PLoS Computational Biology
Public Library of Science
image
Numerical evaluation reveals the effect of branching morphology on vessel transport properties during angiogenesis
DOI 10.1371/journal.pcbi.1008398 , Volume: 17 , Issue: 6 , Pages: 0-0
Article Type: research-article, Article History
Abstract

Blood vessels are important for the transport of various substances, such as oxygen, nutrients, and cells, to the entire body. Control of blood vessel formation is thought to be important in health and disease. In the last decades, various factors which regulate blood vessel branching morphology have been identified. Gene modification of some of these identified factors results in hyper-branching of the vasculature while others cause hypo-branching of the vessel. Given the importance of the transport property of the blood vessel, it is important to examine the effect of these identified factors on the transport property of the affected vascular morphology. In line with these facts, we reconstruct 3D vessel structures from 2D confocal microscopy images. We then numerically simulate blood flow in the structures. Interestingly, our results suggest vessel network complexity negatively affects the blood perfusion efficiency and tissue oxygenation during angiogenesis. Thus, our results highlight the importance of flow analysis considering the detailed 3D branching pattern of the vascular network to quantitatively evaluate its transport properties.

1. Introduction

Upon tissue hypoxia, proangiogenic factors such as vascular endothelial growth factors (VEGFs) are secreted to induce new blood vessel formation from existing vessels, termed angiogenesis [1]. While it is important to promote angiogenic vessel growth for tissue homeostasis, the vessel formation in tumors can give transformed cells better access to nutrients and oxygen [24]. Transformed cells hijack blood vessels for metastasis to a distant tissue [5]. In the last decades, various studies have identified critical factors controlling endothelial proliferation and vascular branching [610]. In line with these observations, anti-angiogenic therapy aimed to starve tumor cells of nutrients and oxygen by reducing tumor vascularization. However, the outcome of the treatment was more limited than expected [11,12]. Reducing blood vessel formation in tumor tissues is thought to enhance ischemia to induce tumor resistance against chemotherapies as well as to restrict drug delivery [6,13].

Recently, numerical simulation of blood flow and associated transport phenomena in biological systems attracts attentions, since it has a potential to clarify the complex dynamics of blood flow within complex capillary network, and also provide quantitative information on local hemodynamic parameters such as wall shear stress, pressure and oxygen concentration, which are difficult to obtain experimentally in vivo . However, attempts to compare transport properties of vascular networks with different morphological features are still quite limited. This could be achieved by artificially removing/adding vessels from a reference structure based on statistical or empirical rules [14,15]. However, the resultant structures are no longer real, and experimental validation is prohibitive.

In the present study, we apply genetic modification techniques to realize hyper- and hypo-branching vascular network in mouse retina. This allows us to systematically analyze blood flow and associated oxygen transport for the real structures with different morphological features. The obtained numerical results are also compared with experimental observation for validation. Interestingly, hyper-branching morphology attenuates effective blood flow at the angiogenic front in silico and fails to improve tissue hypoxia in vivo. In contrast, hypo-branching morphology enhances blood supply at the growing vasculature. Consistently, VEGF expression of the angiogenic front region is efficiently improved, suggesting better oxygen supply at the region. Our results indicate the importance of evaluating branching networks of the vasculature by transport property of the blood vessel.

2. Materials and methods

2.1. Ethics statement

All animal experiments were conducted according to the protocols approved by the local animal ethics committees and authorities (Regierungspräsidium Darmstadt, B2/1073) and institutional regulations.

2.1.1. Mice breeding

As previously described, transgenic Pdgfb-iCre mice were bred into lines of animals containing a LoxP-flanked Prkci [16] and LoxP-flanked Foxo1 [17]. Intraperitoneal injections of tamoxifen (Sigma, MO, T5648) from postnatal day1 (P1) to P3 were used to induce activation of Cre in neonatal mice. The phenotype of the mutant mice was analyzed at P5 and tamoxifen injected Cre negative littermates were used as controls.

2.2. Retina staining

For retina staining, eyeballs were fixed for 20 min in 2% Paraformaldehyde (PFA; Sigma, P6148) at room temperature (RT). Afterwards, retinas were dissected in PBS and fixed for 30min in 4% PFA on ice. Next, they were washed three times with PBS and permeabilized and blocked for 2hr at RT in blocking buffer (BB): 1% fetal bovine serum (FBS; Biochrom GmbH, Berlin, Germany), 3% Bovine Serum Albumin (BSA; Sigma, A2153), 0.5% TritonX-100 (Sigma, T8787), 0.01% Na deoxycholate (Sigma, D6750) and 0.02% Na Azide (Sigma, S8032) in PBS on rocking platform. Then they were incubated with primary antibodies (anti-ICAM-II (BD Pharmingen, 553326, 1:100), anti- Collagen-Type IV (Collagen-IV) (Bio-RAD, 2150–1470, 1:400) and anti-VEGF164 (R&D Systems, AF-493-NA, 1:100) in 1:1 BB/PBS), overnight at 4°C on rocking platform. Retinas were then washed four times for 30 min in PBS/ 0.2% TritonX-100 (PBT) at RT and incubated with Alexa Fluor conjugated secondary antibodies (Invitrogen, 1:500) in 1:1 BB/PBS for 2hr at RT. After another four times of washing with PBT, retinas were radially cut into four lobes and flat-mounted onto slides using Fluoromount-G mounting medium (Southern Biotech, 0100–01).

2.3. Image acquisition, processing and statistical analysis of vascular network

As the models for the hypo- and hyper-branched vasculature, we employed loxP-flanked Prkci [16] and loxP-flanked Foxo1 [17] mice crossed with pdgfb-icre mouse. Endothelial proliferation is modulated via controlling c-Myc expression in both model mice [17,18]. After tamoxifen injection from P1 to P3 to induced effective gene deletion, mouse retina was harvested at P5 and stained with an anti-ICAM-II antibody the marker of the inner lumen of the blood vessels. Stained vasculatures were visualized by confocal microscopy. In the present study, a subset of the retinal tissue, consisting of a single artery as the flow inlet, a single vein as the flow outlet, and the whole in-between capillary bed, from the optic nerve head (ONH) at the center of retinal tissue to the angiogenic front of the vasculature, was extracted for further detailed analyses as shown in S1 Fig. This specific wedge-shaped region between an artery and a vein, which is referred to as ‘A-V region’ hereafter, can be considered as a minimum flow unit of the entire retina vasculature. The validity and limitation of analyzing the minimum flow unit are discussed in S1 Text.

To measure the property of the vasculature, morphometric measurements were introduced (Fig 1A). Consistent with the previous report [18], the retinal vasculature in the Foxo1 EC specific inducible knock out (Foxo1iΔEC ) showed increased vascular density, vessel length density and branching index (Fig 1B, 1D, 1E and 1F), while the same factors in the Prkci EC specific inducible knock out (Prkci iΔEC ) were decreased (Fig 1C, 1G, 1H and 1I).

Characterization of the hyper- and hypo-branching vascular models.
Fig 1
A) Grayscale image of a sample subset of the retinal microvasculature stained for ICAM-II (i), converted into the vessel mask (ii) and used for skeleton analysis (iii); Scale bar represents 30 μm. B) Staining of ICAM-II in control and Foxo1iΔEC mouse retinae at postnatal day 5 (P5); Scale bar represents 200 μm. C) Staining of ICAM-II in control and PrkciiΔEC mouse retinae at P5; Scale bar represents 200 μm. D) Quantification of vascular density (percentage of area occupied by vessels), E) vessel length density (percentage of area occupied by skeletonized vessel) and F) branching index (branch point/ mm of vessel length) in control and Foxo1iΔEC retina at P5. G) Quantification of vascular density, H) vessel length density and I) branching index in control and PrkciiΔEC mouse retinae at P5. Data represent mean ± S.E.M. two-tailed unpaired t-test *p < 0.05, **p < 0.01, ***p < 0.001 (n = 5 for D and G, n = 4 for E, F, H and I).Characterization of the hyper- and hypo-branching vascular models.

In the experiments using KO animal models, data are derived from three independent experiments (three sets of mutant mice and control littermates). The data are presented as mean ± S.E.M. All statistical analyses were carried out using Prism software (GraphPad, CA); p<0.05 was considered as significantly different. Quantification of VEGF-A signal intensity was carried out by measuring mean gray value using ImageJ (1.52o). Morphometric analyses of the retinal vasculature were assessed using Fiji (2.0.0-rc-69/ 1.52n) Vessel Analysis plugin, Fiji Skeletonize plugin and Fiji Skeleton Analyzer. Volocity (Perkin Elmer, MA), Photoshop CS, Illustrator CS (Adobe), ImageJ and Fiji software were used for image processing in compliance with general guidelines for image processing.

2.4. 3D reconstruction of vascular network

The three-dimensional (3D) vascular models were reconstructed from the two-dimensional (2D) confocal images. The obtained RGB images were converted to black and white binarized images (‘vessel mask’) using MATLAB Image Processing Toolbox (Fig 2A), where white and black pixels correspond to the regions of the blood vessel and the surrounding tissue, respectively. The binarized vessel structure was then projected onto a x-y|z = 0 plane in the 2D Cartesian coordinate system (Fig 2B). For all grid points within the blood vessel, the shortest distance to the vessel wall was calculated in the 2D structure, and then a 3D sphere with the diameter obtained at each grid point was placed in a 3D Cartesian coordinate system as schematically shown in Fig 2C. The envelope of all the spheres was then used to define the 3D blood vessel structure (Fig 2D). After reconstructing the 3D vascular structure, a signed distance function, which is commonly referred to as a level-set function [19], was computed at every grid point in both vessel and tissue regions. The obtained level-set function was integrated to an in-house solver for the blood flow.

Image processing and blood flow simulation from the growing mouse retinal vasculature.
Fig 2
A) Generation of binary vessel mask from RGB images. Higher magnification images of indicated areas in the upper panels are shown in the lower panels. Scale bar in the upper panel, 30 μm; 10 μm in the lower panel. B) Schematic of the searching process for the nearest solid wall (black pixel) on each pixel point and defining the local radius r (pixel distribution is projected on Cartesian grid coordinate alternatives x ^ = [x, y]). C) The local radius r for creating 3D vascular structure in a 3D Cartesian coordinate system, x = [x, y, z], extended from alternatives x ^ . The region with ‖x‖<r is defined as blood vessel, while the rest is a surrounding tissue. D) Sample reconstructed 3D model of the retinal vasculature with indication of the inlet and outlet in Cartesian coordinate system. E) Visualization of the velocity amplitude on the central x-y plane (along the z axis) for Foxo1iΔEC and F) PrkciiΔEC mutants and respective controls; Color scale represents the logarithmic form of the normalized velocity (log10u‖/uin).Image processing and blood flow simulation from the growing mouse retinal vasculature.

2.5. Numerical simulation of blood flow

As a suspension of erythrocytes in plasma, blood exhibits shear-thinning properties due to rouleaux formation [20]. Rouleaux, aggregation of red blood cells (RBC), causes increased blood viscosity due to increased effective volume of RBCs [20]. Thus, blood behaves as a non-Newtonian fluid as its viscosity decreases with applied shear. However, under a sufficiently high shear rate (> 150 s-1 , [21]), blood flow can be accurately modeled as Newtonian fluid with a constant viscosity [22]. Nagaoka et al. measured averaged values of 632 to 1539 s-1 for shear rate in human retinal first arteriole and venule branches [23]. Moreover, Windberger et al. [24] measured species-specific effect of shear rate on blood viscosity and erythrocytes aggregation within low shear rates of 0.7, 2.4 and 94 s-1. They reported different viscosity values and RBC aggregation index among different mammalian species: horse, pig, dog, cat, rat, cattle, sheep, rabbit and mouse. In low shear rate regime (0.7 and 2.4 s-1), they found lower shear dependent viscosity enhancement in cattle, sheep, rabbit and mouse, as compared to horse, rat, pig, dog and cat. Compared to other species, erythrocyte aggregation (EA, measured using four different methods) in mouse was found low and, in some methods, undetectable. At high shear rate (94 s-1) they found the EA destroyed and the RBCs orientated to the flow direction. Based on these results, a simple Newtonian behavior is accounted to retinal flow in this model.

In vivo measurements have shown systolic and diastolic flow rate variations in human [25] and mouse [26] retinal vasculature. The influence of pulsation on flow is governed by the Womersley number [27]. This dimensionless number is defined as the ratio between the time-scale for the wall information propagating to the bulk fluid via fluid viscosity and the pulsation period. Consequently, when the Womersley number (α) is sufficiently low (≤1), there is enough time for a velocity profile to develop during each cycle (viscous-dominated flow), so that the resultant flow can be considered as a quasi-steady flow. While, for large values of α (≥10), the transient interaction between pulsation and blood flow plays an essential role, and thereby the unsteady flow analysis is crucial. The Womersley number is defined by the following formula:

alternatives α = d ( ω ν ) 1 2

Where d, ω and ν are the vessel diameter, the angular frequency for a heart rate and the kinematic fluid viscosity, respectively. Using in vivo reported values for mouse as such: d~10−5 [m], ω~1 [s−1], and ν~10−6 [m2 /s] [2830], in neonatal mouse retina, α becomes as low as ~10−2. Hence, the flow in retina model can be considered in viscous-dominated regime and this validates the quasi-steady assumption mentioned above. The other blood rheological properties such as the Fåhræus–Lindqvist effects, and the elastic effects of a vessel wall are not taken into account in the present model. Consequently, the flow is assumed to be incompressible, Newtonian and steady. The governing equations for the blood flow are given by the following Navier-Stokes and continuity equations:

alternatives u j u i x j = 1 R e 2 u i x j x j p x i η u i ϕ ,
alternatives u i x i = 0 .

Here, ui, p and Re denote the velocity component in the i-th direction, the static pressure and the Reynolds number, respectively. The inlet bulk mean velocity ( alternatives U i n * ) and the diameter of the inlet artery ( alternatives D i n * ) were used for non-dimensionalization in Eqs (2 and 3). Here, the superscript of * represents a dimensional value, whereas a quantity without a superscript indicates a dimensionless quantity. The Reynolds number, expressing the ratio of the inertial and viscous forces, is defined as: alternatives R e U i n * D i n * ν * . In this study, Re was set to be 0.1 based on alternatives D i n * and the kinematic viscosity and a typical blood flow velocity in an artery reported in [24,28].

The current code is based on a volume penalization method [31], which is categorized as immersed boundary techniques [32]. The main advantage of the present approach is that an arbitrary 3D structure can be embedded in 3D Cartesian computational grids, so that there is no need to generate boundary-fitted grids for each geometry. Meanwhile, it has relatively slow convergence due to smearing of the fluid-solid boundary, so that we performed grid convergence studies to confirm that the present conclusions are not affected by further grid refinement (see, S2 Text and S2 Fig for the results of the grid convergence study). The last term on the right-hand-side of Eq (2) corresponds to an artificial body force term introduced in the volume penalization method (VPM) in order to realize a no-slip condition at a fluid-solid boundary [31]. Computational grid points were uniformly distributed in space and the spatial grid resolutions were set to be alternatives ( Δ x * , Δ y * , Δ z * ) ( 1.5 μ m , 1.5 μ m , 1.5 μ m ) . The dimensions of the computational domain for each model are listed in Table 1. The dimensions of the computational domain depend on each sample, while the computational resolutions were kept constant for all the cases.

table-wrap
Table 1
Domain size and grid resolution for every model structure.
Name Domain size [μm] alternatives L x * × L y * × L z * alternatives D i n * [μm] Grid size [μm] Number of grid points Nx×Ny×Nz
Foxo1 CTRL-1 800.30 × 1072.10 × 33.22 12.08 1.51 530 × 710 × 22
Foxo1 iΔEC-1 755.00 × 1102.30 × 36.24 12.08 1.51 500 × 730 × 24
Foxo1 CTRL-2 588.90 × 936.20 × 36.24 12.08 1.51 390 × 620 × 24
Foxo1 iΔEC-2 619.10 × 694.60 × 36.24 10.57 1.51 410 × 460 × 24
Foxo1 CTRL-3 694.60 × 1087.20 × 30.20 13.59 1.51 460 × 720 × 20
Foxo1 iΔEC-3 936.20 × 1011.70 × 33.22 13.59 1.51 620 × 670 × 22
Prkci CTRL-1 664.40 × 1208.00 × 36.24 13.59 1.51 440 × 800 × 24
Prkci iΔEC-1 830.50 × 966.40 × 33.22 15.10 1.51 550 × 640 × 22
Prkci CTRL-2 845.60 × 1253.30 × 36.24 13.59 1.51 560 × 830 × 24
Prkci iΔEC-2 543.60 × 1208.00 × 30.20 10.57 1.51 360 × 800 × 20
Prkci CTRL-3 588.90 × 951.30 × 33.22 12.08 1.51 390 × 630 × 22
Prkci iΔEC-3 588.90 × 875.80 × 27.18 10.57 1.51 390 × 580 × 18
Cases with finer resolution for verification of grid resolutions
Foxo1 CTRL-1F 800.30 × 1072.10 × 33.22 12.08 0.76 1060 × 1420 × 44
Foxo1 iΔEC-1F 755.00 × 1102.30 × 36.24 12.08 0.76 1000 × 1460 × 48
Prkci CTRL -1F 664.40 × 1208.00 × 36.24 13.59 0.76 880 × 1600 × 48
Prkci iΔEC-1F 830.50 × 966.40 × 33.22 15.10 0.76 1100 × 1280 × 44

The computational domain is schematically shown in Fig 2D. Uniform velocity profiles were applied at the inlet and outlet boundaries, Uin and Uout, respectively. The velocity at the outlet was determined such that the fluid volume is strictly conserved throughout the structure. At the outer boundaries of the computational domain, a Neumann boundary condition was imposed for the pressure.

For spatial discretization, an energy conservative second-order central finite difference scheme (CDS) was applied for the convection term on the left-hand-side of Eq (2) on the staggered grid [33]. The diffusion term on the right side of Eq (2) was also discretized by a conventional second-order central finite difference scheme. For temporal advancement, the Simplified Marker and Cell, SMAC, [34] method was implemented to decouple the pressure in the Navier-Stokes equations shown in Eq (2). The second-order Crank-Nicolson scheme was employed for the diffusion and VPM terms. The other terms were calculated explicitly with the third-order Runge-Kutta scheme [35].

3. Simulation results

The obtained velocity fields showed distinct flow distributions for hyper- and hypo-branched structures (Fig 2E and 2F). In the present study, we conducted flow simulations for three independent samples for hyper- and hypo-branched structures, respectively. For reference, the results of the other two samples are also shown in S3 and S4 Figs. In general, the flow distribution was un-even in the Foxo1 iΔEC vessel network comparing to that in control. Specifically, the velocity around the peripheral edge (angiogenic front) is drastically attenuated due to the hyper-branching. Conversely, the hypo-branched Prkci iΔEC network results in a more evenly distributed flow throughout the entire network including the peripheral edge.

As shown in Figs 2D and S1, the blood is supplied from an artery and distributed to the azimuthal direction by branching capillary network and then eventually flows into a draining vein toward the outlet. Hence, the spatial distribution of the azimuthal flow rate can be considered as a key quantity to evaluate the transport properties of the vascular network. In order to quantify the amount of flow transported to the peripheral regions, we introduced a cylindrical coordinate system (r -θ ) with its origin at the first branching point from either the artery or the vein near the inlet (Fig 3A). Here, r and θ represent the distance from the origin and the azimuthal direction, respectively. Accordingly, the local flow velocity vector was decomposed into the radial and azimuthal components, i.e., ur and uθ (Fig 3A). Fig 3B shows the spatial distribution of the absolute local velocity ‖u‖, normalized by the inlet velocity uin, while that of the normalized azimuthal flow velocity, uθ/ uin , is depicted in Fig 3C.

Quantification of blood flow at the angiogenic front.
Fig 3
A) Coordinate transformation from the Cartesian system (x, y) to the cylindrical system (r, θ). B) The velocity field transformed from the Cartesian coordinate system to the C) cylindrical one. D) Distribution of the azimuthal velocity uθ between the artery and the vein. E) Schematic of defining the angiogenic front. The entire vasculature is separated into two regions: the inner plexus and the angiogenic front. F) Visualization of the velocity amplitude on the central x-y plane (along the z axis) in Foxo1iΔEC and G) PrkciiΔEC mutant mice and respective controls at P5; Color scale represents the logarithmic form of the normalized velocity (log10u‖/uin). H, I) The normalized azimuthal flow rate at the angiogenic front of the Foxo1iΔEC (H) and PrkciiΔEC mice (I) with control. Data represent mean ± S.E.M. two-tailed unpaired t-test *p < 0.05 (n = 3).Quantification of blood flow at the angiogenic front.

To evaluate the perfusion efficiency of each structure, the entire vascular network was divided into two regions, i.e., angiogenic front and inner plexus (Fig 3D and 3E). The angiogenic front was defined as the region of 0.7Rmax < r <Rmax, where Rmax is the distance of the farthest blood vessel from the origin. The current definition of the angiogenic front is based on the previous observation that c-Myc expression and the proliferation of ECs are active in the region [17,18]. The azimuthal flow rates Qθ for the two regions were separately calculated through volume integration of uθ. This way, the portion of the blood flow transported to the angiogenic front can be quantitatively evaluated. While the vessel density was increased in the Foxo1 iΔEC mice, Qθ at the angiogenic front was significantly decreased in Foxo1 iΔEC mice (Fig 3F and 3H). In contrast, it was increased for Prkci iΔEC mice compared to the control littermates, although vascular branching was decreased (Fig 3G and 3I).

According to Fig 3F and 3G, it can be seen that the blood velocity in the vicinity of the angiogenic front can be four-order-of-magnitude smaller than the inlet velocity (Note that the color contours are logarithmic). However, it does not mean there is no effect of the blood flow at the angiogenic front. Considering the molecular diffusion coefficient of the oxygen, the typical time-scale needed for oxygen to be diffused within a distance of 100 micrometer is in the order of ten seconds. Therefore, if there exists blood flow within a certain distance, oxygen can be supplied to the tissue through the combination of blood flow and molecular diffusion. We can also estimate the significance of the convection by the Peclet number, which is the ratio of the convection and diffusion effects. The Peclet number is around unity even when the blood velocity is 1% of the inlet velocity. This means that the convective effects have similar impacts to the molecular diffusion even when the blood velocity is around 1% of the inlet velocity.

Since the current simulation assumes the blood to be incompressible, the azimuthal flow rate Qθ from the artery to the vein has to be exactly compensated by the decrease of the radial velocity so as to satisfy the mass conservation described by Eq (3). We also confirmed that the distribution of ur is quite similar to that of ‖u ‖ shown in Fig 3B. Specifically, ur is decreased slowly from the inlet artery toward the angiogenic front when Qθ is large, while it decays rapidly when Qθ is small.

We also note that there is a substantial difference in Qθ between the two control populations shown in Fig 3H and 3I. Using two different mouse lines to obtain the hyper- and hypo-branching phenotypes causes variations in litter and animal sizes. During retinal angiogenesis from postnatal day 0 (P0) to P7, the blood vessel expands rapidly and extensively on the surface of retina. In other words, the vasculature dramatically grows even in a couple of hours, which could cause huge variations in vascular morphology across different mouse lines even in wild type animals. Due to the above reasons, meaningful conclusions can be drawn only through comparisons between KO and control animals.

4. Experimental validation

The distinct blood flow distributions for Foxo1 iΔEC and Prkci iΔEC mice should have significant impacts on their transport properties. To validate our numerical results, we observed VEGF expression with an anti-VEGF antibody in the mutant retinae, which reflects the hypoxic status of the tissue. During angiogenesis, VEGF is expressed around the angiogenic front. Once the tissue is vascularized, local hypoxia is improved by oxygen supply with the newly formed blood vessels to downregulate VEGF expression [1].

Since the experiments with each of the hyper- and hypo-branching mutant groups were carried out independently, the basal signal intensity detected for VEGF-A was different across the mutant groups. Therefore, the mean gray value of the VEGF-A signal at the angiogenic front was normalized by the average mean gray values read from the regions showing background noise, e.g. the torn parts of the tissue or the ONH hole in the middle.

Consistent with the decreased flow rate at the angiogenic front in the Foxo1 iΔEC mice, VEGF expression around the angiogenic front is significantly increased compared to that of the control (Fig 4A and 4C). Conversely, VEGF expression in Prkci iΔEC mice around the angiogenic front region was decreased compared to that of the control (Fig 4B and 4D).

The effect of the hyper- or hypo-branching vasculature on VEGF expression during angiogenesis A) Staining of ICAM-II (red) and VEGF-A (green and gray) in the Foxo1iΔEC and B)
PrkciiΔEC mouse retina at P5 and the respective controls; Scale bars represent 200 μm. Higher magnification images of the indicated areas in the left panel are shown in the right panel; Scale bars represent 100 μm. C) Quantification of the normalized VEGF-A signal intensity at angiogenic front in Foxo1iΔEC and D)
PrkciiΔEC mutant mouse retinas and the respective controls at P5. Data represent mean ± S.E.M. two-tailed unpaired t-test *p < 0.05 **p < 0.01 (n = 4).
Fig 4
The effect of the hyper- or hypo-branching vasculature on VEGF expression during angiogenesis A) Staining of ICAM-II (red) and VEGF-A (green and gray) in the Foxo1iΔEC and B) PrkciiΔEC mouse retina at P5 and the respective controls; Scale bars represent 200 μm. Higher magnification images of the indicated areas in the left panel are shown in the right panel; Scale bars represent 100 μm. C) Quantification of the normalized VEGF-A signal intensity at angiogenic front in Foxo1iΔEC and D) PrkciiΔEC mutant mouse retinas and the respective controls at P5. Data represent mean ± S.E.M. two-tailed unpaired t-test *p < 0.05 **p < 0.01 (n = 4).

5. Discussion

While suppression of blood vessel formation has been expected to have a negative impact on blood supply, our results indicate the opposite trend. Namely, the suppression of vessel branching enhances blood flow, and thereby oxygen transport at the angiogenic front. Conversely, the enhancement of blood vessel formation attenuates blood flow at the angiogenic front. The present results underline the importance of detailed flow analysis considering a complex microvascular structure for evaluating its transport properties. In clinical applications, it has been reported that combining anti-angiogenic drug and chemotherapy statistically improves the progression-free survival rate [36]. This suggests that the normalization of the vascular network around tumors could contribute to delivering the medicine to the tumors by enhancing blood supply. Recently, the nanoparticulate drug delivery system has attracted much attention due to its potential to further improve the therapeutic efficacy. Since it is known that the transport of nanoparticles in vasculature is significantly affected by their sizes and shapes [37], the detailed analysis of blood flow and associated mass transfer becomes ever more needed. Computational fluid dynamics of blood flow in vascular networks should be a key for optimizing the shape, size and chemical properties of nanoparticles in drug delivery and even for the design and control of microrobots for future medical applications [38].

Although flows in large vessels such as a coronary artery can be essentially modelled as homogeneous Newtonian fluid, flows in the microcirculation are strongly affected by the complex interactions among plasma flow, interstitial flow, complex geometry of branching patterns and the dynamics of blood cells whose size is close to the blood vessel diameter [39]. As discussed in Sec. 2.5., the Newtonian assumption is considered reasonable, at least, within a single vessel according to existing literatures. However, these evidences have been obtained in relatively straight vessels sufficiently away from branching points [40]. In contrast, near a branching point, the effects of three-dimensionality and axial flow development could be important. However, these effects are neglected in 1D analyses [41,42]. It is also known that red blood cells are not evenly distributed from a parent vessel to daughter vessels. This so-called skimming effect [43] reduces the hematocrit value of a smaller vessel from that of large artery and vein. Therefore, even though the Newtonian assumption is reasonably well for each branch, the effective viscosity could be gradually changed depending on the generation of a branch from the artery or vein.

In order to accurately reproduce such multi-scale and multi-physics phenomena, 3D flow simulation is necessary. Due to its large computational cost, however, most existing studies rely on simplified 1D analysis [41,44], whereas studies applying 3D analysis for vascular network are still limited [45]. In the present study, a new approach to implement a complex structure of vascular network into 3D flow simulation was introduced. By representing an arbitrary complex 3D network structure with the level-set function, the structure was immersed in 3D cartesian coordinate by the volume penalization technique. This has two major advantages: First, the grid generation which is required in flow simulation with a body-fitted coordinate system can be omitted, so that it becomes quite straightforward to simulate flows in different geometries. Second, since the present scheme deploys structured grids in both vessel and tissue regions, the coupling between the blood and interstitial flows and mass transport between blood and tissue can be solved in a unified manner. The latter is particularly important when the transport of oxygen or nanoparticles from blood flow to surrounding tissues will be considered in future work.

Although we assume that the gene modification of Prkci or Foxo1 is the primal cause for the morphologic changes in the vasculature, we cannot exclude a possibility that ischemia induced by the inner shunts discussed in S3 Text causes hyper-branching at the angiogenic front. During vessel growth, it is widely known that ECs are actively proliferating just behind the angiogenic front and around the vein, but not around the artery [46]. Then, ECs migrate toward upstream of the flow and integrated into the artery [47]. c-Myc is one of the major power generators of endothelial proliferation [17]. Previous reports indicate that EC proliferation of Prkci EC specific inducible KO mice results in reduced EC proliferation via compromised c-Myc expression [18], while Foxo1 EC specific inducible KO exhibit enhanced c-Myc expression and increased EC proliferation in both angiogenic front and vein [17,18]. To address the effects of inner shunt formation or arterial/venous formation would be very interesting. For instance, Pard3 EC specific inducible KO mice show reduced vessel branching and increased vessel pruning, although c-Myc expression is not affected [48]. Additionally, EC migration toward blood flow is compromised. Another interesting example is EC specific integrin b1 KO mice, which showed attenuated EC migration [49]. The application of the current approach to those two models would be an interesting avenue in future studies.

Acknowledgements

The authors gratefully acknowledge former graduate students, Mr. Fumiki Mochida, Mr. Chang Cui and Mr. Mingqian Ding, at Institute of Industrial Science, The University of Tokyo for their contributions to developing the software used in this study.

References

Carmeliet P . Angiogenesis in health and disease. Nat. Med. 2003;9(6):65360. doi: doi: 10.1038/nm0603-653

Feigin I , Allen LB , Lipkin L , Gross SW . The endothelial hyperplasia of the cerebral blood vessels with brain tumors, and its sarcomatous transformation. Cancer. 1958;11(2):26477. doi: doi: 10.1002/1097-0142(195803/04)11:2&lt;264::aid-cncr2820110207&gt;3.0.co;2-d

Wood S Jr . Pathogenesis of metastasis formation observed in vivo in the rabbit ear chamber. AMA Arch. Pathol. 1958;66(4):55068.

Warren BA . The ultrastructure of capillary sprouts induced by melanoma transplants in the golden hamster. J. R. Microsc. Soc. 1966;86(2):17787. doi: doi: 10.1111/j.1365-2818.1966.tb05336.x

Carmeliet P , Jain RK . Principles and mechanisms of vessel normalization for cancer and other angiogenic diseases. Nat. Rev. Drug Discov. 2011;10:41727. doi: doi: 10.1038/nrd3455

Folkman J . Tumor angiogenesis: therapeutic implications. N. Engl. J. Med. 1971;285: 11826. doi: doi: 10.1056/NEJM197111182852108

Senger DR , Galli SJ , Dvorak AM , Perruzzi CA , Harvey VS , Dvorak HF . Tumor cells secrete a vascular permeability factor that promotes accumulation of ascites fluid. Science. 1983;219(4587):9835. doi: doi: 10.1126/science.6823562

Ferrara N . Vascular endothelial growth factor. Arterioscler. Thromb. Vasc. Biol. 2009;29(6):78991. doi: doi: 10.1161/ATVBAHA.108.179663

Folkman J . Role of angiogenesis in tumor growth and metastasis. In: Seminars in oncology. pagg: Elsevier; 2002. p.1518. doi: doi: 10.1053/sonc.2002.37263

10 

Folkman J . Opinion: Angiogenesis: an organizing principle for drug discovery? Nat. Rev. Drug Discov. 2007;6:27386. doi: doi: 10.1038/nrd2115

11 

Jayson GC , Hicklin DJ , Ellis LM . Antiangiogenic therapy—evolving view based on clinical trial results. Nat. Rev. Clin. Oncol. 2012;9(5):297303. doi: doi: 10.1038/nrclinonc.2012.8

12 

Kristensen T , Knutsson M , Wehland M , Laursen B , Grimm D , Warnke E , Magnusson N . Anti-vascular endothelial growth factor therapy in breast cancer. Int. J. Mol. Sci. 2014;15(12):2302441. doi: doi: 10.3390/ijms151223024

13 

Ribatti D , Vacca A . Novel therapeutic approaches targeting vascular endothelial growth factor and its receptors in haematological malignancies. Curr. Cancer Drug Targets. 2005;5(8):57378. doi: doi: 10.2174/156800905774932806

14 

Köppl T , Vodotto E , Wohlmuth B , 3D-1D coupled blood flow and oxygen transport model to generate microvascular networks. Int J Numer Method Biomed Eng. 2020;36:e3386. doi: doi: 10.1002/cnm.3386

15 

Secomb TW , Alberding JP , Hsu R , Dewhirst MW , and Pries AR , Angiogenesis: An adaptive dynamic biological patterning problem. PLoS Comput. Biol., 2013; 9(3):e1002983. doi: doi: 10.1371/journal.pcbi.1002983

16 

Nakayama M , Nakayama A , Van Lessen M , Yamamoto H , Hoffmann S , Drexler HCA , et al . Spatial regulation of VEGF receptor endocytosis in angiogenesis. Nat. Cell Biol. 2013;15(3):24960. doi: doi: 10.1038/ncb2679

17 

Wilhelm K , Happel K , Eelen G , Schoors S , Oellerich MF , Lim R , et al . FOXO1 couples metabolic activity and growth state in the vascular endothelium. Nature. 2016;529(7585):21620. doi: doi: 10.1038/nature16498

18 

Riddell M , Nakayama A , Hikita T , Mirzapourshafiyi F , Kawamura T , Pasha A , et al . aPKC controls endothelial growth by modulating c-Myc via FoxO1 DNA-binding ability. Nat. Commun. 2018;9:5357. doi: doi: 10.1038/s41467-018-07739-0

19 

Sussman M , Smereka P , Osher S . A level set approach for computing solutions to incompressible two-phase flow. J. Comput. Phys. 1994;114(1):14659.

20 

Chien S . Shear dependence of effective cell volume as a determinant of blood viscosity. Science. 1970;168(3934):97779. doi: doi: 10.1126/science.168.3934.977

21 

Wells RE , Merrill EW . Shear rate dependence of the viscosity of whole blood and plasma. Science. 1961;133(3455):7634. doi: doi: 10.1126/science.133.3455.763

22 

Bernabeu MO , Nash RW , Groen D , Carver HB , Hetherington J , Krüger T , et al . Impact of blood rheology on wall shear stress in a model of the middle cerebral artery. Interface Focus. 2013;3(2):20120094. doi: doi: 10.1098/rsfs.2012.0094

23 

Nagaoka T , Yoshida A . Noninvasive evaluation of wall shear stress on retinal microcirculation in humans. Investig. Ophthalmol. Vis. Sci. 2006;47:11139. doi: doi: 10.1167/iovs.05-0218

24 

Windberger U , Bartholovitsch A , Plasenzetti R , Korak KJ , Heinze G . Whole blood viscosity, plasma viscosity and erythrocyte aggregation in nine mammalian species: Reference values and comparison of data. Exp Physiol. 2003;88(3):43140. doi: doi: 10.1113/eph8802496

25 

Kimel S , Svaasand LO , Cao D , Hammer-Wilson MJ , Stuart Nelson J . Vascular response to laser photothermolysis as a function of pulse duration, vessel type, and diameter: Implications for port wine stain laser therapy. Lasers Surg. Med. 2002;300(2):1609.

26 

Brown AS , Leamen L , Cucevic V , Foster FS . Quantitation of hemodynamic function during developmental vascular regression in the mouse eye. Invest. Ophthalmol. Vis. Sci. 2005;46(7):22317. doi: doi: 10.1167/iovs.04-0848

27 

Womersley JR . XXIV. Oscillatory motion of a viscous liquid in a thin-walled elastic tube—I: The linear approximation for long waves. London, Edinburgh, Dublin Philos. Mag. J. Sci. 1955;46(373):199221.

28 

Joseph A , Guevara-Torres A , Schallek J . Imaging single-cell blood flow in the smallest to largest vessels in the living retina. Elife. 2019;8:e45077. doi: doi: 10.7554/eLife.45077

29 

Baudrie V , Laude D , Elghozi JL . Optimal frequency ranges for extracting information on cardiovascular autonomic control from the blood pressure and pulse interval spectrograms in mice. Am. J. Physiol.—Regul. Integr. Comp. Physiol. 2007;292(2):R90412. doi: doi: 10.1152/ajpregu.00488.2006

30 

Riches AC , Sharp JG , Thomas DB , Smith SV . Blood volume determination in the mouse. J. Physiol. 1973;228(2):27984. doi: doi: 10.1113/jphysiol.1973.sp010086

31 

Goldstein D , Handler R , Sirovich L . Modeling a No-Slip Flow Boundary with an External Force Field. J. Comput. Phys. 1993;105(2):35466.

32 

Peskin CS . Flow patterns around heart valves: A numerical method. J. Comput. Phys. 1972;10(2):25271.

33 

Ham FE , Lien FS , Strong AB . A fully conservative second-order finite difference scheme for incompressible flow on nonuniform grids. J. Comput. Phys. 2002;177(1):117133.

34 

Amsden AA , Harlow FH . A simplified MAC technique for incompressible fluid flow calculations. J. Comput. Phys. 1970;6(2):322325.

35 

Spalart PR , Moser RD , Rogers MM . Spectral methods for the Navier-Stokes equations with one infinite and two periodic directions. J. Comput. Phys. 1991;96(2):297324.

36 

Pujade-Lauraine E , Hilpert F , Weber B , Reuss A , Poveda A , Kristensen G , et al . Bevacizumab combined with chemotherapy for platinu-resistant recurrent ovarian cancer: The AURELIA open-label randomized Phase III trial. J. Clinical Onocology. 2014;32(13):13028. doi: doi: 10.1200/JCO.2013.51.4489

37 

Wang J , Mao W , Lock LL , Tang J , Sui M , Sun W , et al . The role of micelle size in tumor accumulation, penetration, and treatment. ACS nano. 2015;9(7):7195206. doi: doi: 10.1021/acsnano.5b02017

38 

Nelson B , Kaliakatsos IK , Abbott JJ . Microrobots for minimally invasive medicine. Annu. Rev. Biomed. Eng. 2010;12:5585. doi: doi: 10.1146/annurev-bioeng-010510-103409

39 

Pries AR , Secomb TW , Gaehtgens P . Biophysical aspects of blood flow in the microvasculature, Cardiovascular Research. 1996;32(4):654667.

40 

Logean E , Schmetterer L , Riva CE , Velocity profile of red blood cells in human retinal vessels using confocal scanning laser doppler velocimetry. Laser Physics, 2003;13(1):4551.

41 

Sherwin SJ , Franke V , Peiro J , Parker K . One-dimensional modelling of a vsscular network in space-time variables. J. Engineering Mathematics. 2003;47:21750.

42 

Formaggia L , Lamponi D , Quarteroni A , One-dimensional models for blood flow in arteries. J. Eng. Math., 2003;47(3–4):251276.

43 

Krogh A , Studies on the physiology of capillaries II. The reaction to local stimuli of the blood-vessles in the skin and web of the frog, The Journal of Physiology, 1921;55(5–6):412422. doi: doi: 10.1113/jphysiol.1921.sp001985

44 

Chen Q , Jiang L , Li C , Hu D , Bu JW , Cai D , et al ., Haemodynamics-driven developmental pruning of brain vasculature in zebrafish, PLoS Biology. 2012;10(8):e1001374. doi: doi: 10.1371/journal.pbio.1001374

45 

Bernabeu MO , Jones ML , Nielsen JH , Krueger T , Nash RW , Groen D , et al . Computer simulations reveal complex distribution of haemodynamic forces in a mouse retina model of angiogenesis. J. R. Soc. Interface. 2014;11(99):2014054.

46 

Ehling M , Asams S , Benedito R , Adams RH , Notch controls retinal blood vessel maturation and quiescence. Development. 2013;140:30513061. doi: doi: 10.1242/dev.093351

47 

Pitulescu ME , Schmit I , Giaimo BD , Antoine T , Berkenfeld F , Ferrante F , et al ., Dll4 and Notch signalling coules sprouting angiogenesis and artery formation, Nat Cell Biol., 2017;19(8):915927. doi: doi: 10.1038/ncb3555

48 

Hikita T , Mirzapourshafiyi F , Barbacena P , Riddel M , Pasha A , Li M , et al ., PAR-3 controls endothelial planar polarity and vascular inflammation under laminar flow EMBO Rep., 2018;19:e45253. doi: doi: 10.15252/embr.201745253

49 

Yamamoto H , Ehling M , Kato K , Kanai K , Lessen M , Frye M , et al ., Integrin b1 controls VE-cadherin localization and blood vessel stability, Nat. Commun., 2015;10(10):6429.

17 Nov 2020

Dear Dr nakayama,

Thank you very much for submitting your manuscript "Numerical evaluation reveals the effect of branching morphology on vessel transport properties during angiogenesis" for consideration at PLOS Computational Biology.

As with all papers reviewed by the journal, your manuscript was reviewed by members of the editorial board and by several independent reviewers. In light of the reviews (below this email), we would like to invite the resubmission of a significantly-revised version that takes into account the reviewers' comments.

We cannot make any decision about publication until we have seen the revised manuscript and your response to the reviewers' comments. Your revised manuscript is also likely to be sent to reviewers for further evaluation.

When you are ready to resubmit, please upload the following:

[1] A letter containing a detailed list of your responses to the review comments and a description of the changes you have made in the manuscript. Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

[2] Two versions of the revised manuscript: one with either highlights or tracked changes denoting where the text has been changed; the other a clean version (uploaded as the manuscript file).

Important additional instructions are given below your reviewer comments.

Please prepare and submit your revised manuscript within 60 days. If you anticipate any delay, please let us know the expected resubmission date by replying to this email. Please note that revised manuscripts received after the 60-day due date may require evaluation and peer review similar to newly submitted manuscripts.

Thank you again for your submission. We hope that our editorial process has been constructive so far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Daniel A Beard

Deputy Editor

PLOS Computational Biology

Daniel Beard

Deputy Editor

PLOS Computational Biology

***********************

Reviewer's Responses to Questions

Comments to the Authors:

Please note here if the review is uploaded as an attachment.

Reviewer #1: This manuscript supplies a simulation of blood flow in a murine retinal vascular network obtained from microscopy images, to determine how vascular branching affects blood flow in angiogenesis. Interestingly, results indicate that hyperbranching in the network leads to reduced flow, and hypobranching leads to increased flow, during angiogenesis. This simulation result is implicitly supported by experimental results indicating increased VEGF expression due to lower oxygenation in the angiogenic front of the hyperbranching vasculature.

Minor issues:

* It would be nice to see a paragraph in the Introduction and possibly also in the Discussion about other studies simulating blood flow (and possibly oxygenation) in the microcirculation -- what exactly has been done before? On line 73-74, the manuscript says "the effects of altered vascular branching on blood flow distribution and the oxygen transport property have not been investigated yet", but there are plenty of modeling studies that attempt to predict blood flow in a branching network. It would be nice to have some more details about how these are being improved upon in this study, at least mentioning that this study may be the first to use this particular realistic mouse retinal network during angiogenesis.

* In the first paragraph of the "Numerical simulation of blood flow" section (line 115), it is explained why a Newtonian assumption is made about the blood flow; however, in the Discussion, starting with line 241, there appears to be an argument for why flows in the microcirculation cannot be represented in this way. It would be good to have a reminder in the Discussion as to why it's acceptable to assume constant viscosity in this case, and/or describe the limitations of this assumption on the model predictions.

* The ordering of section titles is a bit difficult to follow; it seems to go Introduction, Results, Discussion, Materials and Methods -- although part of the Results section seems to include some Methods. I would suggest moving Materials and Methods to after the Introduction, and possibly even include all of the following subsections: "Image Acquisition", "3D reconstruction", and "Numerical simulation", since these all appear to describe Methods. Then have "Simulation results" and "Experimental validation" as subsections of a Results section.

Very minor things:

* Page 6, line 134: "define" should be "defined"

* Page 7, line 136: I think there should be a comma after the word "time-scale"

* Page 7, line 138: "pulsations" should be "pulsation's"

* Page 9, line 202: The line "Here, Rmax is the distance of the farthest blood vessel from the origin" can be removed, since it appears just above, in lines 199-200.

Reviewer #2: Review of "Numerical evaluation reveals the effect of branching morphology on vessel transport properties during angiogenesis" by Mirzapour-shafiyi et al.

In summary - I think this is an interesting study, but I don't think the interpretation/claims are yet supported by the evidence provided. I would be interested to see a revised submission addressing these critiques.

1. interpretation of azimuthal and radial blood flow

Because all tissue vascular beds are different, it took me a little while to work out why u_theta is being used as the relevant velocity for delivery of blood flow to the periphery rather than u_r, and if I am correct I think the paper would benefit from further description. I think it is as follows: the simulated tissue wedge includes both an artery and a vein (the two large radial vessels on either side of the wedge); since flow should go from artery to vein, u_theta as the flow "around the circle" measures the 'direct' route from artery to vein via peripheral tissue. I don't think that's explained in the manuscript. Fig 2D and 3A note the artery/vein pair, but the choice of flow unit and interpretation is not really described, as far as I can tell.

But also, let's unpack that: the retinal structure is alternating A-V-A-V around the circle. The flow in an artery does not go only to one vein; there is another vein on the other side to which (presumably) half the flow would go. I don't think it will substantially change the findings here, given the normalization being done for the analysis, but the A-V wedge as chosen is not really a full flow unit.

Given the difference between ||u|| and u_theta, we are left with a question as to what u_r represents. Might it be a metric of heterogeneity of flow delivery to the front (i.e. a metric of how much local branches have to compensate for unbalanced distribution - though one would have to take care with sign of velocity when integrating), or maybe a more 'bulk' interpretation where if azimuthal blood flow is low, radial flow would be higher to try to supply that region from inner regions? It might depend on whether we're looking at u_r in the peripheral, inner, or boundary regions. Note also that u_r is not graphed - perhaps it is too similar to ||u||? If so, this should be stated.

Also, can you explain why the normalized azimuthal flow rate is substantially different between the two control populations (Fig 3G vs 3H)?

2. Modeling approach

The modeling engine itself is reasonably well explained; but some questions remain about the choice of modeling approach, including: does the 3d analysis improve on 1d flow analysis? A 1d analysis should be able to provide u_r and u_theta values (and Q values) based on the orientation and diameters of the vessel segments, so the comparison should be made given that this manuscript is introducing an alternate approach and claiming that it is superior.

Also, describe the calculation of Q. Looking at the velocity graphs, there is a four-order-of-magnitude variation in velocity graphed, and it looks like many values at the lower end are really stagnant flow tubes. Do these unbalance the calculation? They should not play a role in overall blood flow.

It would also be helpful to be able to see the code shared for the various steps here, especially since a different method is being used than is typical.

3. Interpretation of causality

If the azimuthal flow is reduced, then the blood needs an alternate path to get from artery to vein. It looks like (Fig 2E, F) this is probably happening via large (or larger) diameter shunts in the inner network. To what extent is the lower/higher flow in the periphery a result of different network architecture in the periphery vs. different network architecture in the inner network?

Related to this, the paper claims that hyperbranching in the periphery is leading to the lower flow, leading to VEGF increase (from the abstract: "hyper-branching morphology attenuates effective blood flow at the angiogenic front and promotes tissue hypoxia"); I think it *much more likely* that inner shunts lead to lower peripheral flow, leading to VEGF increase, leading to hyperbranching. It seems much more likely to be this order, and matches with our understanding of other retinal syndromes, e.g. ROP. How to tell the difference between these causal orderings? I can think of at least one possibility: hybrid simulations with the inner network of one mouse type and the peripheral network of a different mouse type (e.g. inner control/periphery Foxo1iDEC, and vice versa). I think you also need to consider the impact of the artery and vein architecture themselves - e.g. narrowing of the artery as it moves peripherally would favor shunting.

4. VEGF intensity validation

While the raw intensity numbers in the region of interest at the angiogenic front seem different among the various cases, I think some normalization may be in order. for example, it looks like intensity overall in the Foxo1iDEC case is higher throughout the retina, and intensity throughout is lower in the PrkciiDEC case. Given that the conclusion is about the angiogenic front, it would seem reasonable to also compare other regions of interest, e.g. a more central region away from the front, to see whether the flow there (which is more similar among the cases) correlates with local VEGF production.

Minor

Caption to Fig 3 notes n>=3; Fig 4 notes n>=4; but table 1 lists all models and it looks like n=3 for each group (same is noted in the methods section). Please confirm and use explicit n values for each group.

Fig 3E, 3F are the same networks as Fig 2E, 2F. It would be interesting to see some of the other retinal networks that were simulated.

Is there a reason only one flow region (wedge) was chosen for each mouse? How was this region chosen, both in terms of its size and location? Multiple regions per mouse would improve confidence in the predictions.

**********

Have all data underlying the figures and results presented in the manuscript been provided?

Large-scale datasets should be made available via a public repository as described in the PLOS Computational Biology data availability policy, and numerical data that underlies graphs or summary statistics should be provided in spreadsheet form as supporting information.

Reviewer #1: None

Reviewer #2: No: Code is not provided; digitized networks not provided; images for most retinas/simulated networks not provided

**********

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Figure Files:

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org.

Data Requirements:

Please note that, as a condition of publication, PLOS' data policy requires that you make available all data used to draw the conclusions outlined in your manuscript. Data must be deposited in an appropriate repository, included within the body of the manuscript, or uploaded as supporting information. This includes all numerical values that were used to generate graphs, histograms etc.. For an example in PLOS Biology see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5.

Reproducibility:

To enhance the reproducibility of your results, PLOS recommends that you deposit laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions, please see http://journals.plos.org/compbiol/s/submission-guidelines#loc-materials-and-methods


8 Feb 2021


28 Feb 2021

Dear Dr nakayama,

Thank you very much for submitting your manuscript "Numerical evaluation reveals the effect of branching morphology on vessel transport properties during angiogenesis" for consideration at PLOS Computational Biology. As with all papers reviewed by the journal, your manuscript was reviewed by members of the editorial board and by several independent reviewers. In light of the reviews (below this email), we would like to invite the resubmission of a revised version that takes into account the reviewers' comments.

The biggest concern that of availability of the model code. I agree with Reviewer 2 that you have not provided a strong rationale for not doing so. Furthermore, he reviewer has some additional questions and concerns that you should  consider addressing.

We cannot make any decision about publication until we have seen the revised manuscript and your response to the reviewers' comments. Your revised manuscript is also likely to be sent to reviewers for further evaluation.

When you are ready to resubmit, please upload the following:

[1] A letter containing a detailed list of your responses to the review comments and a description of the changes you have made in the manuscript. Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

[2] Two versions of the revised manuscript: one with either highlights or tracked changes denoting where the text has been changed; the other a clean version (uploaded as the manuscript file).

Important additional instructions are given below your reviewer comments.

Please prepare and submit your revised manuscript within 60 days. If you anticipate any delay, please let us know the expected resubmission date by replying to this email. Please note that revised manuscripts received after the 60-day due date may require evaluation and peer review similar to newly submitted manuscripts.

Thank you again for your submission. We hope that our editorial process has been constructive so far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Daniel A Beard

Deputy Editor

PLOS Computational Biology

***********************

Reviewer's Responses to Questions

Comments to the Authors:

Please note here if the review is uploaded as an attachment.

Reviewer #1: My concerns have been adequately addressed

Reviewer #2: The authors have done a good job replying to the comments and critiques overall. While there are some additional simulations I'd love to have seen for scientific interest reasons, I don't think that their absence stands in the way of publication.

I will just reiterate one of my comments, and clarify another, both of which I will leave to the judgment of the editor:

1. The simulation code should be shared. While commenting and documentation are best practice, their absence should not stand in the way of sharing the code. This would aid review and facilitate reproducibility. It is absolutely standard in experimental sciences that experimental reagents are shared, and codes and models are essentially our reagents.

2. In the abstract, the following statement still seems not fully supported:

"Interestingly, hyper-branching morphology attenuates effective blood flow at the

angiogenic front and promotes tissue hypoxia."

I accept the authors' argument that in the critique I over-interpreted this as claiming that hyperbranching *in the periphery* was reducing flow - rather than the hyperbranched network overall - and the expanded discussion of the impact of inner & outer network regions helps to clarify this. I still think that this sentence in the abstract claims tissue hypoxia effects that haven't been measured here (though highly likely) - and which of course could result in further branching. Perhaps '...attenuates effective blood flow at the angiogenic front, likely promoting tissue hypoxia'?

**********

Have all data underlying the figures and results presented in the manuscript been provided?

Large-scale datasets should be made available via a public repository as described in the PLOS Computational Biology data availability policy, and numerical data that underlies graphs or summary statistics should be provided in spreadsheet form as supporting information.

Reviewer #1: None

Reviewer #2: No: While the authors state that the data will be provided upon acceptance, the code apparently will not.

**********

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Figure Files:

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org.

Data Requirements:

Please note that, as a condition of publication, PLOS' data policy requires that you make available all data used to draw the conclusions outlined in your manuscript. Data must be deposited in an appropriate repository, included within the body of the manuscript, or uploaded as supporting information. This includes all numerical values that were used to generate graphs, histograms etc.. For an example in PLOS Biology see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5.

Reproducibility:

To enhance the reproducibility of your results, PLOS recommends that you deposit laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions, please see http://journals.plos.org/compbiol/s/submission-guidelines#loc-materials-and-methods


28 Apr 2021


29 Apr 2021

Dear Dr nakayama,

We are pleased to inform you that your manuscript 'Numerical evaluation reveals the effect of branching morphology on vessel transport properties during angiogenesis' has been provisionally accepted for publication in PLOS Computational Biology.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. A member of our team will be in touch with a set of requests.

Please note that your manuscript will not be scheduled for publication until you have made the required changes, so a swift response is appreciated.

IMPORTANT: The editorial review process is now complete. PLOS will only permit corrections to spelling, formatting or significant scientific errors from this point onwards. Requests for major changes, or any which affect the scientific understanding of your work, will cause delays to the publication date of your manuscript.

Should you, your institution's press office or the journal office choose to press release your paper, you will automatically be opted out of early publication. We ask that you notify us now if you or your institution is planning to press release the article. All press must be co-ordinated with PLOS.

Thank you again for supporting Open Access publishing; we are looking forward to publishing your work in PLOS Computational Biology. 

Best regards,

Daniel A Beard

Deputy Editor

PLOS Computational Biology

Daniel Beard

Deputy Editor

PLOS Computational Biology

***********************************************************


11 Jun 2021

PCOMPBIOL-D-20-01829R2

Numerical evaluation reveals the effect of branching morphology on vessel transport properties during angiogenesis

Dear Dr Nakayama,

I am pleased to inform you that your manuscript has been formally accepted for publication in PLOS Computational Biology. Your manuscript is now with our production department and you will be notified of the publication date in due course.

The corresponding author will soon be receiving a typeset proof for review, to ensure errors have not been introduced during production. Please review the PDF proof of your manuscript carefully, as this is the last chance to correct any errors. Please note that major changes, or those which affect the scientific understanding of the work, will likely cause delays to the publication date of your manuscript.

Soon after your final files are uploaded, unless you have opted out, the early version of your manuscript will be published online. The date of the early version will be your article's publication date. The final article will be published to the same URL, and all versions of the paper will be accessible to readers.

Thank you again for supporting PLOS Computational Biology and open-access publishing. We are looking forward to publishing your work!

With kind regards,

Agota Szep

PLOS Computational Biology | Carlyle House, Carlyle Road, Cambridge CB4 3DN | United Kingdom ploscompbiol@plos.org | Phone +44 (0) 1223-442824 | ploscompbiol.org | @PLOSCompBiol

https://creativecommons.org/licenses/by/4.0/This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

https://www.researchpad.co/tools/openurl?pubtype=article&doi=10.1371/journal.pcbi.1008398&title=Numerical evaluation reveals the effect of branching morphology on vessel transport properties during angiogenesis&author=&keyword=&subject=Research Article,Biology and Life Sciences,Anatomy,Body Fluids,Blood,Blood Flow,Medicine and Health Sciences,Anatomy,Body Fluids,Blood,Blood Flow,Biology and Life Sciences,Physiology,Body Fluids,Blood,Blood Flow,Biology and Life Sciences,Anatomy,Ocular System,Ocular Anatomy,Retina,Medicine and Health Sciences,Anatomy,Ocular System,Ocular Anatomy,Retina,Biology and Life Sciences,Anatomy,Cardiovascular Anatomy,Blood Vessels,Medicine and Health Sciences,Anatomy,Cardiovascular Anatomy,Blood Vessels,Physical Sciences,Chemistry,Chemical Elements,Oxygen,Biology and Life Sciences,Anatomy,Body Fluids,Blood,Medicine and Health Sciences,Anatomy,Body Fluids,Blood,Biology and Life Sciences,Physiology,Body Fluids,Blood,Biology and Life Sciences,Anatomy,Cardiovascular Anatomy,Blood Vessels,Arteries,Medicine and Health Sciences,Anatomy,Cardiovascular Anatomy,Blood Vessels,Arteries,Biology and Life Sciences,Anatomy,Cardiovascular Anatomy,Blood Vessels,Veins,Medicine and Health Sciences,Anatomy,Cardiovascular Anatomy,Blood Vessels,Veins,Physical Sciences,Chemistry,Chemical Properties,Viscosity,Physical Sciences,Chemistry,Physical Chemistry,Chemical Properties,Viscosity,Physical Sciences,Materials Science,Materials Physics,Viscosity,Physical Sciences,Physics,Materials Physics,Viscosity,