Enhancing engineered vascular networks in vitro and in vivo: The effects of IGF1 on vascular development and durability

Abstract

Objectives

Creation of functional, durable vasculature remains an important goal within the field of regenerative medicine. Engineered biological vasculature has the potential to restore or improve human tissue function. We hypothesized that the pleotropic effects of insulin‐like growth factor 1 (IGF1) would enhance the engineering of capillary‐like vasculature.

Materials and methods

The impact of IGF1 upon vasculogenesis was examined in in vitro cultures for a period of up to 40 days and as subcutaneous implants within immunodeficient mice. Co‐cultures of human umbilical vein endothelial cells and human bone marrow‐derived mesenchymal stem cells in collagen‐fibronectin hydrogels were supplemented with either recombinant IGF1 protein or genetically engineered cells to provide sustained IGF1. Morphometric analysis was performed on the vascular networks that formed in four concentrations of IGF1.

Results

IGF1 supplementation significantly enhanced de novo vasculogenesis both in vitro and in vivo. Effects were long‐term as they lasted the duration of the study period, and included network density, vessel length, and diameter. Bifurcation density was not affected. However, the highest concentrations of IGF1 tested were either ineffective or even deleterious. Sustained IGF1 delivery was required in vivo as the inclusion of recombinant IGF1 protein had minimal impact.

Conclusion

IGF1 supplementation can be used to produce neovasculature with significantly enhanced network density and durability. Its use is a promising methodology for engineering de novo vasculature to support regeneration of functional tissue.

1. INTRODUCTION

Vasculature assures oxygen delivery and nutrient supply for all organs in vertebrates. It allows for carbon dioxide and metabolite removal, which is crucial to maintain health. An injured, flow‐limited, or occluded vascular system can lead to organ malfunction and tissue necrosis. According to the WHO, ischemic diseases are among the leading causes of death globally—for example, ischemic heart disease alone killed 8.8 million people in 2015.1 In the advanced stages, ischemic diseases are life threatening despite well‐established therapies. New strategies need to be developed and implemented to address this situation.

In addition, vascularization of engineered tissue designed for eventual clinical implantation for therapeutic purposes remains a major challenge. Angiogenesis, vasculature formed from pre‐existing host vessels, has a maximum ingrowth rate of approximately 1 mm/day, much too slow to be able to support the significant tissue masses often required. Vasculogenesis, the de novo formation of blood vessels, is now being developed to create tissues ex vivo with pre‐formed vascular networks that rapidly anastomose with host vessels and become functional soon after implantation.

Studies using clinically relevant progenitor sources for pericytes and endothelial cells (ECs) show that they can generate functional microvascular networks.2, 3, 4, 5 These networks, however, are typically immature and may rapidly regress. Vascular development and remodelling is dependent on several growth factors, the best known of which are Vascular Endothelial Growth Factor (VEGF) A, and Fibroblast Growth Factor (FGF) 2.6, 7, 8, 9 Improved engineered vasculature would allow ischemic and traumatic tissue defects to be treated more effectively and efficiently.9, 10, 11, 12, 13

The insulin super‐family is composed of 10 members in mammals, insulin, 2 insulin‐like growth factors, 3 relaxin proteins, and 4 insulin‐like peptides. Sequence conservation is minimal; the major commonality is structural, as they all have 3 characteristic disulfide bonds. They also have diverse physiological roles. Among its many functions, insulin‐like growth factor 1 (IGF1) is recognized as an angiogenic factor, along with specific members of the VEGF, FGF, Bone Morphogenetic Protein (BMP), Tissue Growth Factor (TGF), and Hepatocyte Growth Factor (HGF) families.14, 15, 16, 17 However, the normal regulatory role of IGF1 in promoting angiogenesis is not particularly well understood as most research has focused upon its possible pathogenic participation in promoting aberrant vascularization during retinopathy and tumour growth. In addition, many reports are highly contradictory, perhaps reflecting the importance of the precise context of the studies and the pleiotropic nature of the growth factor and its receptor.18, 19

IGF1 is an effective FDA‐approved therapy for children with short stature as it has a beneficial effect upon the entire musculoskeletal system. In addition to its anabolic properties, IGF1 enhances tissue repair by promoting angiogenesis and attenuating inflammation, necrosis, and apoptosis. Systemic IGF1 is largely synthesized by the liver. Its expression is controlled by Growth Hormone (GH) and mediates much of its growth‐promoting activity. IGF1 is a small peptide and is rapidly cleared by the kidneys (t1/2 = 15 minute) if not in a binary complex with a member of the IGF1‐binding protein family (IGFBP) or ternary complex with an additional acid labile subunit.20 However, being complexed appears to most often reduce the bio‐availability of systemic IGF1 as it cannot bind receptor. In addition, ternary complexes are unable to transit capillary walls. Although total IGF1 levels are typically in the 175‐260 ng/mL in young adults,21 only 0.2‐0.6 ng/mL in serum and 3.6‐10 ng/mL in interstitial fluid is free and bioavailable. Because IGF1 binding proteins are in vast excess to their ligand,21 simply increasing total levels of system IGF1 has minimal impact on free levels. An engineered version of human IGF1, LR3IGF‐I, was generated by inserting the first 11 amino acids of methionyl porcine growth hormone followed by the dipeptide VN and mutating the third codon encoding the mature human IGF1 peptide (E3R); this created a protein that can no longer be sequestered by IGFBPs yet can still bind and activate the IGF1 receptor.22 However, like native unbound IGF1, LR3IGF‐I is also rapidly cleared. IGF1 is also synthesized locally by most tissues through activation of an alternative transcriptional promoter. Local synthesis is GH independent, but is modulated by a variety of other soluble factors. Mural cells produce relatively high levels of IGF1 while the expression level in ECs is usually quite low. However, both cell types have high levels of IGF1 receptor and are fully capable of responding to changes in signalling. Clinical trials evaluating IGF1 on various ischaemia and neuropathies have been somewhat disappointing, perhaps because of their reliance on the systemic application regime developed to augment the GH‐hepatic IGF1 axis in the treatment of short stature. IGF1 that is produced and acts locally could provide a sustained, spatially restricted signal could offer an effective means of treating such pathologies as avascular necrosis.

2. MATERIALS AND METHODS

2.1. Study design

The impact of including exogenous IGF1 on the formation of vascular networks in co‐cultures of human umbilical vein endothelial cells (HUVECs) and human bone marrow‐derived mesenchymal stem cells (MSCs) was examined. IGF1 was provided by including HEK‐293 cells genetically engineered to constitutively secrete IGF1. HEK‐293 cells are frequently used to synthesize biologically active, processed recombinant peptides in the pharmaceutical industry. This co‐culture format allowed for continuous, localized IGF1 supplementation to cells within the core of the dense hydrogel plugs, even after implantation. The dosage was more readily controlled by adjusting the fraction of engineered cells included in the co‐cultures than if HUVECs or MSCs had been engineered to directly express IGF1. Parallel experiments including recombinant IGF1 (rIGF1) protein during the formation of the hydrogel were also performed. Measurement and analysis of network morphometry were performed at several time points throughout the in vitro study period. IGF1‐supplemented co‐cultures were also evaluated in vivo by subcutaneous implantation in immunodeficient mice.

2.2. Vascular cell preparations

Bone marrow samples were obtained from patients undergoing hip replacement surgery with the approval of the Institutional Review Board of Massachusetts General Hospital. hMSC cultures were established essentially as previously described.23 20 mL samples were diluted 1:1 in Hank's Buffered Salt Solution and the mononuclear fraction was isolated on a 10 mL ficoll gradient (Ficoll‐Paque, GE Healthcare Life Sciences, Uppsala Sweden). Cells were plated in 25 mL α‐MEM without nucleotides or β‐glycerol phosphate supplemented with 20% foetal calf serum (Sigma, St. Louis, MO) and incubated at 37°C with 5% humidified CO₂. Non‐adherent cells were discarded in 24 hours. Medium was replaced twice weekly. After 2 weeks, cultures were harvested by trypsin, and cryopreserved at a density of 10⁶ cells/mL in complete medium with 5% DMSO. A thawed aliquot of each isolate was used to verify viability, determine colony forming units (cfu), marker expression, and confirm differentiation potential. Standard culture and assay conditions were used to verify that MSC isolations could be induced into adipocyte‐, chondrocyte‐, and osteocyte‐specific lineages.24 The Human MSC Marker Antibody Panel (R&D Systems, Minneapolis MN) was used to characterize the presence of established surface markers by flow cytometry (FACsort, Becton Dickinson, Franklin Lakes NJ). A single human bone marrow MSC isolation was used for all experiments reported in this study. The isolation yielded 32 cfu/100 cells when plated at a density of 10 cells/cm² and cultured for 2 weeks. Approximately 88% of the cell population was positive for the markers CD90, CD105, CD146, and CD166, and negative for CD45. When plated, the cells stained positive for smooth muscle actin (ACTA2) and PDGF receptor beta (PDGFRβ). Cells were expanded in Mesenchymal Stem Cell Growth Medium (MSCGM) (Cambrex) and used at passage 4.

A pooled culture of HUVECs (Cascade Biologics, Portland, OR) was expanded two passages prior to use. The ECs were cultured in complete Endothelial Growth Medium 2 (EGM‐2, PromoCell, Heidelberg, Germany), which contains 0.5 ng/mL VEGF‐A, during expansion and in all experiments. MSCs were labelled with enhanced green fluorescent protein (eGFP) and HUVECs with tdTomato using standard lentivirus protocols.25

2.3. Construction of IGF1‐expressing cells

The GateWay system was used to create a lentivirus plasmid vector for constitutive expression of IGF1. The protein‐encoding region of a rat IGF1 cDNA26 (kind gift of Derek LeRoith) was amplified by PCR using the oligonucleotides CACCATGGGGAAAATCAGCAGT and CTACATTCTGTAGGTCTTGTTTCCT. The resulting DNA fragment was directionally cloned in the pENTR/D TOPO vector (Invitrogen, Carlsbad, CA) to create an IGF1 entry vector. The leading CACC sequence of the forward oligonucleotide functioned as a ribosome assembly site (Kozak sequence). A Clonase recombination reaction was performed using the plasmid pENTR5′(UbCp) containing the constitutively active human Ubiquitin C promoter, the IGF1 Entry plasmid, and the Destination vector pL6/R4R2V5‐DEST (Invitrogen) to generate the plasmid pLB/UbCp(IGF1). The HEK‐293 derived cell line used expresses the SV40 T antigen gene and is a fast growing subclone (HEK‐293FT, Invitrogen) of the original derived from human embryonic kidney cells.27 The T antigen allowed the transfected plasmid to be maintained as a stabile epitope. The cell line was transduced with pLB/UbCp(IGF1) and selected with the drug blasticidin to generate the IGF1‐expressing cell line, HEK‐IGF1.

2.4. Formation of vascular constructs

The standard protocol for vascularization (control group) was based on previous studies.2, 28 Our prior work indicated a ratio of 2:1 of HUVECS and MSCs was optimal for inducing and maintaining vascular structures in a hydrogel.3 Here, preliminary experiments were performed to confirm conditions for vasculogenesis with our cell preparations and materials, and to verify that inclusion of unmodified HEK‐293FT cells would not be deleterious (data not shown). This information was then used to generate the following protocol. All reagents were maintained on ice prior to use. Each mL of collagen‐fibronectin gel contained 385 μL EGM‐2 medium, 25 μL 1 mol L⁻¹ HEPES buffer, 500 μL 3 mg/mL bovine collagen l (PureCol, Advanced BioMatrix, San Diego, CA), 90 μL 1 mg/mL fibronectin (#33016‐015, Invitrogen, Carlsbad, CA). Co‐cultures studies involving the IGF1 expression vector included 0.125 × 10⁶ HEK‐293 cells in the co‐culture cell suspension prior to pelleting. Three concentrations of HEK‐IGF1 cells were evaluated, 1.25 × 10³ cells (low), 12.5 × 10³ cells (medium), 125 × 10³ cells (high). Additional HEK‐293FT cells were added to the control (no HEK‐IGF1 cells), low and medium levels to maintain constant cell numbers across all conditions. Recombinant protein supplemented cultures contained 0 ng (control [standard]), 4 ng (low), 20 ng (medium), or 40 ng (high) rIGF1 (#100‐11, PeproTech, Rocky Hill, NJ) per 0.5 mL collagen‐fibronectin gel.

Cell suspensions containing 0.5 × 10⁶ HUVECs, 0.25 × 10⁶ MSCs, and optionally HEK‐293 cells, were pelleted at 500 g/10 minute. Cell pellets were suspended in 0.5 mL ice‐cold collagen‐fibronectin gel before transferring to a glass‐bottom 24‐well plate and incubating at 37°C in a 5% CO₂ incubator. The gels were then overlaid with EGM‐2 medium. Medium was replaced twice weekly and cultured for up to 40 days. Although the gel system showed progressive contraction, no conditions induced signs of gel degradation. Confocal imaging and quantification was performed at mid‐depth of the gel, the region of highest cell density. Technical replicates were performed in triplicate for each condition, and the in vitro experiment repeated three times. Time points were measured initiating at day 1 and continued through day 40 (Table 1). The measured outcomes of interest were characteristics density, length, diameter, and bifurcations per image field and normalized for reporting in international units of measurement (eg, μm, or mm²).

Table 1.

Vessel parameters

Time point	Setup	Length of vessels (μm)	± SD	Diameter of vessels (μm)	± SD	Bifurcations (per vessel)	± SD	Vessel density (mm2)
Day 1	Standard	150.23	17.27	14.82	4.38	3.00	1.63	37.47
Day 1	Low IGF‐1 concentration	190.97	29.07	21.68	3.54	4.40	0.71	49.95
Day 1	Medium IGF‐1 concentration	183.28	15.15	24.34	6.11	4.00	0.94	46.83
Day 1	High IGF‐1 concentration	143.77	20.31	20.90	6.32	3.60	1.78	49.95
Day 2	Standard	188.01	13.61	17.91	6.32	2.20	1.03	40.59
Day 2	Low IGF‐1 concentration	196.22	20.12	21.81	3.13	3.50	1.27	53.08
Day 2	Medium IGF‐1 concentration	215.37	52.27	23.24	4.22	3.80	1.23	65.56
Day 2	High IGF‐1 concentration	139.49	14.25	20.26	3.88	2.90	1.10	46.83
Day 3	Standard	223.57	76.20	21.19	6.01	2.70	0.95	53.08
Day 3	Low IGF‐1 concentration	264.36	40.25	23.26	2.96	3.90	1.66	93.66
Day 3	Medium IGF‐1 concentration	188.90	43.30	16.41	2.46	2.50	1.43	124.88
Day 3	High IGF‐1 concentration	225.93	82.05	20.92	1.62	5.60	2.76	84.30
Day 4	Standard	185.70	62.01	22.06	7.12	4.90	1.52	93.66
Day 4	Low IGF‐1 concentration	239.28	57.87	23.38	3.51	6.10	1.20	109.27
Day 4	Medium IGF‐1 concentration	272.63	61.95	27.40	4.79	5.30	0.82	171.72
Day 4	High IGF‐1 concentration	155.52	39.92	13.80	7.84	4.10	1.52	112.40
Day 5	Standard	152.48	32.79	14.21	4.14	3.10	0.86	37.47
Day 5	Low IGF‐1 concentration	240.76	124.51	21.79	5.23	3.10	1.33	40.59
Day 5	Medium IGF‐1 concentration	209.22	52.50	24.12	3.46	4.40	1.88	46.67
Day 5	High IGF‐1 concentration	178.64	55.29	22.80	4.78	4.50	1.98	61.03
After 1 wk = d8	Standard	175.68	61.46	19.24	3.46	3.15	1.88	57.98
After 1 wk = d8	Low IGF‐1 concentration	195.56	87.71	16.69	4.14	1.91	1.08	61.03
After 1 wk = d8	Medium IGF‐1 concentration	240.45	89.26	22.43	4.80	3.21	1.22	72.24
After 1 wk = d8	High IGF‐1 concentration	178.64	55.29	22.80	4.78	4.50	1.98	0.00
After 2.5 wks = d19	Standard	228.19	45.39	18.50	6.82	2.90	0.99	101.14
After 2.5 wks = d19	Low IGF‐1 concentration	254.49	42.74	31.21	6.94	4.90	1.45	98.45
After 2.5 wks = d19	Medium IGF‐1 concentration	256.24	29.41	25.39	6.02	5.10	1.52	101.14
After 2.5 wks = d19	High IGF‐1 concentration	133.48	82.45	28.39	10.74	1.10	0.74	91.89
After 3 wks = d22	Standard	154.00	55.51	14.59	3.99	1.90	0.57	42.95
After 3 wks = d22	Low IGF‐1 concentration	314.22	132.60	20.56	3.38	3.30	1.25	62.58
After 3 wks = d22	Medium IGF‐1 concentration	325.56	71.02	23.33	3.11	3.40	1.17	74.18
After 3 wks = d22	High IGF‐1 concentration	216.59	171.74	17.09	9.45	0.60	0.48	19.56
After 4 wks = d29	Standard	155.65	94.95	17.12	9.30	0.80	0.70	12.21
After 4 wks = d29	Low IGF‐1 concentration	287.51	107.81	24.84	5.06	2.80	1.48	52.89
After 4 wks = d29	Medium IGF‐1 concentration	220.08	60.12	22.40	4.30	4.60	1.78	59.11
After 4 wks = d29	High IGF‐1 concentration	110.15	54.61	14.08	2.73	1.40	0.70	30.52
After 5 wks = d36	Standard	32.65	13.84	12.76	5.41	0.00	0.00	6.10
After 5 wks = d36	Low IGF‐1 concentration	262.62	132.56	15.07	3.71	0.80	0.63	45.77
After 5 wks = d36	Medium IGF‐1 concentration	278.81	168.41	16.05	5.87	1.20	0.79	27.46
After 5 wks = d36	High IGF‐1 concentration	68.23	28.85	20.41	8.73	0.10	0.32	6.10
After 6 wks = d40	Standard	102.86	75.87	22.26	12.10	0.63	0.71	24.41
After 6 wks = d40	Low IGF‐1 concentration	268.75	71.28	38.19	16.61	3.90	1.79	36.62
After 6 wks = d40	Medium IGF‐1 concentration	231.14	135.50	30.51	18.38	1.50	0.71	27.46
After 6 wks = d40	High IGF‐1 concentration	124.23	88.74	30.91	16.38	0.00	0.00	15.26

2.5. Quantification of IGF1 release in vitro

IGF1 synthesis levels by HEK‐IGF1 cells was measured using the Mouse/Rat IGF‐I Quantikine ELISA Kit (R&D Systems, Minneapolis, MN), following manufacturer's protocol. A standard curve was generated using the included murine rIGF1 protein. Standard co‐cultures that included HEK‐293 or low, medium, or high levels of HEK‐IGF1 cells were formed in triplicate using the collagen‐fibronectin gel protocol. After 72 hours, the hydrogels were gently homogenized by stirring with a pipette tip, transferred to a 2 mL microfuge tube, and centrifuged at 16 000 g/2 minute. The supernatant was collected and diluted 10‐ and 100‐fold prior to measuring IGF1 concentrations by ELISA.

2.6. Immunodeficient mouse model

The subcutaneous injection procedure for producing vascularized hydrogel plugs in immunodeficient mice was used to evaluate the effect of IGF1 in vivo,28 with ice‐cold collagen‐fibronectin plugs containing co‐cultures that gelled when warmed to body temperature.29 Providing growth factor by HEK‐IGF1 and as rIGF1 were both examined. All animal procedures were approved by the Institutional Animal Care and Use Committee of the Massachusetts General Hospital and performed according to the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. Control, low, medium, and high concentrations of HEK‐IGF1 and rIGF1 were studied; cell rations and recombinant protein concentrations were identical to those used in the in vitro studies. At least three technical replicates of each of the four conditions for both HEK‐IGF1 and rIGF1 were implanted. The cell compositions of the HEK‐IGF1 were identical to those used for the in vitro studies.

Ten severe immunodeficient mice (NOD.CB17, Jackson Laboratory, Bar Harbor, ME) were implanted, four samples/mouse. Animals were anesthetized with a Ketamine/Xylazine solution (c = 1 mg/30 g mouse) for pain relief for the implant injections. Collagen‐fibronectin gels were created on ice as described above prior to subcutaneous dorsal injection using a 1 mL syringe fitted with a 26 gauge needle. Each mouse received one plug of each of the four conditions; the positions were rotated in each mouse to different quadrants (crossover design). At 2 weeks, the plugs were explanted from sacrificed mice and analyzed via confocal microscopy imaging and histology.

2.7. Imaging measurements using light microscopy

A Nikon A1 confocal microscope for detection and measurements of the vascularized collagen plugs with a 20× objective was used for imaging vascular networks. Representative regions mid‐depth for each sample were imaged. Analysis of images was performed in ImageJ using the Vascular Network Toolkit Plug‐in.30 The red (HUVEC) channel of each image was isolated and the thresholding function used to generate a 2‐bit skeletal vascular map from which the network values were quantified. Node analysis was performed, and distance measurements between nodes and density per unit area obtained. Thickness function was used to determine average vessel radius; this value was doubled and data presented as diameter. Vessel density was determined by determining the fractional area incorporated into vessels in the skeletonized images.

2.8. Histological preparation and analysis

For histological analysis, the collagen‐fibronectin plugs were excised from mice and fixed in 4% paraformaldehyde solution overnight at 4°C. Samples were dehydrated and embedded in paraffin by standard conditions. Transverse sections 10 μm thick were cut and processed for immunohistochemistry using primary antibody mouse anti‐human CD31 (1:25; #M0823, DAKO, Carpinteria, CA), mouse anti‐smooth muscle actin (1:50; DAKO), and anti‐vimentin (1:50; DAKO), detected with the EnVision+ kit (#K4006, DAKO) using 3,3′‐diaminobenzidine, and counter‐stained with hematoxylin and eosin. Sections being probed with the anti‐CD31 antibody were first processed for antigen retrieval by heating at 97°C for 30 minutes in 10 mM citrate, pH 6.0.

2.9. Statistical analysis

To compare four setups at several time points, a non‐parametric repeated measurement ANOVA was used for group comparison.31 The Mann–Whitney‐U Test was used for comparison of two independent setups per time point.31 Group effect, time effect, and the interactions between group and time were investigated. A P < .05 was considered to be significant. Due to the exploratory study design and the weakness of post hoc power analysis, all calculated P values were considered in a non‐confirmatory way. Missing data was accounted for using the last observation carried forward (LOCF) method. All numerical calculations were performed with IBM SPSS Statistics, Ver. 20, and the software package R Project for Statistical Computing, Ver. 3.0.2.

3. RESULTS

3.1. HEK‐IGF1 significantly enhanced in vitro vascular network formation

In order to evaluate the impact of IGF1 on vasculogenesis, HEK‐293 cells were genetically engineered to constitutively express IGF1. The engineered cells were included with co‐cultures of HUVECs and MSCs during the formation of collagen‐fibronectin hydrogels. Three HEK‐IGF1 concentrations were compared to control co‐cultures containing non‐engineered HEK‐293 cells for up to 40 days in vitro. Although many cultures remained viable, compaction of the hydrogel inhibited further analysis. Average network density, vessel length, and vessel diameter were determined periodically throughout the culture period for each of the experimental conditions.

Control co‐cultures with non‐IGF1 expressing HEK‐293 cells largely maintained good viability to almost 3 weeks, after which they rapidly declined (Figure 1). This was in marked contrast to HUVEC cultures without MSCs, which form transient networks that rapidly fell apart in hours (32 and data not shown). Supplementation by HEK‐IGF1 cells dramatically impacted vascular network formation throughout the experimental period as measured by several metrics, with visible differences readily detectable by the third day. Network density was significantly greater for co‐cultures containing low or medium levels of HEK‐IGF1 cells (Figure 2). In contrast, high levels of HEK‐IGF1 produced inferior networks that were largely indistinguishable or worse than control cultures. These differences were maintained throughout the entire experimental time course.

Figure 1.

HEK‐IGF1 promotes durable vascular network formation in vitro. Co‐cultures of HUVECs and MSCs seeded with varying numbers of HEK‐IGF1 cells were incubated for 40 d. Genetic labelling by lentivirus was used to express tdTomato in HUVECs and eGFP in MSCs. HEK‐IGF1 cells were not fluorescently labelled. A, Co‐cultures at 22 d. The networks of low (2) and medium (3) levels of HEK‐IGF1 were denser than those of control cultures (1) and high levels (4). B, Co‐cultures at 40 d. The networks of control cultures (1) and high levels (4) had largely degraded in contrast to those of low (2) and medium (3) levels. Scale bar = 100 microns

Figure 2.

HEK‐IGF1 promotes denser vascular networks in vitro. Co‐cultures of HUVECs and MSCs seeded with varying numbers of HEK‐IGF1 cells were incubated for 40 d. Cultures were imaged at the indicated times and average vessel density was calculated and plotted

Co‐cultures with low or medium levels of HEK‐IGF1 had positive effects upon vessel length relative to controls throughout the experiment (Figure 3A). High levels of HEK‐IGF1 produced networks that were initially similar to both control and medium levels. After day 3, networks became dramatically worse than all other conditions. Vessel lengths were consistently shorter at the highest HEK‐IGF1 levels for the remainder of the study period.

Figure 3.

HEK‐IGF1 promotes vasculature with longer length and greater diameter in vitro. Co‐cultures of HUVECs and MSCs seeded with varying numbers of HEK‐IGF1 cells were incubated for 40 d. Cultures were imaged at the indicated times. Average length (A) and diameter of the vessels (B) were calculated. Box and whisker plots below show the statistical distribution of data for selected subsets of time points. P values for this and subsequent figures: NS = not significant, *<.05, **<.01, ***<.001

Control co‐cultures displayed the greatest average vessel diameter at day 3, after which it gradually declined (Figure 3B). All 3 HEK‐IGF1 concentrations displayed significantly increased average vessel diameter relative to control conditions on day 1. Medium HEK‐IGF1 levels consistently caused a significant and reproducible dip in both vessel length and diameter at day 3, after which the cultures rapidly recovered. Unlike average vessel length, the initial increase in average vessel diameter for HEK‐IGF1 co‐cultures was not maintained throughout the study period by any of the tested conditions. Maximum average diameter peaked on day 19 for all cultures that included HEK‐IGF1, a few days before the period displaying maximum length. The apparent large diameter of the high‐level HEK‐IGF1 co‐cultures at the final time point was the result of the few surviving HUVECs forming clumps or islands rather than tubes. The number of vessel bifurcations occurring under each condition was also determined, but differences were not statistically significant (data not shown).

3.2. Quantification of IGF1 release by HEK‐IGF1

We wanted to determine if medium supplemented with rIGF1 protein could substitute for that produced by HEK‐IGF1. Synthesis levels were assessed using ELISA to measure IGF1 in the medium of co‐cultures containing HEK‐293, or low, medium, or high levels of HEK‐IGF1. Measured IGF1 levels were proportional to a number of HEK‐IGF1 cells in the culture, and corresponded to ca. 1 ng/10³ HEK‐IGF1 cells/day. Negligible levels of IGF1 were detected from HEK‐293 cells in these conditions.

3.3. rIGF1 protein stabilizes in vitro vascular networks

Co‐cultures were prepared with rIGF1 protein replacing HEK‐IGF1. Again, three levels of IGF1 supplementation were compared to unsupplemented control cultures. For low and medium cultures, rIGF1 levels were chosen to match those measured in the corresponding HEK‐IGF1 setups. Because the highest HEK‐IGF1 levels were clearly toxic, a lower concentration of recombinant protein was used in the high‐level cultures. Co‐cultures were followed until compaction of the hydrogel made imaging difficult (36 days). Supplementation with rIGF1 again dramatically impacted vasculogenesis, with visible differences readily detectable throughout the culture period (Figure 4). Although the resulting networks were similar to those prior, rIGF1 supplementation did not precisely mirror that with HEK‐IGF1. For high rIGF1 co‐cultures, the reduced growth factor level was still deleterious, and like unsupplemented co‐cultures, had few surviving ECs. Both low and medium rIGF1 levels promoted network survival to a similar degree. However, the resulting networks were notably less dense with shorter branches (Figure 5A). In addition, vessel diameters were greatly reduced (Figure 5B).

Figure 4.

rIGF1 protein promotes durable vascular networks in vitro. Co‐cultures of HUVECs and MSCs seeded with varying concentrations of rIGF1 protein were imaged at 36 d. Genetic labelling by lentivirus was used to express tdTomato in HUVECs and eGFP in MSCs. The networks of control cultures (1) and high concentration (4) had largely degraded in contrast to those of low (2) and medium (3) concentration. Scale bar = 50 μm

Figure 5.

rIGF1 promotes vasculature with longer length and greater diameter in vitro. Co‐cultures of HUVECs and MSCs seeded with varying concentrations of rIGF1 protein were incubated for 36 d. A, Cultures were imaged at the indicated times and average vessel length was calculated. B, Box and whisker plots show the statistical distribution of data for a selected subset of time points. NS = not significant

3.4. Assessment IGF1 treatment on in vivo vascular networks

To assess potential in vivo functionality of IGF1‐enhanced vascular networks, co‐cultures containing collagen‐fibronectin plugs were implanted subcutaneously in immunodeficient mice. Both supplementations by HEK‐IGF1 and rIGF1 were evaluated. After 14 days the implants were removed for analysis. They were initially imaged by confocal microscopy to assess network survival and morphometry. Vascular networks were readily detectable in control samples for both HEK‐IGF1 and rIGF1 containing plugs (Figure 6A, B). Few surviving HUVECs were found in high HEK‐IGF1 samples; remaining MSCs were numerous but most appeared pyknotic. Networks in both the low and medium HEK‐IGF1 samples were relatively extensive compared to those in controls. Both the average length and diameter of the low HEK‐IGF1 samples were larger than for all other conditions studied (Figure 7A). Although the low HEK‐IGF1 produced vessels that were uniform in appearance, they still maintained a degree of tortuous (immature) character. In contrast, vessels in medium HEK‐IGF1 samples appeared noticeably more linear. Explants containing rIGF1 supplementation produced networks at all levels. No toxicity was seen at high levels of rIGF1. However, supplementation with rIGF1 caused a small but significant increase in the low dosage only; the highest levels were somewhat detrimental. Average vessel diameters were unchanged at all dosages (Figure 7B). Paraffin sections from embedded samples were probed with a human‐specific antibody against the endothelial marker CD31. Numerous HUVEC‐derived capillaries were present in both HEK‐IGF1 and rIGF1 supplemented implants. The vessels were filled with blood cells, providing evidence of the patency of the vascular lumen and anastomosis of the network (Figure 6C).

Figure 6.

IGF1 supplementation promotes extensive vascular networks in vivo. Collagen‐fibrinogen plugs with co‐cultures of HUVECs and MSCs were implanted subcutaneously in immunodeficient mice. Plugs were explanted at 14 d and imaged by confocal microscopy. A, Varying numbers of HEK‐IGF1 cells were included in the co‐cultures: control (1), low (2), medium (3), and high (4). B, Varying amounts of recombinant IGF1 protein were included with the co‐cultures during hydrogel formation: control (1), low (2), medium (3), and high (4). C, A paraffin section from a medium HEK‐IGF supplemented explanted sample was probed with a human‐specific anti‐CD31 antibody to identify implant derived vessels (enlargement on right). Vascular networks were filled with blood cells, indicating that they anastomosed and are patent and functional. Scale bar for (A, B) = 200 μm

Figure 7.

HEK‐IGF1 but not rIGF1 supplementation promotes extensive vascular networks in vivo. Average vessel lengths and diameters were calculated from images of the explanted plugs and plotted for HEK‐IGF1 (A) and rIGF1 (B). NS = not significant

4. DISCUSSION

During angiogenesis and vasculogenesis, interactions between pericytes and ECs promote vessel formation by altering the microenvironment. Both exchange of soluble factors and physical contact are important. In vasculogenesis studies, MSCs are routinely used to serve as mural cells. Indeed, there is evidence that MSCs derive from a subpopulation of mural cells.16, 24 They produce trophic factors that stimulate angiogenesis and subsequent contact significantly stabilizes the resulting networks.2, 3, 32 A number of signalling molecules, metalloproteases, and ECM components have been shown to be critical players in the complex process. In spite of the coordinated involvement of several growth factors during growth of normal, functional vasculature, well‐timed therapy with a single factor such as VEGF‐A or PDGF‐BB can also result in mature vessels.33, 34 This may be because most angiogenic growth factors can induce expression of many of the other factors by the target cells. Although IGF1 can upregulate expression of several growth factors including VEGF‐A and PDGF‐BB, evidence is lacking that any angiogenic factors upregulate IGF1 expression.35 This may place IGF1 in an apex position in hierarchy of angiogenic signalling. Therapeutic rationale for providing angiogenic factors include limiting endogenous concentrations of a growth factor, augmentation of the response by providing supra‐physiological amounts, and overcoming inhibitory conditions.36

IGF1 is an attractive candidate as a potential therapeutic factor to promote angiogenesis. Although its angiogenic properties were first recognized three decades ago,37, 38 even very basic studies involving this factor have been quite limited due to fears of retinopathy and coronary artery disease in clinical use.39, 40 However, extensive and prolonged use of IGF1 to treat short stature and numerous recent clinical studies have not born out such fears. Nevertheless, the failure of many of these clinical trials has clearly indicated that systemic application of the growth factor is often not efficacious for treating localized ailments. IGF1 bio‐availability is tightly regulated by both its binding proteins and turnover; specific local applications may need to be developed in order to take advantage of its properties.

IGF1 can potentially mediate angiogenesis through several mechanisms, as both ECs and mural cells express significant levels of IGF1 receptor.41, 42 IGF1 robustly promotes proliferation of vascular smooth muscle cells.43 Reports of the effect upon EC proliferation vary, and may depend upon the vascular bed being studied.14, 15, 35, 43 IGF1 stimulates migration of both smooth muscle and ECs.37, 38 It promotes EC migration through phosphatidylinositol 3‐kinase (PI3K) activation.44 Importantly, at moderate levels, it is also a major survival factor for a broad range of cell types, and is able to prevent cell death and apoptosis during normal development, stress, repair, and disease.45 The role in modulating inflammatory response is more nuanced, as many studies document inhibition by IGF1 signalling46 while separate studies have demonstrated non‐responsiveness47 or even enhancement48. This is an important consideration as inflammation drives the senescent state where ECs are growth arrested and unresponsive to extracellular signalling. Again, IGF1 inhibits oxidative‐stress mediated senescence, but has been implicated in promoting radiation‐induced senescence.49, 50 In addition, macrophage act as essential cellular chaperones in mediating anastomosis.51

IGF1, like VEGF, also upregulates endothelial nitric oxide synthase (eNOS) activity through PI3K.52, 53 IGF1 synergistically augments the VEGF‐induced angiogenic response through the MAPK pathway.39 IGF1 is a potent anabolic factor, and promotes mitochondrial function.54 IGF1 also directly modulates inflammation response of both macrophage and ECs. Physiological response to the development of tissue ischemia includes the upregulation of angiogenic growth factor activity;55 this may include activation of latent factors sequestered in the ECM as well as de novo synthesis. In in vitro monocultures of ECs, the vascular networks that form rapidly regress. In part, this is due to lysophosphatidic acid (LPA) in the serum that induces destabilization.56 Inclusion of IGF1 prolongs Erk activity, a signalling pathway recently demonstrated to stabilize nascent vessels. LPA may be important for initial destabilization or remodelling of existing vessels in the early stages of angiogenic sprouting, but continued signalling is deleterious. IGF1‐promoted Erk signalling overrides this, at least in the short‐term (hours). Such destabilization does not occur in co‐cultures that include stromal cells such as pericytes or MSCs, as they accelerate LPA metabolism.57

In addition, MSCs secrete IGF1,35 which may also contribute to the vessel stabilization seen in co‐cultures. However, our study demonstrated that augmentation with exogenous IGF1 clearly provided significant additional long‐term vessel stabilization. In our experimental vasculogenesis models, we supplemented the co‐cultures with genetically engineered cells that produce IGF1 or provided recombinant protein in the medium. In normal vasculature, IGF1 is significantly expressed only by mural cells; the low levels of protein detected in ECs is thought to have been sequestered from serum.19, 58, 59 Both cell types, however, are fully capable of responding to IGF1 signalling as they both abundantly express the receptor.41, 42 Furthermore, some of the effects documented in this study may secondary, as IGF1 stimulation can induce synthesis of additional angiogenic growth factors and/or their receptors including VEGF, HGF, and FGF. It is unlikely that IGF1 supplementation functioned directly as a chemoattractant in our studies as it was homogeneously distributed in the cultures. It could, however, have induced secretion of a chemoattractant by a responding cell type that would then present in the requisite gradient.

Both methods of providing IGF1 proved beneficial in vitro, although only HEK‐IGF1 provided a similar effect in vivo. HEK‐IGF1‐containing cultures were also noticeably more robust by most in vitro metrics. The exception was network survival as both methods of IGF1 supplementation caused significant enhanced network persistence. While networks in standard conditions declined precipitously after 19 days in culture, those with either HEK‐IGF1 or rIGF1 were still stabile and viable at 36 days. Inclusion of the growth factor also resulted in denser networks, with those containing HEK‐IGF1 being greater than rIGF1. The average length of the vessels was longer, although the number of bifurcations was similar. Interestingly, the diameters of the resulting vessels of HEK‐IGF1‐containing cultures were noticeably larger as well, often several cells wide. This suggests a possible role for IGF1 in arteriogenesis (enlargement of existing vessels), an important but still relatively poorly understood process.14, 15, 60, 61 Although the mechanism for the apparent toxicity resulting from the highest levels on both MSCs and ECs was not examined in our study, similar concentrations of IGF1 induce apoptosis in other cell types through internalization of the IGF1 receptor and down‐regulation of its associated signalling.62, 63, 64

These results suggest many additional follow up studies to further understand the roles and mechanisms of IGF1 during angiogenesis. From the aspect of therapeutic development of cell therapy, potential benefits from simple pre‐exposure to the growth factor prior to network formation and implantation should be determined. A detailed study of anastomosed vasculature should examine long‐term in vivo persistence and function. Dissecting some of the specific responses to IGF1 by MSCs and ECs could be done in monocultures; others will require co‐cultures expressing mutant or dominant negative IGF1 receptor. In particular, the determination of whether the size of functional vessels can be modulated with IGF1 will be of significant interest.

ACKNOWLEDGEMENTS

This work was sponsored by a grant to CMN from the Stanley H. Durwood Foundation and the Harvard Stem Cell Institute. CCF received a research scholarship from the Charité – Universitätsmedizin Berlin and a scholarship from the Biomedical Sciences Exchange Program of the International Academy of Life Sciences.

CONFLICT OF INTEREST

The authors hereby disclose no conflicts of interest.

Friedrich CC, Lin Y, Krannich A, Wu Y, Vacanti JP, Neville CM. Enhancing engineered vascular networks in vitro and in vivo: The effects of IGF1 on vascular development and durability. Cell Prolif. 2018;51:e12387 10.1111/cpr.12387