A Multitubular Kidney-on-Chip to Decipher Pathophysiological Mechanisms in Renal Cystic Diseases

Abstract

Autosomal Dominant Polycystic Kidney Disease (ADPKD) is a major renal pathology provoked by the deletion of PKD1 or PKD2 genes leading to local renal tubule dilation followed by the formation of numerous cysts, ending up with renal failure in adulthood. In vivo, renal tubules are tightly packed, so that dilating tubules and expanding cysts may have mechanical influence on adjacent tubules. To decipher the role of this coupling between adjacent tubules, we developed a kidney-on-chip reproducing parallel networks of tightly packed tubes. This original microdevice is composed of cylindrical hollow tubes of physiological dimensions, parallel and closely packed with 100-200 μm spacing, embedded in a collagen I matrix. These multitubular systems were properly colonized by different types of renal cells with long-term survival, up to 2 months. While no significant tube dilation over time was observed with Madin-Darby Canine Kidney (MDCK) cells, wild-type mouse proximal tubule (PCT) cells, or with PCT Pkd1 ^+/- cells (with only one functional Pkd1 allele), we observed a typical 1.5-fold increase in tube diameter with isogenic PCT Pkd1 ^-/- cells, an ADPKD cellular model. This tube dilation was associated with an increased cell proliferation, as well as a decrease in F-actin stress fibers density along the tube axis. With this kidney-on-chip model, we also observed that for larger tube spacing, PCT Pkd1 ^-/- tube deformations were not spatially correlated with adjacent tubes whereas for shorter spacing, tube deformations were increased between adjacent tubes. Our device reveals the interplay between tightly packed renal tubes, constituting a pioneering tool well-adapted to further study kidney pathophysiology.

Keywords: ADPKD; hydrogel; kidney-on-chip; microfabrication; tube deformation.

FIGURE 1

Chip microfabrication and cell seeding. (A) Top and (B) side views of the sequential steps for microfabrication and cell seeding: (1) Collagen was poured on top of the tungsten wires placed in a micromold and maintained by PCR tape. (2) Wires were removed. At this stage an additional coating could be applied. (3) Cells were seeded in tubes, and (4) colonized it in a few days.

FIGURE 2

Behavior of MDCK cells in tubes. (A) MDCK-Lifeact-GFP cells were seeded within tubes molded in collagen I at 6 mg/ml, and observed under an optical microscope over time. The temporal evolution at days 9 and 13 after seeding is represented (before and at confluency) is represented. Scale bar: 100 μm. (B) Mean cell density over time (n = 12 MDCK chips). The temporal scale refers to time after cell seeding for colonization curves. Tubes with or without laminin coating were pooled, because of similar behaviors in these two conditions (see Supplementary Figure 8 for separated behavior). The mean curve remains below cell density 1 (corresponding to full colonization) reflecting the fact that some tubes were never fully colonized with cells. Error bars: S.E.M. (C) Kinetic evolution of mean tubes diameter normalized by diameters at seeding as function of the time after tube confluency. The temporal scale refers to time after confluency for curves of tube deformation. A blue horizontal line at D/D₀ = 1, corresponding to no change in diameter, is indicated. Chips with or without laminin coating were pooled. Each time point corresponds to 19–30 tubes. Error bars: S.E.M. (D) Maximum (over time) of the mean normalized diameter. Tubes with or without laminin coating were pooled. Points correspond to individual tube values. Central bar, median; cross, mean; box, values between Q1 and Q3 quartiles; error bars, extreme values [between Q1 - 1.5*(Q3 - Q1) and Q3 + 1.5*(Q3 - Q1)]. (C,D) were computed only for tubes having reached full confluency during the observation period.

FIGURE 3

PCT Pkd1^+/- and Pkd1^-/- organization in tubes. Cells were labeled for F-actin and nuclei, and imaged at confocal both at low and high resolution to study the global cell organization and F-actin organization. Confocal images of tubes labeled with phalloidin-TRITC (red) and Hoechst (blue). (A,B) global organization of Pkd1^+/- (A) and Pkd1^-/- (B) tubes imaged at a low resolution (10× objective). Maximal z projections. Mean tube diameters of tubes in the chips shown, as assessed by z projection, were, respectively, ∼85 and 135 μm for the chips shown in (A,B), in agreement with Pkd1^-/- tube deformation. (C,D) Pkd1^+/- (C) and Pkd1^-/- (D) tubes imaged at a high resolution (40× objective) in steps corresponding to the first early steps of tube dilation for Pkd1^-/- cells. Confocal section (left) and orthogonal projection (right). A median filter (3 pixels) was applied on orthogonal projection images. Scale bar, 100 μm. Mean diameters measured for tubes imaged at high resolution were 85 ± 10 μm for the Pkd1^+/- condition and 95 ± 5 μm for the Pkd1^-/- condition for high resolution images (S.D. are indicated). Here 24 images of Pkd1^+/- tubes and 10 images of Pkd1^-/- tubes were done (performed, respectively, on 8 and 7 chips).

FIGURE 4

F-actin orientation of PCT Pkd1^+/- and Pkd1^-/- cells in tubes. (A–F) F-actin labeling in Pkd1^+/- (A–C) and Pkd1^-/- (D–F) tubes. (A,D) Left: z projection of the inferior half of the tube is shown, scale bar 100 μm. Right, confocal section at the middle of the tube. OrientationJ analysis was performed in a central rectangle corresponding to half of the tube (white rectangle in A,D), in order to get rid of border effects. (B,E) Zoomed part, (C,F) Orientation vector fields (yellow arrows). Magnitude normalized by the strength of orientation (coherency) is represented. Coherency is low in the Pkd1^-/- condition, so that arrows are barely visible in (F). (G) Distribution of F-actin local orientation as assessed by OrientationJ software for PCT Pkd1^+/- (blue) and Pkd1^-/- (red) cells. The analysis was done at a subcellular scale, with a 2 μm local analysis window. Histograms given by OrientationJ are pondered by coherency, meaning that the angle determined for a given window has a more important contribution if there is a clear-cut local orientation. Each histogram is normalized by the size of the analyzed area (in pixels²) before averaging. The analysis was performed on pooled coating conditions (laminin, ECM and collagen), with the majority of tubes corresponding to laminin coating in Pkd1^-/- and Pkd^+/- conditions. (H) Mean coherency (per pixel) for PCT Pkd1^+/- (blue) and Pkd1^-/- (red) cells. ^∗Statistically significant difference with p < 0.05. Each point corresponds to one image.

FIGURE 5

PCT Pkd1^+/- and Pkd1^-/- tube deformation in chips with 200 μm spacing. (A,B) Examples of temporal evolution of tubes with laminin coating, for Pkd1^+/- cells (A) and Pkd1^-/- cells (B). Scale bar:100 μm. Days after seeding: (A) 11, 16 (confluency), 20, (B) 9, 10 (confluency), 14. (C–E) Quantitative analysis, n = 12 Pkd1^+/- chips (black) and n = 14 Pkd1^-/- chips (red), all coatings pooled (see Supplementary Figure 8 for separated behavior). (C) Mean cell density over time. Error bars: S.E.M. (D) Kinetic evolution of mean tubes diameter normalized by diameters at seeding, in function of the time after tube confluency. A blue horizontal line at D/D₀ = 1, corresponding to no change in diameter, is indicated. Each time point corresponds to 8–35 tubes for Pkd1^-/-, 9–20 tubes for Pkd1^+/-. Error bars: S.E.M. (E) Maximum (over time) of the mean normalized diameter. Points correspond to individual tube values. Central bar, median; cross, mean; box, values between Q1 and Q3 quartiles; error bars, extreme values [between Q1 - 1.5*(Q3 - Q1) and Q3 + 1.5*(Q3 - Q1)]. (D,E) were computed only for tubes having reached full confluency during the observation period. ^∗∗ indicates statistically significant difference with p = 0.0002.

FIGURE 6

PCT Pkd1^-/- tube deformation in chips with 100 μm spacing. (A) Example of temporal evolution of tubes seeded with Pkd1^-/- cells at days 2, 4, 10, 17, 23, and 25 after seeding. Scale bar:100 μm. (B,C) The behavior in 100 μm spacing tubes (yellow) was assessed with laminin coating and compared to the behavior in 200 μm spacing laminin-coated tubes (red). (B) Kinetic evolution of mean tube diameter normalized by diameter at seeding, in function of the time after tube confluency. A blue horizontal line at D/D₀ = 1, corresponding to no change in diameter, is indicated. Each time point corresponds to 9–26 tubes for 100 μm spacing, 4–19 tubes for 200 μm spacing. Error bars: S.E.M. (C) Maximum (over time) of the mean normalized diameter. Points correspond to individual tube values. Central bar, median; cross, mean; box, values between Q1 and Q3 quartiles; error bars, extreme values [between Q1 - 1.5*(Q3 - Q1) and Q3 + 1.5*(Q3 - Q1)]. (B,C) were computed only for tubes having reached full confluency during the observation period. ** indicates statistically significant difference with p = 0.001.