Preclinical Models of Adrenocortical Cancer

Simple Summary Adrenocortical cancer is a very rare form of endocrine cancer with dismal prognosis. Preclinical models such as cell lines, organoids, and mouse models are essential for both improving basic understanding of this disease and developing treatments. Herein, we review a currently available model for adrenocortical cancers, with a special focus on adrenocortical carcinoma. Recent developments in in vitro models have included cell and 3D culture models with improved recapitulation of the tumor microenvironment and genetics. We hope to improve visibility and access to these models through this review. Abstract Adrenocortical cancer is an aggressive endocrine malignancy with an incidence of 0.72 to 1.02 per million people/year, and a very poor prognosis with a five-year survival rate of 22%. As an orphan disease, clinical data are scarce, meaning that drug development and mechanistic research depend especially on preclinical models. While a single human ACC cell line was available for the last three decades, over the last five years, many new in vitro and in vivo preclinical models have been generated. Herein, we review both in vitro (cell lines, spheroids, and organoids) and in vivo (xenograft and genetically engineered mouse) models. Striking leaps have been made in terms of the preclinical models of ACC, and there are now several modern models available publicly and in repositories for research in this area.


Introduction
Cancers of the adrenal gland, such as adrenocortical carcinoma (ACC), require a unique approach in treatment, diagnosis, and research, owing to their rarity and the multifunctional environment of the adrenal gland. The adrenal gland comprises two embryologically distinct regions: the medulla and the cortex [1]. Within the medulla, chromaffin cells serve the neuroendocrine function of manufacturing epinephrine (adrenaline), a core stimulant of the fight-or-flight response [1,2]. Tumors of chromaffin cells (pheochromocytomas) are often present with classic symptoms of hypertension, headache, and syncope with an incidence rate of 0.04 to 0.95 per 100,000 [3][4][5]. Within the cortex, three distinct zones are present, the zona glomerulosa, zona fasciculata, and zona reticularis; these zones are responsible for manufacturing steroid-based mineralocorticoids, glucocorticoids, and Despite their value, neither method adequately replicates the tumor microenvironment (TME). Monolayers have demonstrated impaired cytoskeletal activity and lack cellular diversity, while suspensions do not possess adequate extracellular support [41]. Pre-clinical mouse models of ACC, which more closely capture the tumor microenvironment, have been generated; however, they are also limited. Cell line-derived xenografts (CDXs) are established using cell culture that is subcutaneously administered to immunocompromised mice for localized propagation and in vivo tumor growth [42]. Patient-derived xenografts (PDXs) are generated using tumor fragments from human tissue collected during surgical resection and then directly implanted into immunocompromised mice and are further propagated across generational passages [42]. Both CDXs and PDXs present the ability to observe in vivo human tumor progression; however, there are limitations. CDXs present with atypical histologic findings and poorer chromosomal maintenance when compared to PDXs. PDXs are limited by the supply of viable grafts, complex passaging techniques, and issues associated with clinically relevant dosing [42,43]. There are multiple transgenic mouse models as well, which focus on the manipulation of different growth factors and signaling molecules to produce an ACC-like phenotype. While there are many models present, transgenic mouse models are usually limited to one or two genetic modifications that might only partially recapitulate the heterogeneity of human disease.  This review summarizes the currently used cell line and xenograft models of ACC, which have provided valuable insights into the pathogenesis and natural history of human ACCs. We focused especially on elaborating the details of more in vitro models than current mouse models, although all types of models are included for the completeness of the survey. Newer and more accurate models of ACC are critical to furthering the early detection and targeted treatment of this disease, and a holistic understanding of current models and their advantages and deficiencies is the first step to designing better ones.

Cell Lines
Cell lines are summarized in Table 1. Notable primary cultures are summarized in Table 2. The earliest derived ACC cell line for continuous culture was Y-1, initially transplanted from an Itgal −/− mouse line, and noted for their continuous production of progesterone derivates from cholesterol [44]. The other major mouse lines, ATC1 and ATC7, were generated from tumors in transgenic mice with SV40-Tag added under the AKR1B7 promoter. The ATC1 and ATC7 lines have been mostly used in basic endocrine research toward understanding the patterning of and hormonal crosstalk within the adrenal cortex. They have also been used for some therapeutic target discovery, focusing on HOX genes with them [66][67][68].
For decades, the only continuous ACC cell lines available were SW-13 and H295, the latter of which is notable for its sustained steroid secretion even after decades of culture [49,56]. On the other hand, the current consensus is that SW-13, which never produced steroids in culture, is probably derived from a small cell lung cancer metastasis to the adrenal gland, and hence it has fallen out of use in modeling ACC [69,70]. While many more cell lines have been introduced in recent years, H295R remains the most available and heavily used line in current preclinical research [71,72]. In recent years, perhaps not surprising due to its age and passage number, the reproducibility of results across different clones of H295R has been called into question, emphasizing the need not only for new models, but also for avoiding overpassaging models [55]. The ACC HAC15 cell line, first reported in 2008, was later shown to be a subclone of H295R, which had presumably contaminated the attempted culture of a new line [70,73]. Other concerns with H295R cells include their lack of response to ACTH stimulation, which has been attributed to their low expression of its receptor, the melanocortin 2 receptor (MC2R). To remedy this, Nanba et al. used lentiviral particles to introduce the open reading frame of a protein necessary for the surface trafficking of MC2R, the MC2R accessory protein, into H295R cells, generating a strain termed H295RA with the inducible production of 11-deoxycortisol, cortisol, and androstenedione [74].
Excitingly, several more ACC lines have been reported in the last few years, including MUC-1 cell lines in 2016, and CU-ACC1 and CU-ACC2 in 2018 (all three with companion PDX lines for comparison) [55,56]. Recent work has also depended much on primary culture, with more than 40 primary ACC cultures isolated in the last seven years [75,76]. While these cultures have been shared extensively, only the ACC115m clone has been sufficiently immortalized for use as a cell line (now reported as TVBF-7) [40,65]. Hence, H295R remains the only widely available ACC cell line in repositories. Quantitative measures (such as RNA sequencing) of how cell type and function changes in and between models such as tissue culture and PDX vs. in primary tumors are improving, although detailed breakdowns of what changes in particular are present have not yet been developed [56].
Warde et al. recently showed that mitotane sensitivity correlates with intracellular lipid content; while MUC-1 and H295R cells store similar amounts of intracellular lipid droplets, MUC-1 (mitotane resistant) cells are rich in triacylglycerols, whereas H295R (mitotane sensitive) cells are rich in cholesterol esters [77]. Lipid content as measured by Hounsfield units is used in some diagnostic algorithms for ACC; however, these results show that further distinguishing the particular lipids involved may provide more useful clinical information [78,79]. Further investigation of lipid compositions of cell lines may be valuable and can potentially inform the inclusion of such analysis in the future analyses of biopsy and surgical samples to inform the precise treatment of ACC. Recent investigations have also shown crosstalk between adipose stem cells and H295R cells, reinforcing the importance of local lipid metabolism in ACC [80].

Xenografts
While patient-derived xenografts (PDXs) of samples acquired directly from biopsy or surgery into immunodeficient mice are typically recognized as the gold standard for human cancer models, limited PDXs of ACC are available [81,82]. Hence, here, we have included cell line-derived xenograft (CDX) models, which are summarized in Table 3, while PDXs are summarized in Table 4.  The first PDX model was generated from a pediatric patient with ACC and was reported in 2013. No separate cell line of this model has been established [91]. Since then, three new models have been developed, which have companion cell lines [55,56]. Further work has investigated the behavior of one of these models, CU-ACC2-M2B, in a humanized mouse model to better understand the efficacy of checkpoint inhibitor immunotherapy [56,93].
Modern ACC PDX lines not only retain significant molecular similarity (as confirmed by IHC) to their primaries, but also recapitulate the differences between those primaries and some of the heterogeneity of the disease [56,93].

3D Models
Refer to Table 5: Two primary ACC 3D models (one spheroid, one organoid) have been reportedly recently [94,95]. Prior to 2022, 3D models of ACC consisted only of spheroids generated from H295R and SW-13 cells, primarily used in drug-screening protocols [96,97]. One additional H295R-derived spheroid model was also developed last year [98]. As with cell lines, newer models are increasingly moving toward larger-scale biobank models that will enhance the heterogeneity of models available for future research, although these models are not yet publicly available [95]. In addition, a transwell model of ACC co-culture with adipose stem cells showed evidence of crosstalk and worsened disease phenotype induced by the adipose stem cells [80]. Beyond the lipid microenvironment specifics of endocrine cells, co-culture experiments are increasingly important for understanding metastasis and immune response or lack thereof, the latter of which is critical for better improving immunotherapy outcomes.
3D models are particularly promising in the complex microenvironment of the adrenal cortex as an opportunity to better recapitulate tissue zonation. Recent experiments have also looked at interactions between ATC7 cells and human monocytes, showing that activation of intra-adrenal immune cells may play a role in stimulating steroidogenesis or proliferation [66].
A 2022 work by Bornstein et al. on standardized 3D culture techniques has yielded promising results in both replicating H295R and MUC-1 data and establishing additional primary cultures of ACC successfully. Bornstein et al. also worked with bovine and porcine adrenal organoids, but this work was focused primarily on normal tissue working toward transplantation rather than disease. Notably, this comparative work on porcine and bovine organoids also made progress toward the co-culture of medullary and cortical tissue [94].
Although not yet peer reviewed, Dedhia et al. released promising organoid models of ACC, studying metastasis through matrix metalloproteinase experiments in organoids and microfluidic models [99].

Genetically Engineered Mouse Models
While this review will not go into extensive detail about current mouse models, they are summarized here for completeness, and presented in Table 6. A more thorough review which particularly focused on them was recently published by Basham et al. [100]. Relatively many models have been developed to understand adrenocortical neoplasia as opposed to other neuroendocrine, as summarized in Table 5 [101]. Early models mostly focused on the role of IGF2 [102,103]. While it has been confirmed to be involved in the development and progression of tumors, it is no longer seen as likely to be a driver of oncogenesis itself [104][105][106]. Although no longer a central focus of transgenic models, a study continues on elucidating the mechanism of IGF2's role in adrenocortical neoplasia. Pereira et al. showed that its effects on H295R cells could preferentially be inhibited by mTOR pathway inhibition vs. MEK/MAPK/ERK pathway inhibition [103].
Other recent work has focused more on CTNNB1, APC, WNT, ZNRF3, and TP53 [107]. Val and coauthors recently showed evidence that phagocytic macrophages may be involved in the relatively higher prevalence of ACC in women via a conditional ZNRF3 KO model [108].

Discussion
In comparison with ACC explants, several features are important to consider, including genetics, hormone secretion, and growth patterns. We summarize below a synthesis of the processes used for verifying the ACC115m primary culture, associated TVBF-7 cell line, and Bornstein et al.'s spheroid models as a paradigm for an appropriate analysis and confirmation of samples [55,58,94].
To verify the authenticity of primary cultures, cell lines, and xenografts, it is valuable to perform short tandem repeat (STR) profiling in comparison to primary samples. Numerous cases of contamination across cell lines and the overgrowth of lymphocytes or other cells instead of intended tumor cells have reinforced the necessity of such verification.
It is also important for appropriate positive and negative controls to be used in analyzing the secretion and stimulation of endocrine cells, as many popular forms of media, such as Nu-serum, contain hormones such as testosterone [122]. Researchers should ensure that their measurements compare to an appropriate baseline (i.e., of complete media before culture of cells) and use appropriate controls.
In trying to convert less aggressive phenotypes to lines suitable for in vitro study, transgenic models with such genes as SV40-TAg are often used, and ACC models such as ATC1 and ATC7 use this technique. However, we urge caution with such approaches as they may no longer resemble their original less aggressive phenotype. Instead, we encourage more complex culture models that better recapitulate the original environment, such as the standardized spheroid model of Bornstein et al. reported above or other 3D systems. Such systems are also valuable in analyzing the co-culture of ACC with other cell types such as adipose cells or lymphocytes, which are essential to understanding the lipid and immune microenvironment of ACC. As standardized 3D culture systems become a reality, ideally, co-culture techniques will also become more refined and widespread in understanding ACC.

Conclusions
Adrenocortical carcinoma is an aggressive orphan malignancy with limited therapeutic options. Its rarity has slowed clinical research advances. As a result, preclinical models are doubly important in understanding ACC's pathogenesis and potential treatment. Since the development of the first ACC model systems (mouse and human cell lines) in the 1960s and 1970s, much of the research has focused on the use of those now widely available systems. However, many novel systems of various types have been developed since. In particular, biobanking and standardized protocols have led to the generation of more patient-derived models in recent years. Unfortunately, only a few of these have reached wider usage and public availability such as in biobanks and mouse repositories. While there is a critical demand for new ACC model systems, it is just as important that existing model systems are shared and cross-validated across different research groups and between one another.
Using a variety of models is essential to capture the heterogeneity of clinical disease and to compensate for the flaws that different model systems have. No single model system can perfectly recapitulate disease, but the use of multiple models with complementary strengths will bring us closer to that understanding. Moving fast in research is sometimes essential, and simpler in vitro systems (such as monolayer and spheroid culture) perform this admirably. Slower and more relevant in vivo data using mouse or other xenograft models provide some information about how ACC interacts with the rest of the body, but an orthotopic model would be better for showing these interactions than existing flank models. More complicated in vitro systems that incorporate the 3D organization of cells or larger organoid structures can help to bridge the gap between the former simpler and the latter more complex models. In addition, human ACC tumor-bearing immunocompromised mice are useful for the exploration of potential therapeutic approaches, including chemotherapy, targeted therapy, or a combination of these modalities. To accelerate ACC immunotherapy, there is also an urgent need to develop rapid syngeneic mouse models rather than genetically engineered mouse models with slow tumor development.
The concerted application of existing models and the development of new ones to fill gaps will improve preclinical understanding and empower future clinical research on ACC. In particular, the major gaps in current ACC preclinical models are a comparison across newer model systems and the development of better in vitro model systems for organoid or more complex 3D cultures.