Immunoids: Building Immunocompetent Organoids from Human Pluripotent Stem Cells

Abstract

Pluripotent stem cell-derived immunocompetent organoids and assembloids, termed here immunoids, are becoming important human models for studying immune–tissue interactions. Advances across organ systems now include incorporation of mature and immature immune populations into multiple organoid types, multilineage differentiation, and endogenous immune co-development. Defining standards for immunoid fabrication will be essential for improving physiological modeling and clinical translation. In the coming years, as these platforms mature, immunoids are poised to transform research by advancing understanding of human development, disease mechanisms, and therapeutic responses, becoming a cornerstone of precision medicine.

New Approach Methodologies

Despite decades of reliance on animal models to study human development, disease mechanisms, and therapeutic responses, fundamental species-specific differences in genetics, physiology, and immune regulation have limited the translational application of preclinical findings. High attrition rates in drug development^1–3 underscore the limitations of animal systems to reliably predict human efficacy and toxicity, while ethical, cost, and scalability constraints further restrict their use for mechanistic discovery and high-throughput screening. Together, these challenges have driven growing demand for new approach methodologies (NAMs)⁴ that more faithfully capture human-specific cellular behaviors, developmental programs, and pathophysiology. In this context, NAMs complement and help reduce traditional animal use in research, but at present they do not entirely replace the need for animal models due to many limitations. Since last year, however, the FDA has issued general guidance on the use of NAMs in drug development⁵ and the NIH has announced an initiative to prioritize NAM research⁶, both signs of the growing interest these models have generated and indicators of future utilization in basic science and pre-clinical research in the years to come. In this perspective, we will be focusing on self-organizing human organoids and assembloids with immune components, a specific type of NAMs that have experienced enormous growth and raised sharp interest in recent years.

From Organoids to Assembloids

Organoids are 3D cellular constructs that recapitulate, to some significant extent, the structure and function of an organ. Recent progress in organoid technology development was enabled by major advances in three-dimensional (3D) culture technologies, including Bissell and colleagues’ demonstration that extracellular matrix supports epithelial organization and differentiation⁷, followed by seminal work by Clevers and colleagues in establishing stable, long-term epithelial organ-like systems using adult intestinal stem cells⁸. In parallel, advances in the culture of human pluripotent stem cells (hPSCs)⁹ and the invention of iPSCs¹⁰ enabled the establishment of differentiation protocols for most cell types in the human body. Further advances in 3D hPSC culture and differentiation techniques led to the development of hPSC-derived organoids that capture human-specific tissue morphogenesis, multicellular self-organization, and function in vitro¹¹. Assembloids¹², a progression of organoid technology, are multi-tissue 3D models generated by integrating different organoid types (from the same or different organs), or by strategically integrating specialized cell types into developing organoids in ways that give rise to emergent properties not present in the individual components alone. A notable example combined hPSC-derived precursors from each of the three primitive germ layers to create an intestinal assembloid featuring glands, smooth muscle layers, and enteric neurons¹³. Assembloids aim to reconstruct higher-order cellular crosstalk within controlled human 3D systems to model the dynamic inter-tissue communication that facilitates organogenesis, organ maturation, cell migration, cell integration, and disease-relevant phenotypes.

Immunoids: Immunocompetent Organoids and Assembloids

The development of complex organoid models incorporating immune-related cells is the focus of this Perspective and has expanded significantly in recent years, with hPSC-derived immunocompetent organoids or assembloids (termed here “immunoids,” a portmanteau of immunocompetent organoids/assembloids) at the forefront. These systems integrate multiple developmentally distinct hPSC-derived lineages within a single 3D architecture, including at least one immune lineage that exhibits functional immune activity. Immunoids can be generated by mixing immune cells or their lineage-specified progenitors with organoids, or through endogenous co-development of immune cells within organoids. Integrating functional immune cells within human organoids enables immunoids to model immune–tissue crosstalk across development and disease and to explore therapeutic applications directly in humans (Figure 1). It is important to realize that all immunoid models to date are only partially immunocompetent, a fact that should not be surprising given the complexity of the human immune system and current challenges in hPSC immune differentiation protocols. Despite that, immunoids continue to increase in sophistication, complexity and scientific utilitiy, as described later in more detail.

Figure 1.

Schematic representing the applications of hPSC-derived immunoids.

Scope and Other Notable 3D New Approach Methodology Contributions

This Perspective highlights self-organizing approaches for organoids and assembloids composed entirely of autologous human PSC-derived lineages (with the notable exception of cancer immunoids, in which the cancer component is cancer tissue), that demonstrate immune-cell integration, in situ immune function, and physiologically relevant integration or co-development. We prioritized inclusion based on adherence to the self-organization principle^14,15, whereby homogeneous hPSC populations undergo tissue organization and morphogenesis through intrinsic developmental programs, rather than through the post hoc assembly of independently generated diverse cell types¹⁶. For models with endogenous immune-cell co-development, we discussed only tissues recognized as sites of physiological hematopoiesis, excluding some notable studies¹⁷. The literature discussed was selected for its novelty, scientific impact, recency, methodological rigor, and contributions to the field. Our inclusion criteria intentionally define a focused subset of newer models inside a much broader landscape of 3D immune NAMs with a wide variety of other approaches, such as partially hPSC-derived models¹⁸, non-autologous immunoids^19–24, human primary cell-derived immunoids^25,26, mammalian (but not human) organoid models^27,28,29, co-culture systems³⁰, engineered tissue platforms^31,32,33,34, and organotypic models³⁵. Although these approaches are important components of the expanding NAMs landscape, they fall outside the scope of this Perspective, which focuses specifically on self-organizing human PSC-derived immunoids.

Approaches to Building Self-Organizing hPSC-Derived Immunoids

We reviewed the literature from the last 10 years (Table 1), and categorized approaches used to generate autologous hPSC-derived human immunoids (Figure 2): (1) adding terminally differentiated immune cells to an organoid, (2) adding immune cell progenitors to an organoid, (3) mixing organoid progenitors with immune/hematopoietic progenitors and allowing multilineage co-differentiation into an immunocompetent construct, and (4) endogenous co-development of immune cells within organoids. While these four strategies can produce immunoids, their developmental plausibility depends on context. Many tissue-resident macrophage populations, including microglia and macrophages, originate from early embryonic hematopoiesis (yolk sac-derived erythro-myeloid progenitors)^36–38 and are later maintained by local self-renewal rather than continuous replacement from adult bone marrow; therefore, immunoids that introduce immune progenitors into developing organoids may more closely parallel early developmental seeding than late addition of mature immune cells. In contrast, disease models intended to emulate inflammatory recruitment, or acute immune activation, may be better served by introducing terminally differentiated, activation-competent immune cells, particularly when the experimental goal is to capture cytokine-driven pathology or immune-mediated injury responses. Furthermore, homeostatic hematopoiesis occurs in the embryonic yolk sac, fetal liver (primary site), fetal spleen (secondary site), and bone marrow, whereas extramedullary hematopoiesis, which occurs in pathological states, primarily occurs in the spleen, liver, and lymph nodes. Thus, immunoids generated using an endogenous immune cell co-development strategy for tissues where hematopoiesis is known to occur are more biologically relevant than tissues where hematopoiesis is unknown to occur. To date, most immunoids have been generated by adding terminally differentiated immune cells to established organoids.

Table 1.

List of hPSC-derived self-organizing immunoids.

Type	Authors	Immune Cells in situ	Immune Cell Baseline Marker Expression	Integration Strategy	Disease Modeled with Immunoid
Alveolar / Lung	Heo et al.40, 2021	Alveolar Macrophages	CD45, CD11b, CD11c, CD192, CX3CR1, CD14, CD169, CD206	1	Pulmonary Fibrosis via supplementation with TGFβ1
Alveolar / Lung	Seo et al.72, 2022	Alveolar Macrophages	CD45, CD14, CD86	1	N/A
Alveolar / Lung	Kang et al.41, 2025	Alveolar Macrophages	CD45, CD14, CD11c, IL1β, IL6, CD16, CD68, MSR1, MARCO ALOX5, CD163, C1QA	1	Mycobacterium Tuberculosis infection
Bone Marrow	Frenz-Wiessner et al.42, 2024	Macrophages	CD45, CD14, CD11b, LYZ, SPP1, HLA-DRA	4	VPS45 deficiency via CRISPR-Cas9-mediated gene editing
		Neutrophils	CD45, CD11b, CD24, CD49d, CD101, CD35, CD16, S100A8/A9, MPO, CLC, NCF1
		Mast Cells	CD45, KIT, TPSAB1, GATA2, HPGDS
		Megakaryocytes	CD45, CD41, CD42, CD61, CMTM5, PPBP, PF4, TREML1
		Eosinophils / Basophils	CD45, GATA2, CLC, HDC
		Dendritic Cells	CD45, LYZ, CLEC9A, CD1C, IRF8
Brain	Abud et al.39, 2017	Microglia	CD45, CD11b, IBA1, P2RY12, TREM2, CX3CR1, TGFBR1, MERTK, PROS1, ITGB5, CD14, CD16, CABLES1, PROS1, SLCO2B1, PPARD, GPR34, BHLHE41, OLFML3, APOE, SLC7A8, CRYBB1, CSF1R	1	Central nervous system injury via needle puncture
Brain	Ormel et al.73, 2018	Microglia	IBA1, PU.1, CD68, CD11b	4	N/A
Brain	Lin et al.46, 2018	Microglia	IBA1, CD11b, APOE, TREM2	1	Alzheimer’s disease via CRISPR-Cas9-mediated gene editing
Brain	Muffat et al.74, 2018	Microglia	IBA1, PU.1	1	Zika virus infection
Brain	Song et al.75, 2019	Microglia	CD45, CD11b, IBA1, P2RY12, CX3CR1, PGE2, TGFβ1, MMP9, IL34, ITGAM, ENTPD1, GPR34, CX3CR1, TLR4, CD163, CD14, CSF1R, RUNX1, CNR1, CTSD, ITM2B, IRF8, GAS6, SELPLG, MERTK, ITGB5, TMEM119, CD200R1, CD74, TLR2, CD274, CD68, CST3, PROS1, SIRPA, TLR3, CSF1, TYROBP, P2RY11, P2RY14, P2RY1, P2RY4, P2RY2, P2RY6, APOE, PTK2B	1	N/A
Brain	Bejoy et al.76, 2019	Microglia	CD45, CD11b, IBA1, CX3CR1, P2RY12	1	N/A
Brain	Wörsdörfer et al.77, 2019	Microglia	IBA1	3	N/A
Brain	Ao et al.78, 2021	Microglia	IBA1, CD68	1	Neuroinflammation via lipopolysaccharide and opioid exposure
Brain	Fagerlund et al.79, 2021	Microglia	IBA1, PU.1	2	N/A
Brain	Xu et al.43, 2021	Microglia	PU.1, CD45, IBA1, CD11b, TMEM119, CD68, AXL	3	Zika Virus Infection
Brain	Jin et al.80, 2022	Microglia	CD45, TMEM119	3	Down syndrome via patient-derived hiPSCs
Brain	Sabate-Soler et al.81, 2022	Microglia	IBA1, PU.1, TMEM119, CD45, P2RY12	2	N/A
Brain	Park et al.44, 2023	Microglia	IBA1, P2RY12, CX3CR1, CD45, APOE, ABCA1, PLIN2, ABCG1	2	N/A
Brain	Fagiani et al.82, 2024	Microglia	IBA1, TMEM119, CD11b, CD45, TYROBP, HLA-DRA, TREM2, CX3CR1, C3, CSF1R, CD74, RUNX1, C1QB	1	Multiple Sclerosis via administration of cerebrospinal fluid from patients to the organoid culture
Brain	Lange et al.45, 2025	Microglia	IBA1, CD14, CD16, CD11b, CD68, LGMN, C1QC, C1QA, C1QB, ITGAM, LGALS3, CD45, TREM2, ABCA1, ABCG1, APOE, AXL, CD36, CLEC7A, CX3CR1, MRC1, P2RY12, SPP1	1	Toxin-induced demyelination of the central nervous system
Brain	Buonfiglioli et al.83, 2025	Microglia	IBA1, CD45, CD11b, P2RY12, TREM2, CD206−, CD14−	2	Neuroinflammation via administration of either lipopolysaccharide, interferon-γ, interferon-α, or interleukin-6 to organoid medium
Brain	Becerra-Calixto et al.84, 2025	Microglia	CD45, CD11b, TREM2, P2RY12, IL-6, IL-10, TNF-α, IBA1	1	Alzheimer’s disease via patient-derived hiPSCs
Heart	O’Hern et al.48, 2025	Macrophages	CD45, CD163, CD14, CD68, SPP1, CCR2−, MERTK, CSF1R, AIF1, CD74, MRC1, SRGN, MS4A7, C1QA, FCGR3A, CCL4, APOE, LYZ	2	Atrial fibrillation via administration of lipopolysaccharide, interferon-γ, interleukin-1β to organoid medium
Intestinal	Tsuruta et al.49, 2022	Macrophages	IBA1, CD14−, CX3CR1, TLR4	1	Crohn’s disease via patient-derived hiPSCs
Intestinal	Tominaga et al.50, 2025	Macrophages	CD163, CD45, CD14, CD11c, CSF1R, SPI1, C1QA, C1QC, FCGR1A, CD4, FCGR3A, CD68, MRC1, FOLR2, LYVE1, AIF1, ITGAM, NINJ1, F13A1, APOE, CST3, H3F3B, STAB1, CD74, GPNMB, S100A11, S100A4, HLA-C, MS4A7, MAF, ATOX1, CCL2, CCL3, CCL4, IL1B	1 and 2	Intestinal infllammation via Escherichia coli infection and lipopolysaccharide administation to organoid medium
Liver	Ouchi et al.85, 2019	Kupffer Cells	CD68, EMR1, IL-6, CD14, FLT1, TLR4, GPBAR1	4	Steatohepatitis via oleic acid administration to organoid medium and patient-derived hiPSCs with Wolman disease
Liver	Harrison et al.86, 2023	Kupffer Cells	CD68, VSIG4, CD14	4	N/A
Liver	Li et al.51, 2024	Kupffer Cells	CD45, CD14, CD163, CD68, MARCO, LYZ, PLAUR, VSIG4, C1QC, MRC1, C1QB, CPVL	3	Sepsis-associated liver dysfunction via lipopolysaccharide and interferon-γ administration to organoid medium
		Megakaryocytes	CD45, CD41a, CD42b
		Granulocytes	CD45, CD66b
Liver	Lee et al.87, 2024	Kupffer Cells	CD11b, CD14, CD68, MARCO, CD163, FCGR3A	1	Nonalcoholic fatty liver disease via initial hepatitis C viral infection and subsequent fatty acid suplementation to organoid medium
Liver	Rezvani et al.52, 2025	Macrophages	IBA1, CD45, CD14, CD16, CD163, CD68, CD64, CD206, IL-8, CSF1R, CR1, LYZ, CCR2, HLA-DRA, VSIG4, MARCO, FCGR3A, C1QA, SPP1, TREM2	4	Steatotic and lipotoxic injury via oleic acid and palmitic acid administration to organoid medium
		Neutrophils	CD45, CD66b, CD15, IL-8, MPO, PRTN3, CXCR2, S100A9, S100A8, CECAM8, S100P, S100A12, LYZ
		T-Lymphocytes	CD45, CD3, CD5, CD27, CD8α, KLRB1, CD3E, CD7, IL32
		B-Lymphocyte Precursors	JCHAIN, IGLL1, CD79B, MS4A1, IGKC, CD19, LTB
		Megakaryocytes	PF4, ITGA2B, ITGB3, CLEC1B
Ocular	Shiraki et al.88, 2022	Microglia	CD11b, TMEM119, CX3CR1	4	N/A
Ocular	Bartalska et al.89, 2022	Microglia	IBA1, PU.1, CD163	4	N/A
Ocular	Gao et al.53, 2022	Microglia	CD45. CD11b, CD68, IBA1, P2RY12, CX3CR1, TMEM119, IL-8, IL-6, IL-10	1	N/A
Ocular	Usui-Ouchi et al.90, 2023	Microglia	IBA1, CD45, CD11b, TMEM119, C1QC, C1QA, C1QB, TREM2, SPI1	2	N/A
Ocular	Chichagova et al.91, 2023	Microglia	IBA1, CD14, CX3CR1, IL-8	1	Retinal inflammation via lipopolysaccharide administration to organoid medium
Ocular	Schmied et al.54, 2025	Microglia	IBA1, CD45, PU.1, P2RY12, TREM2, MIF, SerpinE1, C1QA, CX3CR1	2	Fetal viral infection via polyinosinic:polycytidylic acid administration to organoid medium
Ocular	Chen et al.92, 2025	Microglia	IBA1	4	Retinal inflammation via lipopolysaccharide administration to organoid medium
Thymic	Ramos et al.55, 2023	T-Lymphocytes	CD45, CD3, TCR, CD4, CD8, CD25, CD117, CD5, CD7	3	N/A
		Natural Killer Cells	CD3−, CD56
		Natural Killer T-Cells	CD3, CD56
Vascular	Chau et al.56, 2024	Macrophages	CD45, CD68, TLR8, TLR4	3	SARS-Cov-2 infection

Figure 2.

Schematic depicting the different approaches for adding or inducing immune cells in an hPSC-derived immunoid.

The Current Landscape of hPSC-Derived Self-Organizing Immunoids

The earliest example of an hPSC-derived immunoid was reported by Abud et al.³⁹ in 2017, who demonstrated that hPSC-derived IBA1⁺ microglia, when added to mature brain organoids, integrated via self-assembly and responded to needle-puncture injury. Although the microglial response to injury was inferred from morphological shifts, this study nonetheless represented the first identifiable example of an hPSC-derived immunoid, establishing proof of concept that hPSC-derived microglia can be incorporated into 3D brain organoids through self-assembly alone. Since then, immunoids have been engineered for alveolar/lung^40,41, bone marrow⁴², brain^43–46, cancer⁴⁷, heart⁴⁸, intestinal^49,50, liver^51,52, ocular^53,54, thymic⁵⁵, and vascular tissue⁵⁶, enabling the modeling of many pathologies (Figure 3).

Figure 3.

Schematic overview of disease models using hPSC-derived immunoids.

Alveolar and Lung

The human lung arises from coordinated interactions between endoderm-derived epithelial compartments and mesoderm-derived stromal, vascular, and immune lineages, resulting in one of the most structurally and cellularly complex organs in the body. This complexity is compounded by the lung’s prolonged developmental timeline, with full alveolar maturation occurring late in gestation and premature neonates frequently requiring respiratory support. Consequently, in vitro lung organoid differentiation protocols⁵⁷, particularly those modeling alveolar structures, are typically longer and more intricate than those for other organs, making the generation of lung immunoids a substantial technical challenge. However, several recent immunoid studies, using a terminally differentiated immune cell-to-organoid strategy, have begun to address this complexity challenge^40,41. Heo et al.⁴⁰ generated macrophage-containing alveolar immunoids by dissociating hPSC-derived alveolar epithelial cell cultures to single cells and co-aggregating them with stage-matched hPSC-derived alveolar macrophages (CD11b/c⁺CD192⁺CD14⁺CX3CR1⁺CD169⁺CD206⁺). This approach generated stable 3D aggregates that retained fluorescently labeled macrophages for downstream assays, and the model was used to capture alveolar macrophage contributions to pulmonary fibrosis and test the efficacy of transforming growth factor beta (TGF-β)–pathway inhibitors. Another contribution last year⁴¹ differentiated hPSCs into alveolar epithelial organoids, dissociated into single cells, and then resuspended within Matrigel domes with mature alveolar macrophages (CD45⁺CD14⁺CD11c⁺). This immunoid model was used to study alveolar macrophage activation in response to lipopolysaccharide (LPS) stimulation and Mycobacterium tuberculosis infection. Enhancing macrophage access to epithelial cells in alveolar compartments, and more advanced systems incorporating epithelial, endothelial, and stromal compartments with a controlled air-liquid interface, remain as challenges. Addressing these would provide human-specific models of neonatal and pediatric lung disease, infection-driven inflammation, and responses to inhaled particulates and environmental toxicants, where alveolar macrophage-mediated clearance, lipid handling, and cytokine signaling are central determinants of pathology and recovery.

Bone Marrow

The bone marrow is the definitive site of hematopoiesis in mammals after fetal development. In 2024, the first complex human bone marrow organoids (BMOs) were differentiated from hPSCs⁴². The team used feeder- and serum-free conditions and a stepwise differentiation protocol, which involves embryoid body patterning through mesoderm and hemogenic endothelium, to generate a self-organized 3D marrow-like microenvironment containing endothelial, mesenchymal/stromal, and hematopoietic compartments within ~3 weeks. The resulting BMOs formed an in vivo–like vascular network with multipotent mesenchymal stem/progenitor populations and endogenous co-development of immune cells, including monocytes/macrophages (CD45⁺CD14⁺CD11b⁺), neutrophils (CD45⁺CD11b⁺CD24⁺CD49d⁺CD101⁺CD35⁺CD16⁺S100A8⁺S100A9⁺MPO⁺), mast cells (CD45⁺KIT⁺TPSAB1⁺GATA2⁺HPGDS⁺), megakaryocytes (CD45⁺CD41⁺CD42⁺CD61⁺), eosinophils/basophils (CD45⁺GATA2⁺CLC⁺HLC⁺), and dendritic cells (CD45⁺LYZ⁺CLEC9A⁺CD1C⁺IRF8⁺). The authors used BMOs to model VPS45 deficiency, which adversely affects intracellular vesicular transport and endosomal sorting of the immune system and bone marrow. They were able to model the inborn errors of hematopoiesis, including myelofibrosis-like stromal changes, myeloid abnormalities, and increased neutrophil apoptosis. Bone marrow organoid-on-a-chip approaches, particularly when coupled to distal tissue organoids, represent a logical next step toward modeling immune cell circulation and systemic immune coordination in a fluidic context.

Brain

In vivo, brain development⁵⁸ and homeostasis are tightly regulated by a complex innate immune niche composed of resident glial populations like microglia, oligodendrocytes, and astrocytes that facilitate neurogenesis, synaptic pruning, circuit refinement, and inflammatory responses⁵⁹. Early brain immunoid studies demonstrated that hPSC-derived microglia could be incorporated into cerebral organoids and influence disease-relevant phenotypes. One example⁴⁶ integrated terminally differentiated microglia (IBA1⁺APOE⁺TREM2⁺) into cerebral organoids and high-resolution imaging confirmed microglial integration into neural tissue. The authors then used isogenic APOE3 and APOE4 induced pluripotent stem cell (iPSC)-derived cells to model Alzheimer’s disease-associated phenotypes, demonstrating that APOE4 microglia-like cells exhibited altered morphology and reduced amyloid-β uptake, accompanied by increased amyloid-β aggregation in APOE4 cerebral organoids. Years later, a different approach⁴³ generated immunoids by combining neural and mesodermal progenitors, allowing the resulting cerebral organoids and microglial populations (PU.1⁺CD45⁺IBA1⁺CD11b⁺TMEM119⁺CD68⁺AXL⁺) to reciprocally influence each other’s differentiation, development, and maturation. In 2023, a brain immunoid model⁴⁴ was established by deriving both neural and microglial lineages from the same patient-derived iPSC source. At matched developmental stages, microglial progenitors were incorporated into cerebral organoids, where they engrafted, matured into microglia-like cells, and persisted over the long term under macrophage colony-stimulating factor (M-CSF)–supported conditions. Integrated microglia (IBA1⁺P2RY12⁺CX3CR1⁺CD45⁺APOE⁺) promoted neuronal maturation and functional development of organoids. Last year, a protocol⁴⁵ integrated terminally differentiated microglia into myelinated brain organoids to model myelin injury and repair. Functional relevance of immune incorporation was demonstrated through robust assays of myelination, demyelination, and remyelination, establishing a role for microglia (IBA1⁺CD14⁺CD16⁺CD11b⁺CD68⁺LGMN⁺C1QC⁺C1QA⁺C1QB⁺ITGAM⁺LGALS3⁺CD45⁺TREM2⁺) in regulating oligodendrocyte dynamics and myelin repair. Microglia-containing brain immunoids could model opportunistic infections in immunocompromised settings, including those caused by cytomegalovirus, Cryptococcus neoformans, and Toxoplasma gondii, by quantifying tropism, microglial activation, cytokine production, neuronal injury, and pathogen persistence. Incorporation of adaptive immune populations could serve as a relevant model of demyelinating diseases such as multiple sclerosis and Guillain-Barré syndrome by directly measuring immune-driven demyelination, oligodendrocyte injury, axonal dysfunction, and remyelination.

Cancer

Recognition that the immune system can both restrain and promote cancer development has revolutionized our understanding of malignancy and enabled immunotherapies that can produce durable and, in some cases, curative responses in patients with advanced disease. As cancer therapy increasingly depends on manipulating complex interactions between tumor cells, immune populations, and the surrounding microenvironment, there is a growing need for experimental systems that faithfully capture these dynamic relationships. Despite not yet being fully hPSC-derived, cancer immunoids integrate 3D tumor organoids with functional immune and stromal components and represent a new generation of human model systems that bridge the gap between reductionist cell culture and human disease⁶⁰. An exemplary case⁴⁷ integrated matched patient tumor cells and peripheral blood mononuclear cells into hPSC-derived human cortical organoids to create an in vitro model that captures the complexity of the in vivo tumor microenvironment. Conversely, cancer immunoids can be generated by reconstituting patient-derived tumor organoids or xenograft-derived organoids with iPSC-derived endothelial cells, fibroblasts, macrophages, dendritic cells, natural killer cells, and other stromal or immune populations. Alternatively, genetically engineered iPSCs carrying clinically relevant oncogenic mutations can be differentiated into tumor organoids, or fused with normal tissue organoids, to study tumor initiation, progression, immune surveillance, and early cancer–niche interactions. Together, these approaches could support mechanistic studies of tumor evolution, immune evasion, therapeutic resistance, and scalable testing of immunotherapies, biomarkers, and combination treatments.

Heart

The heart develops in close coordination with resident immune populations that regulate tissue patterning, electrical maturation, extracellular matrix remodeling, and responses to injury. A human heart–macrophage immunoid⁴⁸ was recently developed by integrating autologous monocytes (macrophage progenitors) into developing human heart organoids. Heart organoids were produced by aggregating hPSCs and sequentially modulating Wnt signaling to drive mesoderm induction and cardiac specification. In parallel, hPSCs were differentiated into monocytes via staged embryoid body–based hematopoietic differentiation. CD45⁺CD163⁺CD14⁺CCR2− monocytes were introduced into heart organoids beginning at early developmental stages through repeated additions, enabling stable engraftment, differentiation into tissue-resident macrophages (CD45⁺CD163⁺CD14⁺CD68⁺SPP1⁺CCR2⁻), and persistence within the cardiac tissue. This study showed that macrophages contributed to the development and homeostasis of the cardiac immunoid by altering extracellular matrix organization, modulating electrophysiology, and participating in cell-cell signaling. Furthermore, mature cardiac immunoids exposed to pro-inflammatory cytokines demonstrated atrial arrhythmias through activation of the NLR family pyrin domain-containing 3 (NLRP3) inflammasome in atrial cells. These immunoids captured immune-electrical coupling and arrhythmia phenotypes, but should be expanded to include mast cells, recruited macrophages, and neutrophils, with the engineering of perfusable cardiac-vascular immunoids to study macrophage-driven lipid handling, endothelial dysfunction, inflammatory remodeling, atherosclerosis, and plaque destabilization.

Intestinal

The gastrointestinal tract is a major immunological organ where immune cells are exposed to numerous environmental antigens and foreign cells; thus, modeling immune–epithelial crosstalk in the gut will yield important insights. Interestingly, the most widely cited and recognized intestinal immunoid publications used donor intestinal tissue specimens, which effectively model the intestinal epithelium and enhance immune cell diversity. Recent work promoted microfold cell formation⁶¹, integrated hPSC-derived macrophages⁶², and autologous T cells⁶³ into intestinal organoids derived from patient intestinal resections. Despite the extensive work modeling adaptive immunity in intestinal organoids, there has also been work generating fully hPSC-derived intestinal immunoids. Earliest autologous hPSC-derived intestinal immunoid studies incorporated macrophages into intestinal organoids and examined their ability to engraft, mature, and respond to inflammatory cues. The first of these models⁴⁹ differentiated hPSCs into suspended intestinal organoids using a heregulin-1β, insulin-like growth factor 1, and fibroblast growth factor 2 differentiation while, in parallel, generating autologous monocyte-like cells that matured into macrophages (IBA1⁺CD14⁻CX3CR1⁺TLR4⁺). Immune incorporation was achieved by microinjecting macrophages into intestinal organoids supplemented with M-CSF, enabling macrophage engraftment within the epithelial tissue. Immune activity was supported by bulk transcriptional evidence of cytokine induction, positioning the system as a preliminary model of intestinal innate immune responses. The most recent intestinal immunoid model⁵⁰ integrated macrophages into human intestinal organoids using two different strategies: (i) integrating erythromyeloid progenitors generated from colonic organoid differentiation into intestinal organoids and (ii) integrating terminally differentiated macrophages into intestinal organoids. Each immunoid displayed stable immune incorporation and maturation of tissue-resident macrophage–like populations (CD163⁺CD45⁺CD14⁺CD11c⁺). Functional molecular, immunophenotypic, and flow cytometric analyses confirmed immune integration, and application of inflammatory stimuli (E. coli infection, LPS, and interleukin-10) induced macrophage-dependent morphological and transcriptional changes. With enteric neuron integration, these systems could also capture neuro–immune–epithelial signaling relevant to irritable bowel syndrome and serotonergic modulation, while additional incorporation of adaptive immune cells will be necessary to model Crohn’s disease, ulcerative colitis, and therapeutic responses.

Liver

The liver is an immunologically active organ enriched in resident innate immune cells, particularly Kupffer cells, that regulate maturation, metabolic homeostasis, and inflammation; it also supports fetal hematopoiesis and extramedullary hematopoiesis under postnatal pathological conditions. Recapitulating these tightly coupled epithelial–immune interactions has driven the development of liver immunoids^51,52. A notable example, utilizing an immune cell progenitor and organoid progenitor integration strategy⁴⁹, differentiated liver organoids by co-seeding hPSC-derived hepatic endoderm, endothelial, and stromal progenitors in 3D micro-dimple plates, enabling rapid self-organization under hepatocyte-supportive culture conditions⁵¹. In parallel, hPSCs were differentiated into erythromyeloid progenitors (KDR⁺CD34⁺CD144⁺CD73⁻) using a Wnt-modulation approach, which were introduced into the developing liver organoid microenvironment and, with M-CSF supplementation, gave rise to hematopoietic populations, including Kupffer cells (CD45⁺CD14⁺CD163⁺CD68⁺CD80⁺), megakaryocytes (CD45⁺CD41a⁺CD42b⁺), and granulocytes (CD45⁺CD66b⁺). LPS and interferon-γ (IFN-γ) exposure induced sepsis-like liver dysfunction, marked by pro-inflammatory Kupffer cell phenotypes, such as increased interleukin-6 and lactate dehydrogenase release, reduced CYP3A4, CYP1A1, and CYP1A2 activity, and impaired ammonia metabolism compared with liver organoids lacking immune cells. Leveraging the hematopoietic capacity of the liver, another group⁵² generated liver immunoids by directing hPSC clusters through concurrent endoderm and hemogenic mesoderm induction using Activin A, BMP4, FGF4, and Wnt signaling, while omitting hepatocyte maturation factors, including oncostatin M, dexamethasone, hepatocyte growth factor, and retinoic acid, to permit parallel hepatic and hematopoietic specification. Embedding clusters in inverted Matrigel domes promoted hepatoblast/epithelial structures alongside KDR⁺CD235a⁺ hematopoietic progenitors and subsequent in situ hematopoiesis. Immunophenotypic profiling identified macrophages (IBA1⁺CD45⁺CD14⁺CD16⁺CD163⁺CD68⁺CD64⁺CD206⁺IL-8⁺) and neutrophils (CD45⁺CD66b⁺CD15⁺IL-8⁺MPO⁺) with transcriptional evidence of B lymphocyte precursors (JCHAIN⁺IGLL1⁺CD79B⁺MS4A1⁺IGKC⁺CD19⁺LTB⁺) and megakaryocytes (PF4⁺ITGA2B⁺ITGB3⁺CLEC1B⁺). The authors performed a T cell induction assay with these immunoids, generating CD7⁺CD5⁺ precursors, 12.6% of which were CD4⁺CD8a⁺. Together, these studies establish liver immunoids as platforms for modeling inflammatory and metabolic liver disease while underscoring the need for improved vascularization, zonation, immune diversity, and drug-metabolizing maturity.

Ocular

Ocular tissues, including the retina and cornea, are highly specialized neuroepithelial tissues whose development and homeostasis are tightly regulated by resident immune surveillance, primarily mediated by microglia that sculpt neural circuits, clear debris, and modulate inflammatory responses. In 2022, co-culture of hPSC-derived retinal organoids and terminally differentiated microglia (CD45⁺CD11b⁺CD68⁺IBA1⁺P2RY12⁺CX3CR1⁺TMEM119⁺) resulted in microglial migration and integration within retinal organoids⁵³. Functional competence was demonstrated through phagocytic E. coli assays, cytokine release, persistence and proliferation within the tissue, and alterations in cell populations in the retinal organoid, highlighting the impact of microglia on hPSC-derived retinal organoids. Recently, these immunoids have been used to model ocular disease pathology. For example, a retinal immunoid⁵⁴ was generated by co-culturing retinal organoids with microglia (IBA1⁺CD45⁺PU.1⁺P2RY12⁺TREM2⁺MIF⁺). This immunoid was used to model fetal viral infection via polyinosinic:polycytidylic acid (poly I:C) treatment and therapeutic responses to ibuprofen, an anti-inflammatory drug. Future work in ocular immunoids should incorporate inflammatory challenge assays, infiltrating immune cells, and vascular or blood–retinal barrier components, in addition to immune cell morphology and localization, to provide a relevant model of retinal disease, ocular toxicity, and drug penetration.

Thymic

Thymus immunoids enable elucidation of human-specific mechanisms of T lymphocyte development and the modeling of adaptive immunity. In 2023, a fully hPSC-derived thymus immunoid was developed⁵⁵ by differentiating hPSCs into splanchnic mesodermal cells, hematopoietic progenitor cells, and thymic epithelial cells in parallel, aggregating them in a 1:1:20 ratio, respectively, and culturing them for 4 weeks. The authors later went on to show the immunoids produced natural killer cells (CD3⁻CD56⁺), natural killer T-cells (CD3⁺CD56⁺), T-cells (CD45⁺CD3⁺TCR⁺CD4⁺CD8⁺CD25⁺CD117⁺CD5⁺CD7⁺) with quantification of different T-lymphocyte types: 44% of organoid cells were CD45⁺, 64.3% of CD45⁺ cells were CD3⁺, and the largest proportion of CD45⁺CD3⁺ T-lymphocytes in descending order were: CD4⁻CD8⁻, CD8⁺CD4⁻, CD8⁺CD4⁺, and CD8⁻CD4⁺ cells. The authors then demonstrated T-cell functionality with anti-CD3⁺/CD28⁺ Dynabeads^™ and noted significant increases in CD4⁺ and CD4⁺/CD25⁺ cells. In the future, thymic immunoids, in combination with other organoid models, could provide a platform for establishing hPSC-derived adaptive immunocompetency in immunoid models.

Vascular

Hematopoiesis is intrinsically vascular in origin, as definitive hematopoietic stem cells emerge directly from hemogenic endothelium via endothelial-to-hematopoietic transition at sites including the dorsal aorta, yolk sac, and placenta⁶⁴. In 2019, Wimmer et al. reported a hPSC-derived vascular organoid that contained highly developed and perfusable vasculature⁶⁵. Building off this advance, in 2024, a vascular immunoid was made⁵⁶ by mixing mesenchymal progenitor cells with endothelial progenitor cells and cultured the aggregates for 21 days. The resulting immunoids contained macrophages (CD45⁺CD68⁺) and CD31⁺ vascular tissue and was used to study the impact of severe acute respiratory syndrome coronavirus 2 infection on the vascular immunoid. Developing vascular immunoids under fluidic conditions will further enable modeling of vascular diseases in a dynamic human-specific context.

Current Limitations of hPSC-Derived Immunoids

Despite rapid progress, hPSC-derived immunoids remain limited in physiological fidelity, scalability, and translational readiness. Macrophages and microglia often express canonical markers and exhibit immune activity but retain embryonic or fetal transcriptional, epigenetic, and metabolic states⁶⁶, limiting models of aging, chronic inflammation, and immune memory⁶⁷. Immune-cell abundance, localization, migration, niche-specific signaling, and appropriate cell-to-cell ratios are also poorly defined. Functional validation remains inconsistent: although some studies demonstrate cytokine release, phagocytosis, or transcriptional activation after inflammatory challenge, others infer activity from morphology, localization, or transcriptional state, complicating cross-study comparisons and attribution of immune-cell effects on development, homeostasis, or disease. Most immunoids also exhibit only partial immunocompetence, typically containing macrophages or microglia but lacking adaptive immune cells, lymphatic structures, and systemic immune trafficking, thereby limiting models of immune coordination, tolerance, and chronic feedback. Finally, variation in hPSC lines, differentiation timing, matrix composition, and immune-cell dosing reduces reproducibility, while long-term stability and compatibility with high-throughput or manufacturing workflows remain unresolved⁶⁸. Current immunoids should therefore be viewed as complementary refinement tools rather than replacements for in vivo models.

Next-generation Immunoids - Maturity, Vascularization, and More Immunocompetence

Next-generation hPSC-derived immunoids should translate current limitations into explicit design goals. Immune ontogeny should match the biological question: developmental models may favor progenitor seeding or endogenous co-development, whereas acute inflammation, infection, or injury models may require mature, activation-competent immune cells. Broader immune complexity beyond macrophage- and microglia-dominant systems will also be needed. Incorporating adaptive immunity may require hPSC-derived thymic epithelial platforms that support T cell specification, positive and negative selection, and functional maturation, together with antigen-presenting cells and major histocompatibility complex-matched tissues to control T cell activation and retention. Vascularization and organ-on-chip approaches will be critical for modeling immune-cell recruitment, circulation, barrier function, and communication with distal tissues. However, increasing complexity will require shared or compartmentalized culture systems that support epithelial, stromal, vascular, neuronal, and immune populations despite differing growth factor, extracellular matrix, metabolic, media, and mechanical requirements. Balancing cell survival, maturation, function, cell–cell interactions, immune responsiveness, vascular perfusion, and long-term stability will remain a major technical challenge.

Design and Standardization Recommendations for hPSC-derived Immunoids

No published standards currently define the generation of physiologically faithful hPSC-derived immunoids. We therefore propose foundational design principles to improve their biological relevance and translational fidelity, including as the field expands into kidney, pancreas, skin, skeletal muscle, and reproductive tissues, for which relevant hPSC-derived immunoids remain limited. Future studies should: (1) prioritize autologous hPSC-derived cells, while permitting biologically justified non-autologous combinations; (2) quantify immune-cell identity, integration, localization, and abundance against benchmarks from native human tissues; (3) validate immune-cell function using established assays appropriate to the population studied; and (4) align immune-cell ontogeny and integration strategy with the biological context, such as incorporating hematopoietic progenitors to model organogenesis or mature immune cells to model acute or chronic organ-specific disease. Addressing these principles may establish immunoids as broadly applicable platforms for studying immune-dependent development, disease mechanisms, therapeutic responses, and personalized medicine (Table 2).

Table 2.

Proposed design principles for next-generation immunoids.

Design axis	Benchmark	Rationale	Suggested evidence or readouts
Autologous hPSC-derived lineages	Generate tissue and immune compartments from matched hPSC sources when possible; justify use of primary or allogeneic cells.	Reduces immune mismatch and unintended inflammatory activation, particularly in developmental models.	• Cell source documentation • Shared genetic background • Lineage identity • Baseline inflammatory markers or cytokines
Immune cell integration	Show that immune cells enter, persist within, and localize to relevant regions of the organoid or assembloid.	Tissue-resident immune cells occupy defined frequencies and anatomical niches that shape function.	• Imaging at both gross and cellular resolution • Flow cytometry • Single cell/nuclei or spatial profiling • Persistence over time • Comparison with human tissue atlases
Immune cell function	Demonstrate that immune cells actively respond to physiological or pathological cues within the tissue context.	Immune cells should function as active participants rather than passive additions.	• Cytokine release • Response to inflammatory challenge • Phagocytosis • Viability • Transcriptional activation • Cellular-resolution imaging
Biologically relevance	Align immune cell source, integration timing, and maturation strategy with the process being modeled.	hPSC-derived systems often resemble embryonic or fetal tissues, so immune incorporation should reflect relevant ontogeny or disease state.	• Justification of use of progenitor versus mature immune cells • Timing based on tissue development • Maturation strategies for adult disease models

Axis 1: Prioritizing the use of autologous hPSC-derived cells and organoids

Cell sources for hPSC-derived immunoids should be selected based on the biological question and the immune mechanisms being modeled. When the goal is to recapitulate normal human physiology with high biological fidelity (Axis 4), autologous hPSC-derived cells should be prioritized because they preserve a shared genetic background across cell types and avoid artificial MHC mismatch, particularly when antigen presentation or donor-specific interactions may influence the phenotype. In contrast, non-autologous combinations create a chimeric context that does not reflect normal physiology and is most biologically relevant to settings such as transplantation. Nevertheless, they may be experimentally appropriate when alloreactivity is unlikely to confound the biology, such as in studies of innate or non-MHC-restricted immune functions, or when the required immune population cannot yet be efficiently generated from hPSCs. Primary or allogeneic cells may also be warranted when specifically required by the experimental question, as in certain cancer immunoid⁶⁹ models. Thus, cell-source selection should be explicitly justified, with autologous hPSC-derived cells serving as the preferred standard for physiological modeling and non-autologous sources used as context-dependent alternatives.

Axis 2: Characterizing immune cells and immune cell integration

Immune cells should not simply be added as a co-culture component; they should migrate into, persist within, and physically incorporate into the tissue, supported by visual and quantitative evidence. Future studies should measure immune cell composition and compare it with relevant human tissue benchmarks using available datasets such as single-cell RNA sequencing atlases. When possible, dissociation-based quantification may provide more precise estimates of immune cell abundance than imaging alone. This is important because immune cells exist at tissue-specific frequencies, and their abundance can shift during pathology through recruitment or expansion. Localization should also be assessed, since tissue-resident immune cells often occupy specific anatomical niches. For example, cardiac macrophages are enriched in regions such as the atrioventricular, nodal and epicardial regions compared with parts of the ventricular myocardium⁷⁰. Imaging-based approaches are therefore needed to determine whether immune cells occupy biologically relevant compartments within the immunoid.

Axis 3: Proving immune cell functionality

Integrated immune cells should show functional activity within the immunoid model rather than acting as passive cellular components. Because immune activation and suppression are central to disease pathology, immunoid models should demonstrate that their immune cells respond appropriately to physiological or pathological cues. If immune cells persist in the tissue but fail to respond to inflammatory stimuli, remain viable, or perform expected in situ functions, then the immunoid model may not be ideal for further study. Several immunoid publications covered in this Review used, either alone or in combination, imaging data, transcriptomics, cytokine release using enzyme-linked immunosorbent assays or western blotting, and phagocytosis assays. Although some studies have used morphological changes as a proxy for immune activity, future work should include more quantitative assays with cellular resolution to demonstrate immune cell function.

Axis 4: Incorporating biologically inspired design

Immunoid design should be guided by developmental, pathological, or tissue-specific biology, including when and how immune lineages normally emerge and interact with target tissues. Because hPSC-derived immunoids most closely resemble embryonic or fetal human tissues, the timing of immune cell incorporation should be considered when modeling organogenesis. For example, tissue-resident macrophages are largely derived from yolk sac and fetal liver progenitors during development³⁶, with limited evidence of an endogenous hematopoietic origin in most organs. Failing to capture the appropriate timing of immune integration may reduce translatability. Similarly, developmental and congenital diseases may be particularly well suited to modeling in current immunoid systems, whereas adult-onset diseases require additional consideration. For example, atrial fibrillation commonly develops later in life, so studies modeling this disease should incorporate maturation strategies⁷¹, as done by O’Hern et al.⁴⁸, to push the model toward a more adult-like state. Future immunoid platforms should therefore align immune cell source, integration timing, and maturation strategy with the biological context being modeled.

O’Hern et al. review advances in modeling immunity and tissue function using self-organizing hPSC-derived immunocompetent organoids, termed immunoids. They outline priorities for advancing the field, including standardization and biologically informed design principles to improve the physiological fidelity and translational value of next-generation models.