Native human collagen type I provides a viable physiologically relevant alternative to xenogeneic sources for tissue engineering applications: A comparative in vitro and in vivo study

2.1 |. Overall experimental approach

As our goal was to compare human collagen-based hydrogels versus animal-derived collagen-based hydrogels, generated from commercially available products, in a broad sense for tissue engineering and regenerative medicine applications, we chose several distinct scenarios in which to test these hydrogel biomaterials. First, we performed a comprehensive set of material characterization studies to compare the material and mechanical properties of human versus animal collagen hydrogels. Next, we employed hMSCs to evaluate collagen hydrogel impact on cell proliferation and viability. Subsequently, we employed two very different application areas to assess the utility of human-based collagen hydrogels versus animal-derived collagen hydrogels. First, to assess these matrices’ utility in a purely in vitro application, 3D tumor constructs–in this case colorectal cancer cells encapsulated in volumes of the hydrogels–were employed in a drug screening study. Second, in a more classical tissue engineering application, these matrices’ capability to support self-assembly of human microvascular networks in vitro and formation of perfused human vessels in vivo was investigated. These different application areas, paired with MSC studies and material characterizations were designed to challenge each hydrogel system in a variety of distinct ways.

2.2 |. Materials

Bovine collagen type I derived from hides (Purecol^®), collagen type I derived from rat tail (RatCol^®), and human collagen type I derived from neo-natal fibroblasts (Vitrocol^®) were purchased from Advanced BioMatrix (San Diego, CA). Human skin-derived collagen type I (HumaDerm) was obtained from Humabiologics Inc (Phoenix, AZ). Human Placenta collagen type I was purchased from Corning (Corning, NY). The source, extraction method and concentration of the different types of commercial collagen are shown in Table 1. Other reagents were purchased from Fisher Scientific (Watham, CA) unless stated otherwise.

2.3 |. Cell culture

2.4 |. Preparation of collagen type I and tissue-derived ECM hydrogels for mechanical characterization in vitro

The initial concentration for different sources of collagen type I was set to 3.1 mg/ml by dissolving (lyophilized solid) or diluting (solubilized liquid) in appropriate acid (i.e., HCl or acetic acid). Collagen hydrogels were synthesized by following the manufacturer’s recommendations. Briefly, bovine, human fibroblasts-derived and human skin-derived collagen type I hydrogels were prepared by mixing eight parts of acid-soluble collagen with one part of 10× PBS and one part of 0.1 N NaOH (pH ~ 7.4), whereas rat tail collagen type I hydrogels were prepared by mixing nine parts of acid-soluble collagen with one part of neutralization solution (pH ~ 7.4), and human placenta collagen hydrogels were prepared by mixing nine parts of acid-soluble collagen with one part of 10× PBS (pH ~ 7.4). Following this step, the neutralized collagen solution was incubated at 37°C for 45 min to induce gelation.

2.5 |. Gelation kinetics

A gelation kinetics assay was performed to determine the fibrillogenesis rate of different sources of collagen type I. Briefly, 100 μl of chilled neutralized collagen solution was casted into individual wells of a 96-well plate at 4°C (N = 4/group) and inserted into M2e Spectramax plate reader (Molecular Devices) preheated at 37°C. Gelation kinetics was determined by measuring the absorbance at 312 nm every 2 min for 120 min. Increase in absorbance value corresponds to increase in the turbidity of the solution during the formation of the hydrogel. All the values were normalized between 0 and 1 by employing a ‘min-max’ normalization technique for each sample of collagen species by using the following equation:

where X–set of absorbance values of x, X’–normalized value of X, min (X)–minimum value in X, max (X)–maximum value in X.

Polymerization rate was calculated by taking the slope of the linear region of the gelation kinetics curve. Polymerization half time was defined as the time needed to reach 50% of the maximum absorbance value.

2.6 |. Swelling and degradation studies

A swelling study was performed to evaluate the water absorption capability of hydrogels prepared using different sources of collagen by following a previously published protocol.^15,16 At first, collagen hydrogels were vacuum dried for 24 h and weighed to measure the initial dry weight $\left(W_d\right)$ of the hydrogel (N = 4/group). Following this step, the hydrogels were incubated in 500 μl of 1× PBS at room temperature for 24 h and weighed again to determine the wet weight $\left({\mathrm{W}}_{\mathrm{w}}\right)$ of the hydrogels. Swelling ratio was calculated by using the following equation:

The stability of collagen hydrogels (N = 4/group) was assessed by performing an in vitro collagenase degradation assay.¹⁷ Hydrogels were prepared using different sources of collagen and kept in ultrapure water for 30 min. The hydrogels were gently blotted using a Kimwipe to remove the excess water and weighed to determine the initial weight $\left({\mathrm{W}}_{\circ }\right)$ of the hydrogel. Each hydrogel was then incubated in individual microcentrifuge tubes containing 500 μl of collagenase solution (5 U/ml collagenase in 0.1 M Tris–HCl buffer and 5 mM CaCl₂; pH 7.4) at 37°C. At periodic intervals, each hydrogel was removed from the collagenase solution, gently blotted on a Kimwipe, and weighed ${\mathrm{W}}_{\mathrm{t}}$ until the hydrogel was fully degraded. The percent residual mass of the hydrogel was calculated using the following equation:

2.7 |. Compression testing

The compressive modulus of different hydrogels (N = 5/group) was assessed using an MT G2 MicroTester (CellScale, Canada). Briefly, hydrogels (diameter 7.5 mm, thickness 2.5 mm) were placed on an acrylic platform in wet condition within the testing chamber and compressed using a tungsten cantilever beam (0.4 mm diameter) equipped with a stainless-steel platen at a rate of 10 μm/s until a displacement of 20% of the hydrogel thickness. Compressive strain was computed by taking the ratio of the platen displacement to the original thickness of the hydrogel. The compression force was calculated using Euler-Bernoulli beam theory.⁶ Compressive stress was determined by normalizing the compressive force with the area of the platen. Since the diameter of the hydrogels (7.5 mm) was greater than the area of the platen (3 mm × mm), the average force calculated from beam theory will be overestimated due to nonuniform stress distribution on the hydrogel surrounding the platen. To correct this overestimation, an in-house computer simulation was performed by modulating the hydrogel diameter for a fixed platen geometry. An indentation depth dependent overestimation correction factor of 2.10 (for 10% strain) was applied to obtain the final compressive stress data. Stress–strain curves were generated, and compressive modulus of each hydrogel was determined by calculating the slope of the 0%–10% region of the stress–strain curve.

2.8 |. Scanning electron microscopy

SEM was performed to assess the fibril microstructure of different collagen type I sources (N = 3/group). To maintain the microstructural features of hydrogels, a rigorous sample preparation protocol was followed as described previously.^6,18 Briefly, the hydrogels were fixed in 2.5% glutaraldehyde solution (in 1× PBS) overnight at 4°C. Following fixation, hydrogels were washed in 1× PBS and dehydrated in a graded series of ethanol solution (20%, 50%, 75%, 90%) for 15 min each step and twice with 100% ethanol for 1 h. The hydrogels were then subjected to critical point drying using Denton DCP-1 dryer, sputter coated with gold, and observed at 5000 × magnification under SEM (JEOL JSM-6380LV).

2.9 |. Cell viability analysis

2.10 |. Cell metabolic activity analysis

The metabolic activity of hMSCs was assessed by performing alamarBlue^™ assay (N = 8/group/timepoint). At periodic intervals, hydrogels were incubated in 10% alamarBlue^™ reagent in a culture medium at 37°C for 2 h. Following incubation, 100 μl aliquots from each well were transferred to a 96-well plate (Greiner) in duplicates and fluorescence was measured at an excitation wavelength of 555 nm and an emission wavelength of 595 nm using an M2e Spectramax plate reader (Molecular Devices). Relative fluorescence units (RFU) are reported as a measure of cell metabolic activity on the hydrogels. Passage-5 cells were used in these experiments.

2.11 |. 3D tumor construct formation

At 80% confluency Caco-2 cells were harvested using 0.05% trypsin and aliquoted into 1.5 ml Eppendorf tubes. Tubes were then centrifuged followed by the supernatant removal and the addition and homogenization of hydrogel components to achieve a cell concentration of 3.0 × 10⁶ cells/ml. Hydrogels were prepared as described above, but with the protein concentrations of 6.0 mg/ml and 3.0 mg/ml for human skin-derived collagen type I and rat tail collagen type I, respectively. After suspending cells in the hydrogel components, 10 μl aliquots (about 30,000 cells) were pipetted into polydimethylsiloxane coated 48-well plates. Following a gelation period of 30–45 min at 37C, 200 μl of DMEM with 10% FBS and 1% Pen/Strep were added to the individual wells.

2.12 |. Chemotherapy screening

Regorafenib (Selleckchem, Houston, TX) and 5-fluorouracil (5-FU; Millipore-Sigma, Burlington, MA) were initially dissolved in dimethyl sulfoxide to a stock concentration of 10 mM. Both compounds were serially diluted in DMEM to final concentrations of 1, 10, and 100 μm. After 24 h following 3D tumor construct formation, the media was aspirated from the wells and 500 μl of regorafenib or 5-FU containing media was added into wells separated into six treatment groups. Tumor constructs maintained in drug-free media were used as a control group.

2.13 |. Drug response evaluation by ATP activity quantification

Luminescent ATP activity detection (N = 4) was carried out before adding regorafenib and 5-FU, and at days 1, 4, and 7 after adding the compounds. For ATP activity quantification, spent media was aspirated and a 50/50 solution of the CellTiterGlo 3D kit reagent (Promega, Madison, Wisconsin) and DMEM was added to the 3D tumor constructs. Well plates were covered in aluminum foil and mixed on a shaker plate for 5 min. The assay reaction was allowed to progress for 25 min at room temperature. Solutions were transferred into 96-well white bottom plate wells and the luminescence in each well was quantified with a plate reader (BioTek Cytation 5).

2.14 |. Synthetic microvessel formation and in vivo implantation

Human endothelial cells (ECs), pericytes (PCs) and fibroblasts (FBs) were embedded in a collagen/hyaluronic acid/fibronectin gel and implanted subcutaneously in the abdominal wall of 6- to 8-week-old female C.B-17 SCID/bg mice. Hydrogels were prepared by suspending 3×10⁶ ECs, 1.5 × 10⁶ PCs and 1.5 × 10⁶ FBs in a volume of 550 μl of either rat tail type I collagen (final concentration 1.65 mg/ml) or human skin-derived collagen type I (final concentration: 4.95 mg/ml) supplemented with 100 μl human plasma fibronectin (final concentration: 0.1 mg/ml, EMD Millipore), 100 μl hyaluronic acid (final concentration: 0.5 mg/ml, Sigma), 50 μl of AB human serum (Sigma), 100 μl of 10× pH reconstitution buffer (0.05 M NaOH, 2.2% NaHCO₃, and 200 mM HEPES), and 100 μl of 10× HAM-F12 medium (Gibco). In three independent experiments, three replicate gels were prepared for each condition by dispensing 300 μl of each solution in a 48-well plate and allowed to spontaneously gel at 37°C for 15–20 min. After gelation, 500 μl of EGM2-MV medium was added to each well and incubated for approximately 18 h. Each gel was subcutaneously implanted into the abdominal wall of a SCID/bg mouse (6 animals per experiment). In this study, 18 animals were evaluated to observe statistical significance. Percentage of gel weight reduction was calculated as follows: gel weight before implantation subtracted by gel weight after harvesting divided by gel weight before implantation multiplied by 100.

2.15 |. Assessment of in vivo perfusion by injection of fluorescein UEA-I

Fourteen days after implantation, a solution of Fluorescein UEA I (Vector Laboratories), which labels perfused human vessels, was prepared in a 1:1 ratio with saline solution. Two-hundred microliters were injected per mouse via tail vein or retro-orbitally and allowed to circulate for 30 min before harvesting gels. Mice were euthanized and gels were harvested and fixed in 10% buffered formalin for paraffinembedding or embedded in OCT, frozen, and cryosectioned (10 μm thick).

2.16 |. Morphological characterization of gel implants

Frozen sections of gel implants were analyzed by H&E, immunofluorescence, and immunohistochemical staining. For immunofluorescence analysis, tissue sections were immersed in cold acetone for 10 min and blocked for 1 h with 10% normal goat serum in PBS. Slides were then incubated overnight at 4°C, in a humid chamber, with primary antibodies against human collagen IV (mouse; 1:200, clone 1043, Novus Biologicals), mouse F4/80 (rat, 1:50, clone BM8; eBioscience^™, a marker of mouse macrophages), rhodamine Ulex Europaeus Agglutinin I (UEA-1; 1:200; Vector Laboratories, a marker of human endothelial cells), NG2-Alexa Fluor 488 (mouse, 1:200, clone 9.2.27, ThermoFisher Scientific, a marker of pericytes), and mouse CD31 (rat, clone SZ31, 1:100, ScyTek Laboratories, a marker of mouse endothelial cells). Slides were then washed with PBS (3×) and incubated for 1 h at room temperature with anti-mouse or anti-rat IgG H&L secondary antibodies conjugated with Alexa Fluor^™ 488 or Alexa Fluor 568 (goat, 1:500; Abcam). Slides were mounted with VECTASHIELD antifade mounting medium containing DAPI (Vector Labs) for nuclear staining and imaged on Leica DM6000 inverted fluorescence microscope (Leica Microsystems). Immunohistochemical analysis of mouse macrophages (F4/80) was performed with donkey anti-rat Biotin-SP AffiniPure IgG (H + L) antibody. Substrate to the secondary antibody was included in the VECTASTAIN ABC Peroxidase Kit and AEC Peroxidase Substrate Kit (Vector Laboratories).

2.17 |. Quantification of perfused human vessels and macrophage infiltration

Microscope images of stained tissue sections were taken using a Leica DMI6000 inverted fluorescence microscope. Eight to twelve randomly selected sections sampling all parts of the tissue block were evaluated. Whole sections were imaged using a 4× or 20× objective, with an image pixel size of 0.444 μm. ImageJ software was used to quantify infused fluorescein ulex+, or F4/80+ area in each 4× or 20× field section, respectively. To quantify infused fluorescein ulex+ area, each green channel image was binarized by defining a threshold equally applied to all images. The human perfused vessel area was defined as the number of infused fluorescein ulex+ (nonzero) pixels in the resulting image multiplied by the squared pixel size. To quantify murine macrophage area, the number of red colored pixels were quantified by defining a color threshold and creating a binarized image of the IHC F4/80 stain. The macrophage area was then defined by the number of nonzero pixels in the resulting image multiplied by the squared pixel size. The calculated average from randomly selected sections for each animal was plotted as a single data point in Figure 5B,D.

2.18 |. Statistical analyses

Results are expressed as mean ± SD. Statistical analysis for material characterization studies was performed using one-way ANOVA method with Tukey post hoc comparisons. Data for alamarBlue^™ assay was combined from two separate experimental runs by normalizing the data from each experimental run with human skin-derived collagen type I (HumaDerm) at day 1 for comparison of different collagen sources. Normalized datapoints from both experiments were then multiplied by the mean value of human skin collagen type I hydrogels at day 1 across experiments and analyzed using two-way ANOVA method with Tukey post hoc comparisons. In studies with 3D tumor constructs and implanted collagen gels in vivo, the significant differences between the means were determined using Studenťs t tests with confidence intervals of 95% or p value<.05. All analyses were conducted with GraphPad Prism v7.0.

3.1 |. Quantification of gelation kinetics, swelling ratio, degradation and compressive modulus of collagen type I hydrogels

To assess gelation kinetics of different collagen type I sources, hydrogels were prepared according to manufacturer’s recommendations. Gelation kinetics curves showed a sigmoidal shape for all collagen hydrogels (Figure 1A). The polymerization rate (slope of the linear region of the gelation kinetics curve) was significantly higher for rat tail collagen type I, followed by human fibroblasts-derived collagen type I, and human placenta collagen type I hydrogels (p < .05; Figure 1B). Polymerization rates of bovine collagen type I and human skin-derived collagen type I hydrogels were comparable, but significantly slower than other sources of collagen (p < .05). Rat tail collagen type I had the shortest polymerization half time (12 min), followed by human placenta collagen type I (20.5 min) and human fibroblasts-derived collagen type I (25 min). The polymerization half time of bovine and human skin-derived collagen type I were the longest, at approximately 30 min. Together, these results suggest that variations in tissue source, species, and extraction method have a significant impact on the rate of polymerization of collagen hydrogels

The swelling ability of hydrogels is an important feature that has been shown to directly impact cell behavior and it determines the diffusivity of oxygen and nutrients into the scaffold and removal of toxic metabolites. We conducted a swelling analysis and compared the respective swelling ratios between dry and hydrated hydrogels across the different collagen type I sources. The swelling ratios of rat tail and human placenta collagen type I were the lowest of all collagen hydrogels, while the swelling ratios of human skin-derived and bovine collagen type were the highest (p < .05; Figure 1C). Lastly, bovine collagen type I showed a significantly higher swelling ratio compared to human fibroblasts derived - collagen type I (p < .05).

Hydrogel-based scaffolds are often designed to degrade within the body following implantation at a rate similar to the rate of tissue formation. Hydrogel degradation at physiological conditions is often an advantage as it allows for the scaffold to remodel as the new cell-mediated ECM is being deposited. However, if hydrogels are to be used in long-term implants, then their stability and susceptibility to degradation need to be controlled. We performed an in vitro enzymatic degradation of collagen type I hydrogels from different sources using collagenase. Human placenta collagen type I hydrogels completely degraded in 3 h (p < .05), followed by human fibroblasts-derived collagen type I (Figure 1D). Hydrogels synthesized using bovine, rat tail, or human skin-derived collagen type I showed the highest resistance to enzymatic degradation, taking approximately 7 h to completely degrade (p < .05). Together, these results indicate that bovine and human skin-derived human collagen type I yield more stable hydrogels, while human placenta and fibroblasts-derived collagen type I are less resistant to enzymatic degradation.

Micro-compression tests were performed to evaluate the stiffness of collagen hydrogels. Figure 1E depicts the typical stress–strain curves for all collagen type I hydrogels. The compressive modulus of bovine, human fibroblast-derived and rat tail collagen type I were comparable: 0.28, 0.3, 0.33 kPa, respectively, (p > .05). Hydrogels comprised of human placenta collagen type I showed a compressive modulus of 0.15 kPa, the lowest of all hydrogels (p < .05). Lastly, the compressive modulus of human skin-derived collagen type I was 0.37 kPa which was comparable to rat tail collagen type I hydrogels, but significantly higher (p < .05) than bovine and human fibroblast-derived collagen type I hydrogels, and two-fold greater than human collagen type I derived from placenta tissue. Together, these results show that human skin-derived collagen type I hydrogels have comparable stiffness to xenogeneic-derived collagen sources.

3.2 |. Analysis of fibril microstructure, cell viability and metabolic activity in vitro

SEM imaging was performed to qualitatively assess the fibrous organization of collagen type I hydrogels. Bovine, human fibroblasts-derived, and rat tail collagen type I hydrogels exhibited a similar morphology of thin and loosely packed fibers (Figure 2A upper row). In contrast, human skin-derived collagen type I displayed a unique morphology of highly dense collagen fibers, while human placenta collagen type I hydrogels showed lower density but thicker fibers. Overall, these results confirm that the organization of collagen fibers in collagen type I hydrogels largely differs between species, tissue source, and extraction method

The viability of human mesenchymal stem cells (hMSCs) seeded on top of collagen type I hydrogels was assessed by live-dead staining assay. Results showed high cell viability in all conditions tested (Figure 2A lower row). Quantitative analyses of the live-dead images revealed cell viability was greater than 95% in all hydrogels ay day 7 (Figure 2B). Quantification of cell number displayed the comparable number of cells/mm² on bovine, human fibroblasts-derived, and human placenta collagen type I hydrogels (Figure 2C; p > .05). Hydrogels synthesized using rat tail and human skin showed significantly higher cell number than bovine, human fibroblasts-derived, and human placenta collagen type I hydrogels (p < .05)

An alamarBlue^™ assay was carried out for fluorometric detection of cell metabolic activity on different hydrogels from day 1 to day 10 (Figure 2D). Results showed that cell metabolic activity increased over time on all hydrogels. Specifically, hMSCs seeded on top of rat tail and human skin-derived collagen type I hydrogels showed a significantly higher metabolic activity than cells cultured on other hydrogels at day 10 (p < .05). At earlier time points, hMSCs metabolic activity on human skin-derived collagen type I was significantly higher than cells cultured on human fibroblasts-derived and placenta collagen type I hydrogels at days 4 and 7 (p < .05). Similarly, cells cultured on rat tail collagen type I hydrogels showed significantly higher cell metabolic activity compared to cells cultured on human fibroblasts-derived and human placenta collagen type I at day 7, and human placenta collagen type I only at day 4. In addition, hMSCs metabolic activity on bovine collagen type I hydrogels was significantly higher than on human placenta collagen type I hydrogels at day 7 and day 10 (p < .05). Higher cell number on rat tail and human skin hydrogels (Figure 2A) suggests that the increase in cell metabolic activity over time may be due to greater cell proliferation on these hydrogels. These results show that hMSCs exhibit highest metabolic activity on rat tail and human-skin derived collagen type I hydrogels.

3.3 |. Three-dimensional hydrogel supported tumor constructs respond consistently to chemotherapy

To test compatibility of the various matrices for use in three-dimensional culture applications and drug screening assays, 3D tumor constructs were fabricated, in a manner as we have described for a variety of cell lines and patient-derived tumor cells,^19–21 by encapsulating Caco-2 colorectal cancer cells in hydrogel matrices (rat tail collagen type I, and human-skin derived collagen type I). Rat tail type I collagen was chosen as it is commonly used for cancer organoids formation and drug screening research. 3D tumor constructs were exposed to multiple concentrations of the chemotherapy agents 5-fluorouracil (5FU), a thymidylate synthase inhibitor that interrupts DNA replication, and Regorafenib, a tyrosine kinase receptor inhibitor that inhibits multiple receptor tyrosine kinases, including epidermal growth factor receptor (EGFR) and vascular endothelial growth factor receptor 2 (VEGFR2). Both drugs are commonly used and clinically approved chemotherapeutic compounds to treat colorectal cancer. ATP activity quantification and live/dead staining of tumor constructs were performed on days 0, 1, 4, and 7. 3D constructs under control conditions (no drug exposure) maintained a consistent ATP activity over the 7-day period in all four hydrogels tested. In agreement, fluorescent microscopy images of control conditions showed consistent expression of calcein AM-positive cells (viable cells), with minor and non-incremental amounts of ethidium homodimer-1-positive cells (dead cells). Conversely, increasing concentrations of Regorafenib or 5FU in treatment groups resulted in dose dependent reduction in ATP activity over time (Figure 3). Interestingly, 3D tumor constructs treated with 100 μm Regorafenib and formulated with rat tail collagen type I, initially showed the highest ATP activity among all hydrogels, however it rapidly declined to comparable levels by day 4. In agreement with these data, live/dead staining images of regorafenib-treated 3D tumor constructs appear to contain more dead cells at later time points. By day 7, all constructs exposed to 100 μm Regorafenib appear to exclusively contain dead cells.

5FU-treated 3D tumor constructs exhibited a similar trend, although less significant. Moreover, while 3D tumor constructs treated with 10 and 100 μm Regorafenib or 5FU showed an expected decline in ATP activity, constructs exposed to 1 μm show little to no decrease in ATP activity (Figure 3). Live/dead staining was consistent with ATP activity data, showing that 3D tumor constructs exposed to 1 μm Regorafenib or 5FU had increasing ATP activity and calcein AM-positive cells over time, while higher concentrations followed an opposing trend (Figure 4). It is likely that a 1 μm dosage may have not been sufficient to effectively kill all cells contained in the construct. Together these results suggest that 3D tumor constructs formulated with collagen type I of different species may respond differently to various dosages of chemotherapy agents. Of note, 3D tumor constructs formulated with human skin tissue-derived collagen type I have shown the most significant differences in ATP activity and cell viability in response to Regorafenib versus 5FU. These differences may determine potential success or failure in the development of 3D tumor models, in drug screening applications, that can faithfully recapitulate in vivo responses.

3.4 |. Comparisons of type I collagen sources for formation of vascularized hydrogels in vitro and in vivo

Cultured human endothelial cells isolated from an umbilical vein (HUVECs) or differentiated in vitro from umbilical cord blood human endothelial colony forming cells (HECFCs) will spontaneously organize into cords that lumenize to form tubes when suspended in hydrogels based on a mixture of rat tail type I collagen and human plasma fibronectin. Inclusion of human dermal fibroblasts in the gel sustains the viability of the endothelial cells (ECs) and inclusion of human placenta pericytes (PCs) stabilizes and constrains the luminal diameter of the tubes. When implanted in a subcutaneous pocket formed by blunt dissection in the abdominal wall superficial to the abdominal musculature of an immunodeficient C.B-17 SCID/bg mouse, these tubular structures spontaneously anastomose at the border of the gel with mouse microvessels in the wound bed and become perfused. The hydrogel may contract overtime which can be reduced by including hyaluronic acid in the gel. Mouse vessels typically do not invade the gel beyond the region beyond the point of interaction with the human EC-lined vessels. Furthermore, such gels do not elicit a significant foreign body reaction and very few mouse macrophages are found within the gel. Here we aimed to determine if the rat tail collagen type I could be replaced by human collagen to eliminate a potential source of xenoantigenicity as part of a program to bioengineer microvascular constructs for clinical use. In a preliminary screening, several sources of collagen type I were tested, as provided by manufacturers, for their ability to promote self-assembly of endothelial networks in vitro. Gelation rate at the time of preparation and hydrogel contraction over a 5-day period were also assessed. With the exception of rat tail collagen type I, hydrogels prepared with collagen products with a final concentration below 3 mg/ml took a longer period of time to complete polymerization, resulting in cells settling by gravity at the bottom of the well. Although all conditions tested were able to support self-assembly of endothelial networks in vitro, the rapid loss of the initial 3-dimensional cell distribution proved to be inadequate for potential tissue engineering applications (Figure S2). These differences allowed us to compare the ability of human and nonhuman type I collagens to promote in vivo formation and perfusion of human vessels in a SCID/bg mouse model. By gross examination at the time of explant, rat tail or human skin-derived collagen type I-containing hydrogels appeared to be highly vascularized (Figure 5A lower row). Measurement of hydrogels weights before implantation and after explant showed that human skin-derived collagen type I-containing gels showed significantly higher weight reduction over the period of 14 days compared to rat tail collagen type I (p < .05), which is a potential indicator of a lower water retention capacity (Figure 5B). H&E staining of formalin-fixed, paraffin-embedded tissue sections of hydrogels confirmed the findings of gross examination and revealed a substantial number of microvessels at 14 days post-implantation in both rat tail or human collagen type I- containing hydrogels (Figure 5C upper row). Quantification of F4/80-stained sections indicated significantly higher infiltration of mouse macrophages in rat tail type I-containing hydrogels compared to human skin-derived collagen type I-containing hydrogels (Figure 5C lower row, 5D). Fluorescein ulex injection 30 min before explant labeled human ECs was used to assess perfusion of human EC-lined vessel segments. Immunofluorescence analysis confirmed that rat tail or human collagen type I-containing hydrogels displayed human CD31⁺ vessels with many surrounded by a secondary layer of human pericytes (NG2-positive cells) and collagen IV, a major component of the microvessel basement membrane. Some microvessels were lined by mouse ECs, suggesting some degree of host angiogenesis (Figure 5E). No significant differences in the number of perfused human vessels were observed between conditions. (Figure 5F). To assess if all human EC-lined had become perfused 14 days after implantation, we stained tissue sections post-harvest with rhodamine ulex. As excepted, all fluorescein ulex-positive vessels co-localized with rhodamine fluorescein ulex staining, however nonperfused EC-lined segments surrounded by collagen IV were also observed, consistent with incomplete perfusion.