3.2.2. Tubule-on-Chip

Most kidney-on-chip studies use proximal tubule cells, as this part of the nephron is the primary site of drug clearance and a critical target for drug-induced nephrotoxicity. Therefore, the in vitro reproduction of proximal tubule function is of major interest in the preclinical assessment of candidate compounds. For this, hollow fibers are used as scaffolds to design a tubular model that resembles the structure of the proximal tubule. Such models were developed by coating the hollow fibers with a hydrogel that served as ECM for human proximal-tube endothelial cells (hPTECs) [148]. This approach enabled the observation of the secretory clearance of albumin-bound uremic toxins and albumin reabsorption [149]. Other studies developed a proximal tubule-on-chip using a polyester membrane to split the main compartment of the device into a luminal-like channel and a basal interstitial-like channel that were populated with PTECs seeded on an ECM coating [90]. This study reports significant changes in PTEC physiology and morphology, such as polarization, display of columnar shape, primary cilia, and transporters due to the simulated FSS induced within the system, and highlights the impossibility to achieve all these under static-culture conditions. These observations were also confirmed by other studies [150,151]. Another strategy in achieving proximal tubule-on-chip systems is bioprinting of tubular 3D architectures [152]. Homan et al. used this technique to obtain functional proximal tubule-on-chips by growing PTECs on a fibrinogen ECM coating the lumen of a structure printed with Pluronic ink [153].

One of the first kidney-on-chip models for nephrotoxicity screening used primary renal tubular cells to assess cisplatin toxicity, and the results most resembled in vivo models [154]. Other nephrotoxicity studies performed on kidney-on-chip models include the study of ifosfamide and acrolein nephrotoxicity assessment performed by Le Clerc et al. [155], the gentamicin kinetics study using MDCK on a porous membrane coated by fibronectin [156], and the toxicological assessment of polymyxins and their nephrotoxic potential on a 3D MPS human model [151].

Lastly, one interesting approach in nephrotoxicity testing is using versatile highthroughput screening platforms such as the OrganoPlate™ system powered by Mimetas for robust and reproducible results in preclinical studies. In this view, several studies report the expression and function of renal transporters and cell-polarization response to cisplatin [157] by modeling the proximal tubule functions in independent chips placed in a standard microtiter plate [158].

#### *3.3. Multiorgan-on-Chip Models That Include Kidney-on-Chip*

Taking the use of OOC models even to the next level, recent studies report the development of multiorgan-on-chip or body-on-chip models as platforms enabling the screening of multiorgan toxicity. In particular, integrating a kidney-on-chip device into a multiorgan system would bring valuable insights regarding secondary toxicity of drugs beyond systemic toxicity and the inflammatory response. Maschmayer et al. created a four tissue coculture (skin, small intestine, kidney, and liver), with 28 days of reproducible capacity, on a microphysiological four-organ-chip model, which is a promising line of research for pharmacodynamic and pharmacokinetic drug parameters and toxic profile [159]. Chang et al. used a combination of liver and kidney cells on a chip platform to study the nephrotoxic effect of aristolochic acid, a nephrotoxic agent that first needs to be activated in the liver [160].
