Tissue engineering of the small intestine
T. Ansari, Northwick Park Institute of Medical Research, UK
S.M. Gabe, Imperial College London, UK
Abstract:
Tissue engineering of the small intestine offers a novel treatment for patients with short bowel syndrome. Its anatomical and functional complexity make this particularly challenging. Current techniques are based on use of synthetic or biological scaffolds with guided tissue regeneration or tissue engineering or both of these. This chapter covers the approaches to tissue engineering of the small intestine, scaffold types and selection, guided tissue regeneration of the small intestine, the different cell seeding sources and the process of combining cells and scaffolds. Finally different growth factors are discussed and the future directions for developments in this field indicated.
Key words
intestinal tissue engineering; guided tissue regeneration; scaffold; organoid units; intestinal stem cells
16.1 Introduction
When the small intestine fails there are limited options for patients. The small intestine is responsible for absorption of nutrients and when the underlying problem is irreversible patients have to go on long-term intravenous nutrition. Generally this is life-long and carries significant risks of infection, central venous thrombosis and liver failure. Furthermore the quality of life is impaired for these patients who have numerous issues to have to deal with. The current approaches include intestinal transplantation or experimental bowel lengthening procedures in some patients. Small intestinal transplantation is a developing field but it is hampered by limited survival rates and quality of life. In recent years the 1-year graft and patient survival rates are 77% and 80%, respectively from the pooled international transplant registry (Garg et al. 2011). However, 3-year graft and patient survival rates are only 59% and 69%, respectively.
The large bowel (colon) is mainly responsible for the absorption of water from the stool and there are conditions that result in colectomy, when the colon is removed. These patients generally have a stoma on the abdominal wall but for some an internal pouch is fashioned from some of their small bowel, enabling them to manage without a stoma. However, this type of surgery uses a significant portion of their native small bowel. In the situation that a pouch has to be reformed, patients can end up with an insufficient length of remaining small bowel to absorb all the nutrients that they require.
Tissue engineering is now becoming established as a treatment for chronic disease of irreversibly damaged or absent tissue. Such treatments have been used successfully in the clinical setting for various tissues and organs such as the trachea (Elliott et al. 2012; Macchiarini et al. 2008) and bladder (Atala et al. 2006). Current research is focused on developing a treatment for intestinal failure by replacement of the small intestine with bioengineered tissue. Such bioengineered tissue would result in an increase in absorptive capacity of the small intestine and allow a patient with a short bowel to be weaned from parenteral nutrition. However, the structural and functional complexities of the small intestine are considerable and this has posed a significant barrier to progress in tissue engineering of the small intestine. This chapter will review progress made to date in the field of small intestinal tissue engineering, highlight the limitations of the current models and techniques and explore possible future research directions.
16.2 Approaches to tissue engineering of the small intestine
Numerous approaches to whole organ tissue engineering are currently being applied and there remains no universally accepted approach to tissue engineering of the small intestine (TESI). Goals of tissue engineering include the development of functional tissue, with appropriate properties (e.g. biomechanical) that can be incorporated safely into the recipient without adverse immunological response. The majority of approaches to tissue engineering involve combination of a scaffold matrix (synthetic or biological) with an appropriate cell source (e.g. mesenchymal stem cells). This is based on the theory that the scaffold matrix provides the ideal three-dimensional structure in which the appropriate cell source can interface and multiply, leading to new tissue regeneration. This model has been applied to the small intestine in the past. Historically, little distinction has been made in TESI studies between ‘tissue engineering’ (implantation of in vitro seeded matrices) and ‘guided tissue regeneration’ (implantation of acellular matrices, repopulated by the host post procedure). In the past, studies have attempted to replicate full thickness intestine, specific layers (e.g. mucosa) or produce a functional absorptive surface without necessarily reproducing the exact anatomical structure of the small intestine.
However, there remain a number of difficulties. The small intestine is anatomically complex, with its morphology reflecting not only its function as an absorptive surface but also its role as a barrier against the external environment (see Fig. 16.1).
The production of innervated muscle layers, vascular and lymphatic networks and appropriate lymphoid tissue in addition to a functional mucosal surface is a considerable challenge. Such structural complexities pose difficulties in choosing appropriate cell type(s) for seeding and producing scaffolds to allow for regeneration.
The majority of studies to date have concentrated on the production of a tissue engineered layer of small intestinal mucosa, often referred to as neomucosa or neointestine (Ansaloni et al. 2006; Binnington et al. 1974, 1975; Choi et al. 1998; De Faria et al. 2004; Kim et al. 1999b; Lillemoe et al. 1982; Lloyd et al. 2006b; Tait et al. 1994b, 1995; Tavakkolizadeh et al. 2003; Thompson et al. 1984). Generation of an intact neomucosal layer is understandably considered to be vital for the manufacture of functional replacement intestinal tissue. However, it must be remembered that the function of the small intestine is also dependent on an adequate vascular supply and lymphatic drainage as well as coordinated peristalsis dependent on correctly innervated muscular layers.
Two main techniques have been used in order to produce small intestinal neomucosa in animal models. The simplest method has been to achieve intestinal lengthening by interposition of a tubular artificial scaffold between segments of healthy small intestine. The scaffold then forms a framework for ingrowth of mucosa from the healthy intestinal tissue and this is an example of guided tissue regeneration (Ansaloni et al. 2006; Chen and Badylak 2001; Hori et al. 2001; Pahari et al. 2006). The alternative technique represents true tissue engineering and involves transplanting intestinal stem cell populations, harvested from neonatal animals, onto denuded bowel, artificial scaffolds or decellularised scaffolds (Avansino et al. 2005, 2006a; Choi and Vacanti 1997; Choi et al. 1998; De Faria et al. 2004; Grikscheit et al. 2004; Kaihara et al. 1999a, 1999b, 2000; Kim et al. 1999b; Lloyd et al. 2006a, 2006b; Tait et al. 1994a, 1994b, 1995; Tavakkolizadeh et al. 2003). All of these techniques, along with their relative strengths and weaknesses, will be discussed in more detail below.
16.3 Scaffold selection
Availability of a suitable scaffold material is vital to the tissue engineering of any organ. A successful scaffold needs to have a range of physical, chemical and biological properties that are tailored to the tissue that it is to support. It must also be biocompatible and not elicit a significant foreign body reaction. Scaffold materials can be either biological or synthetic and a summary of materials used as scaffolds for small intestinal tissue engineering is shown in Table 16.1.
Table 16.1
Materials used as scaffolds for small intestinal tissue engineering
| Type | Structure | Features/Properties | Experimental design | Refs |
| Synthetic | Polyglycolic acid (PGA) | Sheets of non-woven PGA (15 μm fibre diameter and 250 μm average pore diameter) wrapped into a tubular structure and stabilised by coating with 5% polylactic acid (PLA). Improved cellular adhesion demonstrated after coating with type 1 collagen | Stem cell transplantation | Choi and Vacanti (1997), Choi et al. (1998), Duxbury et al. (2004), Gardner-Thorpe et al. (2003), Grikscheit et al. (2004), Kaihara et al. (1999a, 1999b, 2000), Kim et al. (1999a; 1999b), Perez et al. (2002), Ramsanahie et al. (2003), Tavakkolizadeh et al. (2003) |
| Synthetic | Poly(D,L-lactide-co-glycolide) (PLGA) | PLGA foam manufactured by thermally induced phase separation. Radially oriented interconnected pores with large size distribution (50–300 μm). Rolled into tubular structure and opposing edges joined by dissolving in chloroform and opposing | Stem cell transplantation | Lloyd et al. (2006a, 2006b) |
| Natural | Small intestine submucosa (SIS) | Manufacture by mechanical removal of mucosa and muscular layers from porcine small intestine followed by osmotic lysis of remaining cells. Resulting membrane ~ 80–100 μm thick. Greater success with multilayered sheets | Intestinal patch and intestinal lengthening | Chen and Badylak (2001), Z. Q. Wang et al. (2003, 2005) |
| Natural | Surgisis (Cook Biotech Inc.) | Commercially available small intestinal submucosa (SIS)-type scaffold material derived from porcine small intestine | Intestinal lengthening | Ansaloni et al. (2006) |
| Natural | Collagen sponge | Collagen extracted from porcine skin (70–80% type I collagen, 20–30% type III atelocollagen). Fibres whipped, freeze-moulded and freeze-dried. Stabilised by heating to cross-link collagen fibres and subsequent basal coating with polyglycolic acid (PGA) | Intestinal Lengthening | Hori et al. (2001, 2002), Nakase et al. (2006) |
| Natural | Acellular dermal matrix (ADM) | Commercially available ADM (AlloDerm, Cell Life) sutured to created a tubular structure | Intestinal lengthening | Pahari et al. (2006) |
| Natural | Acellular gastric wall | Gastric tissue decellularised by detergent-enzymatic treatment and sutured to create a tubular structure | Intestinal lengthening | Parnigotto et al. (2000) |
| Natural | Decellularised small intestine or colon | Small bowel or colon decellularised by detergent-enzymatic treatment | Intestinal lengthening | Nowocin et al. (2013) |
Synthetic scaffolds have been created from a range of different bioresorbable polymers and co-polymers including polyglycolic acid (PGA), polylactide (PLA) and poly(D,L-lactide-co-glycolide) (PLGA) using a range of different engineering techniques (Maquet and Jerome 1997). A cross-sectional image of an example of a synthetic scaffold is shown in Fig. 16.2.
Experiments attempting to create neointestine have classically transplanted stem cell cultures onto synthetic scaffolds (Choi and Vacanti 1997; Choi et al. 1998; De Faria et al. 2004; Lloyd et al. 2006b). However, there are limitations in these scaffolds and considerable research is now focused on the use of biological scaffolds for full intestinal thickness regeneration.
The three-dimensional structure of the scaffold material is of vital importance in TESI as it will not only influence the mechanical properties of the structure but also affect cell migration and adhesion. Biological and synthetic scaffolds are porous structures with interconnected pore networks. The larger pores allow cellular infiltration and migration and penetration by blood vessels and lymphatics. The smaller pores allow diffusion of oxygen and nutrients inward and waste products outward. A potential advantage of a synthetic scaffold is that it is possible to manipulate the exact physical properties of the material. By altering pore size and pore density it can be possible to change both the physical and the biological characteristics of the synthetic scaffold. It is also easier to control the overall shape of synthetic compared to biological scaffolds. However, the synthetic scaffolds that have been made to date are relatively simplistic and do not replicate the structural complexity of intestinal tissue with the different layers and blood vessel network required to maintain developed intestinal tissue larger than a few centimetres.
Another important property of scaffolds used for tissue engineering is the ability to promote cell adhesion, migration and proliferation. This is heavily influenced by the surface properties of the scaffold material. In general, cell adhesion is enhanced by a rough rather than a smooth fibre surface. This is seen in biological scaffolds such as small intestinal submucosa (SIS) due to the high collagen content. However, the majority of synthetic polymers used to create scaffolds are hydrophobic with a relatively smooth fibre surface. In order to promote cellular interactions a number of surface modification techniques have been employed, including surface coating, chemical modification and plasma treatment (S. Wang et al. 2005). The collagen coating of PGA scaffolds has been shown to improve adhesion of intestinal epithelial cells (Choi et al. 1998) although it is not known how such modified synthetic scaffolds would compare to biological scaffold materials. Biological scaffolds are also believed to confer additional benefit in cell migration and adhesion by the preservation of proteins and growth factors. The exact mechanism of this is poorly understood but there is evidence that the interaction between the extracellular matrix (ECM) and surrounding cells is a dynamic process requiring input from both sources for adequate tissue regeneration (Ansaloni et al. 2006).
Creating the ideal scaffold for small intestinal tissue engineering is considerably more difficult than for other hard and soft tissues such as bone or cartilage due to the increased structural complexity. The ultimate goal is to produce flexible multi-layered, functional tubular intestinal tissue. The production of tubular scaffolds is technically more challenging as the structure must have enough strength to remain patent when initially implanted and allow replacement by intestinal tissue and subsequent tissue expansion. The time taken for certain biodegradable scaffolds to break down is also critical; if it is too rapid then the lumen may collapse but if too slow then the growth of the neointestine may be impeded. The degradation properties of synthetic scaffolds can be modulated by altering the chemical composition and pore structure. Equally, the degradation of biological scaffolds can be modulated by cross-linking the collagen fibres (Hori et al. 2001), thereby affecting collagen degradation.
16.4 Guided tissue regeneration of the small intestine
16.4.1 Tubular scaffolds
A number of groups have attempted to tissue engineer small intestine by interposing artificial tubular scaffolds between sections of healthy intestinal tissue (Ansaloni et al. 2006; Chen and Badylak 2001; Hori et al. 2001; Pahari et al. 2006). Various different materials have been used to produce these artificial scaffolds including freeze moulded collagen fibres (Hori et al. 2001, 2002; Nakase et al. 2006), small intestine submucosa (Ansaloni et al. 2006; Chen and Badylak 2001; Z. Q. Wang et al. 2003, 2005) and acellular matrices derived from gastric wall (Parnigotto et al. 2000) and dermis (Pahari et al. 2006). Our group has used decellularised rat colon to produce neo-intestinal mucosa consisting of all four intestinal cell lineages, with multiple blood vessels and smooth muscle-like cells in the submucosal layer at 6 weeks (unpublished data, Plate X, between pages 354 and 355). The basic concept is to employ the natural regenerative potential of the small intestine to populate an artificial scaffold designed to promote the growth of cells in from adjoining healthy intestinal tissue.
Experiments attempting to lengthen the bowel using tubular scaffolds have developed from earlier studies that demonstrated that small bowel defects could be successfully patched using autologous serosa, abdominal wall muscle or peritoneum (Binnington et al. 1974, 1975; Erez et al. 1992; Lillemoe et al. 1982;Thompson et al. 1984).These experiments demonstrated in-growth of mucosa along the margins of the patch – small intestine anastamosis (Binnington et al. 1974) with formation of neomucosa. Histologically, the morphology of this neomucosa was similar to that of the surrounding small intestine although the central areas of the patch remained only partially covered with immature mucosa at 36 weeks (Binnington et al. 1974). When serosal patches were used, the resulting neomucosa had aminopeptidase, maltase and lactase activities similar to that of the surrounding native intestine although activities of alkaline phosphatase and sucrase were reduced (Binnington et al. 1975; Thompson et al. 1984). The functional potential of this neointestinal patch was further supported by experiments in a porcine model of short bowel syndrome which demonstrated increased weight gain after patching with colonic serosa (Binnington et al. 1975).
Attempts to create tubular scaffolds using colonic serosa have been of limited success with few animals surviving the procedure and only partial coverage of the scaffolds with neomucosa (Thompson 1990). Experimental models using artificial tubular scaffolds have been more successful. Several studies from Japan have employed collagen sponge soaked in autologous blood to produce a tubular scaffold for intestinal lengthening in a canine model (Hori et al. 2001, 2002; Nakase et al. 2006). Collagen sponges were formed by enzyme extraction of collagen from porcine skin followed by whipping and freeze moulding to produce flat sponge-like structures. These were then stabilised by inducing cross-linking between the collagen strands by heating and then further reinforced by application of polyglycolic acid (Hori et al. 2001).
In the initial experiments in beagle dogs a 5 cm section of de-functioned jejunum was resected and replaced with a silicone tube stent. This silicone tube was then wrapped in collagen sponge which had been soaked in autologous blood which was in turn wrapped with omentum. These initial experiments demonstrated the growth of neomucosa inwards from the healthy jejunum, but not the development of a muscular or serosal layer (Hori et al. 2001). In further experiments using the same basic model the collagen sponge was seeded with mesenchymal stem cells derived from bone marrow in an attempt to produce a muscle layer. However, these experiments were essentially unsuccessful with only a very thin muscle layer developing below the mucosal surface (Hori et al. 2002). Further studies from the same group have succeeded in creating an intact smooth muscle layer after 12 weeks by seeding the collagen sponge with autologous smooth muscle cells derived from stomach wall, although only in ileal patch grafts (Nakase et al. 2006).
Our group has recently developed a novel method to produce a biologically derived tubular scaffold with vascular network (Nowocin et al. 2013). Using a vascular perfusion approach, a segment of porcine ileum up to 30 cm long, together with its attached vasculature, was decellularised. The scaffold’s decellularised vascular network was able to be perfused by connection to the renal artery and vein in an anesthetised pig and the mesenteric arcade remained patent throughout the 24 h assessment (Plate XI, between pages 354 and 355). The explanted scaffolds showed signs of cellular attachment, with cells positive for CD68 and CD133 on the vascular luminal aspect. It was therefore possible to decellularise clinically relevant lengths of small intestine, together with the associated vasculature, as a single segment (Nowocin et al. 2013). The next stage will be to cellularise this scaffold in a controlled manner (Patil et al. 2013).
16.4.2 SIS
There has been considerable interest in the use of SIS in the tissue engineering of a range of tissues including urinary tract, tendon and blood vessels. SIS is primarily an acellular collagen-based matrix; it is manufactured from small intestine by mechanical removal of the mucosa and muscular layers followed by osmotic lysis of any remaining cells. Unlike other scaffold materials that have been used for intestinal tissue engineering, SIS has been shown to retain growth factors with properties similar to fibroblast growth factor (FGF) and transforming growth factor-beta (TGF-β) (Voytik-Harbin et al. 1997). Initial experiments showed that porcine SIS could be used as a patch to repair relatively large defects in canine small bowel and that there was migration of cells into the patch, which at 3 months resulted in a mucosal layer, disorganised smooth muscle layers and a serosal layer similar to native small intestine. However, attempts to interpose a segment of tubular SIS between divided small intestinal loops resulted in anastomotic leakage in all animals (Chen and Badylak 2001).
Further experiments in rodents have demonstrated that 2 cm long tubular SIS scaffolds could be successfully interposed into defunctioned small intestine (Z. Q. Wang et al. 2003, 2005). By 12 weeks the entire lumen was covered with neomucosa and by 24 weeks there were also distinct smooth muscle and serosal layers (Wang et al. 2003). Similar experiments using Surgisis, a porcine-derived matrix very similar to SIS, have demonstrated well-organised layers of mucosa, smooth muscle and serosa 24 weeks after interposition of 3 cm scaffolds into defunctioned ileal loops in rodent models (Ansaloni et al. 2006). However, a limitation of using SIS for small intestine reconstruction is that the scaffold contracts in size (Lee et al. 2008; Qin and Dunn 2011).
16.4.3 Limitations
The use of artificial scaffolds to lengthen the small intestine by interposition alone is appealing in its simplicity. The procedure is relatively straightforward and there is no requirement for exogenous biological tissue. Perhaps most importantly, evidence has shown that these techniques can yield well-organised tissue with distinct mucosal, muscle and serosal layers. However, there are drawbacks. In order to prevent anastomotic leakage it is necessary to defunction the loop of bowel into which the tubular scaffold is interposed. In the studies where the interposed scaffold was not taken out of the flow of luminal contents all animals died of peritonitis (Chen and Badylak 2001; Pahari et al. 2006). In patients who already have compromised intestinal function due to short bowel syndrome, the temporary loss of further functional intestine is clearly undesirable. The exact length of time that the scaffold containing loop of intestine needs to be defunctioned for is uncertain; only a single study has reported successful re-anastomosis, and this after 8 weeks (Nakase et al. 2006). It seems very likely that the time required for adequate engraftment of the scaffold will depend on the length of the implant, given that growth of intestinal tissue into the scaffold appears to occur only from the anastomoses. As such it may be difficult and potentially very slow to achieve significant lengthening of the small intestine using interposition of artificial scaffolds alone.
To date, better results in TESI have been achieved by using a ‘tissue engineering’ type approach of combining scaffold (synthetic or biological) with a cell source prior to implantation. Indeed, this approach has been demonstratively successful in a number of tissue engineering projects and the above approach of guided tissue regeneration is now perceived as a less realistic future option. The fundamental reason for this is that, it is becoming apparent that the introduction of a scaffold matrix alone into host tissue without adequate cell source does not lead to regeneration of functional, full thickness intestinal tissue. This does not therefore represent a likely long-term solution in the desire for TESI production.
16.5 Cell seeding sources
16.5.1 Intestinal stem cells
Native intestinal mucosa has an impressive capacity for replication and regeneration both under normal physiological conditions and following injury. This regenerative capacity is dependent on the activity of intestinal epithelial stem cells. Intestinal stem cells are found towards the base of the epithelial crypts. The majority of stem cell divisions are believed to result in a single daughter cell and a single stem cell which retains the original template DNA. These daughter cells then undergo further divisions to produce a population of transit-amplifying (TA) cells. These TA cells are rapidly proliferating and divide and further differentiate to produce the different epithelial cell lines. The concept that a single intestinal stem cell can give rise to all intestinal epithelial cell lines, known as the Unitarian hypothesis, is supported by a significant body of evidence. Stem cell daughter cells and initial TA cells retain their clonogenicity and are able to revert back to stem cells if the crypt is damaged and existing stem cells are lost. However, as the TA cells divide further they lose their capacity for clonal expansion. Enterocytes, goblet cells and enteroendocrine cells undergo further differentiation as they migrate upwards towards the tip of the villus. They then are either shed into the intestinal lumen or undergo apoptosis.
While the hierarchical pattern of cell proliferation and differentiation from crypt to villus is firmly established, the exact number of the stems cells in each crypt is less clear. This was historically due to the lack of reliable molecular stem cell markers, although the discovery of stem cell markers (e.g. Lgr5, Bmi1, Musashi 1) in recent years has greatly improved knowledge. It is believed that there are four to six stem cells in each crypt, which are located in a specific stem cell compartment also known as the ‘stem cell niche’. This niche comprises the intestinal epithelial stem cell, neighbouring proliferating cells, and adjacent mesenchymal cells such as the pericryptal fibroblasts and intestinal subepithelial myofibroblasts. These mesenchymal cells are believed to play an important role in the maintenance of the stem cell population and the control and regulation of proliferation via the secretion of various peptides. There are complex signalling pathways between the different components of the stem cell niche, and understanding of this signalling is increasing rapidly and has been characterised further (Buske et al. 2011, 2012; King and Dekaney 2013).
Recently, human pluripotent stem cells have been directed to differentiate into intestinal tissue in vitro using a series of growth factor manipulations to mimic embryonic intestinal development (Finkbeiner and Spence 2013; McCracken et al. 2011; Spence et al. 2011). The three-dimensional intestinal ‘organoids’ that were developed consisted of a polarised, columnar epithelium that patterned into villus-like structures and crypt-like proliferative zones that expressed intestinal stem cell markers (SOX9, ASCL2, LGR5). Using this culture system it was shown that the combined activity of WNT3A and FGF4 is required for hindgut specification whereas FGF4 alone is sufficient to promote hindgut morphogenesis (Spence et al. 2011).
16.5.2 Organoid units
In the early 1990s Evans and colleagues described a method by which disaggregated intestinal tissue, termed intestinal organoids, were derived from neonatal rat small intestine by partial digestion using a mixture of collagenase and dispase (Evans et al. 1992). As the tryptic activity of the enzymatic solution is low there is not complete dissociation of the epithelial cells. The resulting intestinal organoids are cellular aggregates consisting of polarised intestinal epithelium surrounding a core of mesenchymal cells (see Plate XI, between pages 354 and 355). It is believed that they contain intestinal stem cells, other progenitor cells and epithelium along with the mesenchymal stroma. In vitro studies demonstrated that the intestinal organoids could be maintained in cell culture. Interestingly, more extensive dissociation by prolonged enzymatic action to yield single cell suspensions appeared to inhibit cell proliferation (Evans et al. 1992). This would appear to support the importance of maintaining the stem cell niche by preserving the organoid units but the mechanism of this is not understood.
Subsequent experiments demonstrated that suspensions of organoid units transplanted into subcutaneous pockets in adult rodents could develop into small, short, tubular structures which consisted of a central mucin filled lumen surrounded by a circumferential epithelial layer (Tait et al. 1994b). As with the in vitro experiments, implantation of single cell suspensions did not result in the formation of cysts containing neomucosa (Patel et al. 1996). When intestinal organoids derived from 5–8-day-old rats were implanted into nude mice, 39% developed into neointestinal cysts; when intestinal organoids were implanted into inbred rats, the success rate was 84%. The maximum length of these structures was 5 mm in the nude mice and 8 mm in the rats. As early as 2 weeks after organoid unit transplantation, the epithelial layer had formed crypts and villi and was histologically similar to small intestinal mucosa. This neomucosa was shown to contain all epithelial cell lineages including Paneth cells which were not identified in 6-day-old neonatal small intestine (Tait et al. 1994b). This is of particular relevance as it suggests development of these cells from pluripotent stem cells in the transplanted intestinal organoids rather than simply from multiplication of more differentiated cells already in the transplanted tissue. Lactase, sucrase, aminopeptidase and alkaline phosphatase activity was also demonstrated in the neomucosa, as was sodium-dependent glucose transport (Tait et al. 1995), suggesting that the neointestine had functional potential. In addition to the neomucosal components, smooth muscle-like cells were identified adjacent to the neomucosa although they had not developed into discrete muscle layers (Tait et al. 1994b).
A recent study in mice described a method for the long-term expansion of colonic stem cells in culture (Yui et al. 2012). These culture cells were then reintroduced as cultured colonic organoids into superficially damaged mouse colon. The transplanted donor cells integrated into the mouse colon, covering the area that lacked epithelium as a result of the introduced damage in recipient mice. At 4 weeks after transplantation, the donor-derived cells constituted a single-layered epithelium, which formed self-renewing crypts that were functionally and histologically normal. Long-term engraftment of over 6 months was also seen from a single Lgr5(+) colon stem cell after extensive in vitro expansion.
While the majority of subsequent studies have used rodent tissue, it is noteworthy that Sattar and colleagues (1999) demonstrated that intestinal organoids produced from human foetal small intestine could be successfully implanted subcutaneously into severe combined immunodeficiency (SCID) mice to produce cysts of neomucosa similar to that described in rat models. Tissue was obtained following terminations of pregnancy between 12 and 20 weeks. Successful development of subcutaneously implanted intestinal organoids was achieved in a similar proportion of experiments as in rodent models and, similarly, the resultant neomucosa contained all mucosal cell lines at 50 days after implantation (Sattar et al. 1999). Barthel et al. (2012) have demonstrated that colon can form from human progenitor cells on a scaffold in a mouse host. In their study, organoid units were prepared from human colon waste specimens, loaded onto biodegradable scaffolds and implanted into immunocompromised mice. After 4 weeks, human tissue engineered colon was harvested.
16.6 Combining cells and scaffolds
Experiments by Campbell’s group, following on from their work on the isolation of intestinal organoids, showed that small intestinal mucosa could be regenerated by seeding intestinal organoids onto ascending colon after mucosectomy (Tait et al. 1994a). Fourteen days after implantation onto loops of defunctioned mucosectomized colon, neomucosa had developed in 76% of animals. This neomucosa was histologically similar to small intestine and contained enterocytes, goblet cells, Paneth cells and enteroendocrine cells. No regeneration was seen on control loops of mucosectomised colon, confirming that regeneration was due to proliferation of the implanted intestinal organoids rather than due to incomplete mucosectomy. This works has been repeated more recently as described above by Yui et al. (2012).
Avansino and colleagues (2006a) developed a model in which intestinal organoids derived from the distal ileum of neonatal mice and rats were implanted onto segments of mucosectomised jejunum which had been de-functioned and tied off to prevent loss of implanted organoids. After 2 weeks, neointestine containing all four intestinal epithelial cell lineages had developed on the mucosectomised jejunum. Initial experiments suggested that there was an optimal dose of intestinal organoids for maximal engraftment (Avansino et al. 2006b). The percentage of the mucosal surface covered with neomucosa (as opposed to native jejunal mucosa) was determined by immunohistochemical staining using antibodies against the ileal bile acid transporter (IBAT) and confirmed in further experiments using intestinal organoids harvested from green fluorescent protein (GFP) positive animals (Avansino et al. 2006a). Disappointingly, the maximal covering by neomucosa was only 18% and varied considerable depending on the methods used for mucosectomy. However, additional experiments suggest that these segments of tissue engineered ileum had functional potential. After resection of the native terminal ileum in a rat model, anastamosis of tissue engineered ileum in continuity with native intestine was shown to significantly attenuate bile acid loss compared to animals that did not have neointestine inserted; total bile acid loss in the rats with tissue engineered ileum was similar to animals with an intact native ileum although taurocholate uptake was less than in normal animals (Avansino et al. 2005).
The implantation of intestinal organoids onto artificial scaffolds was pioneered by a research group in the USA headed by Joseph Vacanti. The group employed 1 cm long tubular scaffolds created from PGA fibre meshes stabilised by spraying with poly(L-lactic acid) (PLLA) and a 50/50 copolymer of PLGA (Mooney et al. 1996). In the initial experiments intestinal organoids were seeded onto the polymer scaffold 2 h prior to implantation into the omentum of adult rats; intestinal organoids engrafted on 16 out of 19 scaffolds with the formation of small cyst-like structures with a maximum length of 3.6 mm. Histological analysis of cysts harvested between 2 and 8 weeks confirmed the presence of neomucosa containing columnar epithelium, goblet cells and Paneth cells (Choi and Vacanti 1997).
Further experiments by Vacanti and coworkers demonstrated that intestinal organoid engraftment could be enhanced by coating the scaffolds with type I collagen with cysts forming from 93% of collagen coated scaffolds versus 64% of non-coated scaffolds (Choi et al. 1998). Collagen coated scaffolds also resulted in considerably larger cysts at 6 weeks with a maximal length of 30 mm documented. Histology again demonstrated columnar epithelium and goblet cells in the mucosal layer and smooth muscle-like cells in the submucosa. Immunohistochemistry detected cells staining positively for sucrase on the apical epithelial surface of the neomucosa and Ussing chamber experiments demonstrated a potential difference across the mucosa, although this was significantly less than across normal ileal mucosa (Choi et al. 1998).
Studies employing immunohistochemical staining for CD34, a vascular endothelial marker, have demonstrated vascularization of the tissue engineered intestine (Gardner-Thorpe et al. 2003). Attempts were made to compare capillary density and tissue concentrations of vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF) with native small intestine taken from juvenile and adult rats. Such comparisons are difficult to interpret given the morphological differences between the tissue engineered and native intestines. However, the authors concluded that although VEGF and bFGF were present in the neointestine, the relatively low concentrations suggested that there were other growth factors involved in angiogenesis (Gardner-Thorpe et al. 2003). A separate study analysing similar tissue also reported the presence of lymphatic vessels in the neointestinal cysts although, again, the pattern of lymphangiogenesis appeared different from that seen in native small intestine (Duxbury et al. 2004).
Our group has developed a similar model for small intestinal tissue engineering in which intestinal organoids are implanted onto PLGA scaffolds (Lloyd et al. 2006b). Sheets of PLGA foam are made by a thermally induced phase separation process. This results in a porous structure with radially oriented interconnected pores ranging in size from 50 to 300 μm. The PLGA foam is rolled into a tube and the opposing edges dissolved with chloroform an then pressed together to join (Day et al. 2004). These PLGA scaffolds have been implanted subcutaneously into rats. The PLGA scaffolds were left in situ for several weeks to allow them to become cellularised prior to implantation of intestinal organoids. This facilitated successful development of intestinal neomucosa with fewer implanted organoids than previously described (Lloyd et al. 2006b). Further experiments have demonstrated that the resultant neomucosa contains replicating and differentiated cells and remains viable at 12 weeks after intestinal organoid implantation (Lloyd et al. 2006a).
Vacanti and coworkers developed their model further and have successfully anastamosed neointestine onto native small intestine. Neointestinal cysts were opened longitudinally 3 weeks after initial scaffold implantation and were joined to native jejunum via a side-to-side anastamosis (Kaihara et al. 1999b; Kim et al. 1999b). Examination after a further 7 weeks revealed significantly greater villus number and height as well as greater surface area of the neomucosa that had been anastamosed compared to that which had not (Kaihara et al. 1999b; Kim et al. 1999b). There was also significantly greater expression of the glucose transporter SGLT1 in the anastamosed neomucosa (Tavakkolizadeh et al. 2003). Neomucosal morphology was maintained to 36 weeks (Kaihara et al. 2000). A single study has also demonstrated the feasibility of end-to-end anastamosis with overall patency rates of 78% at 10 weeks (Kaihara et al. 1999a).
More detailed immunohistochemical studies demonstrated cellular proliferation rates in the mucosal layer of anastamosed neointestine to be similar to those seen in native small intestine; proliferation rates were significantly lower in non-anastamosed neointestine. Apoptotic rates were similar in the anastamosed and non-anastamosed neointestine (Tavakkolizadeh et al. 2003). Interestingly, these studies also demonstrated immune cell subsets in the neointestine that has been anastamosed to native intestine with population densities that were similar to native jejunum (Perez et al. 2002). The immune cell populations appeared to develop with time and not to develop in the neointestine that had not been anastamosed to native small intestine, leading the authors to conclude that the development of the neomucosal immune system was dependent both on exposure to luminal content and to the duration of exposure (Perez et al. 2002).
The possible therapeutic potential of the neointestinal cysts created by the Vacanti group has been noted in studies where neointestinal cysts were anastamosed to native jejunum after massive enterectomy (Grikscheit et al. 2004; Kaihara et al. 2000). Compared to animals that underwent enterectomy alone, the animals that had had neointestine anastamosed had a significantly more rapid weight recovery after initial post-operative loss. These animals also maintained serum B12 concentrations in the normal range as opposed to animals that had undergone enterectomy alone (Grikscheit et al. 2004). However, it is not entirely clear whether these effects were due to absorption of nutrients by the neomucosa or to the effect of implanting a segment of immotile bowel on gut transit. Although the neointestine was shown to contain both smooth muscle-like cells staining positively for smooth muscle actin and neuronal cells staining positively for S100, gut transit times were significantly longer in the rats that had had neointestine anastamosed to native jejunum (1825 minutes ± 753 compared to 982 ± 300) (Grikscheit et al. 2004). This may have resulted in prolonged exposure to and hence improved absorption by the remnant native small intestine. Interestingly, myenteric denervation of a segment of ileum has been shown to significantly attenuate weight loss after 80% enterectomy in a rat model (Garcia et al. 1999), suggesting that delaying gut transit has a significant effect on intestinal absorptive capacity.
16.6.1 Limitations
The techniques described above have demonstrated that tissue can be created without having to de-function loops of small bowel and that this neointestine can then be successfully anastamosed with native small intestine. Similar techniques have also been used to create tissue engineered oesophageal, gastric and colonic mucosa (Grikscheit et al. 2002, 2003a, 2003b; Maemura et al. 2004). However, there are some significant limitations at present. The principal problem is the source of donor tissue used to produce the intestinal organoids coupled with the low yield of the tissue engineering process.
Experiments to date have obtained intestinal organoids from either neonatal or foetal small intestine, with studies reporting a yield of about 40 000 intestinal organoids from the small intestine of a single neonatal rat (Choi et al. 1998). Foetal and neonatal tissue is not ideal for the harvesting of intestinal organoids in a clinical scenario, and ultimately an autologous source of tissue for implantation will need to be found in order to avoid the requirement for long-term immunosuppression. Studies by the Vacanti group have implanted up to 100 000 intestinal organoids per 1 cm long biosynthetic scaffold in order to create a single cyst of neomucosa (Ramsanahie et al. 2003) which clearly demonstrates the inefficiency of the tissue engineering process. Implantation onto denuded native jejunum would appear to require fewer intestinal organoids with Avansino and colleagues (2006b) reporting optimal results with the implantation of 10 000 intestinal organoids per 3 cm jejunal segment. It is noteworthy, however, that the overall percentage of successful implantations was considerably lower than that reported by the Vacanti group (Ramsanahie et al. 2003).
In addition to the limitations relating to sourcing intestinal organoids and the yield of the process, the neointestine produced by transplanting intestinal stem cell clusters appears less well developed than that produced by interposition of artificial scaffolds into healthy small intestine. Prolonged follow-up of neointestine created by seeding of synthetic scaffolds with intestinal organoids revealed a well-developed mucosal layer (Kaihara et al. 2000); however, even after anastamosis with native small intestine there were not well-defined muscular and serosal layers as seen in some experiments using biological scaffolds interposed between healthy loops of native bowel (Ansaloni et al. 2006). It remains unclear whether or not further remodelling of the muscular and serosal layers would occur beyond the 36 weeks follow-up period.
16.7 Growth factors
Growth and regeneration of the small intestine are under the control of a wide range of growth factors and cytokines. Attempts have been made to try to speed and enhance the development of tissue engineered small intestine by manipulating these growth factors. Studies of neointestinal development on serosal patches in rabbits demonstrated that infusion of urogastrone resulted in more rapid ingrowth of neointestine from the surrounding native bowel (Thompson et al. 1987). This effect was shown to be dose dependent (Thompson et al. 1988) and prolonged administration resulted in an increase in the amount of neomucosa produced (Thompson et al. 1989).
Using the same model it was shown that 50% enterectomy resulted in greater ingrowth of neomucosa over the serosal patch (Bragg and Thompson 1989). This is unsurprising given the adaptive response seen in individuals after massive small bowel resection and it is interesting to note that the effects of urogastrone are not synergistic with those of massive small bowel resection (Thompson et al. 1990). Experiments using synthetic scaffolds seeded with intestinal organoids also demonstrated increased neointestinal cyst length and diameter in animals that had undergone small bowel resection compared to controls (Kim et al. 1999a). As with native small intestine, contact with luminal contents has also shown to be an important stimulus to neomucosal growth and development. Studies using tubular scaffolds to lengthen the small intestine and those employing transplantation of intestinal organoids onto synthetic scaffolds have consistently demonstrated increased neomucosal growth and maturation when in continuity with the native small bowel (Kaihara et al. 2000).
More recently it has been shown that both glucagon like peptide (GLP)-2 and VEGF have a stimulatory effect on neointestinal development. Parenteral administration of GLP-2 to rodents implanted with scaffolds seeded with intestinal organoids produced significantly greater villus height and crypt depth in the resulting neomucosa, along with increased crypt proliferation and reduced apoptosis. In addition, there was also evidence of apical transporter upregulation (Ramsanahie et al. 2003). These effects are similar to those seen in native small intestine after GLP-2 administration. Taken together, the findings above suggest that the response of neointestine to proliferative stimuli may be very similar to that of native small intestine. Also, studies have been performed using neonatal organoid units from transgenic mouse pups capable of inducible, ubiquitous VEGF overexpression (CMV-Cre/rtTA/tet(0)-VEGF) implanted into non-obese diabetic/SCID mice (Matthews et al. 2011). This group found that VEGF overexpression as associated with a larger construct with increased villus and crypt height after 4 weeks.
16.8 Conclusions and future trends
Tissue engineering of the small intestine offers a novel treatment for patients with short bowel syndrome which avoids the potential complications of long-term parenteral nutrition and intestinal transplantation. Small intestinal tissue engineering is particularly challenging due to the considerable anatomical and functional complexity of the gastrointestinal tract. Current techniques are based on the use of synthetic or biological scaffolds using methods of guided tissue regeneration, tissue engineering, or some combination of both. The appropriate cell source for seeding scaffolds remains unclear at present and while considerable success has been achieved with organoid units, they do not represent a clinically viable solution. The future will certainly lie with human pluripotent stem cells or mesenchymal stem calls as we understand more about how these cells differentiate intestinal tissue. The fields of material engineering and cell biology are evolving rapidly and it seems likely that intestinal tissue engineering will become a viable therapeutic option in the not too distant future.
The progress made in the field of small intestinal tissue engineering is impressive but there are significant limitations to the techniques that have been developed. Intestinal lengthening procedures have produced the most morphologically correct neointestine with well demarcated mucosal, muscle and serosal layers but the process is slow and requires defunctioning of a proportion of the small intestine. Theoretically, it may be possible to combine the technique with transplantation of intestinal stem cells in order to speed the generation of neomucosa and allow earlier reanastamosis, and implantation of multiple cell lines may augment the regeneration of both mucosal and muscle layers.
Transplantation of intestinal stem cells is limited by a lack of suitable donor tissue and the low yield of the process. In order to avoid the problems of tissue rejection and the need for long-term immunosuppression, an autologous source of donor tissue is necessary. It is unlikely that sufficient intestinal stem cells could be harvested from native small intestine given that attempts to expand harvested intestinal tissue in vitro have been unsuccessful to date. However, it may be possible to transplant haematopoietic stem cells harvested from either bone marrow or from the peripheral circulation, and induce transdifferentiation into intestinal stem cells. Studies suggest that bone marrow derived mesenchymal cells can differentiate into pericryptal myofibroblasts in both mice and humans. Studies have shown that in IL-10 knockout mice undergoing bone marrow transplantation, 30% of colonic subepithelial myofibroblasts are of bone marrow origin in normal mucosa, increasing to 45% in inflamed mucosa 3 months after transplantation (Bamba et al. 2005). Transdifferentiated epithelial cells have also been reported in intestinal mucosa several years after bone marrow transplantation (Okamoto et al. 2002). Of note, bone marrow cells have been used successfully in the tissue engineering of vascular tissue (Hibino et al. 2005).
It is likely that it is possible to improve the yield of intestinal tissue engineering by modulating the scaffold properties in order to accelerate and augment neointestinal growth and development. As mentioned above, intestinal organoid engraftment onto denuded jejunum appears more successful than engraftment onto synthetic scaffolds as suggested by the lower numbers of intestinal organoids required. This probably reflects the optimal surface characteristics and pre-existing vascularization of the denuded jejunum. Hybrid scaffolds combining synthetic and biological materials may maintain the flexibility of synthetic compounds while simultaneously gaining the superior surface characteristics of biologically derived tissue. Pre-implantation will allow vascularisation of the scaffolds prior to transplantation of intestinal organoids or other stem cell populations, and the yield of pre-implanted PLGA scaffolds compares favourably to PGA scaffolds that were not pre-implanted (Lloyd et al. 2006b). It may also be possible to improve vascularisation by coating scaffolds either with vascular growth factors such as VEGF (Murphy et al. 2000) or with compounds such as bioactive glass which stimulate endogenous growth factor release and promote blood vessel growth (Day et al. 2005).
A more recent shift in paradigm in tissue engineering has led to the proposal of implantation of a whole organ graft capable of revascularisation. This idea followed from the observation that acellular matrix scaffolds could effectively be produced by decellularisation by perfusion of reagents via existing tissue vascular networks. A significant body of work in this subject has been performed in a variety of organs including the lung, heart and liver. Our group has experimented with large animal intestinal ECM production by perfusion decellularisation and this model has been described in the literature (Totonelli et al. 2012, 2013). The benefits of such a scaffold are that it can be implanted into the host with a vascular component, allowing delivery of oxygen and nutrients to the scaffold via the host circulation. The existing vasculature also provides a route for delivery of cell sources for seeding. While this methods represents an exciting option both for the intestine and other organs, barriers remain in the process of cell seeding of scaffolds. The intestinal scaffolds may require seeding with a variety of cell types and the appropriate method of delivery and cell source remain unclear. In addition, other tissue engineered organs have been seeded ex vivo in a bioreactor type device prior to implantation to aid regeneration and it remains to be seen if such a device may have a role in the development of TESI.