Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 6 (3), 301-319

A High-Throughput Organoid Microinjection Platform to Study Gastrointestinal Microbiota and Luminal Physiology


A High-Throughput Organoid Microinjection Platform to Study Gastrointestinal Microbiota and Luminal Physiology

Ian A Williamson et al. Cell Mol Gastroenterol Hepatol.


Background & aims: The human gut microbiota is becoming increasingly recognized as a key factor in homeostasis and disease. The lack of physiologically relevant in vitro models to investigate host-microbe interactions is considered a substantial bottleneck for microbiota research. Organoids represent an attractive model system because they are derived from primary tissues and embody key properties of the native gut lumen; however, access to the organoid lumen for experimental perturbation is challenging. Here, we report the development and validation of a high-throughput organoid microinjection system for cargo delivery to the organoid lumen and high-content sampling.

Methods: A microinjection platform was engineered using off-the-shelf and 3-dimensional printed components. Microinjection needles were modified for vertical trajectories and reproducible injection volumes. Computer vision (CVis) and microfabricated CellRaft Arrays (Cell Microsystems, Research Triangle Park, NC) were used to increase throughput and enable high-content sampling of mock bacterial communities. Modeling preformed using the COMSOL Multiphysics platform predicted a hypoxic luminal environment that was functionally validated by transplantation of fecal-derived microbial communities and monocultures of a nonsporulating anaerobe.

Results: CVis identified and logged locations of organoids suitable for injection. Reproducible loads of 0.2 nL could be microinjected into the organoid lumen at approximately 90 organoids/h. CVis analyzed and confirmed retention of injected cargos in approximately 500 organoids over 18 hours and showed the requirement to normalize for organoid growth for accurate assessment of barrier function. CVis analyzed growth dynamics of a mock community of green fluorescent protein- or Discosoma sp. red fluorescent protein-expressing bacteria, which grew within the organoid lumen even in the presence of antibiotics to control media contamination. Complex microbiota communities from fecal samples survived and grew in the colonoid lumen without appreciable changes in complexity.

Conclusions: High-throughput microinjection into organoids represents a next-generation in vitro approach to investigate gastrointestinal luminal physiology and the gastrointestinal microbiota.

Keywords: 2D, 2-dimensional; 3D, 3-dimensional; Anaerobic; Barrier Function; CAG, chicken beta-actin promoter with CMV enhancer; CFU, colony-forming unit; CRA, CellRaft Array; CVis, computer vision; EGFP, enhanced green fluorescent protein; FITC, fluorescein isothiocyanate; Fecal Microbiota; GFP, green fluorescent protein; GI, gastrointestinal; HF, hydrogen fluoride; High-Content Sampling; High-Throughput; Microinjection; OUT, operational taxonomic unit; Organoid; PBS, phosphate-buffered saline; PCR, polymerase chain reaction; QIIME, Quantitative Insights Into Microbial Ecology; WT, wild-type; hiPS, Human Induced Pluripotent Stem Cell; rRNA, ribosomal RNA.


Figure 1
Figure 1
Robotically articulated colonoid microinjection maintains atmospheric control facilitating long-term sampling of large batches of colonoids. (A) Organoids grown from adult stem cells in 3D culture form complex monolayers organized around a hollow, mucus-filled lumen cavity analogous to the colon lumen. (B) 3D-printed customized fittings were used to mount robotic microinjection hardware within an atmospheric imaging chamber of an automated imaging system. (C) A 90° bend in the injection needle allows for vertical articulation of the microinjection needle, minimizing hydrogel disruption and needle breaking during injections. (D) Wet etching mechanically pulled borosilicate capillaries produces clean, fine aperture needles capable of injecting large batches of organoids without disturbing monolayer integrity. (E) Computer vision made and measured the area of masks from images of the smallest droplets of fluorescent cargo delivered reproducibly by 5 replicate needles, facilitating volume estimation. (F) Optimized needles produce varying volumes at the same injection duration (black) but similar minimal volumes when the minimal duration reproducibly delivering cargo from each needle is used (blue). (G) Organoid-microbe injections can be performed visually by locating an organoid of interest (1), aligning the needle to the organoid lumen (2), articulating the needle against the organoid monolayer distorting its appearance (3), puncturing the monolayer and injecting cargo (4), and retracting the needle from the organoid lumen (5) to visualize specific transplantation of DsRED-expressing E coli within the lumen (needle tip is marked with an asterisk). EpCAM, Epithelial cell adhesion molecule; Muc2, Mucin 2; Ortho, Orthogonal view; RFP, Red Fluorescent Protein.
Figure 2
Figure 2
Increasing microinjection throughput using computer vision to quantify cargo retention, organoid morphology, and injection success. (A) Inert high-molecular-weight fluorescent cargos fill the organoid lumen with no signal observed in the adjacent areas. (B) Inert cargos are retained long term as the organoid expands in size over 18 hours. (C) CellProfiler computational image analysis pipelines were used to create masks of the fluorescent signal of images collected during the 18-hour time course and quantified the area and mean intensity of the signal. (D) The mean signal intensity observed from fluorescent inert cargos deceased over time to approximately half of the original intensity. (E) The area of fluorescent signal doubled during the 18-hour time course as the organoid expanded in size. (F) Integrated fluorescent signal across the observed area shows relative stability, suggesting that the inert cargo was retained within the organoid lumen. (G) Culture protocols were optimized to grow organoids on 2.5-cm2 microfabricated culture array devices containing retrievable 200-μm2 rafts separated by 50-μm walls regularly addressed to facilitate downstream sampling. (H) Modified CellProfiler image cytometry pipelines can identify DsRED fluorescent organoids and automatically segment identified organoids of varying morphologies into lumen and epithelial monolayer compartments. (I) Automated organoid identification was >95% accurate (n = 1681 rafts), allowing for >96% accurate lumen identification. (J) Organoid cross-sectional area showed a bimodal distribution with an average of 18,511 ± 5916 μm2. (K) Organoid monolayer width also shows a bimodal distribution with an average of 16.4 ± 9.3 μm. (L) The area and morphology of 500 DsRED-expressing organoids was quantified before targeting each for microinjection with 0.36 nL of 7 kilodaltons of FITC-dextran solution, which could be observed specifically within the lumen of successfully injected organoids with fluorescence signal observed outside the organoid of unsuccessful injections. (M) Successfully injected organoids were larger (19,642 ± 8970 μm2) than unsuccessfully injected organoids (9142 ± 8970 μm2). (N) Organoids with wider monolayers were injected in shorter intervals (51.1 ± 35.6 s) than thinner organoids (74.1 ± 35.6 s). (O) Organoids larger than 18,000 μm2 with monolayers >15-μm wide were microinjected with the highest efficiency and accuracy with all targeted organoids of that range successfully microinjected in 39.1 ± 12.5 seconds. Auto, Automated; Int, Integrated; MI, Microinjection; Min, Minimum.
Figure 3
Figure 3
The colonoid lumen forms a discrete compartment compatible with specific microbial growth. (A) GFP-expressing E coli can be visualized after microinjection into DsRED-expressing colonoids and appears to sit in the bottom of the lumen cavity in all colonoids observed. (B) The effects of antibiotics in the colonoid culture media on lumen microbe compatibility was investigated using a CRA device to culture colonoids in 4 discrete reservoirs treated with tetracycline and/or chloramphenicol. Colonoids from each well were targeted for microinjection with a mixed microbial culture of DsRED-expressing E coli resistant to tetracycline and GFP-expressing Y pseudotuberculosis resistant to chloramphenicol. Injected colonoids were monitored over time by live fluorescent imaging before lumen contents were harvested to assess microbial growth by colony formation on conventional agar plates. (C) Fluorescent signal from both microbes could be observed within the lumen of successfully injected colonoids during the entire time course. (D) Antibiotics were essential for preventing off-target growth by excess bacteria delivered to the media during microinjection with no active microbes discovered in culture media treated with chloramphenicol and tetracycline 24 hours after microinjection. (E) Computational analysis showed an increase in integrated DsRED and EGFP fluorescence signal of raft images containing successfully injected colonoids, suggesting an increase in DsRED- and EGFP-expressing microbes. (F) More E coli and Y pseudotuberculosis colonies were recovered from the lumen of colonoids from all media conditions compared with the input injection droplet, suggesting the colonoid lumen protected the injected microbes from chloramphenicol and tetracycline delivered in the culture media (n = 10 colonoids in each condition). Significantly more colonies were recovered from untreated colonoids, correlating with increased integrated fluorescence signal. (G) Fluorescent signal from both microbes as well as inert fluorescent cargo could be observed within the lumen of successfully injected colonoids during the entire time course. (H) More K12 and NC101 colonies were recovered from colonoids collected 24 hours after microinjection compared with those collected immediately after microinjection, suggesting that both microbes grew regardless of the delivered load (n = 5–6 colonoids from each injection duration). (I) Computational analysis showed an increase in integrated DsRED and EGFP fluorescence signal in injected colonoids, suggesting an increase in DsRED- and EGFP-expressing microbes. Computational analysis also showed stable integrated Alexa Fluor 647 signal, suggesting that delivered dextran was well retained. (J) The measured ratio of recovered NC101 colonies to integrated EGFP signal varied significantly between the 2-hour and 24-hour time points, suggesting that integrated EGFP signal cannot be used to directly measure E coli NC101-EGFP microbial load. (K) The measured ratio of recovered K12 colonies to integrated EGFP signal varied significantly between the 2-hour and 24-hour time points, suggesting that integrated EGFP signal cannot be used to directly measure E coli–DsRED microbial load. Chl, chloramphenicol; FU, follow-up evaluation; Int, integrated; Tet, tetracycline; Tx, treatment; Y pseudo, Y pseudotuberculosis.
Figure 4
Figure 4
Monolayer respiration makes the colonoid lumen a hypoxic environment capable of supporting the growth of anaerobic enteric microbes. (A) COMSOL modeling suggests the lumen of the average colonoid (left panel) is maintained in a state of hypoxia (right panel) resulting from respiration by the colonoid monolayer. (B) Modeling suggests that O2 levels decrease rapidly from the basal to the apical colonoid surface to approximately 10% of atmospheric O2 levels (180 mmol/L). (C) Stool filtered using a 5-μm polyethylene glycol membrane (filter 1) was compatible with microinjection needles and contained the majority of species present in the unfiltered or conventionally processed (centrifuged) stool. (D) Stool filtered using a 5-μm polyethylene glycol membrane (filter 1) showed greater similarity to the unfiltered and conventionally processed stool than stool filtered using a 5-μm polypropylene membrane. (E) Colonoid compatibility with human microbial populations was investigated by loading a filtered healthy human fecal sample into a microinjection needle under anaerobic conditions before 30 colonoids were microinjected with approximately 0.2 nL of filtered stool. Injected colonoids were harvested across a 96-hour time course and assessed for growth by colony formation in anaerobic and aerobic conditions. Cent, centrifuged; PCA, principle component analysis.
Figure 5
Figure 5
The colonoid lumen is compatible with patient-derived microbial communities and nonsporulating anaerobes. (A) Few colonies were recovered from colonoids retrieved 6 hours after microinjection grown under anaerobic or aerobic conditions, with increasing microbial loads recovered over time. (B) Significantly more anaerobic and aerobic colonies were recovered from injected colonoids 12 hours after microinjection, with microbial loads peaking 72 hours after microinjection. (C) Samples (10 μL) of the filtered stool collected and cultured under anaerobic conditions grew robustly as expected (top panel), whereas unsuccessfully and noninjected samples produced no colonies in either atmospheric condition (lower panels). (D) The composition of the microbial communities changed quickly after injection but remained stable throughout the 96-hour time course with no significant shifts in 5 dominant phylum. (E) No significant changes were observed in the number of species present in the microbial community after injection or incubation within the colonoid lumen. (F) The even composition of the microbial community remained after injection and did not change significantly during the 96-hour time course. (G) Bifidobacteria, a genus containing nonsporulating anaerobes and aerobes, was a minor member of the healthy stool sample and was detected at increased levels in the communities retrieved from colonoids. (H) Colonoid microinjection compatibility with anaerobic nonsporulating microbes was verified by monoculture microinjection of B adolescentis quantified by 16s quantitative PCR (qPCR) and culturing. (I) An increased abundance of B adolescentis was detected in successfully injected colonoids by targeted qPCR 6 hours after microinjection and was maintained for 48 hours with no increase observed in unsuccessfully injected colonoids. Active B adolescentis was recovered specifically from successfully injected colonoids by anaerobic culturing throughout the 60-hour time course, suggesting that increases in 16S abundance resulted from growth within the colonoid lumen. Bif. adol, B adolescentis.

Comment in

Similar articles

See all similar articles

Cited by 19 articles

See all "Cited by" articles


    1. Gill S.R., Pop M., Deboy R.T., Eckburg P.B., Turnbaugh P.J., Samuel B.S., Gordon J.I., Relman D.A., Fraser-Liggett C.M., Nelson K.E. Metagenomic analysis of the human distal gut microbiome. Science. 2006;312:1355–1359. - PMC - PubMed
    1. Li K., Bihan M., Yooseph S., Methé B.A. Analyses of the microbial diversity across the human microbiome. PLoS One. 2012;7:e32118. - PMC - PubMed
    1. Jones R.M. The influence of the gut microbiota on host physiology: in pursuit of mechanisms. Yale J Biol Med. 2016;89:285–297. - PMC - PubMed
    1. Wu X., Ma C., Han L., Nawaz M., Gao F., Zhang X., Yu P., Zhao C., Li L., Zhou A., Wang J., Moore J.E., Cherie Millar B., Xu J. Molecular characterisation of the faecal microbiota in patients with type II diabetes. Curr Microbiol. 2010;61:69–78. - PubMed
    1. Lau E., Carvalho D., Pina-Vaz C., Barbosa J.-A., Freitas P. Beyond gut microbiota: understanding obesity and type 2 diabetes. Hormones. 2015;14:358–369. - PubMed

Publication types

MeSH terms