L. (c) ∆c(Kyn) values from left to right are 0.40, 0.27 and 0.21 µmol/L. (d) ∆c(IPA) values from left to right are 1.26, 0.19 and 1.27 µmol/L.

  DownLoad: Full-Size Img PowerPoint

  Moreover, the IDO levels in U87 and Caco2 were obtained to analyze the activity degree in the tryptophan-kynurenine reaction. As shown in Fig. S8a (Supporting information), IDO levels in U87 in each situation were 64.87, 56.34 and 33.85 ng/mL from left to right. Similarly, IDO levels of Caco2 were 20.32 and 14.31 ng/mL in Fig. S8b (Supporting information). These results matched ∆c values in the corresponding conditions (Figs. 4b–d). So, conclusions could be drawn that: (1) IDO activity was higher than any other when Caco2, U87 and E. coli were together incubated. (2) When the IDO level was 33.85 ng/mL or less, the tryptophan-kynurenine reaction could not occur since the ∆c values were negative. (3) IDO levels in Caco2 could not support the occurrence of tryptophan-kynurenine reaction in Caco2 cells. Therefore, tryptophan accumulation could be regarded as another key factor of pro-inflammation since the IDO level indicates the inflammation of U87 cells. The accumulation of tryptophan caused an inflammatory response of U87 cell, which is contrary to the known that tryptophan is anti-immunity. This may be attributed that the excess tryptophan, aryl hydrocarbon receptors (AhRs) in target cells are over-activated and may finally result in activation impaired [38].

  Since the tryptophan-kynurenine pathway is closely related to inflammation in GBM, each level of tryptophan and kynurenine has been detected by LC-MS quantification. Theoretically, the tryptophan-kynurenine ratio (Trp/Kyn) indicates the activity of IDO and IDO is a typical pro-inflammatory factor, therefore, low Trp/Kyn implies inflammation. Surprisingly, the results of Trp/Kyn demonstrated in Table S3 (Supporting information) were completely contrary to the prediction. In detail, inflammatory responses were proven to be more severe in co-culture (Caco2 + U87 + E. coli) than in the control group (Caco2 + U87, without E. coli). But the Trp/Kyn ratio comparison of these two situations was opposite to the characterizations (Caco2 + U87 + E. coli: 50.47 > Caco2 + U87: 14.79, according to Table S3). It could be due to the continuous conversion of Trp into Kyn, thereby exacerbating Trp depletion.

  This MCM system has established an in vitro model for studying bacteria-to-cell interaction by recreating the GBA fundamentals. With this method, tryptophan transportation in GBA and the effects of tryptophan catabolites on GBM inflammation have been simulated. In this model, continuous flow can either simulate transportation directly in GBA or provide shear stress for the morphogenesis of epithelium as Caco2 grows. AHMPs are alternatively applied for carrying E. coli, which has solved the problem of bacteria colonization in a continuous flow of the microchip model. The tryptophan accumulation now can be connected to the inflammation of U87, and the edema of GBM cells is here successfully explained by analyzing the kynurenine pathway induced by E. coli. The proposed methods have great potential to construct a more complex structure like the multilayer tissue. Besides, the species richness of bacteria can also be improved due to physiology in GBA is regulated by multiple intestinal microbes. In summary, we have presented a novel method based on MCM including AHMPs and realized the GBA simulation, which system supports the research of short-term co-culture of cells with bacteria. More importantly, the phenomenon of tryptophan accumulation is newly found as an indicator of the inflammation of GBM cells.

  Declaration of competing interest

  The authors declare no conflict of interest.

  CRediT authorship contribution statement

  Gaowa Xing: Writing – review & editing, Writing – original draft, Funding acquisition, Conceptualization. Yuxuan Li: Writing – review & editing, Writing – original draft, Conceptualization. Hongren Yao: Writing – review & editing. Qiang Zhang: Writing – review & editing. Zengnan Wu: Writing – review & editing. Caihou Lin: Writing – review & editing, Supervision. Jin-Ming Lin: Writing – review & editing, Supervision, Funding acquisition.

  Acknowledgments

  This work was supported by the National Key R&D Program of China (No. 2021YFF0600700), the National Natural Science Foundation of China (No. 22034005), Research Projects of Putian University (No. 2024172), and the Startup Fund for Advanced Talents of Putian University (No. 2024046).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.111035 .

  
  
• 1. [1]

     X. Qin, R. Liu, F. Akter, et al., J. Cancer 12 (2021) 2073–2082. doi: 10.7150/jca.53198

  2. [2]

     P. Jain, S. Vashist, B.K. Panjiyar, Cureus 15 (2023) e46089.

  3. [3]

     N. Choudhary, R.C. Osorio, J.Y. Oh, M.K. Aghi, Cancers 15 (2023) 1519. doi: 10.3390/cancers15051519

  4. [4]

     M. Obara-Michlewska, Neurochem. Int. 158 (2022) 105363. doi: 10.1016/j.neuint.2022.105363

  5. [5]

     E. Haroon, C.L. Raison, A.H. Miller, Neuropsychopharmacology 37 (2012) 137–162. doi: 10.1038/npp.2011.205

  6. [6]

     L. Zhai, A. Bell, E. Ladomersky, et al., Clin. Cancer Res. 27 (2021) 6514–6528. doi: 10.1158/1078-0432.ccr-21-1392

  7. [7]

     M. Więdłocha, P. Marcinowicz, M. Janoska-Jazdzik, A. Szulc, Neuro-Psychoph. ´ 106 (2021) 110145. doi: 10.1016/j.pnpbp.2020.110145

  8. [8]

     A. Bosi, D. Banfi, M. Bistoletti, et al., Int. J. Tryptophan Res. 13 (2020) 1–25.

  9. [9]

     A.S. Correia, N. Vale, N. Int. J. Mol. Sci. 23 (2022) 8493. doi: 10.3390/ijms23158493

  10. [10]

      I. Davis, A. Liu, Expert Rev. Neurother. 15 (2015) 719–721. doi: 10.1586/14737175.2015.1049999

  11. [11]

      D.H. Munn, A.L. Mellor, Trends Immunol. 37 (2016) 193–207. doi: 10.1016/j.it.2016.01.002

  12. [12]

      R. Schwarcz, T.W. Stone, Neuropharmacology 112 (2017) 237–247. doi: 10.1016/j.neuropharm.2016.08.003

  13. [13]

      M. Platten, M. Friedrich, D.A. Wainwright, et al., Curr. Opini. Immunol. 70 (2021) 57–66. doi: 10.1016/j.coi.2021.03.005

  14. [14]

      J.P. Dewulf, M. Martin, S. Marie, et al., Sci. Rep. 12 (2022) 9959. doi: 10.1038/s41598-022-14292-w

  15. [15]

      Q. Wang, Y.N. Sun, C.M. Zou, et al., Behav. Brain Res. 422 (2022) 113764. doi: 10.1016/j.bbr.2022.113764

  16. [16]

      A. Heylen, Y. Vermeiren, I.P. Kema, et al., Pharmaceuticals 16 (2023) 16040615.

  17. [17]

      S.G. Teixeira, P. Houeto, F. Gattacceca, et al., Toxicol. Lett. 388 (2023) 1–12. doi: 10.1016/j.toxlet.2023.09.011

  18. [18]

      L. Zhang, Z. Liu, Q. Deng, et al., Angew Chem. Int. Ed. 60 (2021) 3469–3474. doi: 10.1002/anie.202012487

  19. [19]

      J. Li, W. Bi, Y. Gao, et al., Sensor Actuat. B: Chem. 408 (2024) 135526. doi: 10.1016/j.snb.2024.135526

  20. [20]

      Y. Yin, Z. Li, H. Gao, et al., Nano Lett. 24 (2024) 1081. doi: 10.1021/acs.nanolett.3c03580

  21. [21]

      T. Roodsant, M. Navis, I. Aknouch, et al., Front. Cell. Infec. Microbiol. 10 (2020) 272. doi: 10.3389/fcimb.2020.00272

  22. [22]

      J. Puschhof, C. Pleguezuelos-Manzano, A. Martinez-Silgado, et al., Nat. Protoc. 16 (2021) 4633–4649. doi: 10.1038/s41596-021-00589-z

  23. [23]

      C. Aguilar, M. Alves da Silva, M. Saraiva, et al., Exp. Mol. Med. 53 (2021) 1471–1482. doi: 10.1038/s12276-021-00629-4

  24. [24]

      P. Rider, Ž.P. Kačarević, S. Alkildani, et al., J. Tissue Eng. 9 (2018) 1–16. doi: 10.1117/1.jmm.17.2.020901

  25. [25]

      J. Groll, J.A. Burdick, D.W. Cho, et al., Biofabrication 11 (2018) 1758–5090.

  26. [26]

      C. He, F. Lu, Y. Liu, et al., Heliyon 10 (2024) e23504. doi: 10.1016/j.heliyon.2023.e23504

  27. [27]

      C. Shao, Y. Yu, X. Lei, et al., Chem. Eng. J. 460 (2023) 141739. doi: 10.1016/j.cej.2023.141739

  28. [28]

      W. Shin, H.J. Kim, Nat. Protoc. 17 (2022) 910–939. doi: 10.1038/s41596-021-00674-3

  29. [29]

      S. Jalili-Firoozinezhad, A. Bein, F.S. Gazzaniga, et al., Methods Mol. Biol. 2373 (2022) 69–85. doi: 10.1007/978-1-0716-1693-2_5

  30. [30]

      J. Wu, B. Zhang, X. Liu, et al., Trend. Food Sci. Technol. 134 (2023) 1–12. doi: 10.1016/j.tifs.2023.02.013

  31. [31]

      W. Shin, C.D. Hinojosa, D.E. Ingber, et al., iScience 15 (2019) 391–406. doi: 10.1016/j.isci.2019.04.037

  32. [32]

      S.Y. Chen, Y.X. Li, Y.R. Zhang, et al., J. Pharm. Anal. 14 (2024) 100967. doi: 10.1016/j.jpha.2024.100967

  33. [33]

      H. Yao, Y. Li, Y. Zheng, J.M. Lin, Anal. Chem. 95 (2023) 10999–11006. doi: 10.1021/acs.analchem.3c01134

  34. [34]

      Y. Hou, Y. Zheng, X. Zheng, et al., Lab Chip 23 (2023) 2654–2663. doi: 10.1039/d3lc00251a

  35. [35]

      K.L. Rock, H. Kono, Annu. Rev. Pathol. 3 (2008) 99–126. doi: 10.1146/annurev.pathmechdis.3.121806.151456

  36. [36]

      L. Du, R. Qi, J. Wang, et al., Int. J. Mol. Sci. 22 (2021) 12435. doi: 10.3390/ijms222212435

  37. [37]

      S.K. Seo, B. Kwon, Exp. Mol. Med. 55 (2023) 1371–1379. doi: 10.1038/s12276-023-01028-7

  38. [38]

      S. Trombetti, R. Sessa, R. Catapano, et al., Antioxidants 10 (2021) 1603. doi: 10.3390/antiox10101603

• 

  Figure 1  Illustration of the influence of tryptophan metabolism among cells and the manipulation for GBA simulation on MCM. (a) Three types of changes were noted in three colors: blue for chemical reactions; green for morphological responses of cells; and red for the direction of tryptophan transportation. (b) Seeding protocol: step 1, Caco2 and U87 were seeded in Ch1 and Ch2, respectively; step 2, delivery of E-AHMPs (E. coli) into Ch1 after cells adhesion; step 3, continuous flow by the direction of 3a to 3b while the other four inlets were blocked.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  Sustainability of cells and E-AHMPs in MCM. (a) Images of viable/dead cells characterized by Calcien-AM/PI at t₀ + 24 h and t₀ + 36 h. (b, c) Histograms of cell viability at t₀ + 24 h and t₀ + 36 h. Significance was calculated by unpaired t-test: P = 0.3629 for Caco2 and P = 0.6541 for U87, both non-significant results. (d) Images of viable E. coli in AHMPs characterized by SYTO9 and no PI signal were captured during the process (E. coli concentration of 1 × 10⁴ CFU/mL).

  下载: 全尺寸图片 幻灯片
  

  Figure 3  Parameters describing the morphological alterations of U87 cells before and after the co-culture (control referred to Caco2 + U87, co-culture referred to Caco2 + U87 + E. coli). (a) Graph of U87 cells after being transformed into black and white 2D graphics in ImageJ software, where the yellow border in the right panel indicates the external ellipse of the cell, and a and b are the half-length and half-short axes of the external ellipse, respectively. (b-d) Polarization rate (P = 0.3233, P < 0.0001), cell area (P = 0.9763, P = 0.3328), and roundness (P = 0.5017, P = 0.0018) of U87 cells in the control groups versus the co-culture groups, respectively. (e) Fluorescence views of Caco2 and U87 with or without E-AHMPs. Nucleus stained with DAPI and F-actin labeled with ATTR. (f) Quantification of average F-actin level of Caco2/U87 cells. Significance was calculated by a two-way t-test (P = 0.0103 for U87 and P = 0.5041 for Caco2).

  下载: 全尺寸图片 幻灯片
  

  Figure 4  Tryptophan metabolism results analyzed by LC-MS. (a) Two metabolic pathways of excess tryptophan produced by E. coli in AHMPs. (b) ∆c(Trp) values from left to right are 26.62, 17.32 and 20.23 µmol/L. (c) ∆c(Kyn) values from left to right are 0.40, 0.27 and 0.21 µmol/L. (d) ∆c(IPA) values from left to right are 1.26, 0.19 and 1.27 µmol/L.

  下载: 全尺寸图片 幻灯片
•