Designer Cellular Spheroids with DNA Origami for Drug Screening

doi:10.21203/rs.3.rs-3555194/v1

Download PDF

Article

Designer Cellular Spheroids with DNA Origami for Drug Screening

https://doi.org/10.21203/rs.3.rs-3555194/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

In vitro models are crucial for drug screening, yet they often fail to accurately reflect human physiological responses. While 3D cell cultures aim to simulate human tissues, many lack the detailed complexity and interaction of various cell types found in actual tissues. Additionally, integrating these models with high-throughput drug screening remains a challenge. Current models can't strike the balance between capturing the complexity of human diseases and being suitable for large-scale drug tests. Here we introduce a method that uses self-assembling Nucleic-Acid-nanostructures-decorated-living-Cells, termed NACs, to create spheroids with a customizable 3D layout. To demonstrate its uniqueness, our method effectively creates designer 3D liver spheroids by combining hepatocytes with different non-parenchymal cells, leading to improved drug sensitivity and detailed modeling of complex chronic diseases and immune-stromal interactions. Our approach achieves a high level of biological detail while being standardized and straightforward to construct with the potential for large-scale drug discovery applications. By combining the precision of DNA nanotechnology with advanced cell culture techniques, we're streamlining human-centric models, balancing complexity with standardization, to boost drug screening efficiency.

Physical sciences/Nanoscience and technology/DNA nanotechnology/Organizing materials with DNA

Biological sciences/Drug discovery/Drug screening/High-throughput screening

Standard cell cultures and animal models exhibit limited translational fidelity in predicting human drug responses, underscoring the need for advanced in vitro model systems that more precisely emulate the profound multicellular complexity and architectural sophistication of human tissues^1,2. Current 3D culture platforms aim to reconstitute tissue-like structures under physiological conditions^3–5. A common approach employs carefully designed biomimetic scaffolds to direct cell organization through localized biochemical and biophysical cues presentation^6–8. For instance, hydrogels with defined mechanical properties and adhesive ligand display can guide stem cells and cancer cells to self-assemble into organoid-like architectures^9–12. However, recapitulating the exquisite compositional heterogeneity of native tissues comprising multiple interacting cell types in a physiologically accurate spatial context remains an outstanding challenge^13,14. While emerging micro-physiological system techniques aim to enhance physiological relevance^15–17, excessive architectural intricacy frequently impedes integration with automated, cost-effective, high-throughput drug screening pipelines that are central to preclinical drug discovery. Therefore, new biomimetic tissue engineering strategies are imperative that balance model sophistication to emulate a disease state with amenability to reproducible large-scale pharmacological analysis^18,19.

Conventional cellular spheroid models generated through simple cell aggregation present feasible high-throughput screening platforms^20–22. However, they fail to recapitulate the compositional heterogeneity of multiple cell types and the architectural intricacy of native tissues, severely restricting accuracy in modeling complex multifaceted diseases^23,24. This motivates the development of advanced structured spheroids with programmed spatial control over cell organization to balance model relevance and screening compatibility^25,26. Exciting advances in structural DNA nanotechnology provide promising routes to prescribe spatial heterogeneity within synthetic tissues with nanoscale precision^27–32.

Here, we present a novel methodology leveraging self-assembling Nucleic Acid nanostructures to decorate living Cells, termed NACs, to generate designer spheroids with programmable 3D architecture. Our approach rapidly produces viable spheroids with reproducible size uniformity. Crucially, DNA nanostructures allow for the introduction of tissue-like complexity, such as controlled positioning of multiple cell types, to emulate native microanatomy, owing to their programmable and highly specific sequence-based interactions that enable precise spatial arrangement. We demonstrate the generation of heterotypic liver spheroids comprising parenchymal hepatocytes and non-parenchymal stromal cells. Further, we recreate complex pathological features of progressive non-alcoholic fatty liver diseases. These in vitro models successfully recapitulate unique fibrotic morphologies and distinct developmental phases from steatosis to inflammation and fibrosis. To enhance physiological relevance, we integrate immune cells into spheroids to establish tumor microenvironments with stromal interactions. Our approach promises reproducible generation of architecturally-defined spheroid models with precision spatial control; such advanced in vitro systems that integrate tissue-like complexity and drug screening throughput hold exciting potential to transform preclinical evaluation and personalized medicine.

NACs strategy for scale-up production of uniform spheroids

Standardized, scalable 3D cell culture models are vital for consistent, high-throughput drug screening. Size uniformity is particularly crucial, as spheroid size strongly influences cellular functions and assay reproducibility³³. To address this challenge, we developed a NACs strategy, enabling rapid and standardized production of 3D models with uniform size (Fig. 1a). Unlike traditional spheroid models that rely on inherent homotypic cell adhesion, our technique employs DNA origami nanostructures to modify cell surfaces, thereby mimicking "cell adhesion molecules". These structures are termed "Nucleic Acid Nanostructure Cell Linkers" or NAC-Linkers for short (Supplementary Figs. 1–3). This methodology equips cells from various germ layers with innovative programmable interaction forces through sequence-specific DNA hybridization, thus allowing for the rapid assembly of a multitude of cell types into 3D spheroids. Demonstrating the prowess of our approach, we successfully assembled millions of human liver primary cells into a 3D spheroid inside a droplet. Cells equipped with matched NAC-Linkers conjoined quickly, culminating in the formation of entire spheroids in just 24 hours (Fig. 1b, Supplementary Fig. 4). On the other hand, unaltered cells or those modified with just one variant of the NAC-Linkers exhibited rough exteriors and voids (Fig. 1b). We recorded a noteworthy expansion in the spheroid's diameter over the culture duration, from day 1 to day 7 (Fig. 1c, Supplementary Fig. 5). Such an observation suggests that cells housed within the spheroids retain their heightened proliferation capabilities. Traditional techniques using low-adhesion plates face hurdles like inadequate gravitational aggregation for sparse cell numbers and necrosis in overly dense spheroids. Our strategy markedly bolsters cell-to-cell interactions, bypassing cell number restrictions, and ensures efficient modeling for a range from 50 to 1,000,000 cells (Fig. 1c, d). When we varied the hepatocyte count (50 − 10,000), the spheroid's diameter showed a strict correlation with the cell count, and post-assembly proliferation remained consistent (Supplementary Fig. 5). We realized standardized mass production of 3D spheroids with a size variability less than 3.7%, and reaching sizes up to 3,000 µm (Fig. 1d, e). These findings underscore the efficiency of our standardized spheroid production using NACs.

Highly biomimetic liver spheroid models with NACs strategy

In vitro biomimetic models aim to replicate the native physiological functions of target cells. Using the NACs strategy, we rapidly reconstituted differentiated cells into 3D niches ex-vivo. Demonstrating our approach, we incorporated hepatocytes from the inner germ layer with three non-parenchymal cells from the middle germ layer: liver sinusoidal endothelial cells (LSECs), hepatic stellate cells (HSCs), and Kupffer cells (KCs) (Fig. 2a). This integration ensures the stable maintenance of hepatocyte secretion and metabolism, closely simulating the in vivo niches of hepatocytes. We conducted a comprehensive assessment of the liver spheroid morphology, functionality, metabolism, and receptor expression over time. Remarkably, the combination of these four cell types led to the formation of 3D NAC-Liver spheroids (NAC-Livers) within 24 hours, which exhibited stable growth in vitro for over 28 days, culminating in tight cellular junctions (Fig. 2b). Our observations confirmed that these liver spheroids consistently secreted high levels of albumin and urea (Fig. 2c, d). In contrast, 2D monolayer-cultured primary human hepatocytes (2D PHHs) showed rapid deterioration of their characteristic hexagonal binucleation by day 7. This deterioration is attributed to non-physiological microenvironments, resulting in substantially lower albumin secretion and urea synthesis when compared to their 3D spheroid counterparts (Fig. 2b-d). Crucially, the spheroids demonstrated enhanced drug responsiveness and maintained elevated drug-metabolism enzyme (DME) activities compared to 2D cultures (Fig. 2e, f). Further gene expression profiling indicated that the spheroids retained Phase I/II DME and Phase III transporter expression more efficiently than freshly isolated primary human hepatocytes (F-PHHs) (Supplementary Figs. 6–7). Although there was a modest decrease in the expression of coagulation factors (Factors II and X) over time, the levels observed were still superior to those in F-PHHs (Fig. 2g, h). Furthermore, the expression of the functional hepatitis B virus (HBV) receptor, sodium taurocholate co-transporting polypeptide (NTCP)³⁴, was on par with F-PHHs. This suggests the potential of these spheroids for modeling HBV infection in vitro (Fig. 2i). Our findings highlight the capacity of multi-cell spheroids to mimic hepatic functions in vitro over extended periods. Leveraging these physiologically advanced spheroids for in vitro drug safety assessments, we evaluated drug-induced liver injury (DILI) for four compounds^35,36 (Compound details in Supplementary Table S1). Notably, three hepatotoxic drugs displayed reduced TC₅₀ values in 3D spheroids as opposed to 2D PHHs. Conversely, the non-hepatotoxic drug exhibited no cytotoxicity. This underscores the sensitivity of 3D spheroids in detecting toxic drugs and their respective concentrations when compared to 2D cultures (Fig. 2j).

Enhancing 3D disease modeling with NACs strategy

Chronic diseases present unique challenges for 3D in-vitro models due to their multifaceted nature. Achieving realistic representations, especially of conditions like liver fibrosis, demands intricate interplays between various cell types and spatial configurations³⁷. Existing models often fall short in recreating fibrous structures typical of chronic liver diseases, affecting drug delivery and treatment efficacy^23,38. Addressing this gap, we harnessed the NACs strategy. This approach programs spatial heterogeneity within 3D models by controlling the addition of distinct cell populations. Our foundational setup involved a core mimicking liver tissue, assembled from PHHs, LSECs, and KCs using NAC-Linkers. Surrounding this, activated HSCs formed a protective shell, leading to the creation of what we term "package NAC-Livers" (pNAC-Livers) to model the effects of pseudo-lobule structures on hepatocytes (Fig. 3a-b, Supplementary Fig. 8a). Upon implementing this approach, we made several notable observations. As the major collagen producer in fibrosis, varying HSC numbers modulated collagen deposition and shell thickness, confirming pNAC-Livers histologically modeled differing fibrosis stages (Fig. 3c). HSC activation and proliferation were observed in our fibrosis mix models constructed with blended PHHs, LECs, KCs, and HSCs (termed mNAC-Livers) (Supplementary Fig. 8b, Fig. 3b, d), with upregulation of inflammatory (IL-2, IL-8, CCL3, TNFα) and fibrogenic (TGFβ) factors over time (Supplementary Fig. 9). Comparing pathological features between the two fibrosis models constructed with equal HSC numbers showed comparable hepatic stellate cell activation gene expression (Timp1, TGFβ, Col1a1, Col3a1) (Supplementary Fig. 10a). pNAC-Livers secreted higher inflammatory cytokine levels compared to mNAC-Livers, indicating enhanced Kupffer cell-driven inflammation (Supplementary Fig. 10b). Moreover, greater hepatocyte damage occurred in pNAC-Liver cores compared to mix models, likely from the collagenous deposition barrier impeding oxygen and nutrients (Fig. 3e), analogous to exacerbated fibrosis forming pseudo-lobules that isolate hepatocytes from nutrients.

Non-alcoholic steatohepatitis (NASH) involves severe fibrosis progression, where pseudo-lobule formation heavily impacts drug delivery efficiency³⁸. We employed NACs strategy to imitate NASH pathogenesis and progression, first treating pNAC-Livers cores with free fatty acids (FFAs) to generate steatotic disease, then assembling HSCs around these fatty livers to form fibrotic barriers, versus uniform FFA exposure in mNAC-Livers (Fig. 3f, Supplementary Fig. 11). Both NASH models exhibited lipid accumulation and hepatocyte ballooning, recapitulating hallmarks of steatotic degeneration (Fig. 3g, Supplementary Fig. 12). Compared to controls, inflammatory cytokine secretion was enhanced in NASH mNAC-Livers and pNAC-Livers, confirming steatotic hepatocytes can further activate KCs and HSCs to induce subsequent inflammation and fibrosis (Fig. 3h). By continuous modeling from benign fatty liver through steatotic degeneration, inflammation, and fibrosis with FFA treatment, we successfully constructed disease models exhibiting NASH pathological transformation (Supplementary Fig. 12). Applying the same drug GFT505³⁹ to both NASH models revealed minimal improvement in lipid accumulation, inflammation and fibrosis in NASH pNAC-Livers (Fig. 3g-i), highlighting the impact of fibrotic barriers on drug delivery, consistent with clinical trial failures of GFT505 in late-stage NASH patients (fibrosis stage F2-3)⁴⁰. In contrast, GFT505 applied to NASH mNAC-Livers showed weakened fibrosis, likely because this model only exhibits early fibrosis features (Fig. 3j). Our NACs strategy enabled modeling complex chronic disease pathogenesis and constructing pathological structure models, providing more accurate prediction of drug efficacy for clinical trials.

NACs strategy for immune microenvironment models

Studying the immune microenvironment is increasingly important for elucidating pathogenesis and drug development^13,41. Leveraging our innovative NACs strategy, we've crafted two distinct models to closely replicate immune cell interactions with specific target tissues/ organs. Our first endeavor was to create a 3D model integrating resident immune cells within liver tissues, which we term the "immune integration model". This assembly involves functionalized immune cells, hepatic parenchymal cells, and non-parenchymal cells (Fig. 4a). Our experimental results highlighted an intriguing spatial distribution: T cells predominantly settled on the periphery, while macrophages distributed more evenly throughout the structure (Fig. 4b). This spatial distribution mirrors physiological livers where T cells are found around portal triads and macrophages reside within hepatic sinusoids⁴². Moreover, when subjecting the liver-immune model to acetaminophen (APAP) injury, our observations showed a pronounced increase in the TC₅₀ value in the presence of macrophages. This suggests macrophages play a protective role by efficiently metabolizing and clearing the drug/toxin, thereby reducing liver damage (Fig. 4c). This phenomenon corroborates previous findings that underscore the pivotal role of resident Kupffer macrophages in curtailing APAP-triggered liver injuries^42,43.

Our second model introduces circulating immune cells to tumor spheroids, aiming to faithfully recreate the tumor microenvironment (TME) and simulate the process of immune cell infiltration into tumors. We commenced by assembling tumor NAC-spheroids from functionalized tumor cells, later cultivating them alongside immune cells to construct our tumor-immune infiltration models (Fig. 4d). Interestingly, the typically pro-tumor growth and metastatic characteristics of M2 macrophages within the TME can be reversed by certain immune therapeutics⁴⁴. When we treated tumor spheroids with the immunotherapeutic agent BLZ945, we noticed a marked decrease in macrophage infiltration, indicating reduced tumor homing post-treatment (Fig. 4e). Furthermore, qPCR analyses revealed an upswing in M1 markers and a dip in M2 markers, reinforcing the BLZ945's role in shifting macrophages towards an antitumoral M1 phenotype (Fig. 4f). This observation aligns seamlessly with prior in vivo studies that reported similar outcomes⁴⁴. The incorporation of NACs, especially without the reliance on matrix gels, ensures robust interactions between immune cells and organs. This novel approach allows for extended monitoring of immune cell homing and targeting. Fundamentally, our NACs strategy sets the stage for engineered interactions amongst immune components, resulting in the creation of microenvironment models that not only emulate but also accurately represent physiological immune contexts.

High-throughput drug screening with NAC-spheroids

NACs strategy enables mass-producing tissue-complex and standardized 3D models without complex equipment, providing a low-cost, efficient drug screening platform. We developed a target- and phenotype-driven drug screening strategy encompassing target identification, artificial intelligence (AI) screening of molecular candidates, drug treatment, high-content analysis, data analysis, and ultimate functional molecule discovery (Fig. 5a). We validated the NAC-spheroid for drug research and development (R&D) using a liver fibrosis model. Three target inhibitors were first selected for target screening, all inhibiting activated HSC proliferation in 2D culture as expected (Supplementary Fig. 13a). However, only the PDK1 inhibitor OSU03012 reduced HSC proliferation in 3D fibrotic pNAC-Livers (Supplementary Fig. 13a, b). Decreased transaminase levels and fibrogenic genes further demonstrated the anti-fibrotic effects of the PDK1 inhibitor without liver toxicity (Supplementary Fig. 13c-e). In contrast, the ALK4 inhibitor SB505124 and Smad3 inhibitor SIS3 HCl increased transaminase secretion to varying extents, indicating their poor therapeutic effects and potential hepatotoxicity (Supplementary Fig. 13a-e). We further tested the three inhibitors in DDC-diet mice, a widely used animal model for studying liver fibrosis and injury (Supplementary Fig. 14a). Histological staining revealed significantly reduced HSC activation and collagen deposition after PDK1 inhibitor treatment (Supplementary Fig. 14b-d). Serum transaminase measurements also confirmed the PDK1 inhibitor's ameliorative effects on hepatocellular injury (Supplementary Fig. 14e, f). The other two inhibitors showed no anti-fibrotic efficacy, and the Smad3 inhibitor even caused severe necrosis and liver injury with increased serum aspartate aminotransferase (AST) levels (Supplementary Fig. 14b-f). The high consistency between drug efficacy and safety profiles in 3D models versus animal studies demonstrates the potential of 3D models to bridge traditional 2D cultures and animal experiments.

After identifying PDK1 as the target for the following phenotype screening, a small molecule library of FDA-approved drugs and internal compounds was first constructed. AI and machine learning were then utilized to select the top 100 molecules with the highest binding specificity and affinity to PDK1 (Fig. 5a). Considering diversity and availability, 33 were chosen as candidates (#1–33) along with 4 known PDK1 inhibitors as positive controls (#34–37, details in Supplementary Table S3). The 37 candidates were then tested on fibrotic pNAC-Livers assembled from GFP-labeled PHHs and RFP-labeled HSCs (Supplementary Fig. 15a). High-content imaging quantified GFP/RFP fluorescence intensity in each 3D model post-treatment (Fig. 5b). Reduced GFP intensity indicates potential hepatotoxicity causing PHH death, while lower RFP intensity signifies stronger HSC growth inhibition and better anti-fibrotic efficacy. Hence, candidates eliciting weak GFP intensity were first excluded during analysis, and then the remaining drugs were ranked by increasing RFP intensity (Fig. 5c, d). The top three candidates with the lowest RFP intensity (#37, #4, #35) were analyzed for PDK1 targeting. Finally, axitinib (#4) was discovered for its anti-fibrotic effects and PDK1 targeting properties (Fig. 5e). Reduced fibrogenic genes and transaminases in axitinib-treated pNAC-Livers validated the high-content imaging results (Supplementary Fig. 15b, c). In vivo experiments further demonstrated axitinib significantly decreased α-SMA levels and collagen deposition in fibrotic mouse livers (Fig. 5f-h), along with reduced serum transaminases (Fig. 5i, j), confirming its anti-fibrotic efficacy. The animal results verified the utility of 3D models for accurate, effective drug screening. From target identification to high-throughput screening, we achieved rapid drug development within 30 days.

The designable NAC-spheroids presented here enable introducing tissue-like complexity while maintaining critical advantages of spheroid models like scalability and uniformity. This novel approach has significant potential to advance in vitro tissue modeling. Firstly, this approach can be further expanded to construct more complex tissue architectures. The DNA-directed spatial organization enables the precise positioning of different specialized cell types in defined geometries to recreate the intricate heterocellular interactions underlying tissue-level physiology. This could enable the culturing of designer spheroids with emergent functions to emulate those of complex native tissues and organs. Additionally, the modularity of this system could allow to assemble multiple spheroids into higher-order tissue constructs such as assembloids⁴⁵, providing new avenues to engineer 3D organoid-like structures with capabilities approaching those of true organs. Secondly, the rapid and uniform generation of these designer spheroids positions the technique for integration with high-throughput drug screening pipelines, balancing model sophistication and feasibility of large-scale analysis. By enhancing control over 3D architecture and compositional heterogeneity, the system can potentially improve predictive accuracy for complex disease models without sacrificing compatibility with automated platforms for pharmacological testing. Finally, generating substantial standardized datasets using these architecturally defined spheroids provides a means to synergize with state-of-the-art deep learning methods. Sufficient high-quality human-relevant data remains a key limitation in applying AI to problems like in vitro toxicity assays. Our approach represents a promising route to produce tailored training sets for developing enhanced in silico predictors. This novel platform technology may find diverse utility in areas such as drug discovery, toxicology screening, and development of personalized therapeutics.

Data availability

The data supporting the findings of this study are available within the article, supplementary information, and source data file. Source data are provided with this paper.

Acknowledgements

This work was supported by the National Science Foundation of China (Grant No. 81700550, 22074041, 22274029 and U22A20340), the Shanghai Science and Technology Committee (STCSM) (Grant No. 20ZR1411200, 22ZR1419800, 22ZR1412000, 22140900700 and 23ZR1411300).

Author contributions

H.P., Y.Z. and G.S. provided the concept, designed the study and provided the conceptual framework for the study. X.S., Q.Y., L.L. and T.Z. provided the concept and the financial support. J.W., Y.S., H.W., Y.Z. and T.Y. performed experiments and analyzed results. T.Z. and C.Z. performed the artificial intelligence analysis. Q.L., W.L., Z.D. and B.W. provided conceptual evaluation of the project. J.W., Y.S., H.W., L.L., G.S., Y.Z. and H.P. wrote the manuscript with inputs from all authors. All authors reviewed and commented on the manuscript.

Competing interests

G.S. is an employee of Puheng Technology Co., Ltd. The other authors declare no competing interests.

Horrobin, D. F. Modern biomedical research: An internally self-consistent universe with little contact with medical reality? Nat. Rev. Drug Discovery 2, 151-154 (2003).
Pound, P. & Ritskes-Hoitinga, M. Is it possible to overcome issues of external validity in preclinical animal research? Why most animal models are bound to fail. J. Transl. Med. 16, 1-8 (2018).
Lancaster, M. A. & Knoblich, J. A. Organogenesis in a dish: Modeling development and disease using organoid technologies. Science 345, 1247125 (2014).
Shamir, E. R. & Ewald, A. J. Three-dimensional organotypic culture: Experimental models of mammalian biology and disease. Nat. Rev. Mol. Cell Biol. 15, 647-664 (2014).
Hofer, M. & Lutolf, M. P. Engineering organoids. Nat. Rev. Mater. 6, 402-420 (2021).
Kratochvil, M. J. et al. Engineered materials for organoid systems. Nat. Rev. Mater. 4, 606-622 (2019).
Li, Q. et al. Extracellular matrix scaffolding guides lumen elongation by inducing anisotropic intercellular mechanical tension. Nat. Cell Biol. 18, 311-318 (2016).
Nikolaev, M. et al. Homeostatic mini-intestines through scaffold-guided organoid morphogenesis. Nature 585, 574-578 (2020).
Liang, Y. et al. A cell-instructive hydrogel to regulate malignancy of 3D tumor spheroids with matrix rigidity. Biomaterials 32, 9308-9315 (2011).
Gjorevski, N. et al. Designer matrices for intestinal stem cell and organoid culture. Nature 539, 560-564 (2016).
Li, S. et al. Hydrogels with precisely controlled integrin activation dictate vascular patterning and permeability. Nat. Mater. 16, 953-961 (2017).
Prince, E. et al. Biomimetic hydrogel supports initiation and growth of patient-derived breast tumor organoids. Nat. Commun. 13, 1466 (2022).
de Souza, N. Organoids. Nat. Methods 15, 23-23 (2018).
Takebe, T. & Wells, J. M. Organoids by design. Science 364, 956-959 (2019).
Kang, H. W. et al. A 3D bioprinting system to produce human-scale tissue constructs with structural integrity. Nat. Biotechnol. 34, 312-319 (2016).
Zhang, Y. S. et al. Bioprinting 3D microfibrous scaffolds for engineering endothelialized myocardium and heart-on-a-chip. Biomaterials 110, 45-59 (2016).
Urciuolo, A. et al. Hydrogel-in-hydrogel live bioprinting for guidance and control of organoids and organotypic cultures. Nat. Commun. 14, 3128 (2023).
Kingwell, K. 3D cell technologies head to the R&D assembly line. Nat. Rev. Drug Discovery 16, 6-7 (2017).
Zhao, Z. et al. Organoids. Nat. Rev. Methods Primers 2, 94 (2022).
Friedrich, J., Seidel, C., Ebner, R. & Kunz-Schughart, L. A. Spheroid-based drug screen: Considerations and practical approach. Nat. Protoc. 4, 309-324 (2009).
Kunz-Schughart, L. A., Freyer, J. P., Hofstaedter, F. & Ebner, R. The use of 3-D cultures for high-throughput screening: The multicellular spheroid model. SLAS Discovery 9, 273-285 (2004).
Cui, X., Hartanto, Y. & Zhang, H. Advances in multicellular spheroids formation. J. R. Soc. Interface 14, 20160877 (2017).
Sacchi, M., Bansal, R. & Rouwkema, J. Bioengineered 3D models to recapitulate tissue fibrosis. Trends Biotechnol. 38, 623-636 (2020).
Lovett, M. L., Nieland, T. J. F., Dingle, Y.-T. L. & Kaplan, D. L. Innovations in 3D tissue models of human brain physiology and diseases. Adv. Funct. Mater. 30, 1909146 (2020).
Fennema, E., Rivron, N., Rouwkema, J., van Blitterswijk, C. & de Boer, J. Spheroid culture as a tool for creating 3D complex tissues. Trends Biotechnol. 31, 108-115 (2013).
Astashkina, A. & Grainger, D. W. Critical analysis of 3-D organoid in vitro cell culture models for high-throughput drug candidate toxicity assessments. Adv. Drug Delivery Rev. 69-70, 1-18 (2014).
Akbari, E. et al. Engineering cell surface function with DNA origami. Adv. Mater. 29 (2017).
Ge, Z. et al. Programming cell-cell communications with engineered cell origami clusters. J. Am. Chem. Soc. 142, 8800-8808 (2020).
Wu, Z. et al. Nucleic acid-based cell surface engineering strategies and their applications. ACS Appl. Bio Mater. 5, 1901-1915 (2022).
Sun, Y. et al. Nanoscale organization of two-dimensional multimeric pMHC reagents with DNA origami for CD8+ T cell detection. Nat. Commun. 13, 3916 (2022).
Xiao, M. et al. Programming receptor clustering with DNA probabilistic circuits for enhanced natural killer cell recognition. Angew. Chem. Int. Ed. 134, e202203800 (2022).
Sun, Y. et al. DNA origami–based artificial antigen-presenting cells for adoptive T cell therapy. Sci. Adv. 8, eadd1106 (2022).
Costa, E. C. et al. 3D tumor spheroids: An overview on the tools and techniques used for their analysis. Biotechnol. Adv. 34, 1427-1441 (2016).
Yan, H. et al. Sodium taurocholate cotransporting polypeptide is a functional receptor for human hepatitis b and d virus. eLife 1, e00049 (2012).
Levy, G. et al. Long-term culture and expansion of primary human hepatocytes. Nat. Biotechnol. 33, 1264-1271 (2015).
Chen, Y. et al. Assessment of long-term functional maintenance of primary human hepatocytes to predict drug-induced hepatoxicity in vitro. Arch. Toxicol. 95, 2431-2442 (2021).
Paradiso, A. et al. In vitro functional models for human liver diseases and drug screening: Beyond animal testing. Biomater. Sci. 11, 2988-3015 (2023).
van Os, E. A. et al. Modelling fatty liver disease with mouse liver-derived multicellular spheroids. Biomaterials 290, 121817 (2022).
Staels, B. et al. Hepatoprotective effects of the dual peroxisome proliferator‐activated receptor alpha/delta agonist, gft505, in rodent models of nonalcoholic fatty liver disease/nonalcoholic steatohepatitis. Hepatology 58, 1941-1952 (2013).
Shan, L. et al. New drugs for hepatic fibrosis. Front. Pharmacol. 13, 874408 (2022).
Zhou, Z. et al. Harnessing 3D in vitro systems to model immune responses to solid tumours: A step towards improving and creating personalized immunotherapies. Nat. Rev. Immunol., 1-15 (2023).
Triantafyllou, E., Woollard, K. J., McPhail, M. J., Antoniades, C. G. & Possamai, L. A. The role of monocytes and macrophages in acute and acute-on-chronic liver failure. Front. Immunol. 9, 2948 (2018).
Gilboa, D. et al. Erythropoietin enhances Kupffer cell number and activity in the challenged liver. Sci. Rep. 7, 10379 (2017).
Ngambenjawong, C., Gustafson, H. H. & Pun, S. H. Progress in tumor-associated macrophage (tam)-targeted therapeutics. Adv. Drug Delivery Rev. 114, 206-221 (2017).
Pasca, S. P. Assembling human brain organoids. Science 363, 126-127 (2019).

Mice

All animal experiments were performed according to approved protocols by the Institutional Animal Care and Use Committee (IACUC) of Zhongshan Hospital, Fudan University. WT male C57BL/6J mice (aged 6-8 weeks) were obtained from Cavens Biological Technology (Jiangsu, China). The mice were housed in specific pathogen-free (SPF) rooms, and the temperature was maintained at a stable 25℃ with a 12-hour-light/12-hour-dark photoperiod. The feed of mice underwent irradiation with ultraviolet light, and their drinking water and bedding were changed every other day.

Materials and reagents

All oligonucleotides were synthesized and purified by Sangon Biotech Co., Ltd. (Shanghai, China). M13mp18 DNA was purchased from Bayou Biolabs (Los Angeles, USA). Type I collagenase and rat tail collagen for primary cell extraction and culture were purchased from Sigma-Aldrich. Roswell Park Memorial Institute (RPMI) 1640 medium was obtained from KeyGen Biotech. Co., Ltd. (Jiangsu, China). Dexamethasone and recombinant human insulin were purchased from Solarbio Science & Technology Co., Ltd. (Beijing, China). TGFβ for activating HSCs was bought from Abcam. All small molecules used in hepatotoxicity tests and high-content drug screening were purchased from TopScience. The comprehensive list of these small molecules and their relevant information can be found in Supplementary Table S1 (hepatotoxicity tests) and Supplementary Table S3 (high-content drug screening), respectively. The reagent Nile Red was bought from Yeasen BioTechnologies Co., Ltd. The primary antibodies of rabbit anti-human αSMA (1:100, cat. #19245) and rabbit anti-human Collagen I (1:100, cat. #72026) were purchased from Cell signaling technology, rabbit anti-human Collagen III (1:100, cat. #22734-1-ap) were bought from Proteintech.

Cell lines

The PHHs, HSCs, LSECs, and macrophages were obtained from Liver Biotechnology (Shenzhen) Co., Ltd. Cells were cultured in the Liver-Organ Physiological medium (Puhengtec, cat. #M0001). The live cell trackers used for cell labeling were Calcein-AM (green, Dojindo, cat. #C326) and CMTPX (red, Yeasen, cat. #40717ES50). The cells were incubated with each dye for 30 minutes at 37 °C, and successful staining was confirmed by fluorescent microscopy.

Fabrication and characterization of DNA origami nanostructures (referred to as NAC-Linkers).

The process of producing NAC-Linkers was as follows: M13mp18 ssDNA scaffold, short DNA staple strands, strands for cell-cell connection, and strands modified with cholesterol were mixed at a ratio of 1:5:5:5 in 1×TAE-Mg²⁺ buffer (40 mM Tris, 20 mM acetic acid, 2 mM EDTA, and 12.5 mM magnesium acetate, pH 8.0). The mixture was slowly annealed from 95 °C to 20 °C. After annealing, the NAC-Linkers were filtered three times using Millipore Amicon Ultra 100 kDa centrifuge filters to remove the extra strands. The concentration of the purified NAC-Linkers was estimated by its absorption at 260 nm (Agilent cary60 UV-Vis spectrophotometer, USA). The purified NAC-Linkers were then stored at 4 °C for further use. For fluorescence-labeled conjugation, Alexa Fluor 488 (AF488)-labeled single DNA strands and NAC-Linkers were mixed at a ratio of 10:1 and combined through base pairing.

To characterize the fabrication of NAC-Linkers, we employed agarose gel electrophoresis (AGE) electrophoresis and atomic force microscope (AFM). For AGE analysis, each sample (10 μL, 5 nM) was mixed with 6× loading buffer (2 μL). The 1% agarose gel was prepared with 1×TAE-Mg²⁺ buffer and Gel-Red (Solarbio, Beijing, China). Electrophoresis was performed in 1×TAE-Mg²⁺ buffer at 100 V for 90 min in an ice bath. Subsequently, the agarose gel was visualized using a gel imager under UV light (Tanon Science & Technology Co., Ltd., China). For AFM imaging, 5 μL of diluted NAC-Linkers (2 nM) was deposited onto freshly prepared mica and allowed to incubate for 3 minutes. After rinsing the samples with ultrapure water and drying them under a nitrogen atmosphere, the samples were imaged using a Multimode 8 (Bruker, USA) in ScanAsyst-Air Mode. The acquired data were then analyzed with Nanoscope analysis software.

NAC-Linker-based cell surface modification and cell-cell interaction

Before initiating cell surface modification, cells were collected and washed with PBS. To optimize the incubation conditions, flow cytometry analysis (FlowSight, Germany) was used to assess the cell surface fluorescence intensity after incubation with different concentrations of AF488-labeled NAC-Linkers and varying incubation times. Based on the results obtained from this optimization process, 1´10⁵ cells were re-suspended in complete culture medium with 20 nM NAC-Linker A or B and incubated for 30 min at room temperature to achieve efficient membrane modification. After the incubation period, unbound NAC-Linkers were effectively removed by centrifugation at 1,000 rpm for 3 minutes. To establish cell-cell connections, the cell mixture previously modified with matched NAC-Linkers (NAC-Linker A and B) underwent an additional 30-minute incubation at room temperature.

The aggregation of cells into spheroids by the hanging-drop method

We utilized the hanging-drop method to establish the spheroid cultures without the need for an ultra-low attachment cell culture plate. Firstly, 2 mL PBS was added to each well of a 6-well plate for moisturizing. The mixed cells incubated with matched NAC-Linkers were then supplemented with complete culture medium, with each drop containing a volume of 20 μL. Next, we inverted the plate cover and added 20 μL cell suspension per drop onto the plate cover. The spacing between the two drops was maintained to be more than 0.5 cm. Finally, we turned the plate cover over and allowed the cells to aggregate with gravity, forming a spheroid in the center of each drop. After the formation of the spheroid, 50% of medium was refreshed daily with complete culture medium.

Isolation and cultivation of mouse parenchymal and non-parenchymal cells

Primary mouse parenchymal cells were isolated from 8-week-old C57BL/6J wild-type mice using collagenase perfusion method as described below. Firstly, perfusion fluid I contained Krebs Hensleit's solution (Procell) with 100 μM ethylene glycol tetraacetic acid (EGTA). Perfusion fluid II contained Krebs Hensleit's solution with 3 μM CaCl₂ and 37~42 mg collagenase I. Both perfusion fluids were preheated to 37 °C. Before the procedure, the mice were anesthetized with 50 mg/kg pentobarbital via intraperitoneal injection. 50 mL of perfusion fluid I and 30 mL of perfusion fluid II were perfused within 10 minutes. After the perfusion process, the gallbladder was removed using sterilized scissors, and the liver was quickly transferred to pre-cooled RPMI 1640 containing 5% FBS to terminate collagenase digestion. After removing the outer layer of the liver using sterile tweezers, the liver was transferred onto a 70 µm filter membrane placed over a 50 mL centrifuge tube. The liver cells were then gently dispersed using RPMI 1640 medium, allowing them to pass through the filter membrane and collect in the centrifuge tube. Subsequently, the collected cell suspension was centrifuged at 50 g for 2 min at 4 °C. As a result, the hepatocytes formed a pellet at the bottom of the tube, while the supernatant contained the non-parenchymal cells. The supernatant was transferred to another tube and further centrifuged at 1,000 rpm for 3 minutes at 4 °C. After harvesting the hepatocytes (parenchymal cells) and non-parenchymal cells through the earlier centrifugation process, each cell type was individually resuspended in serum-free RMPI 1640 medium. The resuspended cells underwent 3-4 times of additional washing and centrifugation steps with RMPI 1640 medium at 1,000 rpm for 3 minutes, to remove impurities and debris until the supernatant became clear. Subsequent cell culture, both for 2D and 3D models, was performed using RPMI 1640 medium supplemented with 10% FBS, insulin, and dexamethasone. For 2D in vitro models, 6-well plates were coated with 50 mg/mL rat tail collagen the day before cell seeding. The coated plates were then placed in the incubator at 37 °C for 1 hour and subsequently transferred to 4 °C overnight to allow for the formation of the collagen-coated surface.

Cell viability assay

To assess 2D cell viability, we utilized CCK-8 kit (40203ES60, Yeasen) and CellCounting-Lite 2.0 Luminescent Cell Viability Assay (DD1101, Vazyme) following the manufacturer's instructions. The measurements of optical density (OD) at 450 nm and luminescence values were recorded using a multifunctional reader (MD FlexStation3). 3D spheroids were cultured in an ultra-low attachment 96-well plate. To assess 3D cell viability, we measured the adenosine triphosphate (ATP) content of each designer spheroid using the CellTiter-Glo® 3D Cell Viability Assay (G9681, Promega) following the manufacturer's instructions. Luminescence values were then recorded using a multifunctional reader. To assess the biocompatibility of NAC-Linkers (Supplementary Figure 3b), the experimental groups were treated with different concentrations of NAC-Linkers, while cells cultured with physiological medium served as the control. For evaluating the therapeutic efficacy of OSU03012, SIS3 HCl, and SB505124 on fibrotic pNAC-livers in vitro (Supplementary Figure 13a), the experimental groups were treated with different candidate molecules (OUS03012, SIS3 HCl, and SB505124), and spheroids cultured with physiological medium were used as the control. For the blank control, cell-free medium was used. Percentage of viability = (Experimental group _{OD 450 nm or luminescent} - Blank _{OD 450 nm or luminescent}) / (Control _{OD 450 nm or luminescent} - Blank _{OD 450 nm or luminescent}) × 100%.

Urea detection

Urea levels in the culture medium were measured using urea assay kit (Nanjing Jiancheng Bioengineering Institute) following the manufacturer’s instructions. In brief, 4 μL of the culture medium was mixed with reaction regent (a mixture of diacetyl oxime solution and acidic reagent at a ratio of 1:1) provided in the kit. The mixture was then subjected to a reaction at 100°C for 15 minutes, followed by immediate cooling. The OD value of the resulting solution was measured at 520 nm. Cell-free culture medium was used as the blank control for baseline correction.

Paraffin‑embedding of designer spheroids

Designer spheroids were first washed with PBS and then fixed with 4% paraformaldehyde at 4°C overnight. After fixation, the spheroids were embedded in agarose and further processed for paraffin embedding. Sections of 4 μm thickness were obtained for various analyses, including H&E staining, Sirius Red staining, immunofluorescent (IF), and immunohistochemistry (IHC).

Frozen section of designer spheroids

To prepare frozen sections of the designer spheroids, we collected 3-6 spheroids into a 1.5 mL centrifuge tube and fixed them with 4% paraformaldehyde at 4 °C overnight. Next, the spheroids were moistened twice with PBS and dehydrated using a 30% sucrose solution at 4°C for 24-48 hours until they sank to the bottom of the tube. The dehydrated spheroids were then embedded in OCT Cryomount (Tissue-Tek, Sakura Finetek) within 7 mm×7 mm×5 mm specimen molds (Tissue-Tek, Sakura Finetek) and rapidly frozen using liquid nitrogen. The frozen spheroids were sectioned into 7 µm-thick slices using a cryostat (Leica CM1860).

Sirius Red staining of designer spheroids

For Sirius red staining of the designer spheroids, the tissue slices were first heated at 60 °C for 30 minutes to ensure proper adhesion to the slides. After de-waxing and hydration steps, the slices of fibrosis spheroids were subjected to Sirius Red staining (G1340, Solarbio) at room temperature for approximately 60 minutes and washed with PBS. Next, the nucleus was counterstained with hematoxylin (G1101, Solarbio) at room temperature for 3 minutes, in accordance with the manufacturer's instructions.

Immunofluorescent (IF) and Immunohistochemistry (IHC) of designer spheroids

The IF and IHC of designer spheroids and spheroid sections involved the following steps. Firstly, all samples were incubated with 5% donkey serum at room temperature for 30 minutes. After washing with PBS containing 0.1% Tween-20, all samples were incubated with primary antibody at 4°C overnight. Following another round of washing, the samples were incubated with secondary antibody labeled with Alexa Fluor or horseradish peroxidase (HRP) at room temperature for 1 hour. For IHC, HRP was detected by using diaminobenzidine (DAB) as a chromogen, followed by additional staining with hematoxylin to visualize the nuclei. In the case of IF, the cell nuclei were specifically stained using 1 mg/mL of 4',6-diamidino-2-phenylindole (DAPI) for 10 minutes. Bright-field and fluorescence images were obtained using an inverted fluorescence microscope (IX73, Olympus) and a confocal microscope (TCS SP8, Leica).

RNA isolation and qRT-PCR

Total RNA was isolated from both 2D cells or 3D designer spheroids using TRIzol reagent (Invitrogen) and qRT-PCR was conducted as previously described. The samples for qPCR analysis in Fig. 4f contained both designer spheroids and immune cells. To ensure high-quality RNA samples, designer spheroids with the same treatment were pooled together as one sample, and the total cell amount was adjusted to 1´10⁶. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) served as an internal reference gene. Relative mRNA levels in each sample were calculated based on their threshold cycle (Ct) values and -2^-ΔΔCT method was used. The primers used were listed in Supplementary Table S4.

Detection of aspartate transaminase (AST) and alanine aminotransferase (ALT)

AST and ALT were detected using the AST/ALT detection kit (Nanjing Jiancheng Bioengineering Institute) based on the Reitman Frankel method in accordance with the manufacturer's instructions. For detection, briefly, 50 μL of culture supernatant or blood serum was collected at specific time points, and cell-free medium was used as the blank control. Standard curves of AST and ALT were generated using the standard solutions provided in the kit. The absolute OD value was calculated as follows: Absolute Value_{OD510 nm}=Experimental group Value_{OD510 nm}-Blank group Value_{OD510 nm}. Then, the absolute OD value was substituted into the standard curve to calculate the AST and ALT content (Carmen's unit). Relative level of AST or ALT secretion=Experimental group AST or ALT content/Mean control group AST or ALT content.

Detection of inflammatory and fibrotic cytokines secretion

The levels of inflammatory and fibrotic cytokines in the samples were quantified using the enzyme-linked immunosorbent assay (ELISA) method. The ELISA procedures were performed according to the instructions provided in the respective ELISA kits, including CCL2 (abs510026, Absin), CCL3 (abs510029, Absin), IL-6 (RK00004, ABClonal), IL-8 (RK00011, ABClonal), TNFα (abs510006, Absin), and TGFβ1 (abs510032, Absin). The absorbance of each well was measured at a specific wavelength, by the instructions provided by the respective ELISA kits, using a multifunctional reader.

Acquisition of drug candidates by artificial intelligence technology

The crystal structure of PDK-1 in complex with the inhibitor Compound-8i (PDB code: 3IOP)⁴⁶ was obtained from the Protein Data Bank. The structure was prepared before docking in order to add hydrogen atoms, optimize hydrogen bonds, and remove atomic clashes. A small molecular library containing FDA-approved drugs and in-house small molecules was constructed. The deep learning-based molecular docking software GNINA⁴⁷ was used to perform the virtual screening. And the 300 molecules with the highest score were kept. Then molecular dynamics simulation and the molecular mechanics Poisson-Boltzmann surface area method (MM-PBSA)⁴⁸ were combined to calculate the binding affinity of these 300 molecules with PDK-1. The Amber18 software⁴⁹ was used. The force field parameters for small molecules were generated with the general AMBER force field, and the AM1-BCC method was used to calculate the atomic point charges. The Amber ff14SB force field was used for the protein. 100 molecules with the highest binding free energy were selected. Considering the diversity of molecular structures as well as their accessibility, 33 out of 100 molecules were selected for the experiments.

High-content imaging

A total of 33 small molecules screened by AI and 4 known PDK1 inhibitors were tested on designer spheroids at concentrations of 1 μM and 10 μM. The solvent used for all drugs was DMSO, and the final concentration of DMSO in the medium was maintained at 0.1%. After 96 hours of drug exposure, high-content imaging (Nikon W1 SoRa spinning disk confocal microscope) was employed to measure the fluorescence intensity in each designer spheroid. To ensure accurate imaging, the imaging system was first calibrated using the four vertex holes of the 384-well plate, positioning the spheroids at the center of the field of view. The focal length was then adjusted to determine the Z-axis scanning range, with scans set at 10 μm intervals to acquire Z-axis stacked images in 3 planes. From these images, 2D projections were generated. The intensity of Calcein-AM and CMTPX fluorescence signals in each scan plane was recorded, and the average fluorescence readings were used as the output values. The drug liver injury-related properties were assessed based on the Calcein-AM intensity, while the antifibrotic effect was assessed based on the CMTPX intensity.

High-throughput sequencing

High-throughput sequencing was conducted to analyze the gene expression profiles of 2D and 3D cultured primary mouse parenchymal cells at 24 hours and 96 hours. For each sample, a total of 2×10⁵ cells were collected. Both mouse liver tissue and the 2D/3D culture samples were frozen in 500 μL TRIzol and stored at -80℃. Mouse liver tissue samples were used as the control group. The total RNA extraction and RNA-Seq library construction was performed by Lianchuan Corporation. To visualize the gene expression patterns, heat maps were generated using the Fragments Per Kilobase Million (FPKM) values of selected genes. Additionally, Principal Component Analysis (PCA) analysis was performed using ClustVis website based on the differentially expressed genes in each group.

Animal treatment

C57BL/6J mice were acclimated and housed for two weeks before being treated with a 0.1% DDC diet for three weeks to induce liver injury. After three weeks of continuous DDC diet feeding, the mice were treated with specific small molecule inhibitors (OSU03012, SIS3 HCl, SB505124, Axitinib) administered at equimolar concentrations through intraperitoneal injections on alternate days. The doses of each small molecule inhibitor are detailed in Supplemental Table S2. Throughout the treatment period, the DDC diet was maintained to preserve the liver injury model. After a total of six weeks from the start of the DDC diet, the mice were euthanized, and blood samples and liver tissues were collected for further analysis.

Statistically analysis

Statistical analysis was performed using GraphPad 8.4.3. The analysis was carried out using Student’s t-test (n.s. means not significant, * means P < 0.05, ** means P < 0.01, *** means P < 0.001, **** means P < 0.0001). The H score and Sirius Red Area statistics were performed by Image J.

Method references

46 Nittoli, T. et al. The identification of 8, 9-dimethoxy-5-(2-aminoalkoxy-pyridin-3-yl)-benzo [c][2, 7] naphthyridin-4-ylamines as potent inhibitors of 3-phosphoinositide-dependent kinase-1 (pdk-1). Eur. J. Med. Chem. 45, 1379-1386 (2010).

47 McNutt, A. T. et al. Gnina 1.0: Molecular docking with deep learning. J. Cheminf. 13, 1-20 (2021).

48 Wang, E. et al. End-point binding free energy calculation with mm/pbsa and mm/gbsa: Strategies and applications in drug design. Chem. Rev. 119, 9478-9508 (2019).

49 Nakano, M. & Champagne, B. Nonlinear optical properties in open‐shell molecular systems. Wiley Interdiscip. Rev.: Comput. Mol. Sci. 6, 198-210 (2016).

Yes there is potential Competing Interest. G.S. is an employee of Puheng Technology Co., Ltd. The other authors declare no competing interests.

SINAC.docx
SI

Download PDF

Version 1

posted

You are reading this latest preprint version

Designer Cellular Spheroids with DNA Origami for Drug Screening

Status:

Version 1

Abstract

Figures

Introduction

Results

NACs strategy for scale-up production of uniform spheroids

Highly biomimetic liver spheroid models with NACs strategy

Enhancing 3D disease modeling with NACs strategy

NACs strategy for immune microenvironment models

High-throughput drug screening with NAC-spheroids

Discussion

Declarations

References

Methods

Additional Declarations

Supplementary Files

Status:

Version 1