An Overview of In Vivo and In Vitro Models for Autosomal Dominant Polycystic Kidney Disease: A Journey from 3D-Cysts to Mini-Pigs

Autosomal dominant polycystic kidney disease (ADPKD) is the most common inheritable cause of end stage renal disease and, as of today, only a single moderately effective treatment is available for patients. Even though ADPKD research has made huge progress over the last decades, the precise disease mechanisms remain elusive. However, a wide variety of cellular and animal models have been developed to decipher the pathophysiological mechanisms and related pathways underlying the disease. As none of these models perfectly recapitulates the complexity of the human disease, the aim of this review is to give an overview of the main tools currently available to ADPKD researchers, as well as their main advantages and limitations.


Introduction
Today, biomedical researchers have a plethora of different tools in hand to study disease mechanisms and investigate potential therapeutic strategies in preclinical drug development. The choice of a specific tool is often the result of a compromise influenced by different factors, not all of them being directed by primary research interests: suitability regarding the specific research hypothesis under study, 2.
relevance to the human pathology in terms of causative mechanisms and symptoms, 3. availability and ease of manipulation (e.g., tools already available and well established in the laboratory conducting the research vs. introduction of new technologies), 4. economic factors and rapidness of result obtention.
In preclinical drug development, a combination of different methods is commonly used to identify potential drug targets and therapeutic approaches, beginning often with a cost-and time-effective high-throughput primary screen using in vitro technologies followed by an in vivo evaluation of positive hits in different disease specific animal models (orthogonal screenings and preclinical test). As it is the case for a large number of human pathologies, numerous tools useful for, or specific to, polycystic kidney disease research have been developed over recent decades.
Autosomal dominant polycystic kidney disease (ADPKD) is the most common monogenic inherited kidney disease in humans and the leading inheritable cause of end stage renal disease (ESRD).
Its main symptoms are the progressive formation and enlargement of fluid filled renal cysts which gradually impair kidney function until renal replacement therapy, dialysis, or kidney transplantation becomes necessary [1][2][3][4][5]. ADPKD patients present highly enlarged kidneys, which can reach a total volume exceeding 3500 mL for both kidneys combined [6] (versus a mean single kidney volume of 196 mL in healthy individuals [7]). Important kidney enlargement can sometimes be observed even before a significant decline in kidney function. Importantly, ADPKD patients account for about 6% of all patients under renal replacement therapy [8]. In addition, numerous extra-renal manifestations of the disease, such as pancreatic and liver cysts, increased blood pressure, and a higher risk of cerebral aneurisms have been reported [9]. As of today, the selective vasopressin V 2 receptor antagonist tolvaptan (Jinarc ® (EU, UK, Canada), Jynarque ® (USA), Samsca ® (Japan)) is the only available treatment that can slow cyst growth in ADPKD patients [10,11]. Through the inhibition of the binding of the antidiuretic hormone arginine vasopressin (AVP) to its receptor, tolvaptan diminishes cAMP production and thus cAMP-associated cyst cell proliferation and cyst fluid secretion. However, this treatment is only approved in adult patients at risk or showing evidence of rapidly progressing disease. Furthermore, it is not suitable for all patients due to its risk of hepatotoxicity and some other adverse effects, like polyuria and increased thirst, that can significantly lower the quality of life of people under treatment [12,13].
At the molecular level, ADPKD is caused by mutations in either of two genes, PKD1 or PKD2, which encode respectively polycystin-1 (PC-1) and polycystin-2 (PC-2) [14][15][16]. A third mutated (monoallelic) gene, DNAJB11, was recently identified to trigger ADPKD [17]. Mutations in PKD1 are present in about 85% of ADPKD patients and are related to a more severe disease progression with earlier onset of end stage renal disease [1]. No human case of homozygous PKD gene mutations is known and studies in animal models suggest that the PKD genes play a crucial role during embryogenesis as homozygous knock-out mice present severe developmental defects and die before birth. Even though the precise pathophysiological mechanisms of ADPKD still remain elusive, extensive research has offered valuable insights regarding the function of the polycystin proteins ( Figure 1, for review see [18]). Hence, PC-1 and PC-2 have been shown to interact at the primary cilium of renal epithelia cells, where they form a mechanosensitive cation channel, and ciliary defects appear to be implicated in ADPKD pathology. However, the polycystin proteins are also found in other subcellular locations and seem to be implicated in numerous cell signaling pathways regulating gene expression, cellular differentiation and proliferation, as well as apoptosis [19,20].
The level of functional polycystin proteins produced in the cell seems to be a crucial factor for renal cyst formation and decreased expression of the PKD genes can lead to ADPKD development in humans as well as animal models of the disease [27,28]. Importantly, renal injury seems to accelerate disease progression through the activation of repair mechanisms [29][30][31]. However, much about the pathogenesis of ADPKD remains to be learned and the aim of this review is to give an overview of the most important in vitro and in vivo tools available for ADPKD research, some of the most relevant progress achieved through the use of those tools as well as an update on the emerging technologies ( Figure 2). Hence, this article should hopefully help researchers to choose the best suited models for their specific studies and possibly to design the most efficient approaches to investigate their hypotheses. extensive research has offered valuable insights regarding the function of the polycystin proteins ( Figure 1, for review see [18]). Hence, PC-1 and PC-2 have been shown to interact at the primary cilium of renal epithelia cells, where they form a mechanosensitive cation channel, and ciliary defects appear to be implicated in ADPKD pathology. However, the polycystin proteins are also found in other subcellular locations and seem to be implicated in numerous cell signaling pathways regulating gene expression, cellular differentiation and proliferation, as well as apoptosis [19,20].  The level of functional polycystin proteins produced in the cell seems to be a crucial factor for renal cyst formation and decreased expression of the PKD genes can lead to ADPKD development in humans as well as animal models of the disease [27,28]. Importantly, renal injury seems to accelerate disease progression through the activation of repair mechanisms [29][30][31]. However, much about the pathogenesis of ADPKD remains to be learned and the aim of this review is to give an overview of the most important in vitro and in vivo tools available for ADPKD research, some of the most relevant progress achieved through the use of those tools as well as an update on the emerging technologies ( Figure 2). Hence, this article should hopefully help researchers to choose the best suited models for their specific studies and possibly to design the most efficient approaches to investigate their hypotheses.

In Vitro Research Tools
Numerous in vitro models have been developed in order to unravel ADPKD disease mechanisms and carry out drug screenings with the aim to identify potential therapies for ADPKD. Those models have evolved from simple monolayer cultures of diverse kidney epithelia cell lines to 3Dimensional (3D) cystic structures and then to PKD (polycystic kidney disease) specific kidney organoids. While these models clearly represent valuable tools to unravel cellular mechanisms and genetic interactions relevant to ADPKD, as well as to carry out initial large-scale drug screens, one needs to bear in mind that these models are not representative of whole organism physiology, notably

In Vitro Research Tools
Numerous in vitro models have been developed in order to unravel ADPKD disease mechanisms and carry out drug screenings with the aim to identify potential therapies for ADPKD. Those models Int. J. Mol. Sci. 2020, 21, 4537 4 of 34 have evolved from simple monolayer cultures of diverse kidney epithelia cell lines to 3Dimensional (3D) cystic structures and then to PKD (polycystic kidney disease) specific kidney organoids. While these models clearly represent valuable tools to unravel cellular mechanisms and genetic interactions relevant to ADPKD, as well as to carry out initial large-scale drug screens, one needs to bear in mind that these models are not representative of whole organism physiology, notably due to the absence of their natural environment. Thus, even though considerable progress is made to develop more representative in vitro models, when interpreting results of such essays researchers have to consider that the cellular models used today are lacking important aspects, such as tissue interactions or complex drug metabolization. Therefore in vivo validation of results is still a mandatory step for understanding ADPKD pathophysiological mechanisms and drug development. Furthermore, and concerning specifically ADPKD research, renal cell culture systems, or even kidney organoids can only mimic the nephrological aspects of the pathology. Even if this aspect is predominant in the disease, one wants to keep in mind that rather frequent extrarenal manifestations of the disease-such as vascular defects-exist that cannot be taken into account in these various models. A broad overview of the main (general and ADPKD-specific) advantages and limitations of the different in vitro tools is presented in Table 1. For more detailed information about the different in vitro tools the readers are referred to the references used to establish the Table 1, in particular to a recently published book chapter describing the applications of kidney organoids for PKD research and containing several detailed protocols by Jinghua Hu and Yong Yu [32] and to an excellent review on human cell-based models used in ADPKD research published by Weydert et al. in 2019 [33].

Human
• Culture of cells extracted from a diseased organ is possible and permits the study of particular pathological cellular mechanisms • Interspecific differences in cell or organ functions do not need to be considered • ADPKD patient specific primary cells are generally representative of an advanced disease stage and no control cells with an identical genetic background are available • Differences in genetic background can complicate the reproducibility and interpretation of results • Possible ethical concern depending on the cell type (e.g., embryonic stem cells) [33,34,39] Animal • The culture of cells extracted from well characterized transgenic disease models can provide valuable insights in disease mechanisms as a maximum of factors can be controlled (e.g., animal husbandry environment, cell culture conditions) and results can be related to specific in vivo phenotypes • Lesser ethical concern than for cells of human origin

•
The phenotypes displayed in culture may differ from those of human cells [34,39]

Cultured Monolayers of Kidney Cells and Three-Dimensional Cysts
Diverse immortalized kidney cell lines have been used to study ADPKD disease mechanisms and carry out drug screening assays. The Madin-Darby canine kidney (MDCK) cell line was isolated by Madin and Darby in 1958 from an adult female cocker spaniel [42,43]. This cell line retains differentiated kidney properties. When cultured on a solid surface, MDCK cells form a continuous and fully polarized epithelial sheet and vesicles appear, the formation of these vesicles is due to fluid transport in an apical-to-basolateral direction where the accumulation of fluid lifts the cell sheet away from the culture dish [44]. In 1987, McAteer et al. described a method permitting to initiate renal epithelial cyst formation with MDCK cells when cultured in a three-dimensional (3D) collagen matrix [45]. The fluid-filled cysts formed by these cultured MDCK cells have the basolateral surface directed towards the collagen matrix, fluid flow thus occurs in the opposite direction from the flow responsible for the formation of vesicles when the cells are cultured on a two-dimensional (2D) surface and therefore better mimics fluid transport in cysts of ADPKD patients [46,47]. In 1989, Mangoo-Karim et al. showed that renal epithelial cells of a different origin, especially normal human kidney cells and cells extracted from ADPKD kidneys, are also able to form fluid-filled cysts when cultured in a 3D collagen matrix [48]. Diverse renal epithelial cell lines in 2D or 3D culture systems, as well as excised cysts from ADPKD kidneys have been extensively used since the 1980s to study the molecular mechanisms at the basis of cyst formation and growth in ADPKD and to identify potential therapeutic molecules. These studies notably helped to identify dysregulated fluid and solute transports as promoters of cyst formation and growth. They also revealed the crucial role of cAMP in cyst enlargement through its implication in fluid secretion and cell proliferation. In addition, studies using these in vitro tools showed that diverse molecular transporters as well as receptors seem to be abnormally located in ADPKD cyst-lining epithelia, and that diverse signaling pathways are implicated in ADPKD pathogenesis [48][49][50][51][52][53][54][55][56][57][58][59][60][61][62][63][64][65]. A detailed protocol for in vitro cyst formation was recently published by Sharma et al. [47]. Furthermore, a high-throughput screening platform using 3D cultured cysts was described by Booij et al. in 2017 [66]. This platform was developed in order to identify compounds affecting cystogenesis, thus allowing to screen for potential therapeutic molecules which prevent or slow down cyst formation, and to unravel new molecular pathways involved in cyst swelling.
In addition to normal kidney cells, any epithelial kidney cell line isolated from transgenic or non-transgenic ADPKD animal models can be used to study the cellular mechanisms of ADPKD pathology. As an example, Felekkis et al. used, among other cell lines, primary tubular epithelial cells from a transgenic PKD rat model expressing a truncated polycystin-2 protein, to investigate p57/Cdk2 contribution to increased cell proliferation observed in ADPKD cyst lining epithelial cells [67]. Finally, a very appealing high-throughput approach for ADPKD drug screening has very recently been published by Asawa et al. [68]. In this paper, a comprehensive screening of a large compound library (~8000 compounds) is described. It is based on high throughput viability and 3D cyst growth assays performed on primary human renal cells and immortalized murine cell lines, comparing, for each compound, activity between Pkd1 deficient and wild-type cells in order to identify molecules which preferentially affect Pkd1-mutant cells. The initial screening approach was carried out based on a viability assay performed on Pkd1-null (MEK-null) and wild-type (MEK-wt) murine embryonic kidney collecting duct cells, as well as Pkd1 null (PN24) and Pkd1 heterozygous (PH2) murine adult proximal tubule cells. While only a minority (~18%) of active compounds presented differential activity in both murine cell pairs (Pkd1 deficient vs. wild-type), a vast majority presented differential activity in embryonic but not postnatal kidney cells, thus underlining the importance of the origin and initial developmental stage of cell lines. Then, the compound's efficiency in cyst reduction was evaluated thanks to a 3D high throughput cyst swelling assay based on PN24 cells (no cyst formation could be reached with MEK cells). Finally, in order to evaluate whether the identified compounds were relevant to the human disease, a viability assay, comparing primary normal human kidney (NHK) and cyst-lining cells from ADPKD patients (unfortunately, no information on cyst origin and mutation type was provided) was used and identified 21 molecules, among those some compounds that have previously been evaluated in the context of ADPKD pathogenesis, which further validates this approach. In addition, this study underlines the importance of the choice of the cell lines used for such an assay, as different results can be obtained when using cells originating from two distinct albeit similar murine ADPKD models, and that a validation of the results with cells of human origin is mandatory. The variability that was observed by Asawa et al. with ADPKD cells of human origin further underlines the importance of the specific genetic background of the cells. Hence, detailed information about their exact origin and the specific ADPKD mutation present should be gathered, if possible.

Embryonic Kidney Culture
Metanephric organ culture of mouse embryonic kidneys in vitro has been developed to study kidney organogenesis [69]. Such wild-type or transgenic mice embryonic kidney cultures have been used to study mechanisms of cyst formation and decipher the implicated molecular pathways. These studies helped to reveal the implication of CFTR (cystic fibrosis transmembrane conductance regulator), a cAMP regulated chloride channel, in cyst enlargement. Indeed, in 2008, Yang et al. showed that CFTR inhibitors are capable of slowing cyst formation in vitro. They used embryonic kidney cultures of wild-type mice in which 8-Br-cAMP was employed to initiate cyst-formation combined with an in vivo approach based on a Pkd1 kidney-specific knock-out strain [70]. Magenheimer et al. used embryonic kidney cultures from wild-type, Cftr knock-out, Pkd1 knock-out or combined Cftr and Pkd1 knock-out mice and demonstrated that Cftr knock-out completely suppressed 8-Br-cAMP-as well as Pkd1 knock-out-induced cyst formation [71]. The role of aquaporin-1 in retarding cyst growth was demonstrated by Wang et al. in 2015 [72]. They designed an elegant study combining in vitro and in vivo approaches, among which the culture of embryonic kidneys from aquaporin-1 null mice as well as from aquaporin-1 expressing mice after induction of cyst formation by 8-Br-cAMP.

Stem Cell Approaches and Kidney Organoids
As stated by De Souza, "an organoid is a 3D multicellular in vitro tissue construct that mimics its corresponding in vivo organ, such that it can be used to study aspects of that organ in the tissue culture dish" [73]. The development of organoids has been made possible thanks to the observation that stem cells (pluripotent embryonic or induced stem cells and adult stem cells) are able to assemble into complex tissue-like structures under specific culture conditions. This means providing appropriate differentiation signals in the case of embryonic stem cells (ESC) and induced pluripotent stem cells (iPSC), or providing signals inducing tissue repair or maintenance in the case of adult stem cells. In a 2013 publication, Takasato et al. described the successful generation of kidney organoids from human embryonic stem cells (hESC) [74]. Since then, protocols for kidney organoid production from stem cells have been refined and several methods have been developed and published, thus revealing ever more potential medical and research applications for this technology (for review see [75,76]).
In 2011, Thatava et al. described for the first time the generation of iPSC starting from cells originating from an ADPKD kidney of a patient carrying the W3842X mutation in exon 41 of the PKD1 gene [77]. The team of Benjamin S. Freedman at the University of Washington School of Medicine is specialized in the use of iPSC and CRIPSR/Cas9 technology in order to model human kidney disease with the aim to develop "clinical trials in a dish" [78]. In 2013, Freedman et al. used iPS cells derived from ADPKD kidneys in order to study whether the level of ciliary polycystin-2 is influenced by a reduced level of functional polycystin-1 [79]. Two years later they demonstrated that kidney organoids derived from iPSC cells can be used to study ADPKD specific pathophysiological mechanisms. They induced a PKD1 or PKD2 knock-out in hESC cells using the CRISPR/Cas9 technology and observed tubular cyst formation in derived kidney organoids [80]. In 2017, this team further showed that kidney organoids are capable of modifying their microenvironment and that this ability partly depends on PC-1 [81]. In addition, their results indicate that PKD patient derived iPS cells show a high variability in their capacity to form organoids and that variations in tubular morphology were observed when organoid formation was successful. They concluded that CRISPR-generated mutant hPSC carrying either a PKD1 or PKD2 knock-out are, at the current state of technology, more suitable to study PKD specific effects than patient derived iPSC. Finally, using hPSC derived kidney organoids, the Freedman Lab developed the first platform for automated organoid differentiation allowing high-throughput screening (HTS) [36]. In addition, they demonstrated that this HTS system is suitable for disease modeling and drug screening by showing that it produces reproducible results when used with kidney organoids derived from PKD knock-out hPS cells. In a study published in 2018, Boreström et al. also reported the development of a refined protocol for kidney organoid generation with high-throughput applicability [37]. By using the CRISPR/Cas9 technology they generated kidney specific reporter lines that permit the monitoring of kidney differentiation, glomerular maturation, and podocyte health in living cells.
In addition, patient derived iPS cells and kidney organoids are also used to study other specific aspects of ADPKD pathology. In line with this utilization, the differentiation of ADPKD patient derived iPS cells into vascular endothelia cells was used to study cardiovascular complications associated with ADPKD. Using this approach, Ameku et al. revealed altered Ca 2+ entry and gene expression in ADPKD patients-derived endothelia and were able to identify MMP1 as a potential novel risk factor for intracranial aneurysms in ADPKD patients [82]. Here again, we refer the reader to the recently published chapter describing several protocols and numerous valuable resources for the application of kidney organoids in ADPKD research, in the book Polycystic Kidney Disease edited by Jinghua Hu and Yong Yu [32].
Even though kidney organoids are organ-like structures composed of different kidney specific cell lineages, they nevertheless lack crucial characteristics of a complete functional kidney as organoids are not vascularized and not subject to the kidney specific microenvironment and fluid flows present in vivo. In response to this problem, an emerging technology is the development of cell-based microphysiological systems, the so-called organs-on-a-chip [83]. Culturing cells in microchannels and microfluidic systems permits the recapitulation of in vivo tissue-and organ-level functions in an ex vivo system. However, while this technology is very promising for the field of ex vivo drug screening and testing, and while on-a-chip models of isolated functional components of the kidney have successfully been created (for review, see [83]), there is no bona fide kidney-on-a-chip available today. Hopefully, the ongoing research and refinement of protocols for organoid production and organ-on-a-chip technology will finally lead to such a valuable tool in the near future.

In Vivo Research Models
A large number of in vivo models are available for the study of ADPKD-specific disease mechanisms and drug testing, the most common being rodent models. Animal models used in ADPKD research either recapitulate human pathology induction through the introduction of mutations in the corresponding PKD gene orthologues, or present a phenotype closely mimicking the hallmarks of ADPKD pathology, such as kidney and liver cyst formation and slow disease progression. The knock-out of PKD orthologous genes and the observation of a consecutive ADPKD-related phenotype in various animal models can help gain insights into the molecular mechanisms of pathology development.
Invertebrate or lower vertebrate organisms like Xenopus or the zebrafish can also be valuable tools, in particular for high-throughput in vivo drug screening [84,85]. Another advantage of these organisms for biomedical research is that they do not fall under the same strict directives on the protection of animals used for scientific purposes in the European Union, at least until a defined developmental state, which in the case of vertebrates is the independently feeding larva [86]. Whereas invertebrates present a fast generation time, which permits the rapid generation of models and results, the generation time of lower vertebrate such as zebra fish or xenopus is comparable to those of rodent models (sexual maturity is reached 3-4 months post-fertilization in D. rerio and at about 4-6 months of age in Xenopus frogs) [85,87]. An overview of the most important advantages and limitations of the invertebrate and lower vertebrate models is presented in Table 2. Mammals, however, are genetically and physiologically more similar to humans than the aforementioned organisms, thus studies conducted in mammals might better reflect human ADPKD disease mechanisms and permit more reliable observations, in particular regarding determination of treatment efficiency and adverse effects. However, as compared to invertebrates and lower vertebrate models, higher maintenance costs, less abundant progeny, and slower generation time do not easily permit the use of mammalian models for large scale drug screening, and thus limit their use to the in-depth analysis of efficiency and toxicity of otherwise identified drug candidates and to their preclinical validation before clinical testing.

Invertebrates
Even though one would not suppose that invertebrates, that do not have a excretory system as complex as the mammalian kidney, could find an application in the study of human chronic kidney disease, research carried out on the nematode Caenorhabditis elegans and the fruit fly Drosophila melanogaster have helped gain valuable insight in ADPKD disease mechanisms.
For example, the group of Maureen M. Barr has focused on the study of human disease mechanisms in the C. elegans worm model. In their landmark paper from 1999 Maureen M. Barr and Paul Sternberg described the C. elegans homologues of PC-1 and PC-2, LOV-1 and PKD-2 respectively, which are expressed in the worm ciliated neurons and locate to cilia, though they are not required for their assembly [88,89]. They further report that both, lov-1 and pkd-2 mutations induce impaired mating behavior in C. elegans males and suggest that LOV-1 and PKD-2 might act in a common sensory signaling pathway. Since then, C. elegans has become a tool to study protein interactions and molecular signaling pathways in the cilium as well as potential disease mechanisms associated with ciliary dysfunction (ciliopathies), including polycystic kidney disease (for review see [90][91][92][93]). Several direct interaction partners or indirect modifiers of LOV-1 and/or PKD-2 expression, localization, and activity have been identified in C. elegans and human orthologous genes exist for many of those [94][95][96][97][98][99][100]. Thus, these works based on C. elegans contributed important insights into the molecular functions of polycystins and hinted to potential pathophysiological mechanisms and modifier genes for ADPKD. We also want to refer our reader to the recently published chapter describing several protocols and numerous valuable resources for the application of C. elegans in ADPKD research, in the book "Polycystic Kidney Disease" edited by Jinghua Hu and Yong Yu [101].
As for the fruit fly Drosophila melanogaster, ADPKD research has mainly focused on defects of Amo, the fly homologue of PKD2. It has been shown, that similarly to C. elegans, defects in the polycystin-2 homologue impair fruit fly male fertility. In D. melanogaster, Amo is localized at the tip of the sperm flagella, a motile ciliary structure, suggesting a role in signal transduction upon transfer to the female reproductive tract [102][103][104][105]. Furthermore, a D. melanogaster model producing an ADPKD phenotype was described in 2017 by Gamberi et al. [106]. They described that a knock-out of the fruit fly translation regulator Bicaudal C (BicC) causes progressive cystic degeneration of renal tubules that can be reduced by rapamycin treatment. As BicC expression has been found to be downregulated in PKD patients, the authors of the study suggested that BicC acts downstream of PKD1 in a conserved molecular pathway. Furthermore, the C-terminal cleavage product of mouse polycystin-1 was over-expressed in D. melanogaster, revealing that PC-1 might be implicated in regulation of the mitochondrial function [107].

Lower Vertebrates
Extensive ADPKD research has also been conducted in zebrafish and Xenopus models. In Xenopus, extensive studies of polycystin protein function and the molecular pathways implicated in polycystin expression, trafficking, and complex formation were performed. Several important properties of the polycystins as Ca 2+ permeable cation channels [108,109] as well as the capacity of the PC-1 C-terminal cytoplasmic tail to induce Clcurrents in transfected oocytes [110] were revealed through in vivo studies using Xenopus.  [111][112][113]. Furthermore, thanks to the use of specific morpholino-oligos raised against one amongst several genes, a PKD-like phenotype could be induced in Xenopus, thereby leading to valuable insights into the signaling network underlying ADPKD pathology. Indeed, the single knock-out of Xenopus Pkd1 [112,114], Pkd2 [115], Bicc [116], or Gnas (Guanine nucleotide-binding protein G subunit α isoforms (short)) [114] induces this PKD-like phenotype that includes edema, due to impaired kidney function, as well as dilated pronephric tubules and ducts. This work also revealed the implication of Bicc and the miR-17 micro RNA family in the regulation of Pkd2 mRNA translation. Furthermore, the interaction between the C-terminal tail of polycystin-1 and G-protein α-subunits was shown to play a critical role for correct kidney function and to be an important inhibitor of aberrant G-protein β/γ signaling.
In the last few decades, the zebrafish Danio rerio has been increasingly used for drug screening and as a relevant and convenient system to model various human kidney diseases (for review see [117,118]). As for ADPKD, this model has led to important insights into the pathways modulated by the polycystin proteins. The C-terminal cytoplasmic tail of polycystin-1 was shown to be implicated in Wnt signaling [119] and in the STAT6/p100 pathway [120] thanks to in vitro experiments and overexpression of the PC-1 C-tail in zebrafish embryos. In addition, several genes implicated in cilia assembly or function have been shown to induce kidney cysts and, interestingly, body curvature (ventral or dorsal) when mutated in zebrafish lines or in specific knock-outs induced via morpholino injection [22,121,122]. Of those models, Pkd2 and Pkd1a/b (the two Pkd1 isoforms with different expression patterns identified in zebrafish [22]) and bicC morpholino-mediated knock-outs seem to be the most relevant for ADPKD research. Indeed, they all develop kidney cysts and dorsal body curvature without any apparent defects in cilia assembly or structure, while other models presenting kidney cysts generally develop ventral body curvature [121]. Furthermore, a comparative study of cyst formation and cilia function between locke, swt, and curly mutants (the three developing renal cysts and ventral body curvature) and Pkd2 morphants suggests that the mechanisms underlying cyst formation in Pkd2 morphants differ from those observed in these three mutant strains [123]. Importantly, the phenotypes observed in these zebrafish ADPKD models can most often be rescued by injection of mouse or human mRNA encoding the corresponding protein orthologue [122,124]. Furthermore, Pkd2 morphants present left-right patterning defects similar to those observed in Pkd2 knock-out mouse embryos [125,126]. In 2008, Fu et al. used this Pkd2 morpholino model to demonstrate that polycystin-2 is probably involved in different cellular functions depending on its intracellular localization. Indeed, various and specific phenotypes of the Pkd2 zebrafish model were rescued by injection of a mutant PC-2 mRNA targeted to various and specific subcellular localizations, supporting the hypothesis that polycystin-2 exerts tissue-specific functions depending on its distinct subcellular localizations [127]. Furthermore, Pkd1a morpholino injection induces the development of liver cysts in zebrafish embryos [128]. In 2006, Low et al. made the interesting observation that overexpression of the PC-1 cytoplasmic C-terminal tail in zebrafish leads to kidney cysts in the absence of body curvature [120]. In addition, polycystin-mediated defects in collagen expression during zebrafish embryogenesis could be revealed in a morpholino-induced Pkd2 knock-out model. The authors suggested that polycystins are implicated in a negative feedback signaling pathway regulating the expression of several collagen genes in response to mature crosslinked extracellular matrix (ECM) formation [22]. Also, in 2014, Sussman et al. identified two zebrafish homologues of the phosphodiesterase 1A (PDE1A), an enzyme interesting in the context of ADPKD because of its capacity to hydrolyze cAMP and as the unique PDE isoform regulated by Ca 2+ [129]. They demonstrated that PC-2 and PDE1A might function in a common pathway and that ADPKD pathology might be modulated via PDE inhibition. Here again, we want to refer our reader to the recently published chapter describing several protocols and numerous valuable resources for the application of D. rerio in ADPKD research, in the book "Polycystic Kidney Disease" by Jinghua Hu and Yong Yu [130].

Drug Screening Approaches in Invertebrate and Lower Vertebrate Models of ADPKD
Invertebrate and lower vertebrate models present a number of advantages that make them attractive for drug screening approaches (Table 2). These advantages include notably the relatively fast and easy construction of transgenic animals and the numerous progeny that allow to produce a large number of (transgenic) animals in a short time frame. In addition, the ex utero development and very small size (at least at some developmental stages) allow to perform high throughput screening assays relatively easily.
Nevertheless, as of today, and to the best of our knowledge, no ADPKD specific drug screening assays have been carried out in C. elegans or Xenopus models of ADPKD. In contrast, a handful of studies have employed Drosophila melanogaster models to carry out drug screening assays or to study effects of specific drugs. Hofherr et al. studied the effect of a patient specific PKD2 mutation (PKD2 D511V ) and described the use of D. melanogaster as a readout for polycystin-2 channel function in vivo [104]. Following in vitro assays with cellular models, they suggest that the PKD2 D511V mutation compromises PC-2 channel folding, thus promoting lysosomal degradation. In accordance with this, treatment with the FDA-approved lysosomal inhibitor chloroquine was able to increase protein levels of the mutant PC-2 channel in vitro. In order to carry out an in vivo validation of these results, the patient specific mutation (Amo D627V , equivalent to human PKD2 D511V ) was introduced in Amo deficient male flies (Amo -/-). The double transgenic animals (Amo -/-; Amo D627V ) showed slightly increased fertility compared to the Amo deficient male flies, indicating that the mutated protein presents some residual function in vivo. Administration of chloroquine in the food led to a partial rescue of the reduced fertility only in the double transgenic animals, which correlated with an increased expression of the mutant protein in sperm flagella of chloroquine-treated Amo -/-;Amo D627V flies. This study thus demonstrates that thanks to the time-and cost-effective production of transgenic animals, D. melanogaster could represent a valuable model to study the effects of patient specific PKD2 mutations in vivo and to develop screening assays that aim at identifying compounds specifically suppressing the effect of these mutations. This system is however limited by the absence of a PKD1 homologue in the fly, thus permitting only to evaluate PKD2 specific mutations. In addition, interactions between the polycystin proteins, which are observed in humans and other organisms and are implicated in the regulation of their function and trafficking to specific subcellular localizations can therefore not be studied in this model. This represents a major limitation as defects of protein trafficking are thought to be implicated in ADPKD pathogenesis. In 2019, Millet-Boureima et al. tested the effect of Smac (second mitochondria-derived activator of caspases) mimics, of which one was previously shown to suppress the pathological phenotype in an ADPKD rat model, in the BicC mutant fly model of ADPKD and found important cyst reduction upon Smac treatment [134]. Hence, the BicC model might be useful to conduct high throughput drug screening aimed at isolating potential therapeutic molecules in ADPKD. However, as BicC is thought to act downstream of PKD1 in ADPKD patients, and as a PKD1 homologue has not been identified in D. melanogaster, potential differences in the signaling pathways implicated in the development of the renal cystic phenotype between human ADPKD patients and the BicC fly model might limit its relevance, both as a model for the human disease and for drugs screening.
The vast majority of drug screening studies using invertebrate or lower vertebrate models of ADPKD employed the zebrafish D. rerio. Cao et al. identified histone deacetylase (HDAC) inhibitors as candidates for PKD treatment using zebrafish Pkd2/hi4166 (which does not develop a cystic phenotype but instead a dorsal body curvature) and ift172/hi2211 mutants (developing ventral body curvature and kidney cyst formation), zebrafish Pkd2 morphants and a murine kidney specific Pkd1 knock-out model [140]. In 2016, Wu et al. employed a multimodal approach, using a 3D cyst assay with human renal epithelial cells (immortalized and normal), zebrafish Pkd2 morphants and the Han:SPRD rodent model, which led to the identification of a beneficial effect of resveratrol on PKD progression, as well as of NF-κb as a potential therapeutic target [141]. Furthermore, Chang et al. demonstrated that metformin-which has previously been shown to reduce cyst growth and fluid secretion in 3D cell culture assays, murine embryonic kidney cultures, and rodent models of ADPKD-also reduces pronephric cyst formation and dorsal body curvature in zebrafish Pkd2 morphants [142]. The dorsal axis curvature is often used as an indicator of efficiency as its severity is dose-dependent (morpholino injection) and, in contrast to the presence of cysts, it is the most penetrant and consistent feature observed in zebrafish ADPKD models [22]. An automated imaging pipeline in 96-well plates using larvae of zebrafish in which the developing pronephros is highlighted by GFP (Tg(wt1b:EGFP)) has been developed by Westhoff et al. and validated for nephrotoxicity screening. They further demonstrated that this system can be used to detect phenotypic abnormalities (cyst development and ventral curvature) caused by morpholino-induced ift80 and ift172 knock-out [143]. In future studies, this system could thus potentially be employed for drug screening assays with Pkd specific morphant embryos. In addition, a recent study identified ALK5 kinase (or TGFßR1 -transforming growth factor, beta receptor I) and non-canonical androgen receptors as potential targets for ADPKD drug development using a high throughput screening approach based on a zebrafish Pkd2 mutant (using tail curvature as a primary readout) followed by a secondary validation in 3D cyst assays based on MDCKII and a human PKD1 patient-derived cystic cell line (OX161c1) [138]. However, as significant phenotypic variability has been described between zebrafish Pkd2 morphants models and/or Pkd2 mutant models, standardized protocols should be developed to ensure comparability and reproducibility of the results when carrying out drug screening assays based on this model. Finally, zebrafish contains two Pkd1 paralogs with distinct expression patterns and the single knock-out of Pkd1a, or the simultaneous knock-out of Pkd1a and b, induces kidney cysts only in a small fraction of embryos (not exceeding 20%) and even though signs of kidney failure like severe edema are observed more frequently, a hydrocephalus and jaw defects seem to be the primary developmental phenotype caused by the knock-out of these genes [22]. This could indicate that Pkd1 zebrafish morphants or mutants might not be very adequate models for ADPKD as a kidney disease.

Rodent Models
As small mammals, the mouse and the rat present several advantages for biomedical research, that include the fact that their physiological and genetic structures are closer to those of humans than the aforementioned models, together with a relatively short life cycle and affordable housing costs [144]. In addition, in particular thanks to the generalization of the CRISPR/Cas9 genetic editing system, the generation of specific transgenic mouse strains has been significantly facilitated and accelerated. Two useful resources to gather information about specific murine models are the interconnected databases MouseMine (http://www.mousemine.org/mousemine/begin.do) and Mouse Genome Informatics (MGI (Mouse Genome Informatics, a resource for laboratory mouse hosted by the Jackson Laboratory, http://www.informatics.jax.org/) [145]. A large number of murine models are commercially available, and distributers can easily be found on the website of the "International Mouse Strain Resource" (IMSR, http://www.findmice.org/). Even though transgenesis is more complicated in the rat than in the mouse, technical advances have recently been made and, as a result, the number of transgenic rat models for human disease will certainly significantly increase in the coming years [146].
Hence, numerous rodent models have been developed for the study of ADPKD (Figure 3). These models can roughly be divided in two categories: those arising from spontaneous genetic mutations and those produced through genetic engineering [28]. In addition, there are models recapitulating human disease initiation via modified expression (hypomorphic alleles, conventional heterozygous knock-out alleles, conditional and potentially inducible knock-out alleles, overexpression) of either the human or the rodent PKD genes on the one hand and, on the other hand, models recapitulating more or less precisely an ADPKD phenotype due to mutations in various other gene families (e.g., cell cycle, developmental regulation, cilia) ( Table 3). While the second type of models is certainly important to reveal underlying mechanisms and regulatory pathways involved in some aspects of the pathological phenotype, and therefore potential modifier genes, their use in drug discovery is debatable as they do not mimic human disease initiation [147]. For more details, the reader is referred to several excellent and comprehensive reviews on rodent models used in PKD research Int. J. Mol. Sci. 2020, 21, 4537 16 of 34 that have been published over the last two decades [28,[147][148][149][150][151][152]. The objective of the present review is not to provide an exhaustive list of all existing models but rather to give a broad overview of the various types of models that can be used to study ADPKD. Thus, in this paragraph, we will briefly introduce only the rodent models that seem to us most closely related to human ADPKD.  Figure 3. Schematic overview of the different types of rodent ADPKD models (adapted from [28]). *: homozygous germline knock-out embryonically lethal Figure 3. Schematic overview of the different types of rodent ADPKD models (adapted from [28]). *: homozygous germline knock-out embryonically lethal. Table 3. Examples of rodent models frequently employed in ADPKD research, in which a phenotype closely resembling ADPKD is caused by mutations in genes other than the PKD genes. Models developing a phenotype rather resembling ARPKD than ADPKD, such as the bpk mouse in which the polycystic phenotype is caused, as in the jcpk model, by a mutation in the Bicc1 gene [153] or the orpk mouse (oak ridge polycystic kidney) carrying a hypomorphic allele of Ift88 [148,154] have not been included in this

Myc proto-ongogene protein
Progressive polycystic kidney disease with atypical plasmacytic infiltrates, anemia, premature death due to renal failure at 2 weeks to 4 months of age [156,157] Spontaneous mutation jcpk mouse Bicc1 Single base-pair change (AG→AA) in the splice acceptor site of exon 3 causing a frameshift resulting in a premature stop codon

Nek8 (NIMA (never in mitosis-A) related kinase)
Slowly progressive polycystic kidney disease, decreased fertility in males from 15 weeks of age, premature death at 20 to 25 weeks of age [159,160] kat and kat 2J mouse Nek1 kat: internal deletion kat2J: single base-pair insertion causing a frame shift resulting in a premature stop codon

NPHP3
(associated with nephronoph-thisis) Nephrocystin-3 Slow progressive polycystic kidney disease with cerebral aneurysms, premature death and chronic inflammatory infiltrates in advanced stages, with a mean age at death of 6.5 months (females) and 8.2 months (males) [163,164] Han:SPRD cy/+ rat (Cy-rat) Missense point mutation resulting in an amino acid substitution (R→W) ANKS6 Ankyrin repeat and SAM domain-containing protein 6 Slowly progressing renal enlargement by cyst formation, phenotype more severe in males than in females, premature death (males from 6 months of age, later for females) [165][166][167][168] Even though rodent models based on the engineering of PKD1 and PKD2 genes seem to be an obvious choice to study ADPKD pathophysiological mechanisms and for drug screening, these models present several limitations. First, as supposed to be the case in humans, homozygous loss of function of the murine Pkd1 or Pkd2 genes by conventional knock-out is embryonically lethal, whereas animals heterozygous for those mutations only develop very mild cystic disease. In humans, the pathology slowly progresses over decades before it leads to end-stage renal disease, a feature that fits with the 'two-hit' hypothesis that postulates that cyst formation is initiated when a somatic mutation occurs in the wild-type allele of the PKD gene [169]. Hence, according to this hypothesis, the lifespan of the heterozygous Pkd knock-out murine models might simply be too short to effectively recapitulate human ADPKD. A murine model supporting this hypothesis is the Pkd2 WS25 model in which a mutant exon 1 has been introduced in tandem with the wild-type exon 1 in the Pdk2 locus, resulting in an unstable allele [170]. This allele is prone to homologous recombination resulting either in the reversion to the wild-type allele or a true null mutant allele of Pkd2. When the unstable WS25 allele is combined with a true null allele of Pkd2 (Pkd2 WS25/-), animals escape embryonic lethality but show decreased survival with a median age at death of 65 weeks. Furthermore, Pkd2 WS25/mice develop bilateral kidney cysts as well as liver cysts in a variable manner at 10-11 weeks of age and, in some animals, pancreatic cysts could be observed at 3 months of age [170,171]. However, the 'two-hit' hypothesis for cyst initiation has today been expanded to a gene dosage hypothesis where the level of functional polycystin proteins play a crucial role for cyst formation. It follows that, if the gene dosage hypothesis does not exclude a 'second hit' via somatic mutations, it further includes other factors that could modify PKD expression [18,27]. This hypothesis is supported by animal models displaying reduced or increased expression of the Pkd genes. A mouse model carrying a hypomorphic Pkd1 allele induced by aberrant splicing of intron 1, thereby reducing the quantity of functional polycystin-1 to 13-20%, was presented by Leeuwen et al. in 2004 [172]. In this so-called Pkd1 nl model, overall survival was decreased but disease progression was very variable. Animals showed bilateral kidney cysts, mild cystic disease in the liver and pancreas, and development of aortic aneurysms, thus exhibiting a large spectrum of the human ADPKD symptoms. Furthermore, in 2012 Katharina Hopp et al. described a new murine model carrying an hypomorphic Pkd1 allele, the Pkd1 RC/RC mouse [173]. This model was developed through the knock-in of a clinically relevant and incompletely penetrant PKD1 allele (p.R3277C). This particular allele was first identified in a consanguineous family and, of note, homozygous carriers of this allele are viable and present mild to typical cystic disease, supporting the gene dosage hypothesis. Pkd1 RC/RC mice develop a slowly progressing polycystic kidney disease with embryonic cyst initiation, present cystic lesions in the liver and elongated primary cilia in collecting ducts, while heterozygous Pkd1 RC/null show a more severe disease progression and significantly reduced survival when compared to Pkd1 RC/RC homozygous animals. The slowly progressive nature of the disease in this model, as well as the clinical relevance of the introduced mutation make it particularly interesting for the study of ADPKD. Its validity as a preclinical model for ADPKD drug efficiency studies was further established by the observation of a positive effect of tolvaptan in this model. This model also served to show the activity of pasireotide, a synthetic analogue of somatostatine which, as tolvaptan, indirectly reduces adenylyl cyclase 6 (AC6) activity and has hence proven effective in slowing cyst progression. In this study, pasireotide or tolvaptan alone were active, and their combination led to an additive effect [174]. A model in which Pkd1 is overexpressed, the Pkd1 TAG mouse, was presented in an elegant study by Kurbegovic et al. in 2010 [175]. They injected a Pkd1 TAG construct (murine Pkd1 carrying a silent point mutation in order to distinguish it from the endogenous gene) in murine oocytes and analyzed three of the resulting mouse lines carrying either 2, 6, or 15 copies of the transgene. They observed a copy number-dependent increase of both Pkd1 mRNA level and polycystin-1 expression. Furthermore, all hallmark symptoms of human ADPKD were evaluated in this study: Pkd1 TAG mice present a decreased lifespan presumably due to renal failure, develop kidney as well as hepatic cysts, cardiac abnormalities and, in some cases, increased blood pressure was observed. Importantly, the severity of the ADPKD disease phenotype developed was shown to be dependent on the number of copies of the transgene integrated in the genome, with the Pkd1 TAG line 26 (Pkd1 TAG 26, bearing 15 copies of the transgene) presenting the most severe phenotype and the shortest lifespan. This model somewhat supports and expands the gene dosage hypothesis through the experimental validation that a dysregulation of polycystin-1 through overexpression of the protein can induce an ADPKD specific phenotype in rodents, in addition to significantly reduced polycystin-1 expression as it has previously been shown. This is further supported by the observation of an overexpression of polycystin-1 in end stage ADPKD renal cystic epithelia, even though it has been suggested that this constitutes only a secondary phenomenon reflecting the differentiation stage of end-stage cystic epithelia, thus being only a consequence of the mechanisms initially triggering cyst development [176].
The development of conditional knock-out models has also enabled prevention of the embryonic lethality of homozygous null mutations of the Pkd genes displayed by conventional germline knock-out mice. The widely employed Cre/loxP technology permits to control transgene expression in a tissue specific or inducible manner [177]. Briefly, two specific, short nucleotide sequences, the loxP sites, are introduced in either side of the target gene (or of part of it) in such a manner that they do not interfere with normal gene expression. To modify the expression of the so-called "floxed" (flanked by loxP sites) gene, another transgene, the cre gene, must be present in the genome. Cre encodes a recombinase, which specifically recognizes the loxP sites and can excise any nucleotide sequence comprised between two of those sites. If cre expression is controlled by a tissue-specific promoter, inactivation of the floxed gene only takes place in tissues where the recombinase is expressed. One example of such a strain used in ADPKD research is the Pkd1 flox :Ksp-Cre model presented in 2008 by Shibazaki et al. [178]. In this model, exons 2 to 4 of the Pkd1 gene are floxed and recombinase expression is restricted to renal tubules via the Ksp-cadherin promotor [179], permitting kidney specific inactivation of Pkd1. Pkd1 flox :Ksp-Cre animals show a rapid progression of polycystic kidney disease and die, between the postnatal days 14 and 17, presumably due to kidney failure.
The availability of systems in which recombinase activity is inducible by the administration of a specific substance to the animals, for example tamoxifen, permits to produce animals presenting a ubiquitous Pkd knock-out while avoiding embryonic lethality. In 2007 Piontek et al. used the CAGGCre-ER TM strain (Jackson Laboratory stock number 004682) in which the recombinase expression is induced by tamoxifen, in combination with a floxed Pkd1 allele and demonstrated that the timing of Pkd1 disruption had a significant influence on disease severity and progression [180]. After tamoxifen induced Pkd1 knock-out, all the animals develop renal as well as hepatic cysts, thus presenting two hallmark features of ADPKD. In addition, disease progression was observed to be strikingly faster when tamoxifen was administrated before postnatal day 13.
Today, numerous Cre strains carrying cre alleles with tissue-specific and/or inducible recombinase expression have been created in various genetic backgrounds and a large number are commercially available (Cre strains and corresponding expression data and references can be found on the MGI website, an international database resource for laboratory mouse hosted by the Jackson Laboratory, http://www.informatics.jax.org/home/recombinase). In combination with strains carrying floxed alleles of ADPKD relevant genes, these various Cre strains allow the creation of a large number of models to study specific features of ADPKD pathology and to elucidate the underlying molecular pathways and their timing. As it would exceed the scope of this broad review to describe the specific transgenic Pkd1 and Pkd2 murine models, here again we refer our reader to the recently published chapter describing those models in detail, as well as the divers techniques used to assess the PKD phenotype in those animals, in the book Polycystic Kidney Disease edited by Jinghua Hu and Yong Yu [181].
In addition to these models created by targeting the orthologues of the human PKD genes, rodent models displaying phenotypical similarity to ADPKD pathology resulting from mutations in other genes have been used in ADPKD research. A brief summary of the most frequently employed models is presented in Table 3. These models have helped to gain important insights in the molecular pathways and gene networks relevant to polycystic kidney disease. For example, the SBM mouse developed by Trudel et al. in 1991 overexpresses the proto-oncogene c-myc predominantly in the kidney and develops a renal phenotype similar to ADPKD with kidneys being the sole or predominantly affected organ [156,157]. Thus, the SBM murine model underlines that the dysregulation of tubular epithelial cell proliferation and apoptosis is sufficient to induce renal cyst formation and further supports the idea that elevated c-myc expression plays a major role in the pathogenesis of human PKD. Lin et al. demonstrated that kidney specific inactivation of the KIF3A subunit of kinesin-II, a protein essential to cilia formation, results in large cystic kidneys by postnatal day 28 (P28) and the complete replacement of renal parenchyma by large cysts by P35 [155]. This study thus strengthens the hypothesis that ciliary defects play a crucial role in PKD disease development. It is worth noting that a number of other rodent models developing renal cysts and other symptoms similar to ADPKD exist, in which dysregulation of cilia-associated proteins, proto-oncogene expression, or the mutation of genes related to other human diseases (e.g., Pkdh1 in the Pck rat, autosomal recessive polcystic kidney disease, ARPKD) are responsible for the observed phenotypes [151,152,182]. This suggests that common molecular pathways are causative for renal cyst formation and that the polycystin proteins participate, directly or indirectly, to various major regulatory networks implicated, among others, in cell proliferation and apoptosis [150].
Importantly, in view of the large number of PKD rodent models available, of which each presents some specific features, advantages and limitations, Hester Happé and Dorien J. M. Peters proposed, in their review of animal models used in ADPKD research, that at least two different models should be used in preclinical testing for potential therapies [28]. This point of view is strongly supported by the fact that different therapeutic interventions do not display the same efficiency in different rodent models of PKD [150].

Other Mammalian Models
In Persian cats, a spontaneous mutation arose, which is inherited in an autosomal dominant pattern and causes an adult onset ADPKD phenotype, with progressive renal cyst formation leading to renal failure later in live (>7 years), biliary fibrosis and hyperplasia, sporadic hepatic cysts and slightly increased blood pressure [183][184][185]. The gene responsible for ADPKD in Persian cats is the feline PKD1 gene, and a C to A transversion resulting in a premature stop codon inducing the loss of~25% of the protein's C-terminus was identified in affected cats by Lyons et al. [186]. Even though this model is rarely used in ADPKD research, its striking genetic and phenotypic similarities to human ADPKD together with its significantly longer lifespan than rodent models make it an attractive tool to study ADPKD disease progression and to test therapeutic approaches. In 2019, in an elegant study, Torres et al. presented the therapeutic potential of a ketogenic diet which was evaluated in two rodent and the feline models of ADPKD [187]. The results of their study suggest that a high-fat, very-low-carbohydrate diet might significantly slow down disease progression without any pharmacological intervention, thereby representing a promising therapeutic approach which deserves further investigations.
Porcine species are also increasingly gaining attention as animal models of human disease in biomedical research and the remarkable physiological and functional resemblance between porcine and human kidneys make swine very good models for human kidney diseases [188,189]. In addition, physiological and structural similarities between human and porcine organs make this species the most likely candidate for the production of organs intended for human organ replacement therapy. In line, the possibility of xenotransplantion of organs from pig into human is currently investigated as a possible solution to organ shortage [134]. In 2011, a research team of the State Key Laboratory for Agrobiotechnology at the China Agricultural University of Beijing directed by Ning Li characterized the porcine PKD1 and PKD2 genes. They found strong sequence similarities between the porcine and human PKD genes and proteins, as well as in their expression patterns [190,191]. In 2015, the same research team reported that a mini-pig model of ADPKD had been developed by mono-allelic knock-out of the PKD1 gene [192]. In this model, PKD1 expression was reduced at a transcriptional and translation level, macroscopically visible kidney cysts were detectable from 5 months of age and by 24 months of age these cysts began to deform the normal kidney shape. Furthermore, liver cysts were reported to be present in all investigated animals. This model was first used in a study by Lian et al. published in 2019 [193], in which the effect of single and combined treatments with 2Deoxy-D-glucose (2DG, an inhibitor of glycolysis reported to reduce cysts growth in mouse PKD models [194,195]) and metformin (an AMP-activated protein kinase inhibitor) on ADPKD progression was investigated. They found that all drug treatments significantly inhibited cyst growth in the ADPKD mini-pig model, the combined therapy being the most effective. Furthermore, none of the treatments led to apparent side effects. The research team from the China Agricultural University of Beijing also developed two further potential transgenic mini-pig models of ADPKD, one via the induction of c-myc overexpression and the other via the induction of PKD2 overexpression [196,197]. However, no kidney cysts or signs of impairment of renal function were observed in these models after 6 months and 12 months respectively. The authors suggested that the lack of symptoms might be due to the slowly progressing nature of ADPKD pathology and stated that both models continued to be observed. However, no update has been published since the original publications from 2013.

Discussion
This current review illustrates the vast number of tools available today for the study of ADPKD, as well as some promising emerging technologies, like the 'kidney-on-a-chip'. This large choice can, however, be confusing and, since no perfect model of ADPKD exists to date, the suitability of a chosen model in view of a specific research hypothesis must be thoroughly considered. This issue is currently partly addressed by the multimodal approach most research teams choose. Importantly, the choice of such an approach is the result of a balance between a rigorous scientific strategy, with the final aim to validate results in several distinct experimental conditions, ethical concerns, as well as economic considerations, not to mention the need to obtain publishable results in a reasonable timeframe. Most in vitro techniques present the advantages of being relatively rapid, cost-effective, and independent from any ethical issues (as compared to in vivo models). Hence, a large number of reproducible results can be rapidly obtained through standardized protocols and manipulation techniques. Even though the 'kidney-on-a-chip' might represent an attractive solution to a certain extent, in vitro models in use today are studied by essence in a complete absence of any physiological context. This may be viewed as an advantage for the investigation of isolated cellular mechanisms, but those models cannot give any indications on complex in vivo tissue or organ interactions, compound metabolization or potential off-target in vivo toxicity of a parent compound or of its metabolites. Small and widely-used invertebrate models, like the worm Caenorhabditis elegans, and non-mammalian vertebrate models, like the zebrafish Danio rerio, may be considered as they represent an interesting alternative and compromise between in vitro models and complex in vivo models [85,87]. Indeed, their housing costs are relatively low, their numerous, and by definition ex utero developing embryos are, up to a certain developmental state in the case of vertebrates, of no concern to the legislation on the protection of animals used for scientific purposes [86]. In addition, as the embryos and larvae of those animals are often transparent, they allow direct observation of physiological processes and facilitate the development of high-throughput technologies to test the toxicity or efficacy of a large number of different compounds. Furthermore, an increasing number of tools for the genetic manipulation of such organisms are available as of today. The genomes of these small model organisms have been sequenced, which allows the study of protein expression and function in a physiological context and thereby to create human disease models in these very distant relative species. Especially in biomedical research, high-throughput screening using in vitro or small in vivo models is often the first step to identify potential therapeutic molecules. For ADPKD research, the high-throughput platforms using 3D cysts [66] or kidney organoids derived from PKD knock-out hPS cells [36], as well as the automated imaging pipeline using zebrafish larvae [143] in combination with Pkd1a/b or Pkd2 morpholino oligos could be particularly useful to detect promising candidates while limiting false positives. The efficiency and lack of toxicity of the potential therapeutic molecules identified in such a way must then be validated in animal models more closely related to humans, such as the commonly used rodent models. Those species have again the advantage that housing costs are relatively low compared to larger mammals such as dogs or mini-pigs, while producing a relatively high number of offspring [144]. Furthermore, those organisms are very well studied and the scientific community has a large number of different transgenic or non-transgenic strains readily available. However, even though rodents are more closely related to humans than the aforementioned organisms and the fact that a considerable number of transgenic mouse models of human diseases exists today, results obtained in those models are not always fully transposable to humans. Even though murine models have proven reliable models of aging, in ADPKD research notably, the short lifespan of those animals might be a limiting factor. The majority of human ADPKD patients present a heterozygous mutation in one of the PKD genes and the disease progresses over decades before renal function is impaired to a point that renal replacement therapy is necessary; while in murine models, heterozygous Pkd mutations-which better recapitulate the human disease-cause a mild and very variable disease phenotype. Somatic mutations of the second non-mutated PKD allele are supposed to play a role in the initiation of cyst formation in ADPKD patients, and even though somatic mutation rates may differ between organisms, the short lifespan of rodents might thus explain the only very mild phenotype observed as a consequence of heterozygous Pkd mutations [18,198]. In this sense, the ADPKD cat and mini-pig models seem to recapitulate human ADPKD pathogenesis more accurately than rodent models. Porcine models of human disease seem especially promising in view of their strong physiological similarity to humans. However, up to now, only few research facilities have the required equipment, staff experienced in handling, and an adequate husbandry to house those species, not to forget that the housing costs of such models are high. However, in the long term, increasing use of specific mini-pig models of human disease in biomedical research might eventually decrease these expenses. In addition, the significant financial loss due to the clinical trials of molecules which seemed effective and promising in preclinical rodent models, but that later turned to be ineffective or to present important adverse effects in humans may also be in favor of the use of cat and mini-pig models for ADPKD. Example of such molecules that showed promising results in preclinical studies carried out on rodent models of PKD are the mTOR inhibitors sirolimus (rapamycin) and everolimus. While both molecules efficiently slowed down disease progression in the Han:SPRD cy/+ rat [199][200][201] and rapamycin further proved to be efficient in the bpk mouse and the orpk-rescue mouse [202], no beneficial effect of rapamycin or everolimus treatment could be observed in clinical trials with ADPKD patients [203,204]. In this context, it may be interesting to note that none of the PKD rodent models employed during the preclinical evaluation of those compounds was based on mutations in the Pkd genes.
Finally, we share the view of Hester Happé and Dorien J. M. Peters [28] that ADPKD treatment strategies should be tested in at least two different animal models, of which at least one should be based on a PKD gene mutation. Vasopressin V 2 receptor antagonists OPC-31260 and its derivative tolvaptan (OPC-41061) have for example shown effectiveness in several distinct rodent models of PKD, among those the PCK rat, the pcy mouse, the Pkd2 WS25 mouse, and a conditional kidney specific Pkd1 knock-out model [205][206][207][208], as well as in the cpk mouse [209], a model of autosomal recessive PKD (ARPKD), before and while clinical testing for its effectiveness in ADPKD treatment was conducted. We would also argue in favor of the integration in studies investigating potential disease treatments of larger mammals like the ADPKD cat or the recently developed ADPKD transgenic mini-pig. However, before tests are carried out in animals, potential therapeutic compounds have to be identified in adequate in vitro models in order to limit the wasteful use of laboratory animals. In ADPKD research, transgenic hPSC-derived kidney organoids are potentially the most promising in vitro model available today. A second validation of identified candidates might then be performed in zebrafish models of ADPKD induced by Pkd2 or Pkd1a/b morpholino injection before mammalian models are employed. In any case, the choice of specific models must be anticipated according to the chosen strategy and we hope that this review will provide some helpful information for this process of decision-making. Finally, the InterMOD project (MOD: Model Organism Database), which "aims to make it easier to carry out cross-MOD comparative analysis and to relate MOD data to medical research" [210] should also be mentioned in this place. Different MODs, among those FlyMine (https://www.flymine.org/flymine/begin.do), ZebrafishMine (http://www.zebrafishmine.org/begin.do), MouseMine (http://www.mousemine.org/mousemine/begin.do), RatMine (http://ratmine.mcw.edu/ ratmine/begin.do) and HumanMine (https://www.humanmine.org/humanmine/begin.do), participate in the project and thus permit cross-species analysis of biological data. Therefore, this project constitutes a helpful and interesting tool for the integration and translation of model organism research into medical applications, as well as for the comparison of model organisms in view of a specific research project.

Conclusions
The existence of such numerous tools and models to conduct ADPKD research results from the rigorous and important research that has been carried throughout the last decades. The valuable findings which have been collected through the use of those tools help to better understand this complex disease and have made the discovery of the first drug treatment permitting to slow down ADPKD progression possible. Furthermore, the existence of several in vitro and in vivo models for this human disease provides researchers with the flexibility to choose among the diverse models those which best suit their specific research questions in order to obtain more reliable and reproducible results. However, as highlighted in this article, as of today no model perfectly recapitulates the human disease and thus special attention has to be paid to the specific characteristics of each model when establishing a research protocol. We therefore hope that this review will provide ADPKD researchers a comprehensive overview of the existing and emerging models used to study this disease, together with their main advantages and limitations.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.