Organoid models of the tumor microenvironment and their applications

Abstract A small percentage of data obtained from animal/2D culture models can be translated to humans. Therefore, there is a need to using native tumour microenvironment mimicking models to improve preclinical screening and reduce this attrition rate. For this purpose, currently, the utilization of organoids is expanding. Tumour organoids can recapitulate tumour microenvironment that is including cancer cells and non‐neoplastic host components. Indeed, tumour organoids, both phenotypically and genetically, resemble the tumour tissue that originated from it. The unique properties of the tumour microenvironment can significantly affect drug response and cancer progression. In this review, we will discuss about various organoid culture strategies for modelling the tumour immune microenvironment, their applications and advantages in cancer research such as testing cancer immunotherapeutics, developing novel approaches for personalized medicine, testing drug toxicity, drug screening, study cancer initiation and progression, and we will also review the limitations of organoid culture systems.


| INTRODUC TI ON
In cancer research, conventional animal models and cell culture systems have several problems, in the in vitro 2D cell cultures, the cancerderived cell lines may acquire considerable genetic mutations and fail to recapitulate the cancer genetic heterogeneity that originated from it. 1 In addition, the absence of stromal compartments and the lack of normal tissue-derived cell lines as control are another limitation of 2D cell culture systems. 1 in vitro 2D cell cultures and in vivo xenografts models are used in the pharmacological intervention, viral transduction and multiplexed drug screening studies. Genetically engineered animal models provide the dynamic context of tumour tissue vasculature and structure. 2 The generation of genetically engineered animal models is time-consuming, and it is clear that these animal models do not truly recapitulate pathogenic processes in human. 3 Both conventional in vivo and in vitro models inefficiently recapitulate the complex immune microenvironment of native tumours. 2 Humanized immuno-oncology models are generated by transplanting patient-derived xenografts (PDXs) into humanized immune system mouse models that bearing human immune cells, but time, cost, throughput and complete immunocompatibility remain challenges. 4,5 Indeed, patient-derived tumour xenografts (PDTXs) mimicking the human tumour microenvironment (TME) much better than in vitro 2D culture systems. PDTXs are generated by engraftment of freshly patient-derived tumour tissue fragments orthotopically or subcutaneously into immunodeficient mice. 1 Low reproducibility rates of results obtained from animal models, except PDX models that organoids are of human origin, in humans is one of the disappointing problems with cancer therapy development, indeed less than 10% of findings observed in these models can be translated to humans. 6 Therefore, using human physiological mimicking models are vital to reduce this attrition rate and to improving the preclinical screening.
Organoids are 3D in vitro cultures of tissues with multiple cell lineages, comprising differentiated cells and stem cells, and tissue native construction in vitro. 7,8 in vitro human organoid culture is a new approach to studying tumour immunobiology and cancer modelling. The large-scale 3D patient-derived organoids (PDOs) culture permits the establishment of large tumour biobanks that represent the histological and the genetics of their original malignancies. [9][10][11] In addition, in forward genetic strategy studies, organoids from induced pluripotent stem cells (iPSCs) or normal tissues can be genetically engineered to gain specific tumour suppressor or oncogene mutations. [12][13][14][15] Tumour organoids can recapitulate tumour (immune) microenvironment that are including neoplastic cells and nonneoplastic host components. These properties of the tumour microenvironment play a critical role in tumour behaviour such as carcinogenesis, tumour progression and metastases. 2 Importantly, these studies have shown that tumour-derived organoids, both genetically and histologically, be similar to the tumour tissue that originated from it. Currently, a large tumour organoids (3D PDO) biobank and internationally accessible for the research community has been created by cooperation of the Wellcome Sanger Institute and the foundation Hubrecht Organoid Technology, Cancer Research UK, the US National Cancer Institute (NCI) and Human Cancer Models Initiative (HCMI).
In this review, we will discuss about various organoid culture strategies for modelling the tumour immune microenvironment, their applications and advantages in cancer research such as testing cancer immunotherapeutics, developing novel approaches for personalized medicine, testing drug toxicity, drug screening, study cancer initiation and progression, and we will also review the limitations of organoid culture systems.

| ORG ANOID CULTURE SYS TEMS FOR MODELLING THE TUMOUR IMMUNE MICROENVIRONMENT
There are various organoid culture strategies for modelling the tumour immune microenvironment including (a) Reconstitution approaches, reconstituted tumour microenvironment immune components, like submerged Matrigel culture, (b) Holistic approaches, native TME immune components, like microfluidic 3D culture, and air-liquid interface (ALI) culture ( Figure 1; Table 1). These methods will be fully explained below. The term TME refers to the complex cellular milieu surrounding cancer epithelium, including mesenchymalderived cells such as fibroblasts and pericytes, blood vessels, innate and adaptive immune cellular network and extracellular matrix (ECM). TME immune cells include lymphocytes, myeloid-derived suppressor cells (MDSCs), macrophages, dendritic cells (DCs), natural killer (NK) cells, mast cells, eosinophils and populate the cancer tissue and can be infiltrated from secondary lymphoid organs, or derived from tissue-resident cell components. In the tumour tissue, the cellular and humoral components and diverse inflammatory responses of TME support the tumour progression. [16][17][18] Depending on the organoid culture strategy, TME complex cellular milieu may or may not be preserved in the organoid structure. In approaches where these cells are not preserved, exogenous cellular components can be used. Which are discussed in the following sections. Recapitulation of the vascular system and hypoxia conditions of native TME also are other important aspects of organoid culture. 19,20 Several studies in recent years addressed the issue of organoid vascularization. Takebe et al 21 indicated that condensation of mesenchymal cells, endothelial cells, and specific parenchymal cell types leads to the formation of vascularized complex organ buds. After transplantation of these vascularized buds into a mouse, the vasculature was connected to the host circulatory system and the blood was perfused through it.
Successful adaptation of this protocol has been reported for generating human complex tissues. 22 Instead of using a mixture of terminally differentiated cell types, multi-layered human blood vessels can also be generated via self-organization from iPS cell-derived mesodermal progenitor cells (MPCs), which can differentiate into all cell types of blood vessel wall. 20, 23 Wörsdörfer et al 24 have also described a method to incorporate stromal components in organoids generated from stem cells (that do not have the major components of the organ stroma) by co-culturing with induced pluripotent stem cell-derived mesodermal progenitor cells.

| Reconstitution approaches: Submerged matrigel culture
In submerged Matrigel culture system, tumour cells obtained from tumour tissue that dissociated enzymatically and physically, culture underneath tissue culture medium in mixed with a flat or dome gel of 3D Matrigel. Depending on the types of cancer tissue, various pathway inhibitors and/or growth factors are added to the culture in this approach. 13,25,26 Based on tumour type and histology, culture situations can be adapted and customized, but mostly contain some additives, such as R-spondin (RSPO), WNT3A, epidermal growth factor (EGF) and bone morphogenetic proteins (BMP) inhibitor Noggin, which help the stem cells to maintain their ability of differentiation and self-renewal. 27 These additives have also been used for the ALI culture system. 28 In tumour organoid, the niche factor requirements are mainly determined by genetic mutations and help to local tumorigenicity. 26 In contrast, during the development of advanced cancer, changing biological behaviours such as the acquisition of TGFβ/BMP resistance, were observed independently of genetic mutations. 26 Finally, advanced genetic analysis will detect new genetic mutations accounting for the cancer progression. 26 In organoids, MAPK signalling pathway mutations were detected. In contrast, no mutation in the RAS/MAPK pathway was observed in the other 6 EGF-independent colorectal cancer organoids, and 2 of these organoid lines were associated with Epiregulin overexpression. 26 Fujii et al 26 showed that 29 CRC organoid lines propagated strongly in the absence of TGFβ inhibitor. Of these, the known TGFβ pathway mutations were not detected in 18 CRC organoids. 26 Similarly, dependency for BMP inhibition was acquired partially   26 Cancer organoids have been grown in rich conditions supplied with niche factors including EGF, WNT, R-spondin and other factors, whereas, by the alternation of niche factors in the culture medium, functional selection of CRISPR-induced oncogenic mutations becomes possible, 13,14 and different cancer subtypes can be grown for establishing cognate PDOs from a mixture of different cancer subtypes. 29 For instance, TP53-mutant organoids can be selected using a medium containing the MDM2-P53 complex inhibitor Nutlin-3, whereas to select APCmutant organoids, they need to be cultured in a medium without WNT/R-spondin. 13 Studies show that PDOs in submerged Matrigel systems can facilitate drug screening and cancer modelling by simulating not only the phenotypic and genetic complexity of cancer tissues, but also by potentially modelling functional individual responses to drug and clinical treatment. 10,11,[43][44][45][46] It should be noted that typical submerged Matrigel PDOs particularly enrich epithelial tumour cells but lose their stromal components and immune cells. 27 Thus, tumour (immune) microenvironment modelling in this approach requires co-culture of PDOs with exogenous immune components such as peripheral blood mononuclear cells (PBMCs), primary leukocytes, tumour-associated macrophages (TAMs), and DCs. Therefore, one of the intrinsic limitations of submerged Matrigel culture is the lack of immune cells, blood vessels and stroma. In many studies, researchers have utilized exogenous stromal cells such as cancer-associated fibroblasts (CAFs) for the investigation of the tumour microenvironment in this technique. 25,[47][48][49] Co-culture of human pancreatic ductal adenocarcinoma (PDAC) organoids with CAFs showed that WNT produced by CAF can drive organoid growth in WNT-nonproducing PDAC subtypes. 25 Co-culture of pancreatic stellate cells, a precursor population of CAFs, with PDAC organoids provides evidence for CAF heterogeneity and reveals two distinct CAF subtypes from pancreatic stellate cells: high αSMA-expressing myofibroblast-like CAFs that located closely adjacent to tumour cells, and IL-6 and additional inflammatory mediators secreting CAFs activated by paracrine factors produced from neoplastic cells. 49 Biffi et al 47  which promote CAF heterogeneity and induce distinct myofibroblast and inflammatory CAF subtypes, respectively. Understanding the CAF heterogeneity mechanisms is essential for the development of new methods that selectively target tumour-promoting CAFs. 47 Reconstitution of organoids has also been performed by vari- T cells studies 54 have been also used.

| Holistic approaches: Microfluidic 3D culture
In holistic approaches, the small fragment of tumour tissue is pre- filters, respectively, to obtain three separate fractions including S1 (>100 μm), S2 (40-100 μm), and S3 (<40 μm). Then, the S2 fraction is pelleted in ultra-low-attachment plates and mixed in collagen to be inoculated into the microfluidic device.

| Holistic approaches: Air-liquid interface culture
In this system, in the first stage, the bottom layer of the collagen gel matrix in the inner dish is prepared. 61 For the preparation of primary tissues, tissues are removed from other parts and immediately are immersed in ice-cold medium. 61 After rinsing the tissue, the tissue is minced into small fragments and then the minced tissue is mixed into collagen solution. 61 The tissue-containing collagen gel is poured onto the inner dish with bottom layer gel matrix. 61 The completed inner dish is placed in a new empty outer dish. 61 The covered outer dish is transferred to a 37°C incubator and is allowed the gel of the inner dish to solidify. 61 After solidifying the top layer tissue-containing gel, media is added to the outer dish that can diffuse into the inner transwell dish through a permeable membrane, and the top layer of tumour fragments-collagen mixture is exposed directly to air via an ALI, allowing tumour organoids to supply their own oxygen efficiently. 7,15,61 In this method, in contrast to

| APPLIC ATIONS OF TUMOUR ORG ANOIDS
Organoids have many applications in cancer research such as testing cancer immunotherapeutics, developing novel approaches for personalized medicine, testing drug toxicity, study cancer initiation and progression, etc, some of which will be explained in more detail below.   76 which are available through institutions such as the HCMI.

One of the main limitations of organoids in immunotherapy inves-
tigations is that the epithelial-only patient-derived organoids are widely available, but their absence of immune cells hinders immunotherapy studies, such as the response to ICIs.

| Tumour organoids in immune checkpoint inhibitor studies
In 3D microfluidic cultures, the in vivo therapeutic resistance and sensitivity to PD-1 blockade can be recapitulated by MDOTS/ PDOTS for short duration cultures through the evaluation of TIL cy- Tumour organoid technologies will need prospective validation and correlation with clinical outcome but provide a considerable opportunity for clinical translation through identifying the cohorts significantly responsive to immunotherapies.

| Tumour organoids in adoptive cell immunotherapy studies
ACT immunotherapy can be a suitable alternative to ICI therapy.
In the ACT, researchers generally utilize bulk autologous TILs or alternatively genetically manipulated T cells such as CAR T-cell or high-affinity TCRs recognizing tumour-specific antigens. In these treatment strategies, antitumour lymphocytes are expanded ex vivo and then the cells are injected into the patient's body. 72 PDOs, the organoid-immune cell co-culture strategies, could be utilized to evaluate CAR T cell-mediated tumour-specific cytotoxicity in cancer and normal organoids. 79 However, CD19-targeted CAR T Cells display striking tumour cell cytotoxicity in haematological malignancies, such as acute lymphoblastic leukaemia 80 and B-cell lymphoma, 81  TA B L E 2 Overview of the currently available human-patient-derived tumour organoid (PDO) biobanks in CRC. 82 Reconstitution strategies, on the one hand, can improve the reproducibility of experiments by long-term preservation of the epithelial cells; on the other hand, the co-culture of single immune cell types with organoids might not completely recapitulate the complex interactions among various immune cell populations following treatment with immunotherapy agents, either alone or in combination.
Recently, it has been shown that peripheral blood can be used as an easily accessible source of tumour-reactive T cells, an alternative to TILs. 54 For this purpose, cancer organoids are stimulated with IFNγ to increase antigen presentation and then co-cultured with autologous T lymphocytes. 54 Treatment with IFNγ also induce the expression of PD-L1, a negative regulator of effector T cells, and to eliminate the inhibitory effect of PD-L1 on the effector T cell, it is necessary to adding blocking antibodies to PD-1. 54 To support T-cell expansion and to provide co-stimulation, anti-CD28 and IL-2 should be added to culture. 54 It should be noted that the dependency of the induced T-cell responses on IFNγ should be investigated by the untreated IFNγ control group, as well as the specificity of the response to tumour antigens should be assessed by evaluating the stimulation of T cells with organoids of autologous normal tissue. 54 It has been reported that T helper cell reactivity is not limited to tumour-derived organoids but, in some cases, is also stimulated against normal tissue-derived organoids. 54 As cross-reactivity to normal tissue-derived organoids was observed only for T helper cells and not for cytotoxic T cells, it was suggested that this could be directed against foreign antigens which are present in the culture medium. 54 Because organoids are grown in murine basement membrane matrix (Geltrex), therefore mouse antigens can be presented to immune cells. 54 The T helper cell reactivity is only observed in organoids grown in Geltrex or Geltrex-loaded DCs, but not observed in organoids grown with DC that exposed with healthy or tumour organoids or irradiated cells. 54 Of note, recently organoids cultures have expanded in synthetic matrices 83 and they can be used to escape the stimulatory properties of animal antigens.
In immunotherapy, it is crucial that cancer cells show adequate immunogenicity to provoke an appropriate immune response. [84][85][86] For cancer cells, the mutational load of a tumour, which represents the amount of neo-antigens expression, determines the rate of immunogenicity and immune responses. 84,85,87 In most cases, the potency of immune response triggered by neo-antigens is insufficient. and activating mutations in KRAS are introduced to healthy organoids, tumour growth was independent of the TME factors noggin, R-spondin-1, WNT and EGF. 13,14 It has been displayed that combined inactivating mutations in Indeed, when the same CRC organoids were transplanted into the caecal epithelium of mice by an orthotopic approach, spontaneous metastases are seen in the lungs and liver. 90,91 This orthotopic transplantation approach was also utilized to show that the loss of dependency on specific stem cell niche signals is essential to the ability to metastasize to distant sites, 90 thus approving former observations in colorectal cancer organoids. 26 Similar colorectal cancer progression models were generated using RNAi-based technologies in ALI colonic and mouse small intestinal organoids. 15 The current revolution in genome-editing strategies such as

In vitro immune cells expansion and activation
RNAi and CRISPR-Cas9 has made it possible to 'repairing' the ge-

| TUMOUR ORG ANOIDS IN DRUG TOXI CIT Y S TUD IE S
The ability of organoids generation from both tumour and healthy tissues is one of the main advantages of using organoid culture in drug development studies, which provides a powerful tool for selecting drugs that target cancer cells specifically but do not damage the healthy cells. As a result, toxicity in patients is likely reduced.
Drug-induced hepatotoxicity is the main reason for the failure of the translation of the drugs to clinical trials. 94 So recently, liver organoids culture has developed that could facilitate the preclinical screening of the hepatotoxicity of drugs. 95,96 The major mechanism of drug-induced liver injury is mediated through cytochromes P450 (CYPs), and it is suggesting that expression of these enzymes in liver organoids be close to physiological levels upon induced differentiation. 33,96 Similarly, iPSC-derived cardiac organoids could be exploited for cardiac toxicity testing, 97,98 and iPSC-derived kidney organoids were recently utilized for nephrotoxicity screening. 99 Researchers also can investigate the exciting possibility of assessing the potential cytotoxicity of healthy donor-derived T cells on patient-derived tumour organoids after the selection of neo-antigen-specific T cells derived from healthy blood donors. 100

| ORG ANOIDS IN PER SONALIZED C AN CER TRE ATMENT AND DRUG DE VELOPMENT
Although high-throughput 2D cell line screening platforms have provided major insights into the genetic background of experimental components response, 101

| Pure tumour material
Tumour tissue-derived organoids do not usually grow faster than their matching healthy organoid tissues, and, unexpectedly, in many cancers the organoid growth rate is even slower, probably due to higher rates of the mitotic failure process and following cell death. 13,109 Therefore, the overgrowth of tumour organoids can be occurred by remaining healthy tissue fragments in tumour biopsy samples, which should be avoided. So, it is necessary that tumour organoids are cultured using either pure cancer tissue materials or grow the tumour specimens under selective culture conditions. For instance, in the vast majority of CRCs, the Wnt signalling pathway proteins demonstrate gain-of-function mutations. 110 In this type of cancer, pure tumour organoid material for culture can be achieved by using WNT and R-spondins free culture media, 41 which these factors are needed for the growth of healthy tissue-derived organoids.
Similarly, cancers with activating mutations in the EGF receptor signalling pathway can be selected by EGF-free culture medium. 13,26,111 Nutlin-3, which inhibits the interaction between p53 and its negative regulator MDM2 by blocking the p53-binding domain of MDM2, has been exploited to remove healthy fragments from TP53-mutant tumour organoids. 9,13 When such selection approaches are not accessible, using pure tumour materials is a prerequisite.

| LIMITATI ON S AND PER S PEC TIVE S
Although organoid technology is promising at first glance, but organoids have their limitations as well. For instance, compared with 2D culture systems, organoid-based approaches require huge time, materials and reagent. The lack of immune cells, stroma and blood vessels is also one of the intrinsic restrictions of organoid systems. 29 The requirement for mouse-derived ECM and foetal calf serum (which is essential for the production of WNT conditioned medium in some organoids 112,113 ) is required for the organoid culture which as undefined external factors can influence the test results. 25 Another imaginable limitation may be that advanced cancers derived organoids often grow more slowly than healthy tissue-derived organoids, which probably leads to the overgrowth of tumour organoids contaminated with healthy tissue materials. 13,109 This low growth rate of tumour organoids could be due to a much higher rate of mitotic failure and subsequent cell death.
Despite these restrictions, organoid cultures can be efficiently generated from individual-patient-derived tumour tissue, making them as a more physiologically mimicking model for translational applications and the development of personalized cancer medicine.
It will be crucial that organoids can be generated and expanded efficiently to allow drug testing in a clinically meaningful time window. Although the use of organoid culture in the diagnosis of cancer relatively has been shown, the predictive value of tumour organoids in drug responses will have to come from ongoing trials.
The finding of a study comparing drug responses of the patients in the clinic with the responses of gastrointestinal tumour-derived organoids are very promising. 43 Optimizing drug testing strategies in terms of robustness and sensitivity will be crucial before organoidbased precision medicine can be implemented in the clinic.

ACK N OWLED G EM ENTS
This work was supported by the Zhejiang Provincial Science and Technology Projects (No. LGF21H160032 to TX).

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing not applicable to this article as no datasets were generated or analysed during the current study.