Direct visualization and quantitative analysis of water diffusion in complex biological tissues using CARS microscopy

Abstract

To date, it has not been possible to measure microscopic diffusive water movements in epithelia and in the interstitial space of complex tissues and organs. Diffusive water movements are essential for life because they convey physiologically important small molecules, e.g. nutrients and signaling ligands throughout the extracellular space of complex tissues. Here we report the development of a novel method for the direct observation and quantitative analysis of water diffusion dynamics in a biologically organized tissue using Coherent Anti-Stokes Raman Scattering (CARS) microscopy. Using a computer simulation model to analyze the CARS O-H bond vibration data during H2O/D2O exchange in a 3D epithelial cyst, we succeeded in measuring the diffusive water permeability of the individual luminal and basolateral water pathways and also their response to hormonal stimulation. Our technique will be applicable to the measurement of diffusive water movements in other structurally complex and medically important tissues and organs.

Figure 1. Experimental set-up and analysis model.

(A) Set-up for the H₂O/D₂O exchange experiment on the CARS microscopy stage. (B) Multi-shelled three compartments model for analyzing the H₂O/D₂O exchange data. LM and BL are the numbers of the most-outside shell compartment representing lumen and cell. See the supplementary method for more detail. (C) Property of CARS signal obtained from mixtures of H₂O and D₂O. (i) Raw CARS signal intensity obtained from H₂O/D₂O mixtures with a various different ratios. n = 3. (ii) The calibration curve for the relationship between the H₂O concentration and the normalized CARS signal intensity. Derived from the raw CARS signal intensity shown in Fig. 1C–i.

Figure 2. Complete cyst formation by WT-MDCK cells.

(A) Cross section and 3D projection images of immuno-histochemical stained cysts from WT-MDCK cells (Day 5). (green) α-tubulin; (red) ZO-1; (blue) Hoechst33258. The scale bars indicate 10 μm. (B) Representative example of a leak test in WT-MDCK cysts. (i) A fluorescent image of a WT-MDCK cyst 30 min after the application of 1 ml 4-kDa FITC-dextran (5 mg/ml) with the intensity profile of the FITC fluorescent signal across the MDCK cyst. (ii) Time-course of the FITC signals in bath and lumen after the FITC-dextran application.

Figure 3. Direct observation of the H₂O/D₂O exchange process in a MCDK cyst using CARS microscopy.

(A) CARS x-y scanning images of a MDCK cyst during the exchange (i) from H₂O to D₂O and (ii) from D₂O to H₂O. Scale bar: 10 μm. 512 × 512 pixels. Also see Supplementary Video 1. (B) CARS line scanning data in the MDCK cyst during the exchange from H₂O to D₂O. 266 × 2015 pixels. 0.103 μm/pixel; 1.6 ms/line.

Figure 4. Procedure for estimating the water permeability from the CARS [H2O] signal intensity data.

(A) Normalization of the sequential CARS [H₂O] signal intensity data at representative points in the bath (black), cell (blue) and lumen (red) in the MDCK cyst following the H₂O to D₂O perfusion switching. (i) Raw and (ii) normalized sequential CARS [H₂O] signal intensity data are shown. (Top panel) CARS line scanning data. 120 × 10385 pixels. 0.301 μm/pixel; 1.3 ms/line. See the supplement for the method for normalizing the data in detail. (B) Determination of water permeability from the CARS signal intensity data by the analysis program based on the multi-shelled three compartments model. (i) Input: the normalized CARS signal intensity input to the analysis program. (ii) Output: the normalized H₂O concentration data calculated from the normalized CARS signal intensity data input. The best-fit curves generated by the multi-shelled three compartments model were superimposed with the experimental luminal and cellular [H₂O] data. See Supplementary Methods for more detail.

Figure 5. Reproducibility of the H₂O/D₂O exchange experiment.

A continuous recording of the CARS signal intensity obtained during the three repeat H₂O/D₂O exchanges in the lumen (red), cell (blue) and bath (black) compartments (upper panel). 130 × 31750 pixels. 0.414 μm/pixel; 4 ms/line. The insets (i) – (iii) show the expansions of the [H₂O] decays in luminal, cell and bath superimposed with the best-fit curves generated by the analysis program. P_d(l): (i) 4.3 × 10⁻³ cm/s, (ii) 4.2 × 10⁻³ cm/s, (iii) 4.3 × 10⁻³ cm/s. P_d(bl): (i) 4.7 × 10⁻³ cm/s, (ii) 4.4 × 10⁻³ cm/s, (iii) 4.9 × 10⁻³ cm/s.

Figure 6. Comparisons of the water permeability values estimated from the H₂O to D₂O switch and a sequential D₂O to H₂O switchback in the same cyst.

(A) Raw CARS signal intensity data recorded on switching from an H₂O-based solution to a D₂O-based (i) solution and then subsequently switching the D₂O-based solution back to the H₂O-based solution (ii). (B) Normalized signal decays induced by (i) the H₂O to D₂O switch and (ii) the sequential D₂O to H₂O switchback. (C) Estimation of the water permeabilities (Pd) from the signal decays in (i) the H₂O to D₂O switch and (ii) the subsequent switchback shown in Fig. 6B. Note that the best-fitted curves are superimposed. (i) P_d(l): 0.85 × 10⁻³ cm/s; P_d(bl): 1.3 × 10⁻³ cm/s; P_d(p): 10⁻⁷ cm/s. (ii) P_d(l): 0.66 × 10⁻³ cm/s; P_d(bl): 1.9 × 10⁻³ cm/s; P_d(p): 10⁻⁷ cm/s. In the switch back experiment (Fig.S6C–ii), the whole [H₂O] data were used for the best-fit. (D) Summary data for the water permeabilities (Pd) obtained from the H₂O to D₂O switch and the sequential D₂O to H₂O switchback (performed on the same cyst). Mean ± SEM, n = 12 cysts. p = (luminal) 0.012 and (basolateral) 0.41 by a paired Wilcoxon rank test.

Figure 7. Effects of AQP4 expression on water movement in MDCK cyst.

(A) Cross section and 3D projection images of immuno-histochemical stained cysts from MDCK cells stably transfected with AQP4 (Day 5). (green) α-tubulin; (red) ZO-1; (blue) Hoechst33258. The scale bars indicate 10 μm. (B) Expression of AQP4 in AQP4-MDCK cysts. (i) Immnuo-staining of AQP4 in an AQP4 MDCK cyst. (green) AQP4; (red) F-actin; (blue) Hoechst33258. Scale bar: 10 μm. (ii) EGFP (left) and CARS (right) signals obtained from an AQP4 MDCK cyst on the CARS microscope. Note that the intracellular fluorescent signal comes from EGFP co-transfected with AQP4 by the mAQP4-M1-pIRES2-EGFP construct, not a leak in of FITC-dextran. Scale bar: 20 μm. (C) Representative example of a leak test in AQP4-MDCK cysts. (i) A fluorescent image of an AQP4-MDCK cyst 30 min after the application of 20-kDa FITC-dextran (5 mg/ml). Scale bar: 20 μm (ii) Time-course of the FITC signals in bath, cell and lumen after the FITC-dextran application. Note that the AQP4-WT cells also expressed EGFP and its fluorescent signal was also detected in the cell layer of the cyst. (D) Decay of the normalized H₂O concentration in bath, cell and lumen during the H₂O/D₂O exchange in an AQP4-transfected MDCK cyst. (Top panel) CARS line scanning data. 176 × 10385 pixels. 0.207 μm/pixel; 1.36 ms/line. (E) Summary data for the water permeability of the luminal and basolateral membranes in WT and AQP4-transfected MDCK cysts. Mean ± SEM, n = 15 (WT) and 8 (AQP4) separate cysts. p = (luminal) 0.9 and (basolateral) 0.005 by unpaired Wilcoxon rank test (**: p < 0.005).

Figure 8. Effects of vasopressin on water movement in an M-1 cyst.

(A) Immuno-histochemical staining of AQP2 in M-1 cysts in control and after incubation with 100 nM arginine vasopressin (AVP) for 30 min at 37°C. (green) AQP2; (red) F-actin; (blue) Hoechst33258. (B) Leak test for M-1 cysts using FITC-dextran. (i) Fluorescent images of an M-1 cyst before and 30 min after the application of 1 ml 20-kDa FITC-dextran (5 mg/ml). Scale bar: 20 μm (ii) Time-course of the FITC signals in bath, cell and lumen after the FITC-dextran application to the M-1 cyst. (C) Plots of the normalized H₂O concentration in the bath, cell and lumen in the M-1 cyst following the H₂O to D₂O perfusion switch (i) before and (ii) after an application of vasopressin. P_d(l): (i) 1.4 × 10⁻³ cm/s, (ii) 4.1 × 10⁻³ cm/s. P_d(bl): (i) 1.9 × 10⁻³ cm/s, (ii) 2.0 × 10⁻³ cm/s. (D) Summary data for the water permeability of luminal and basolateral membranes in M-1 cysts before and after an application of vasopressin to the same cyst. Mean ± SEM, n = 8 cysts. p = (luminal) 0.0078 and (basolateral) 0.84 by a paired Wilcoxon rank test (*p < 0.01).