First we sought to determine the cellular mechanism by which PDGF-BB co-expression prevents VEGF-induced aberrant angiogenesis, i.e., whether by switching vascular growth from splitting to sprouting, or rather modulating the initial morphogenic events of VEGF-induced vascular enlargement and splitting. In order to rigorously control the in vivo dose of VEGF and PDGF-BB, we took advantage of well characterized libraries of monoclonal populations of retrovirally transduced murine myoblasts that we previously generated and that homogeneously express specific levels of VEGF
164 alone (V clones) or together with PDGF-BB at a fixed ratio of 1:3 (VIP clones), or PDGF-BB alone (P clones) [
6,
7,
11]. As every cell in each monoclonal population has the same copy number and genomic integration sites of the retroviral vector, they all secrete the same amount of the factors and therefore the use of such populations enables the control of their dose in the microenvironment around each cell [
6,
7]. Myoblast clones were selected that secrete a high level of VEGF alone (
Vhigh, causing aberrant angiogenesis) or together with PDGF-BB (VIP
high, causing instead its normalization into physiological capillary networks) [
6] and implanted them into limb muscles of mice (tibialis anterior and gastrocnemius). Myoblasts that do not express either VEGF or PDGF-BB were used as control (ctrl). The kinetics of initial vascular morphogenesis was analyzed 2, 3, 4, and 7 days after cell implantation by corrosion casting, the gold standard to identify intussusception [
12], and immunofluorescence confocal microscopy, which effectively detects endothelial sprouting. Control myoblasts did not induce angiogenesis at any time point (Fig.
1a–d, ctrl). As previously observed [
9], after 2 days high levels of VEGF alone did not yet perturb the pre-existing vessels (Fig.
1a,
Vhigh) and enlarged vascular structures appeared by 3 and 4 days (Fig.
1b, c,
Vhigh). Vessel casts of these structures were pierced by numerous small holes (Fig.
1b, c,
Vhigh, and arrowheads in Fig.
1e, g), which are the hallmark of the initial formation of transluminal pillars [
5,
9]. Again consistently with previous observations, by 7 days signs of pillar formation had disappeared from the aberrant bulbous angioma-like structures induced by high VEGF (stars in Fig.
1d,
Vhigh). On the other hand, co-delivery of PDGF-BB caused vascular enlargements already by 2 days, whose casts also showed signs of numerous intraluminal pillars being formed (Fig.
1a, VIP
high and f). By day 3, vascular casts displayed both pillar formation and evidence of longitudinal splitting (Fig.
1b, VIP
high and dashed rectangle in Fig.
1h) into new vascular segments. Surprisingly, with co-delivery of high levels of VEGF and PDGF-BB vascular remodeling was already complete by 4 days, yielding networks of normal capillaries (Fig.
1c, VIP
high), which were similar to and denser than networks visible in control samples (Fig.
1a–d, ctrl), with no further change in morphology by day 7 (Fig.
1d, VIP
high). The occurrence of intussusceptive angiogenesis was quantified by measuring the numerical density of pillars, defined as the total number of pillars per mm
2 of vascular surface area. Since VIP conditions displayed an anticipated and faster kinetics of vascular remodeling, we sought to compare equivalent biological stages of vessel development rather than the same time points. Therefore, we defined two different stages to represent equivalent biological states (Fig.
1e–h), i.e., stage 1 to be the first day of vascular enlargement (corresponding to day 3 for
Vhigh and day 2 for VIP
high conditions) and stage 2 to be the second day of remodeling (corresponding to day 4 for
Vhigh and day 3 for VIP
high). As shown in Fig.
1i, quantifications confirmed the abundant and similar pillar formation during the two different stages in both conditions (day 3
Vhigh = 1345 ± 195 vs. day 2 VIP
high = 1091 ± 131,
p = n.s.; day 4
Vhigh = 868 ± 74 vs. day 3 VIP
high = 843 ± 241,
p = n.s.), suggesting that PDGF-BB co-delivery did not influence the timing and frequency of pillar formation.
On the other hand, corrosion casting only shows the lumen of vascular structures and is not optimal to detect endothelial sprouts, which are devoid of lumen in their migrating tips [
13]. Therefore, we investigated the evidence for sprout formation also by confocal microscopy and 3D reconstruction of thin optical sections after immunostaining for endomucin, which homogeneously stains all endothelial structures, including the filopodia on sprouting tip cells, and laminin, which labels the basal lamina and defines the external boundary of vessels [
9]. Careful analysis of the enlarged vessel walls revealed that the endomucin-positive cells of enlarged vascular structures were completely contained within the respective basal lamina and no endothelial extensions could be seen protruding outside of it, thereby confirming the lack of abluminal sprouting (Fig.
2a–c, d–f). Remarkably, in the enlarged vascular structures it was possible to observe pillar formation both as small holes in the endomucin-positive endothelium (arrows in Fig.
2a, d, high-magnification panels), compatible with those displayed in the corrosion casts, as well as endothelial filopodial extensions (arrowheads in Fig.
2d, high-magnification panel) projecting inside the vascular lumen (marked with stars). The abluminal side of both the holes (inner part) and of the intraluminal filopodia-bearing endothelial cells (external part) showed positive laminin signal, as expected for a forming intraluminal pillar [
14].
Taken together, these data show that, compared to high levels of VEGF alone, PDGF-BB co-delivery: (a) did not promote endothelial sprouting, but rather modulated VEGF-induced splitting angiogenesis; (b) anticipated the initiation of vascular enlargement; (c) did not affect the timing and frequency of transluminal pillar formation; and (d) accelerated the splitting of enlarged vessels and the completion of their remodeling.