Bioengineered vascular graft grown in the mouse peritoneal cavity

Abstract

Background

We tested the hypothesis that the mouse peritoneum can function like a bioreactor to generate directed bio-engineered tissues such as those used for bypass grafting. Additionally, we reasoned that the mouse animal model would allow us to elucidate the underlying cellular and molecular mechanisms that are responsible for the generation of tissue in peritoneal cavity.

Methods

Plastic tubes (2 tubes/mouse) were implanted into the peritoneal cavity of 3 strains of mice (C57BL/6, BALB/c, and MRL). The tubes were harvested, tissue capsule surrounding the tubes was removed, and analyzed by immunostaining (5 capsules/5 mice/strain) and microarray (3 capsules/3 mice/strain). In addition, the tissue capsules that were harvested from MRL mice (n=21) were grafted into abdominal aorta of the same mice as autografts. The patency of all grafts was monitored by micro-ultrasound and their functionality was assessed by Laser Doppler Imaging of blood flow in femoral arteries. Venous (n=13) and arterial isografts (n=11) were used as positive controls. In a negative control group (5 mice/strain), the abdominal aorta was occluded by double ligation with 9-0 silk.

Results

The implanted plastic tubes required at least 8 weeks of incubation in the peritoneum of the 3 strains of mice in order to generate useful grafts. No vascular cells. were found in the tissue capsules. Microarray analysis of tissue capsules revealed that the capsular cells express a gene expression program that is vastly shared among the 3 strains of mice and the cells exhibit high degree of plasticity. The micro-ultrasound analysis of the grafts showed that 62% of autografts remained patent compared to 77% of venous isografts and 91% of arterial isografts. The Laser Doppler Imaging analysis showed that blood flow dropped by 40% and 35% in the autografts and vein isografts, respectively, one day after surgery. The flow, however, rebounded to the level of arterial isografts one month post surgery and remained unchanged among all grafts for the next 4 months. Immunostaining of the autografts showed a thick vessel wall with endothelial cells that lined the lumen and smooth muscle cells that constituted the graft wall.

Conclusion

The mouse peritoneal cavity of mice has the ability to function like a bioreactor to generate bio-engineered tissues. The tissue capsules harvested from peritoneal cavity of a mouse are composed of nonvascular cells that display phenotype of progenitor cells. After grafting, however, the capsule auto-grafts become arterialized and remained patent for at least 4 months after surgery, similar to venous or arterial iso-grafts.

Introduction

Vascular bypass grafting is the mainstay of revascularization for ischemic heart disease and peripheral vascular disease, and in the US alone 1.4 million arterial bypass operations are performed annually. However, 30% of patients who require arterial bypass procedures do not have saphenous veins suitable for use, because of previous harvest for bypass surgery, varicose degeneration, or inadequate diameter or length. [1]. Campbell et al showed that implantation of silastic tubing into the peritoneal cavities of dog, rabbits and rats, led to the formation of free-floating avascular tissue tubes over 2 weeks, with few intestinal adhesions [2–4]. Investigations into the underlying molecular mechanisms of tissue capsule generation in the peritoneal cavity and their application as a functional graft have been hampered by the non-murine nature of animal model.

The objective of the present study was to refine a mouse model allowing the development of a peritoneal-derived capsule graft which would facilitate a study of the genetic and molecular characteristics of the graft. We hypothesized that the mouse peritoneum can function like a bioreactor generating directed bio-engineered tissue that can be used for grafting in a similar way to venous or arterial grafts. By employing 3 strains of mice, we were able to analyze gene expression program underlying development of tissues in mouse peritoneum. Micro-ultrasound and Laser Doppler Image analyses were used to evaluate the function of grafts.

Materials and Methods

8 mm. long of plastic tubes were implanted into the peritoneal cavity of 3 strains of mice. The tissue capsule that formed over the tube during the incubation was used for microarray analysis and immunostaining. The tissue capsules harvested from MRL mice were grafted by end-to-end anastomosis into the abdominal aorta of the same mice as auto-grafts. Venous and arterial isografts were used as positive controls. Immunohistochemical staining and expression profiling were used to characterize the cellular composition of tissue capsule. Micro-ultrasound Imaging and laser Doppler Imaging assessed the functionality of the capsule grafts. Detail methodologies are described in the Supplement.

Results

To test the hypothesis that the mouse peritoneum can function like a bioreactor to generate capsule grafts, two 8-mm. long sialistic tubes (outer diameter 0.96 mm.) were implanted into the peritoneal cavity of C57BL/6 mice and harvested after 2 weeks, as per Campbell’s [3] recommendation (Fig. 1, panel A). However, only two ends of the tubes were found to be covered with proteineous material and the midsection of tube remained uncovered (Fig. 1, panel B, the arrow indicates the areas that are not covered). Subsequently, we found that an incubation period of least 8 weeks was required to generate a tissue capsule that was suitable for grafting (Fig. 1, panel C). The tissue capsule that formed around the tubes was carefully separated from the tube (panel D) and used for grafting.

Fig 1.

Generation of tissue capsule in the peritoneal cavity of a mouse. Siastic tubing (panel A) was inserted into peritoneal cavity of C57BL/6 mice and removed 14 days after implantation. Panel B shows that the two ends of the tube are covered with tissues, whereas the midsection of the tube remained uncovered (arrows). Panel C shows the implanted tube that was harvested after 8 weeks which is completely covered with tissue. Panel D illustrates the removal of the tissue capsule from the tube by cutting one end of the capsule. Panel E shows the harvested plastic tube and the tissue capsule that was used for auto-grafting into the abdominal aorta of a mouse. Similar results were obtained with implanted polyethylene tubing. For detail on the number of tubes and Methodology please see the Method section in the Supplement.

Since our goal was to use the tissue capsule for bypass grafting, we asked whether the harvested capsules exhibit a phenotype of blood vessel conduits. To explore this, the harvested capsules were stained by antibodies that are specific to smooth muscle α-actin and an endothelial cell-specific marker CD31. We found that the capsule contains cells that are unevenly distributed; however, these cells did not express either smooth muscle marker (Fig. 2A) or endothelial cell marker (Fig. 2B). This suggests that the tissue capsules harvested from the mouse peritoneum lack cells with a vascular cell phenotype. In addition, few macrophages were observed in only one of the capsules prior to grafting (Fig. 2C, arrow), suggesting that inflammation may not be a major player in the formation of tissue capsule. Collagen staining revealed that the capsules are composed of collagenous material (Fig. 2D). To characterize the phenotype of cells that were found in the tissue capsule, total RNA was extracted from freshly harvested capsules from 3 BALB/c mice, 3 C57BL/6 mice, and 3 MRL strains of mice and total RNAs were analyzed by microarray. The detailed analysis of quality of RNA isolated from tissue capsules of each strain of mice, their intra-group reproducibility-scatter plot analysis, and the gene lists are included in the supplement. Since we were able to grow complete tissue capsules in the peritoneal cavity of C57BL/6, BALB/c, and MRL mouse strains, we reasoned that the process of capsule generation in the peritoneum of the 3 strains of mice is likely controlled by a gene expression program that is shared among the 3 strains of mice. In addition, the use of overlapping genes among the 3 strains of mice allowed us to filter out non-relevant, strain-specific, genes. To explore this idea, gene expression profile of tissue capsules harvested from the 3 strains of mice were compared and the shared gene expression program was identified. In addition to strain-specific genes, we identified a subset of genes commonly enriched among the 3 strains of mice (Figure 3A). The probability of observing such an overlap by chance as estimated using hypergeometrical distribution [5] is extremely low (P=10⁻¹¹). The distribution of genes within the shared transcriptome across functional categories is shown in Fig. 3B. Molecules thought to be involved in development, morphogenesis, and differentiation tend to be overrepresented in the shared gene expression program of tissue capsule cells. These results suggest that there is a gene expression program in the peritoneal-derived tissue that is shared among the 3 strains of mice and that this program is indicative of a high degree of plasticity in the phenotype of capsular cells.

Fig 2.

Representative immunohistochemical analysis of tissue capsule. The frozen section of tissue capsule harvested at 8 weeks from the peritoneum of MRL mice were stained with 1:50 dilution of anti-smooth muscle α-actin antibody (panel A), 1:50 dilution of anti-CD31 antibody (panel B), and 1:50 dilution of anti-moma-2 antibody (panel C). The vast majority of the sections from the 3 strains of mice were negative for Moma-2 expression and only one section from one mouse that was positive for moma-2 expression (arrow). Panel D shows collagen staining. We stained 5 tissue capsules that were harvested from each strain of mice. See method section in the Supplement for details.

Fig. 3.

Expression profiling of freshly-harvested tissue capsules isolated from the 3 strains of mice. Total RNA was isolated from freshly-harvested tissue capsule from the 3 strains of mice (3 mice/strain), and after pooling and quality control evaluation, they were analyzed by microarray using mouse Affymetrix GeneChip MOE-430 that contains 22,000 cDNA and ESTs. Panel A. Venn diagram showing the number of genes enriched in each strain-specific cell population and their overlaps. Note the high overlap between the 3 strains of mice. In a small percentage of the cases, the same gene is recognized by more than one probe set in the Affymetrix array; hence, the 669 probe sets enriched in all peritoneal cells actually correspond to 628 unique genes. Panel B shows the distribution within the shared transcriptome for gene products with known or putative functions.

Next, we evaluated the ability of tissue capsules to function as a bypass auto-grafts when compared to venous or arterial iso-grafts. For this evaluation, we used only MRL mice because they are 50% larger than other 2 strains of mice, making surgical manipulations easier. The tissue capsules harvested from 21 MRL mice were grafted into abdominal aorta of the same mice by an end-to-end anastomosis, as autografts. The photograph of tissue capsule immediately after grafting shown in Fig. 4, panel A, arrow illustrates that the surface of the capsule grafts are smooth. The capsule auto-grafts were harvested 4 months after grafting (panel B, arrow). Two positive controls were used: a venous iso-graft (panel C, arrow), and an arterial iso-graft (panel D, arrow),(the donor and recipient mice are MRL mice, i.e., the grafts are isografts). For the number of grafts and patency rate see Table 1. For the negative control group, the abdominal aorta of MRL mice was occluded by double ligation with 9-0 silk. All mice from the negative control group lost their ability to move shortly after surgery and had to be euthanized; whereas majority of the mice receiving capsule auto-grafts and the two positive control groups of mice receiving venous or arterial iso-grafts remained mobile, vigorous and viable.

Fig. 4.

Grafting of tissue capsule into abdominal aorta of the same MRL mice. Panel A, freshly harvested tissue capsule was everted and grafted into abdominal aorta of the same mice, i.e., the capsule grafts are autografts. Please note the smoothness of the surface of the graft (arrow). Panel B shows the capsule autograft 4 months after transplantation. Panels C and D show venous isografts (arrow) and arterial isografts (arrow) 4 months after grafting, respectively. The number of grafts, patency rate, and animal survival rate are shown in table 1.

Table 1.

All grafting were performed in MRL mice. Tissue capsules harvested from the peritoneum of 21 MRL mice were grafted into the same animal. For vein and arterial grafts that were used as positive controls, vena cava and thoracic aorta harvested from donor MRL mice were grafted into abdominal aorta of the recipient MRL mice, i.e., these grafts are isografts. The total number of vein grafts performed was 13 animals, and for arterial grafts, the total number is 11 animals. We used these numbers in order to obtain at least 5 surviving animal (patent grafts) at 4 and 16 weeks post surgery. The grafts were harvested from each group of animals 4 weeks and then 16 weeks post surgery and used for analysis.

Graft type	4 weeks (patent grafts)	16 weeks (patent grafts)	Total Exp.	Survival rate
Capsule grafts (autograft)	8	5	21	13/21= 61.9%
Vein grafts (isograft)	5	5	13	10/13= 76.9%
Arterial grafts (isograft)	5	5	11	10/11= 90.9 %

We used three methodologies to assess patency of the grafts. The graft was considered successful if it met the following criteria: at the time of implantation and harvesting it was strongly pulsating and fully distended, the animal had normal motor activity, and the graft was patent as indicated by a good blood flow through the graft, as well as through the both femoral arteries. In addition, subsequent immunostaining examination of the sections showed no evidence of clot formation. Conversely, grafts that were not patent were flaccid, retracted and had no detectable pulsation. Microscopically, they clearly showed a clot in their lumen. Using these criteria, MRL mice with successful grafts were retained for investigations. The viability rate/patency rate among the recipient of capsule autografts, venous isografts, and arterial isografts was 60%, 76%, and 90%, respectively, (Table 1). A representative micro-ultrasound image of 4 month-old capsule graft in the abdominal aorta of MRL mice (Fig. 5) indicates that the capsule grafts remained distended with average length of 7.38 mm and diameter of 1.23 mm as shown in panels A and B, respectively. Pulse Wave Doppler shown in panel C illustrates good blood flow consistent with the results of Laser Doppler Imaging.

Fig. 5.

Representative micro-ultrasound analysis of the recipient of the capsule graft. The micro-ultrasound was used to evaluate patency of the capsule graft 4 months after grafting in MRL mouse. Panels A and B show that the capsule autografts remained distended with average length of 7.38 mm and diameter of 1.23 mm. Panel C show blood flow as measured by pulse wave Doppler.

In addition to micro-ultrasound, we used Laser Doppler Imaging to assess the ability of grafts to perfuse the lower extremities of mice. As shown in Fig. 6A, blood flow to the femoral artery was significantly reduced 1 day post surgery in the capsule autograft recipients (panel II) or venous isografts recipients (panel V) compared to base line (before surgery, panels I and IV, respectively); whereas the perfusion of recipients of arterial isografts (panel VIII) appeared to be greater than that of other grafts. The blood flow in all types of graft recipients was similar at one month post grafting (panels III, VI, and IX).. Quantitative analysis of blood flow in the recipient mice shown in Fig. 6B revealed that the average blood flow in the recipients of either capsule autografts or vein isografts groups dropped by 40% (2.19±0.13 vs. 1.29±0.24) and 35% (2.21±0.01 vs. 1.43±0.27), respectively, at one day post surgery. In contrast, no statistically significant difference was found in the blood flow of mice that received arterial isografts (2.2±0.09 vs. 1.79±0.1). At four weeks post grafting, however, the flow rebounded in the recipients of both capsule autografts and venous isografts. Similarly, no significant differences in the blood flow were detected in all three groups of graft recipients at 4 months after grafting.

Fig 6.

Representative Laser Doppler Images from the bilateral hind limbs of the recipients of capsule, venous and arterial grafts. Fig. 6A shows the LDI images that were recorded at three different times: one day prior to grafting (baseline), one day, and one month after grafting in MRL mice. Images taken of hind limbs with the capsule graft (panels II and III) were compared to those of base line (panel I). Similarly, venous graft (panels V and VI) as well as arterial grafts images (panels VIII and IX) were compared to their respective baseline images (panels IV and VII), respectively. Representative color-coded images reflect red blood cell velocity (red is highest velocity, green intermediate, and blue, lowest velocity). Fig. 6B shows cumulative results of LDI derived blood flow of the bilateral hind limbs are shown over the experimental time course. Each data point represents the mean value ± SD. Venous graft (VG) compared to arterial graft (AG) P < 0.05, and the capsule graft (CG) compared to arterial graft (AG) P < 0.01 at one day following transplantation. For detail number of grafts and the patency rates please see Table 1.

To assess the cellular composition of the capsule autografts, the mice were sacrificed at 4 months (n=5), the grafts were harvested, snap- frozen, embedded in OCT, sectioned, and analyzed by immunohistochemical staining (Fig. 7). Movat staining revealed that the wall of capsule autografts was thick with multiple layers of cells that were distributed throughout the graft (panel A). In contrast to tissue capsule before grafting (Fig. 2), we did not find any acellular regions in the capsule autografts. The cells that formed thick wall of the autografts expressed proteoglycans (panel A, greenish color). Unlike the staining of freshly harvested tissue capsules (Fig. 2), the autografts strongly expressed markers of smooth muscle α-actin that are concentrated in the thickened wall (panel B). The expression pattern of smooth muscle cell markers in the autografts was similar to those of positive control vein isografts that are known to be arterialized when exposed to arterial pressure (panel C). The autografts expressed endothelial cell markers, suggesting that endothelial cells lined the lumen of the capsule grafts (panel D). The autografts expressed little, if any, macrophage markers (panel E), suggesting that the grafts were not inflammed. These data suggest that the harvested tissue capsules were basically acellular before grafting; however, they displayed phenotype of vascularized blood vessel conduit when they were grafted into abdominal aorta for at least 4 months.

Fig 7.

Representative immunohistochemical analysis of the capsule graft harvested 4 months after transplantation in MRL mice. Frozen sections harvested from 5 MRL mice tissue capsule autografts were analyzed by histochemical staining and immunostaining. Panel A shows Movat staining. Panel B and C show smooth muscle α-actin staining of autografts and venous isografts, respectively. Panel D shows staining of capsule autograft for the expression of CD31 endothelial cell marker. The presence of macrophages was evaluated by anti-moma-2 staining (panel E). We stained 5 tissue capsule autografts, 5 venous isografts, and 5 aortic isografts.

Discussion

The peritoneal cavity of several different animal species has been used to develop artificial grafts [4,6]; however, those studies utilized a non-murine model that precluded analysis of these grafts at the genetic and molecular level. To overcome this limitation and to test the hypothesis that the mouse peritoneal cavity has an environment that is conducive to the formation of artificial blood vessels, we have developed a mouse model where the artificial conduit is grown within the peritoneal cavity. To accomplish this goal, we initially followed the Campbell’s protocol [3] by harvesting the implanted silastic tubes 2 weeks after its implantation into the mouse peritoneum; however, the harvested tubes were found to be incompletely covered with the tissues that were not suitable for grafting. This was a consistent observation among the 3 strains of mice examined, suggesting that the incomplete formation of tube under this condition is independent of mouse strain. Further studies revealed that by extending the incubation time of the tubes in the mouse peritoneum to 8 weeks, tissue capsules that were generated was suitable for grafting. Immunohistochemical analysis revealed that the tissue capsules isolated from peritoneal cavity of BALB/c mice, C57BL/6 mice, or MRL mice contained few cells and these cells did not express vascular cell markers.

To generate a hypothesis about the nature of cells that contribute to the formation of a tissue capsule in the mouse peritoneum, we examined the expression profile we established from total RNA harvested from the freshly-harvested capsules following an 8-week incubation in the mouse peritoneum. As all 3 strains of mice were capable of generating tissue capsule, we reasoned that the peritoneal cavities of C57BL/6, BALB/c, and MRL mice share transcriptome signatures that allow these mice to generate tissue capsules. Additionally, this hypothesis of using shared signature gene expression program allowed us to filter out strain-specific gene expression program and exclude non-relevant transcriptome that is unlikely to be involved in tissue capsule formation. Using a Venn diagram, we noted that the vast majority of genes found in the tissue capsules derived from the 3 strains of mice displayed extensive overlap, suggesting that all the tissue capsules generated are very similar to one another and likely represent the results of a conserved gene expression program. The profiling data also revealed that the tissue capsules shared a transcriptome profile that is similar to progenitor cells. Among the 669 up-regulated genes that are shared among the capsule grafts, the vast majority of them are related to the development, differentiation, and morphogenesis; a genotype that is similar to those of progenitor cells. These data suggests that the peritoneal cells that contributed to the formation of tissue capsule may exhibit a high degree of plasticity which allow these cells to adopt the phenotype of vascular cells when exposed to conditions that is conducive to the formation of a vascular cell phenotype, such as grafting of tissue capsules into arterial circulation. If this proves correct, one could speculate that the peritoneum most likely contains progenitor cells with the ability to home in, cover the plastic tubes, and generate tissue capsules with the ability to function as blood vessel conduits. In addition, expression profiling suggests that the cells that constitute tissue capsule are not differentiated cells, a finding that is supported by immunohistochemical analysis of the capsule grafts. Further, the shared transcriptome profile does not support the concept that the capsular cells exhibit phenotype of fibroblasts. Overall, the microarray data coupled with immunostaining analysis of tissue capsules point to the plastic nature of capsular cells. Studies are underway to explore this hypothesis in depth.

We found that when the tissue capsules harvested from MRL mice were grafted into abdominal aorta, blood flow dropped approximately by 40% in the first day after surgery, as compared to arterial grafts. A similar drop, 35%, in the blood flow is also observed when venous grafts were used. Over a period of 4 weeks, however, no statistically significant difference was found in the blood flow among all the recipient groups. The initial drop in the blood flow among the recipient of capsule grafts and vein grafts is likely related to the delay in flow-induced remodeling of the grafts, i.e., arterialization. The differences in the venous and arterial pressures are thought to be responsible for the remodeling of vein grafts inducing proliferation of pre-existing venous smooth muscle cells that foster a thick media [7]. While this notion is applicable to remodeling of vein grafts that contain vascular cells, it would not be applicable to capsule autografts because immunostaining and microarray analyses found no vascular cell phenotype characteristics in the tissue capsule before grafting. We believe that the experimental arterialization that has been described for vein iso-grafts in animal models is likely different from the arterialization of tissue capsule autografts, for at least two reasons. First, unlike isografts, capsule autografts do not generate a significant immune response to the grafts; therefore, the contribution of inflammatory-derived cells/factors is not a prominent factor/s in the remodeling of capsule grafts.. Second, the vein grafts contain endothelial cells and smooth muscle cells that can proliferate in response to arterial pressure and develop into artery-like structure. Such a possibility does not exist for capsule grafts because there are no vascular cells in the capsule before grafting. Therefore, we feel that the capsular remodeling is unique and most likely related to the adaptation remodeling of progenitor cells to the arterial pressure, the consequence of which is formation of capsule grafts which function like those vein grafts and arterial grafts. The origin of vascular cells in the capsule grafts remains unclear at this time. The cells may have originated from local progenitor cells residing within the local capsular autograft vessel wall or from circulating progenitor cells. While some studies have shown that endothelial progenitor cells contribute to the repair of the damaged blood vessels, foster bioengineering of vascular prosthetic grafts and stents [8,9], and contribute to the remodeling of graft in a parabiotic mouse model [10], other studies found no evidence for such a contribution of bone marrow-derived cells [11][12]. The reason for this divergent data remains unclear; however, reliable identification of the phenotype and origin or lineages of cells requires careful examination and localization of cell markers to a single cells. With a thickness of less than few microns, co-localization studies involving single cells are particularly demanding and may not be reliable. Further studies will be needed to address this fundamental question.

In summary, we have developed a mouse model for generation of artificial grafts that can be used effectively for grafting. The grafts remained patent when grafted into abdominal aorta and they become arterialized. The tissue capsules before grafting were primarily composed of nonvascular cells and collagen; however, endothelial cells were found to line the lumen and multilayer smooth muscle cells were found in the “wall” after auto-grafting of the tissue capsule. Gene expression profiling showed that the tissue capsule cells express a phenotype that exhibits a high degree of plasticity. This animal model will provide us an opportunity to investigate the cellular and molecular basis of graft remodeling, specifically the contribution of resident peritoneal cells and circulating progenitor cells in this process.