Laser ablation of the DSFC Microcirculation
The typical mouse DSFC microcirculation contains a main artery and vein pair (Fig. 1A, B, solid green arrowhead) and smaller artery - vein pairs (open green arrowheads). There are multiple arcade/collateral vessels that connect the arteries to other arteries and veins to other veins (a few are indicated by red and blue arrowheads, respectively in (A). These arcading vessels provide vascular redundancy by allowing redistribution of blood flow. Arteries have significantly smaller diameters than the paired veins with tighter concentric layers of smooth muscle cells (red and yellow in Fig. 1, Pre-ablation 1–3 and Post-ablation 1–3).
The laser ablation was performed at three major locations (Fig. 1A and B, two artery/vein pairs in regions 1 and 3, and an artery in region 2) in the center of the window to maximize blood flow redistribution and to allow long term observation of the developing vascular changes (as some drifting of the tissue occurs within the DSFC over two weeks). The ablated vessels experienced rapid vasoconstriction upstream and downstream from the ablation site (Fig. 1, Post-ablation 1–3). There was complete blood flow interruption in segments just distal and proximal from the ablations (Supplementary videos 1A, 2A and 3A). The laser ablation procedure was focused only on the target vessel, effectively cauterizing them while having little effect on the surrounding tissue. The brown scar tissue located in the muscle fascia subsides at later time points (Supplementary videos 1B, 2B and 3B). Note that in region 2, the ablation of the artery had no effect on the diameter of the adjacent large vein or the blood flow in that vessel (Fig. 1, Post-ablation 2).
Time-course of vascular network remodeling
By day 6 after ablation, there was clear evidence of vascular remodeling throughout the network. Vessel segments associated with the ablated vessels had reduced diameter at day 6, while there was increased diameter in a number of collateral vessels (regions 4, 5, 6, 7 in Fig. 2). By day 13, vessel diameters had nearly returned to pre-ablation values for much of the network. This was due to remodeling of collateral vessels, which allowed an increase in compensatory flow entering tissue regions previously supplied by the ablated vessels. There were also large increases in diameter in a few small vessels that restored flow through the veins by bypassing the ablation sites (arrowheads in Fig. 2, D13, and Supplementary Video 1A, 2A). These structures formed from sequences of smaller microvessels that were part of the original vascular bed. The increased flow through these small bypass channels likely caused the expansion of vessel diameter which eventually matched that of the original vein, similar to previous observations in the mouse gracilis muscle2. Some branches from this small network were deleted/pruned, once this segment became part of the large vein. There was no visible evidence of extensive angiogenesis contributing to the regeneration of the network or restoration of flow in these veins. We did not observe direct reconnection of the veins through the original ablation sites but rather through dilation of small bypass channels, even at day 30.
In contrast, the process of re-routing through pre-existing microvessels was not observed on the arterial side of the network where vessels reconnected through the original ablation site. This is easily visible in ablation region 1 from day 13 on (Fig. 2 and Supplemental Video 1B-D). In ablation regions 2 and 3, the arterial segments appear to reconnect over the ablation site at later time points, but a full connection cannot be clearly traced by the end of the observation period (Supplemental videos 2C and 3C and D). This may be due to the higher resistance to increased blood flow pressure of terminal arterioles vs. a more elastic and compliant vessel wall in venules. This difference in vessel wall structure and hemodynamic response likely spares the capillaries from high arterial pressures, but also slows down the direct bypassing a site of injury. Instead, on the arterial side, the high-pressure flow is redistributed to more distant pre-existing collaterals which are responsible for compensating for the reduced flow downstream of the ablation in the initial stage (until day 20). However, on day 20, there was evidence of angiogenesis on the arterial side, as the artery ablated in Region 1 (Fig. 2) reconnected (arrowhead, D20; Supplemental video 1B). As this new vessel segment grew, original flow through the artery was restored, and the diameters of the major compensating collaterals decreased. The arterial flow in region 3 was re-established by day 30 but via smaller vessels than the original artery (Fig. 2, D30 arrowheads), with blood flow evident via doppler OCT at day 14 (Fig. 5, D14b) and intravital BF imaging at later time points (Supplemental Video 3D). The artery in region 2 did not achieve reconnection by the 30-day time point although some small flow pathways can be traced (Supplemental Videos 2C and 3C).
Angiogenesis at the ablation sites
Because of the endogenous reporters expressed by the mice, we were able to visualize endothelial cells (TIE2-GFP - green) and smooth muscle cells (aSMA-dsRed - red) longitudinally at the ablation sites. In vivo laser confocal imaging of the regions 2 and 3 in Fig. 1 revealed migration of the endothelial and smooth muscle cells through the ablation sites (Fig. 3). In region 3, the vascular pathway was re-established, and blood flow was observed (Fig. 3D). Both endothelial and smooth muscle cells migrated into the damaged region and appeared to establish a connection by day 30, based on doppler OCT imaging (see Fig. 5). A similar process was observed for the other artery, which was ablated at location 2 in Fig. 1 (Fig. 3A, B), although this vessel did not reconnect by the end of our observation period. Angiogenesis was not observed in the large vein that remodeled in region 3, but the remodeled region acquired a covering of smooth muscle cells (Fig. 3C). After day 30, the relevant vessels had shifted out of the window chamber and were no longer observable.
Time course of diameter remodeling
Overall, both arteries and veins changed their diameters over time (Fig. 4). Immediately after the ablations, a number of smaller arteries dilated to increase the fraction of vessel diameters in the 50µm range. Some larger vessels also constricted, moving from the 100µm to the 50 µm range. Starting at day 16, this trend reversed, as more small and large vessels emerged. By Day 30, the arterial diameter distribution returned nearly to pre ablation values. The vein diameters decreased at early time points, but then increased again at days 16 and 20 and returned closer to the baseline distribution at later time points (days 28 and 30, Fig. 4). Overall, there was less diameter distribution changes on the venous side than the arterial side possibly reflecting more adaptability of veins to changes in blood flow.
We next focused on individual vessels to determine how specific vessels contributed to the flow redistribution. Using quantitative flowmetry OCT methods based on amplitude-decorrelation which can be used to estimate flow rate as well as lumen diameters 36,37, we analyzed a number of segments distal and proximal to the ablations sites before and following the ablations (Fig. 5). We also used intravital BF microscopy to determine flow directions (see Supplemental Videos 1–3). In the intact network, the blood flows from left to right in the large artery (#2, Fig. 5) to its branches (#4, 6 and 9). The blood flows from the venous branches (#3, 5, 7, 8 and 10) towards the main vein (#1). Following ablation, the blood flow stopped in the ablated segments, but both upstream and downstream arteries continued to be perfused by arcading vessels from adjacent vascular trees (#2,4,6 and 9). Immediately after and at day 2 post ablation, the segments near the ablations were not perfused. The main vein (#1 and 3) significantly decreased its diameter at day 2 but by day 14 the main vein and its small branch (#10) as well as a contiguous series of capillaries became enlarged to match the size of the vein.
We observed and measured the diameter of the relevant vessels in the ablated area to understand their relationship (Fig. 6). The pattern of the diameter remodeling differs among arteries. The arteries upstream from the ablation (#s 2 and 4) have a decreased diameter and flow velocity during the first two weeks post ablation while the more peripheral arteries (#s 6 and 9 with reversed flow as observed experimentally) increased their diameters from day 2 post ablation and through day 14 suggesting that they are largely responsible for the compensatory flow being rerouted from the parallel arteries (which are outside of the field of the window). Flow velocity decreased in all observed arteries but combined with an increase in diameter, arteries #6 and 9 could deliver a higher volume of blood flow to the remodeled area.
The veins remodeled more significantly than arteries. While the branches of the main vein decreased their diameter initially and after day 14 returned close to pre-ablation values, some side branches remodeled upward more dramatically to re-route the flow in the vein. This was most evident in the microvessels that grew by more than 400x to match the main vein diameter.
Computational model simulation
We next investigated flow patterns in the network before and after the ablations. To do this, we used a computational approach to estimate flow in each segment. The first step in computational modeling is extraction of the network topology and characterization from bright field images taken with the stereo microscope (Fig. 7). The venous network roughly parallels the arterial network with visibly larger diameter vessels. The direction of the flow for each segment was observed from the live BF microscopy recordings and marked on the network map (Fig. 7a and b).
We then used a simulated annealing method to estimate flow rates and pressures throughout the network (see Methods). Guesses are made for the terminal segment pressures and the flows are calculated based on topology and measured vessel diameters. The predicted flow direction in each segment is compared to the observed direction, and an error function is calculated based on the number of incorrect directions. The error is used to scale a set of new guesses for the pressures, which is also subjected to a random function (this is the basis for the simulated annealing method). The process is then repeated to minimize the number of incorrect flow directions in individual segments. Using this method, we find that the pressures are relatively insensitive to flow direction in some vessel segments. Thus, the flow directions in these segments are relatively uncertain, suggesting that these vessels can readily serve to redirect flow in either direction if necessary.
First, the flow distribution of individual vessels was optimized based on network topology and flow directions in the normal non-ablated state for vessels with different levels of uncertainty/flow levels (Fig. 8). Before ablation, the larger arteries have low uncertainty, suggesting that they rarely change flow direction (Fig. 8, blue and yellow color vessels). For example, the vessel fragment in Fig, 8, panel C has a low level of uncertainty (indicated by blue color on the vessel map) and the relative values of the volumetric flow rate are mostly around 20% of that in the largest vessel (which is assumed at a value of 1000). The areas with the highest uncertainty are located between arteries near the center of the network, between the top and the bottom feeding vessels (Fig. 8, red and yellow color vessels). To illustrate this point, vessel fragments in Fig. 8, panels A, B and D have a higher uncertainty (yellow and red on the vessel map) and therefore a wider range of possible values. Note that the segments in panels A and B stabilize at zero or close to zero values which reflects a low priority for these collateral vessels prior to ablation.
Using this method, we estimated flow through the network before (Fig. 9A and C) and after ablation (Fig. 9, B and D) for arteries and veins, respectively. The venous network had more segments with higher flow rate preablation (Fig. 9, C vs. A). In arteries, after ablation, flow tends to be reversed in vessels with a high uncertainty index in the pre-ablation model close to the site of ablation (Fig. 9). There was no flow reversal in the vein network although the flow magnitude was slightly changed in many vessel fragments.