Figure 1.
Mouse submandibular salivary gland organ structure and cleft formation during branching morphogenesis.
Brightfield images of (a) an embryonic day 12 (E12) submandibular salivary gland (SMG) organ explant and (b) an SMG explant harvested at E12 and grown for 24 hours ex vivo with epithelium (E) and mesenchyme (M) labeled. Scale, 200 µm. Single confocal images of E12 SMGs following ICC to detect epithelium (E-cadherin, red) and mesenchyme (PDFGR, cyan) captured at (c) cleft initiation and (d) a late stage of cleft progression. Progressing clefts are indicated (white arrow head). Scale = 50 µm. (e) Diagram depicting cleft transitions.
Figure 2.
Cellular and cytoskeletal organization in developing salivary glands.
Epithelial cells express E-cadherin (red) during organ development and organize as polarized outer columnar cells (OCCs) and non-polarized inner polymorphic cells (IPCs) at (a) E12 and (b) retain this organization after 24 hours of growth. Cortical F-actin localization occurs during cleft formation at (c) E12 and (d) E12+24 hrs. Epithelial proliferation occurs in both outer and inner cell compartments shown with phospho-histone H3-labeled nuclei (red) relative to total nuclei with SYBR green (blue) at (e) E12 and (f) E12+24 hrs. Scale = 20 µm.
Figure 3.
Construction of a GGH model of cleft formation and scope of modeling.
A six cell deep single cleft was designed having 36 pixels as the total cleft depth with predefined cleft cells (dark and medium blue). The local cleft simulation shows the other epithelial areas as polarized OCCs (dark green) and non-polarized IPCs (light green) with mitotic cells (yellow). The mesenchymal compartment (cyan) has been simplified to a single large cell. FPP links in the OCCs are shown as white lines. Spatial conversion: 1 µm = 1.06 pixels. Temporal conversion: 1 MCS = 48 sec. Single cleft model at (a) 0 Monte Carlo steps (MCS) and (b) 1500 MCS (Scale = 50 µm). Time lapse images of a mesenchyme-free E12 epithelial rudiment at (c) time 0 hr and (d) time 20 hrs with cleft measurements under 200× magnification (Scale = 20 µm). Average cleft depth = 36.2 µm. (e) Since cleft depth reaches a maximum value at 1500 MCS, this value was selected to represent the end of cleft progression. (f) The cleft depth distribution over time for the base case condition showing 34.1 pixels cleft depth after 1500 MCS, corresponding to a 20 hr growth period of a pre-defined initiating cleft through the end of cleft progression.
Table 1.
CC3D parameters that were varied in the model and their biological significance in branching morphogenesis and cleft formation.
Figure 4.
Quantitative analysis of cleft formation.
A MatLab function was created for tracing the border of the local cleft. (a) The two cleft extremes were labeled in green and the cleft tip in red. A MatLab tracing of a successful cleft at 1500 MCS shows a (b) high cleft depth and (c) low spanning angle (red lines). (d) MatLab tracing of a non-progressive cleft at 1500 MCS shows (e) low cleft depth and (f) high spanning angle.
Figure 5.
An organ-level simulation containing three ideally localized clefts is shown at (a) 0 MCS and (b) 1500 MCS (Scale = 50 µm). The results were quantified with (c) cleft depth and (d) spanning angle. A slight decrease in the average cleft depth and increase in the average spanning angle was observed relative to the single cleft model run under the same basal conditions.
Figure 6.
Validation of the single cleft model by comparison of predictions with experimental results for manipulation of ROCK.
SMGs were cultured ex-vivo and treated with Y27632 (ROCK inhibitor) and blebbistatin (inhibitor of actomyosin contractility that lowers the affinity of myosin-actin interaction) for 24 hours. Cleft depths were measured using brightfield time-lapse confocal imaging and compared with cleft depths in GGH simulations of ROCK I knockdown (KD). ROCK I KD simulation consisted of a reduced FPP λ value (1) along with a reduced proliferation rate (0.05%). The simulation mimicking blebbistatin action was achieved using only a reduced FPP λvalue (1). Similar trends were observed in the simulations and the experimental results, indicating that the model effectively simulated the cellular effects of inhibition of signaling molecules.
Figure 7.
Effect of varying mitosis location on cleft progression.
The location of proliferating epithelial cells was varied, with 25%, 50% or 75% mitosis occurring in the OCC population, while the MR rate was also varied from 0.5–5%. Results were quantified as (a) cleft depth and (b) spanning angle. The location of proliferating cells had an effect on cleft formation with 50–75% proliferation in the OCCs generally being more effective than 25%. (c) Image segmentation and analysis of confocal images acquired from pHH3 ICC and SYBR green-stained explants at four time points shows temporal changes in mitosis rate that produce an average epithelial mitosis rate of 0.99% between 2–24 hours of ex vivo culture. Dotted line denotes average mitosis rate (0.99%), ANOVA *P<0.05, n = 6.
Figure 8.
Effect of varying cell contractility, cell-cell adhesion, and cell-matrix adhesion strength on cleft progression.
FPP λ values were varied between 1–30, and effects on cleft progression were quantified with (a) cleft depth and (b) spanning angle. FPP λ values lower or higher than 10 are detrimental to cleft progression. Cell-cell (CC) adhesion strength in the cleft region was manipulated by increasing the cell-cell contact energy to mimic E-cadherin-based adhesions and evaluated with (c) cleft depth and (d) spanning angle. Decreasing cell-cell adhesion in the cleft region (increased contact energy) resulted in deeper clefts with lower spanning angles. Modulation of cell-ECM (CM) junctional strength by increasing cell-matrix contact energy was monitored by (e) cleft depth and (f) spanning angle. Increasing cell-matrix junctional strength affects clefts quality marginally with greater cleft depth and lower spanning angle.
Figure 9.
(a) Distribution of cleft measurements from SMG organ explant image data. Classes for parameter combinations were assigned as follows: failed, <17.8 µm; non-progressive, 17.8–30.5 µm; progressive, 30.5–40.7 µm; super-progressive, >40.7 µm. (b) Cross-validated classification accuracy on test set for the full parameter combination (CC, CM, FPP and MR), and exclusion of a single parameter CC, CM, FPP, or MR. A larger decrease in accuracy indicates greater importance. (c) Parameter value distributions within the progressive class. Peaks indicate higher importance for that value in progressive cleft formation.
Table 2.
Classification accuracy table.