Xenotransplantation: Progress Along Paths Uncertain from Models to Application

Abstract

For more than a century, transplantation of tissues and organs from animals into man, xenotransplantation, has been viewed as a potential way to treat disease. Ironically, interest in xenotransplantation was fueled especially by successful application of allotransplantation, that is, transplantation of human tissue and organs, as a treatment for a variety of diseases, especially organ failure because scarcity of human tissues limited allotransplantation to a fraction of those who could benefit. In principle, use of animals such as pigs as a source of transplants would allow transplantation to exert a vastly greater impact than allotransplantation on medicine and public health. However, biological barriers to xenotransplantation, including immunity of the recipient, incompatibility of biological systems, and transmission of novel infectious agents, are believed to exceed the barriers to allotransplantation and presently to hinder clinical applications. One way potentially to address the barriers to xenotransplantation is by genetic engineering animal sources. The last 2 decades have brought progressive advances in approaches that can be applied to genetic modification of large animals. Application of these approaches to genetic engineering of pigs has contributed to dramatic improvement in the outcome of experimental xenografts in nonhuman primates and have encouraged the development of a new type of xenograft, a reverse xenograft, in which human stem cells are introduced into pigs under conditions that support differentiation and expansion into functional tissues and potentially organs. These advances make it appropriate to consider the potential limitation of genetic engineering and of current models for advancing the clinical applications of xenotransplantation and reverse xenotransplantation.

Introduction

Few, if any, subjects of research and fields of medical practice provoke as much excitement and as much controversy as xenotransplantation. The excitement stems from the prospect that lethal and debilitating diseases might be conquered by replacing sick or damaged organs with healthy organs using an inexhaustible supply provided by animals and from progressively improving results of experimental xenotransplants, some surviving for a year or more, suggesting that xenotransplantation could soon emerge from the laboratory and enter (actually reenter) the clinic.^1–3 The controversies stem from the limited supply of human organs that impels consideration of xenotransplantation, from the apparent need to introduce human genes into the germline of animals to facilitate acceptance of foreign tissue grafts, and from fears about yet-unknown organisms that might originate from the clinical application of xenotransplantation.

Below we summarize the rationale for pursuing xenotransplantation, the obstacles to success, and recent advances in overcoming those obstacles. We deliberately avoid consideration of fundamental advances that are not pertinent to practical applications. These advances may prove the most enduring legacy of research in xenotransplantation, but they distract from addressing such questions as what advances are needed for application xenotransplantation in preference to alternative therapies. We emphasize the evolution of technologies of genetic engineering, as they might be applied to the animals that likely would serve as sources of xenografts. We do so not to focus on or endorse particular genetic manipulations but rather to provide a sense about how problems yet undetected will be approached, that is, by modifying source of a graft in preference to treating a patient. We also focus on the ways animal models provide sound inference about what would occur if the tissue or organ from an animal were transplanted into a patient and on some of the ways animal models might misinform about the likely outcome of xenografts.

Definitions

Xenotransplantation refers to the deliberate transfer of living cells, tissues, or organs from individuals of one species to individuals of another. Cells, tissues, or organs so transferred are called xenografts.⁴ Allotransplantation/allograft refers to transfers between individuals of the same species. An isograft refers to a transplant between genetically identical individuals, an autograft to a transplant from an individual to itself. The term “heterograft” was previously used to refer to xenografts and the term homograft to allograft, but in some instances heterograft was applied to both xenografts and allografts. Today, xenotransplantation is usually taken to refer more narrowly to the transplantation of cells tissues or organs from animals into humans or to animal models that represent such clinical transplants. There is also interest in what we shall call “reverse xenotransplantation” (Figure 1) or the transplantation of human cells and tissues into animal hosts for the purpose of either expanding the human cells or developing humanized tissues and organs. The term xenotransplantation is not usually applied to accidental exchanges of living cells between species or to infestations by parasites. Xenogeneic cells introduced into animals to investigate properties of the transferred cells (eg, cancer cells implanted in immunodeficient mice) and devitalized structures such as xenogeneic heart valves are sometimes called xenografts, but these applications are not commonly taken to represent xenotransplantation. Here we shall summarize the current understanding of the biological barriers to xenotransplantation of cells, tissues, and organs of animals into humans and the extent to which these barriers are represented in models commonly used today. We will also briefly consider reverse xenotransplantation as it applies to large animals models.

Figure 1.

Reverse xenotransplantation. The term “reverse xenotransplantation” is used to refer to the transplantation of human cells into animals. Reverse xenotransplantation has been explored as an approach that might be used to expand a population of mature human cells or to coax the differentiation of human stem cells to generate mature human cells, or a tissue or organ for transplantation as an autograft into the individual who provide the original human cells. The figure illustrates several examples of reverse xenografts. As one example, mature cells such as fibroblasts might be harvested from an individual with organ failure. The fibroblasts would be converted to induced pluripotent stem cells. These stem cells would be treated to begin tissue or organ-specific differentiation and then transplanted into a mature pig or the undifferentiated stem cells might be transplanted into a fetal pig. In the mature or fetal pig, the stem cells would undergo further differentiation and begin organogenesis. Depending on the organ or tissue needed, the maturing human cells, tissue, or primordial organ would be harvested from the pig and then implanted into the individual with organ failure. Reverse xenotransplantation might offer biologically more efficiently and less costly ways to use stem cells for replacement of tissues and organs. Not illustrated in the figure but discussed in the text are various genetic changes that might be introduced in pigs to facilitate engraftment and differentiation of human cells.

Insights from Early Experiences in Xenotransplantation

Clinical xenotransplantation (and allotransplantation) of skin has been performed throughout history. The skin grafts were usually to provide covering of burn or traumatic wounds (to prevent excess loss of water, scar formation, etc.).^5,6 Skin allografts taken from amputated limbs or cadavers were found to remain in place for days to weeks but most often failed with time. Skin xenografts, sometimes used when human skin was not available, were usually reported to behave like allografts, effectively covering wounds for days and sometimes weeks but also ultimately failing.^5,6 Although some claimed that xenografts were comparable to allografts, the impression emerged that xenografts were less enduring than allografts and allografts between unrelated individuals were less enduring than allografts between closely related individuals. In contrast to allografts and xenografts, autografts usually survived permanently. Microscopic examination of skin xenografts, allografts, and autografts also suggested relatedness of the transplant and the recipient determined histologic integrity.^7,8 These experiences, however, did not deter use of xenogeneic skin as temporary covering for wounds in the past and proposals for such use today.^9,10

The experience in transplanting organs within and between species was quite different. Development of techniques to allow the surgical joining of the cut ends of blood vessels (the vascular anastomosis) sparked attempts to transplant intact organs.^11,12 The first “successful” vascularized kidney allografts were performed in dogs in 1905.^13,14 These successes led almost immediately to several attempts to use the technique to treat patients with kidney failure. Because it was not then clear that human organs could be obtained even from deceased individuals (because some reasoned that the presence of living cells during the hours after death precluded ethical harvesting of human organs), the first clinical kidney transplants were performed using kidneys harvested from pigs and sheep.^13,15 One of these first clinical xenografts did not function, the other issued a few drops of urine and then it too ceased to function. These results and presumed failure of a clinical kidney allograft widely reported to have been performed¹⁶ discouraged all but a few experimental attempts at clinical kidney transplantation. In the 1960s, when immunosuppressive agents had been developed, transplantation resurfaced as a potential approach to treatment of failure of the kidneys, liver, and heart.^17,18 In that era, as before and since, availability of human organs was considered the preeminent limitation to the application of transplantation for treatment of disease. On a few particularly urgent settings, animals—monkeys or chimpanzees—in lieu of humans were used as the source of organs for transplantation.^19,20 With the recipients receiving immunosuppressive agents then available, most clinical kidney xenografts from chimpanzees functioned for ~2 months and one functioned 9 months; clinical kidney xenografts from baboons functioned days to weeks (Table 1).

Table 1.

Experience in Clinical Xenotransplantation of the Kidney^a

Source (Author)	Number	Outcome	Reference
Chimpanzee (Reemtsma)	12	1 immediate failure 11 function 2–9 months Infection, not rejection, caused most deaths	294
Baboon (Starzl)	6	Function 10 days–2 months 2 functioning but failing xenografts removed when allografts available; 2 fully  ceased function; 2 rejected; 2 regrafted, the regrafts failing at death from sepsis.  1 death from pneumonia, 1 from multiple pulmonary emboli	20

^aAdapted from¹²⁷ and references listed.

The early experiences in experimental and clinical xenotransplantation within and between species fueled some controversies that are still unsettled and pertinent for models and potential clinical applications of xenotransplantation today. One controversy concerned the cause of graft failure. Some believed allografts and xenografts, particularly cancers but also normal tissues, evoke immune responses that destroy the transplants.^21,22 Others believed that biochemical incompatibilities between individuals within a species and between different species, but not immunity per se, cause the failure and destruction of grafts.⁸ Today we understand that immunity was then and is still the main obstacle to successful transplantation between different individuals, and the clinical practice of transplantation today is predicated on the continuous provision of immunosuppressive agents (and on availability of antimicrobial agents to address toxicities imposed by immunosuppression).

Although immunity is the most important barrier to successful transplantation, it is not the only barrier. Despite the availability of powerful and highly effective regimens of immunosuppression, up to one-half of all allografts ultimately fail over time. Which grafts are likely to fail and why some fail and some persist are subjects of intense research. One possibility is that the diversity of individuals within species and between species creates incompatibilities that are not amenable to immunosuppression, and it is these that determine the fate of grafts. Below we shall discuss emerging evidence that when immunosuppression is optimized, properties of organ transplants other than antigens and properties of recipients other than the capacity to respond to antigens may determine whether and how well an organ transplant functions. Modeling these determinants especially in large animals poses a considerable challenge but also an opportunity because it presently represents the main cause of failure of grafts.

Rational and Applications for Xenotransplantation

Organ Failure

Transplantation is the preferred treatment for severe failure of the heart, kidneys, liver, and lungs. Although organ transplantation can dramatically reverse the pathophysiology of organ failure, the impact of organ transplantation on public health is limited by a severe shortage in the supply of human organs available for transplantation (Figure 2). The limited supply of human organs and tissues for transplantation remains the preeminent rationale for developing xenotransplantation as an alternative to allotransplantation. Advances in therapeutics and preventative medicine might decrease the incidence of organ failure and lessen the demand for organ transplantation for a period of time after advances are introduced. However, advances in medicine that affect longevity are likely to eventually increase the prevalence of organ failure owing to increased prevalence of diseases of aging. For example, increased attention to blood pressure, cholesterol, and lifestyle and the advent of statins undoubtedly helped to limit the prevalence of cardiac disease and accentuated the relative contribution of cancer among causes of death.²³ But, as advances in cancer treatment further increase longevity, heart and kidney failure will take on renewed significance. Accordingly, we speculate that advances in medicine and public health ultimately increase the prevalence of diseases of aging, including failure of the heart and kidneys, and hence the potential impact of xenotransplantation.^24,25 Xenotransplantation might also find favor in cultures that eschew organ donation and in areas that lack the infrastructure needed to support use of artificial organs. We can also imagine that lower costs we expect to be associated with xenotransplantation could fuel some demand.

Figure 2.

The shortage of human organs for transplantation. Displayed are the number of persons on the waiting lists for transplantation and the number of transplants performed in the United States during the time periods shown. The data are drawn from 2016 Annual Data Report of the Scientific Registry of Transplant Recipients http://srtr.transplant.hrsa.gov/annual_reports/Default.aspx.

Models for Evaluating Impact of Transplantation for Organ Failure

Organ failure significantly affects the outcome of clinical transplantation, increasing the risk of infection, early graft failure, and other complications. Unfortunately, few if any of the preclinical (ie, large animal) models used to investigate transplantation faithfully represent this impact. Generally, healthy animals with or without acute organ failure are used to represent conditions that over periods of months or years eventuate in organ failure. These relatively healthy recipients of organ transplants bypass comorbidities, such as atherosclerosis, chronic changes in vascular resistance, kidney disease, autoimmunity, cancer, etc. that limit the success of organ allotransplantation. However, the limitations of animal models used to study application of allotransplantation for treatment of organ failure have not hindered preclinical testing of novel drugs and regimens for clinical allotransplantation, and there is no reason to think the experience in xenotransplantation will differ. However, the absence of models representing acute and chronic failure of the liver and insulin dependent (type 1) diabetes have slowed development of therapies in general and could particularly limit testing the efficacy of xenotransplantation as a treatment for these conditions. For example, after experimental xenotransplantation of the liver, incompatibilities between the complement and coagulation systems of the liver of the foreign species appear to amplify (rather than resolve) insufficiencies in these systems caused by liver failure. As a result, xenotransplantation of the liver in animal model systems is quite difficult physiologically. However, patients with liver failure often have baseline insufficiencies of complement and coagulation systems, and treatments used to secure survival of experimental xenografts could be more toxic than experimental work would suggest. In treatment of autoimmune diabetes, xenogeneic islets conceivably could pose a lower or higher hurdle to success; if residual “autoimmunity” did not target xenogeneic islets the hurdle would be less, if heighted inflammation associated with xenotransplantation amplifies autoimmunity the impact could be greater if epitopes were similar between species. In the absence of suitable models, the impact of xenotransplantation on liver failure and diabetes might thus be difficult or impossible to predict.

Preemptive Transplantaion

Rapid advances in diagnostics, including molecular profiling, genomics, and molecular imaging, expand the opportunities to detect disease before clinical manifestations appear and to identify individuals at high risk for development disabling or lethal disease. These diseases include cardiac malformations and arrhythmias, inherited defects and deficiencies of metabolic pathways of liver and other organs, and cancer of various types. Identification of individuals with incipient or early-stage disease encourages consideration of preemptive therapies, including transplantation. The benefits versus risks of early diagnosis and the weighing of preemptive therapy versus “watchful waiting” are topics of great interest in medicine. Although preemptive transplantation is practiced,^26,27 practice and investigation of benefits versus risks is limited for the most part to kidney transplantation for which living donors can provide organs.^28,29 Obviously, xenotransplantation would make it possible to introduce preemptive transplantation of other organs and other settings. For investigation of xenotransplantation as a preemptive therapy, physiologically normal recipients, such as those used today, likely provide a reasonable model.

Metabolic Disease

Transplantation of the liver, hepatocytes, pancreas, or islets is performed to correct metabolic diseases. Investigation of xenotransplantation for these conditions has focused mainly on immunological hurdles, and for that purpose physiologically normal recipients provide a reasonable model. Some metabolic diseases have been modeled in mutant mice; however, weighing the potential efficacy versus risks of allo- or xenotransplantation versus other therapies requires development large animal models.

Genetic Engineering for Xenotransplantation

Rationale

One important and sensational rationale for xenotransplantation and reverse xenotransplantation (Figure 1) is the opportunity to engineer the genome of the animal used as the source of the transplant or the host for human cells. Genetic engineering of pigs was first proposed for suppression of complement-mediated injury^30,31 and later for eradication of antigen.³² The first transgenic pigs generated for this purpose expressed human complement regulatory proteins at low levels but still evaded the immediate complement-mediated injury thought to preclude clinical xenotransplantation.³³ During the 20 years since then, genetic engineering of pigs has been appreciated as a key strategy for advancing xenotransplantation toward clinical practice (see Tables 2 and 3 and³⁴ and¹ for examples). Genetic engineering of the sources of xenografts potentially decreases the need to administer toxic agents to recipients and, if modifications are stably represented in the germline, allows the extension of favorable characteristics by breeding rather than by manipulation of individual animals. Before the rationale for specific manipulations of the genome is discussed, it is helpful to consider some merits and limitations of approaches used to modify the genome of large animals that could be used as sources of xenografts or as hosts for human cells (Table 3).

Table 2.

Some Outcomes of Experimental Pig Organ Xenografts in Nonhuman Primates^a

	Target of Genetic Modification			Outcome (Survival)	Reference
	Ag	C Reg	Coag & Hemost Reg
Heart Xenograft (n)
6	α1,3GT KO	Hu CD46	Hu TM	159–945 days	295
5	α1,3GT KO	Hu CD46		42–236 days	295
8	α1,3GT KO			23–179 days	296
Kidney xenograft (n)
1	α1,3GT KO	Hu CD46 Hu CD55	Hu TM EPCR CD39	136 days	297
5	α1,3GT KO	Hu CD55		6–133 days	298
5				24–229 days	299
7	α1,3GT KO			18–83 days	300

Abbreviations: Ag, antigen; C, complement; Coag & Hemost Reg, coagulation and hemostasis regulation; α1,3GT KO, α1,3-galactosyltransferase knockout; Hu, human; TM, thrombomodulin transgenic; EPCR, endothelial protein c receptor transgenic.

^aThe table shows results from the references cited. Most recipients were baboons. Recipients received various regimens of immunosuppression, some designed to induce tolerance. Some recipients were treated with cobra venom factor to inhibit complement. The results should not be taken to indicate the genetic modifications were mainly responsible for the results but rather to indicate range of responses observed. The significance of this range is discussed in the text.

Table 3.

Approaches to Genetic Modification of Animals for Xenotransplantation

Method	Target Cell	Selectable Marker	NHEJ	HDR	Reference
Pronuclear injectiona	Zygote	No	0.9%	No	301
Random insertion and SCNT	Somatic cell	Yes	10−3–10−4	No	302,303
Conventional HR and SCNT	Somatic cell	Yes	10−3–10−4	10−5–10−7	46,73,304
Gene editing and SCNT	Somatic cell	No	1–50% MA1–30% BA	2–5%	73,305 b
Gene editing and direct embryo injection	Zygote	No	10% MA 100% BA	3–80%	76,78,306

Abbreviations: BA, biallelic; HDR, homology directed repair; HR, homologous recombination; MA, mono-allelic; NHEJ, nonhomologous end-joining; SCNT, somatic cell nuclear transfer.

^aEfficiency per embryo injected and transferred (combination of 20 projects).

^bOf the many manuscripts in this area, those selected report results in multiple loci using multiple targets/loci and as such represent what can be expected.

Approaches to Genetic Engineering of Large Animals

Genetic engineering of pigs for xenotransplantation initially relied on pronuclear injection of DNA constructs in early zygotes and was restricted to gain-of-function modifications (see³⁵ for review). These approaches were costly and inefficient and could not be used for targeted inactivation of genes. Thus, although complement might be suppressed by expressing heterologous complement regulatory proteins, suppression of antigen production depended on expression of proteins that could hinder (via competition for substrate) synthesis of the carbohydrate of interest.³⁶

Still, the possibility of directly targeting the synthesis of antigenic targets was enabled when the seminal work of Smithies and Cappechi^37,38 proved homologous recombination could introduce mutations in precise regions of the genome and set the stage for gene targeting. This advance and successes in generating gene “knock out mice” sparked the first proposals to target the enzyme responsible for the synthesis of the carbohydrate antigen that had been identified as the initial target of immunity in xenotransplantation.^32,39 However, the low efficiency of homologous recombination precluded targeting of genes in mature animals or embryos. One potential avenue to targeting of genes in animals was to perform gene targeting and selection in embryonic stem (ES) cells in culture and then introduce the manipulated ES cells into primitive embryos, that is, generating germline chimeras, some of the offspring of which transmit the trait to subsequent generations.^40,41

Availability of ES cells of mice enabled the generation of lines of gene-targeted mice that have played an essential role in biomedical research. The advances in mice spurred efforts to generate ES cells that could be used for gene targeting in large animals, especially pigs.^42,43 However, despite over 20 years of research in many laboratories worldwide, no ES cell line that could be used for generating gene-targeted pigs was found. As a result, generation of complex transgenic pigs for xenotransplantation was slow and limited to a few research groups.

In 1997, however, Wilmut and Campbell⁴⁴ reported that nuclei of somatic cells from sheep removed and inserted into an enucleated egg underwent full reprogramming and could generate a living animal (Dolly), the cells of which, including the germ cells, had the chromosomal DNA of the somatic cell. Thus, somatic cell nuclear transfer (SCNT) could generate animals, cloned from a mature cell, and genetic modification of animals might be undertaken without ES cells or the inefficiencies of microinjection of DNA.

This approach was soon applied to other mammalian species, including swine.⁴⁵ The ability to generate offspring from somatic cells meant that ES cells could be bypassed and living animals generated after genetic modification of the somatic cells in vitro. SCNT thus had a major impact in pig transgenesis and xenotransplantation because it enabled the generation of the first α1,3-galactosyltransferase knockout pigs.⁴⁶ The combination of conventional homologous recombination and SCNT allowed the generation of multiple transgenic pig lines (reviewed in⁴⁷); however, the low rate of recombination in somatic cells^48,49 limited the progress that could be made in developing complex transgenic animals.

The application of zinc finger nucleases (ZFN), transcription activator-like effector nucleases (TALENs) and clustered regularly interspaced short palindromic repeat (CRISPR)-Cas9 to gene editing in cultured cells provided the efficiency and specificity needed to generate complex genetic changes. The 3 systems increased rates of targeted modification several orders of magnitude beyond conventional homologous recombination. Even biallelic inactivation and targeted insertions/gene replacements were now achievable at high efficacy (reviewed by^50,51). The frequency in both cases can range between 10% and 80%, making identification of the correct event a simple task. With these tools, multiple groups have now reported the ability to simultaneously generate mutations in more than one locus.^52–55 These technologies also allow gene replacement and knock-in (placing a gene into a preselected genomic region).^56,57 CRISPR-Cas9, in particular, has shown wide applicability and ease of use.⁵⁰ Initial concerns regarding high frequency of off target effects (OTE) persist but may be addressed in part by generation of Cas9 enzymes with greater fidelity⁵⁸ and in part by improvement in approaches to detecting OTE.⁵⁹ Still, the impact OTE on the functioning of organ xenografts could be subtle, and the possibility should be considered when genetic manipulations fail to achieve expected improvements in outcome, as later discussed.

Gene Editing Applied to Pigs

The gene editing technologies have been applied to pigs. The initial gene-edited pigs were generated using ZFN,^60,61 including development of IL2RG KO pigs,⁶² but high costs and complex rules of assembly and target selection limited its wide applicability. The advent of TALENs^51,63 and novel assembly methods⁶⁴ enabled rapid application of the technology to pigs.^65–69 However, CRISPR-Cas9, with its simplicity of use and lower costs, rapidly eclipsed ZFN and TALENs as the method of choice for generating transgenic pigs. Since then, multiple gene-edited pigs have been generated using CRISPR-cas9 combined with SCNT.^70–74 A recent search of Pub Med yielded >60 reports, including several for xenotransplantation.⁷⁵

The CRISPR-Cas9 system is now being used to rapidly modify pigs by direct injection in zygotes. For reasons still unclear, the efficiency of gene editing in zygotes is even higher than the efficiency in somatic cells, sometimes yielding frequencies of 100% bialleic modification and even multi locus modification.^76–78 Although zygotic injection results in very effective gene editing, the use of SCNT makes it possible to carry out multiple rounds of mutations in pigs without the need for breeding and increases the ability to generate multi-transgenic animals carrying both gene inactivation and gene replacements. Use of SCNT drastically reduces generational intervals and costs associated with breeding multi-transgenic animals where independent segregation can lead to complex litters. In pigs, genetically modified fetuses obtained at day 32–42 of gestation can be used for the next round of modification. We have successfully performed 3 sequential SCNT rounds yielding viable offspring (J. Piedrahita, unpublished observations), and other groups have performed 6 rounds in cattle and 25 rounds in mice.^79,80 Thus, the combination of CRISPR-Cas9 gene editing of somatic cells, direct injection into zygotes and SCNT allows the rapid and efficient generation of essentially any genetic modifications needed in pigs.

Reverse Xenotransplantation

Rationale

Xenotransplantation of human cells to animals (Figure 1), “reverse xenotransplantation,” has been envisioned as a way to generate and expand human cells, tissues, and organs for transplantation.^81–83 Reverse xenotransplantation offers certain obvious advantages, histocompatibility, and physiologic compatibility of the human cells upon return to the stem cell source. As we envisioned it, stem cells from an individual needing transplantation (or stem cells generated from differentiated cells or nuclei that after transfer to an egg would be reprogrammed to yield stem cells) might be introduced into fetal animals, the microenvironment of which would coax differentiation or development into mature immunodeficient animals and in these environments the human stem cells might differentiate and grow to human primordia. The primordial could be harvested and implanted in the individual from whom the stem cells were generated and undergo organogenesis. For some purposes, for example, to generate hepatocytes, islets, or hematopoietic cells, the human stem cells would be allowed to fully differentiate in the animal host whereupon the mature cells or tissues could be harvested and transferred to the patient. Generation of human stem cells, transfer to fetal animals, and differentiation and development in the xenogeneic hosts has been accomplished and extensively studied in mice (humanized mice).^84–87 Although essential to progress in many fields, humanized mice will not be considered here. We shall consider the status of transplanting human cells into large animals (humanized pigs) for generation of sizable masses of human cells, tissues, and organs that for clinical purposes might be transplanted into humans.

Seminal work in sheep demonstrated that human hematopoietic stem cells administered to the fetus establish multi-lineage hematopoietic chimerism.^88–90 However, reverse xenotransplantation in pigs would offer certain advantages, including cost, multi-parity, and the large body of knowledge regarding biological barriers to engraftment. However, experience in human-pig reverse xenotransplantation is quite limited. Human stem cells transferred to fetal pigs have been shown to contribute to formation of some nephrons in kidney, segments of skin, and to the thymus⁹¹ and hematopoietic system.^91–93 Mature pigs generated from these fetuses have some human T cells selected and matured in the chimeric thymus that can generate human restricted responses to antigen.⁹³ The potential of reverse xenotransplantation to address clinical problems, however, remains to be determined. Application could well depend on optimizing the sources and types of human cells, the approach to delivery (eg, devising approaches to deliver cells to mature rather than to fetal pigs), and on minimizing hurdles posed by immunity and biological incompatibility. Below we discuss some of the model systems in which progress is being made.

Intra-Uterine Stem Cell Transplantation (IUSCT) to Achieve Xenogeneic Engraftment

The first approach used for delivery of human cells to animals involved IUSCT. Introduction in utero averts rejection and provides a more nurturing microenvironment. Sheep were initially preferred as hosts for IUSCT because the fetus would tolerate manipulation and the size made surgical intervention easier. Transplantation of allogeneic hematopoietic stem cells early in gestation of wild-type sheep fetuses yielded sustained multi-lineage hematopoietic chimerism.^90,94 The approach was used for other types of stem cells.^95–97 However, engraftment was quite low (<1%), making this model impractical.^88,89

As mentioned, size, anatomic, genetic, and physiological similarity to humans and extensive information already assembled about zoonosis and compatibilities make the pig a more attractive host for IUSCT intended to have clinical applications. Thus, others⁹² and we^93,98 successfully performed IUSCT, introducing human hematopoietic stem cells in porcine fetuses and detecting mature progeny years after birth. The introduction of human stem cells in the porcine fetus facilitated development of tolerance by the host.⁹³ Still application of IUSCT in pigs was limited by the low level of human cell engraftment.

Although the IUSCT host animals exhibited immune tolerance, engraftment is potentially limited by innate immunity (NK cells), a niche that is less than optimally supportive of xenogeneic cells and incompatibility of growth factors between species. Some of these hurdles can be overcome by delivering more human cells to the fetus, by administering human growth factors with the transplant, and/or by depleting some of the porcine cells that compete with the transplant for growth factors. In mice, success has been most readily achieved by genetic engineering (see^99–101 for review), and that is the approach others and we have pursued in pigs.

Genetic Engineering to Achieve Xenogeneic Engraftment

An absolute requirement for engraftment of human cells in other individuals of the same or disparate species is suppression or elimination of adaptive immunity. This barrier was minimized by using the fetus at a gestation that precedes development of mature lymphocytes, particularly T lymphocytes, as a recipient of foreign cells, introduced by IUSCT, as described above. However, to avoid IUSCT, foreign cells could be introduced into animals that were immunodeficient. For decades mice with naturally arising immunodeficiency, such as nude mice, have been used to harbor and study malignant human cells.¹⁰² However, full and enduring engraftment of normal cells was never achieved. Hence, with the advent of genetic engineering, efforts were made to more completely remove immune barriers to engraftment posed by NK cells and adaptive immunity. The greatest success achieved by targeted disruption of IL2RG and either RAG-1 or RAG-2 (see¹⁰³ for review). In addition to averting innate and adaptive immunity, optimal and enduring engraftment of human cells in mice was achieved by providing human growth factors (eg, IL-3, hM-CSF, GM-CSF, thrombopoietin) and a phagocytosis suppressor SIRP-α (see¹⁰⁴ for review) and by limiting the competition of murine stem cells (see¹⁰⁵ for review) or mature cells.¹⁰⁶

For reasons discussed above, we and a few others have begun to use genetic engineering to generate immunodeficient pigs as potential hosts for reverse xenografts. Transgenic IL2RG^−/y pigs exhibit some features of X-linked severe combined immunodeficiency syndrome, including marked decreases but not complete absence of T cells and NK cells in peripheral blood and spleen (~2.3% of normal) but normal B cell numbers.^62,107 The pigs accept grafts of semiallogeneic but not human hematopoietic stem grafts and therefore are not likely to prove useful for reverse xenotransplants. RAG-1^−/− and RAG-2^−/− transgenic pigs have a hypoplastic thymus and significantly decreased numbers of T cells and B cells in the circulation and in spleen, although some CD3 + cells, likely NK cells, are detected in spleen.⁶⁸ Biallelic RAG-2^−/− pigs have been reported to have a phenotype similar to that of pigs deficient in both RAG-1 and RAG-2 and to accept transplants of human induced pluripotent stem cells, developing teratomas, and transplanted allogeneic trophoblast cells.¹⁰⁸ Whether the pigs would accept normal cells remains unknown. Pigs with targeted biallelic disruption of genes encoding RAG-2 and IL2RG have been reported.⁷⁸ As might be expected, the pigs have a ~100-fold decrease in circulating T cells and B cells but a small decrease in NK cells, reflecting some residual IL2RG function and inability to clear norovirus. Whether the pigs accept foreign grafts is unknown.

We have generated pigs with targeted disruption of RAG2, RAG1, and IL2RG (J. Piedrahita, unpublished observation). The pigs accept allogeneic stem cells and in so doing reconstitute the immune system. The pigs also accept xenogeneic cells; however, our experience indicates, perhaps not surprisingly, that hurdles beyond innate and adaptive immunity limit xenogeneic engraftment. We expect advances in gene editing discussed above will allow us to overcome this limitation in the near future.

Animal Species as Sources of Xenografts

Nonhuman Primates

When transplantation was introduced into clinical practice at a few academic centers and donated organs were scarce, xenotransplantation was seen as a reasonable alternative “in certain rare circumstances”¹⁷ and nonhuman primates, because of taxonomic and physiologic proximity to humans, were used as the source of most organs used for clinical xenografts.¹⁹ Nearly all of the xenografts functioned at least briefly, but none provided enduring support and all patients died either because of infection or rejection of the transplant. The results of some renal xenografts from nonhuman primates to human patients are summarized in Table 2.

Certainly better results and perhaps enduring function could be achieved today. Yet, nonhuman primates have been excluded as potential sources of organs in part for reasons of ethics, but especially because nonhuman primates are too scarce to have any meaningful impact on the shortage of human organs. There is also concern that transplantation might convey lethal infection. Furthermore, although tissue physiology of nonhuman primates may resemble that of humans, the smaller size of chimpanzees and monkeys limit the physiologic impact the organs would have as xenografts in mature humans. On the other hand, nonhuman primates are commonly used to model human xenograft recipients, as discussed below.

Pigs

During recent decades the pig has received universal acclaim as the preferred source of xenografts.^30,109,110 Pigs are plentiful enough to fulfill any conceivable need. Early in life the size of pigs overlaps with human. Pigs can be genetically engineered and owing to sizable litters, readily bred, as described below. Because pigs have long existed in proximity to humans, the susceptibility of infectious diseases and potential for transmission to humans is understood well enough to formulate detailed approaches to screening and prevention.^111,112 As discussed below, experience and investigation have also tempered some concerns that use of pigs in xenotransplantation might generate exotic microorganisms.³

Because present interest focuses almost exclusively on pigs as sources of tissues and organs for clinical xenotransplantation, modeling of clinical xenotransplantation today also generally uses pigs as a source and primates as recipients. Therefore we shall focus mainly on xenografts in which pigs are used as a source. Still, experimental xenografts between various combinations of species (eg, guinea pig-to-rat, rat-to-mouse, pig-to-dog) have contributed to the body of knowledge about xenotransplantation. Where appropriate, we shall refer to these models without offering detailed review.

Biological Barriers to Xenotransplantation

Introduction

The biological barriers to xenotransplantation include the immune response of the recipient against the graft, physiological and biochemical incompatibility between the graft and the recipient, and the potential for transmission of infection between the graft and the recipient and the consequences thereof including potential generation of novel microorganisms.^113–116 Although typically these barriers are investigated independently, sometimes using divergent models, the elements of the barriers intersect in origin, pathogenesis, and manifestations (Figure 3). For example, ischemia-reperfusion injury associated with transplantation of a pig organ into a nonhuman primate incites activation of complement, the control of which is thwarted by incompatibility of complement regulatory proteins.¹¹⁷ At the same time natural antibodies of nonhuman primates directed against Galα1-3Gal³⁹ and antibodies others directed against neoantigen on ischemic cells¹¹⁸ increase the extent of complement activation,¹¹⁹ which in turn amplifies B cell¹²⁰ and T cell^121,122 responses to foreign antigens. The inflammatory and immune environment associated with ischemia-reperfusion injury and innate immunity modifies the physiology of parenchymal cells and endothelium, potentially effacing control of viral latency^123,124 but also potentially circumscribing infectious agents in the recipient or carried with graft.^125,126 The intersection of immunity, physiologic incompatibility, and infection underscore the importance of taking account of the limits of simplified experimental systems, such as cell cultures and small animals, in predicting the impact of the various barriers as they would be manifest in swine-to-human transplants.

Figure 3.

Integration of pathogenesis of ischemia-reperfusion injury with pathogenesis injury caused by xenoreactive antibodies and xenoreactive phagocytes. Ischemia reperfusion injury and innate immunity directed at xenografts converge to amplify complement-mediated early graft injury. Activation of complement (C) by ischemia (through several pathways) and/or by xenoreactive antibodies increases the amount and kinetics of membrane attack complex (MAC) assembly, which increases membrane injury. C activation generates C3a and C5a, which activate leukocytes and C3bi, thereby tethering phagocytes to endothelium. C activation also causes smooth muscle contraction, decreasing blood flow and increasing the extent and duration of ischemia. C activation causes shedding of heparan sulfate (HS) from cell surfaces, compromising barrier functions; impairing regulation of complement, coagulation, and platelet activation; and hindering control of oxidants. Leukocytes, platelets, and endothelial cells release proteases that expose neoantigen on endothelium, heightening the reaction. Ischemia and innate immunity also amplify adaptive immunity (not shown).

Immunity as a Barrier

The immune response of the recipient to a xenograft has been considered the most daunting barrier to xenotransplantation. The importance of the immune barrier emerged from repeated failures to achieve permanent engraftment organs from between disparate species, such as pig organs in nonhuman primates.¹²⁷ Recent reports of long-term (>1 year) survival of some heterotopic cardiac xenografts and kidney xenografts might suggest, however, that the hurdle posed by immunity can be overcome and clinical trials might soon commence.^1,3,128 These promising results were achieved, however, using immunosuppressive agents and regimens more severe than those typically used for clinical transplantation. We briefly discuss some aspects of the immunology of xenotransplantation pertinent to animal models currently used and some of the limitations inherent in those models. More detailed reviews of the immune response to xenotransplantation can be found in other publications.^129–131

Innate Immunity

Xenotransplantation potentially recruits every facet of innate immunity through the response to ischemia-reperfusion injury and through recognition of the graft as “foreign by natural antibodies, complement, and phagocytes (Figure 3). The 2 processes are truly synergistic because each compromises resistance to the other and each promotes smooth muscle contraction, amplifying ischemia (Figure 3). Both are further amplified by relative ineffectiveness of controls of complement, coagulation, and platelet activation. Hence the barrier posed by innate immunity is particularly significant in xenotransplantation.

Complement

The most dramatic example of dysregulation of innate immunity is the early pathogenic impact of activation of the complement system. Complement can be activated by one or more of the several distinct initiating mechanisms: (1) the classical pathway, typically initiated by binding of complement-fixing antibodies; (2) the alternative pathway, typically initiated generation of C3b and association with factor B exceeds the control exerted by circulating (factor H) and membrane associated (CD46) complement regulators; (3) the lectin pathway, initiated by the binding of mannose binding lectin or ficolin with mannose-binding lectin associated serine proteases 1, 2, and/or 3; and (4) the properdin pathway, initiated by the binding of properdin directly to a target. These canonical pathways belie a much larger number of mechanisms that can initiate the complement cascade (eg, antibodies bound to a surface can activate the alternative and/or properdin pathway, and C1q can attach directly to injured cells).

Upon reperfusion of organ xenografts (or introduction of xenogeneic tissue into blood^132–134) ischemia-reperfusion injury and binding of xenoreactive antibodies activates the complement system. The extent and the kinetics of complement activation are governed at key steps by regulatory proteins in blood, such as factor H; on cell membranes, such as CD46, CD55, and CD59;^135,136 and by the condition of cell surfaces.¹³⁷ Some complement regulatory proteins may function more effectively in homologous than in heterologous systems,^{30,117,138,139} although some have challenged this concept based on work using isolated cells.¹⁴⁰ Although challenges to the concept of homologous restriction of complement are welcome, heterologous complement has always appeared more active than homologous complement cell lysis assays.¹⁴¹ More to the point, observation obtained over decades in transplanting organs between various combinations of disparate species (and hence the principle of in vivo veritas) provides compelling support for the importance of species specificity of complement regulation. For example, heart and kidney xenografts from pigs into unmanipulated nonhuman primates invariably undergo hyperacute rejection triggered by anti-Galα1-3Gal antibodies, the concentrations and functions of which resemble isohemagglutinins,¹⁴² leading to activation of complement. In contrast, ABO-incompatible allografts rarely undergo hyperacute rejection.¹⁴³ Consistent with this concept, expression of small amounts of human complement regulators in transgenic pigs can prevent this type of rejection,³³ and pigs developed as sources of organs for xenotransplantation often incorporate one or more transgenes for expression of such proteins.³⁴

Perhaps more important than the cell-associated complement regulators is factor H, a plasma protein that regulates the alternative pathway of complement (by facilitating dissociation C3bBb complexes and by acting as a co-factor for factor I-mediated cleavage of C3b). To exert its function, factor H attaches to acidic moieties on cell surfaces (eg, heparan sulfate and sialic acid), and it is the interaction with cells surfaces that may limit the activity of the protein on foreign surfaces.¹⁴⁴ If factor H fails to control complement on surfaces (eg, rat factor H fails on guinea pig cell surfaces), immediate and severe complement-mediated injury, that is, hyperacute rejection, ensues.¹⁴⁵ Fortunately, human (and nonhuman primate) factor H regulates human complement on porcine cells and hence this limitation is not often discussed. However, factor H might sterically compete with properdin (a protein that promotes the alternative complement pathway)^146,147 and other proteins for binding to cell surfaces, and it is conceivable that novel mixtures of proteins in xenograft recipients could hinder the regulation of complement.

Natural Antibodies

The natural antibodies of greatest interest in xenotransplantation are natural antibodies specific for Galα1-3Gal.^148,149 Galα1-3Gal is the product of a galactosyltransferase (α1,3-galactosyltransferase) that is produced by New World monkeys and lower mammals, including the pig, but not by humans, apes, and Old World monkeys.¹⁵⁰ Mammals lacking Galα1-3Gal produce natural antibodies specific for that saccharide, much as humans lacking blood group substances A or B produce isohemagglutinins directed at the corresponding substances.¹⁴² When a porcine organ is transplanted into a nonhuman primate with natural antibodies specific for Galα1-3Gal, the binding of those antibodies triggers immediate, complement-mediated rejection of the organ¹¹⁹ and (if immediate rejection is avoided) antibody-mediated rejection (also called acute vascular rejection).¹⁵¹ The importance of antibodies specific for Galα1-3Gal in xenotransplantation led to the generation of pigs with targeted (α1,3-galactosyltransferase) (Gal KO pigs).^152,153 It should be noted, however, that the ability of anti-Galα1-3Gal to trigger immediate rejection and even antibody-mediated rejection depends very much on failure of complement regulation, as the presence of even low level of human complement regulatory proteins in a xenograft thwarts immediate rejection³³ and temporary removal of natural antibodies against blood groups prevents hyperacute and antibody-mediated rejection.¹⁵⁴

Human natural antibodies against structures other than Galα1-3Gal might initiate rejection of xenografts.¹⁵⁵ The significance of these natural antibodies in xenotransplantation remains uncertain because the antibodies have been studied in systems in which antibody interaction with Galα1-3Gal cannot occur (eg, when Gal KO organs are transplanted into nonhuman primates). Some of the antigens might be “neoantigen” produced in the absence of α1, 3-galactosyltransferase and some antigens might be recognized by natural antibodies of only a fraction of nonhuman primates and humans. Therefore, it is difficult to know a priori whether targeting of the corresponding glycosyltransferases will confer more benefit than harm.

Still another type of natural antibody, the polyreactive antibody, could have an effect on xenotransplants. Polyreactive antibodies recognize multiple antigens, as the name indicates, including autoantigens and are produced by a distinct subset of B cells.^30,156 Polyreactive antibodies have been implicated in the pathogenesis of ischemia-reperfusion injury.^118,157 In this setting, the antibodies can attach to neoantigen formed by degradation or oxidation of normal molecules or to antigens exposed by injury of cell membranes. Polyreactive antibodies also can initiate the repair of injured cells and tissues^158–160 and thus potentially benefit a transplant.¹⁵⁴ Polyreactive antibodies bind xenogeneic to endothelial cells in culture and can be found in xenografts.¹⁶¹ What impact polyreactive antibodies have on the fate of xenografts is unknown, but it is not unreasonable to think that impact is exaggerated over the effect exerted in allotransplants¹⁶¹ and the “autoreactivity” of the antibodies should be considered when in evaluation of the specificities in serum.¹⁶²

Cellular Innate Immunity

Leukocytes (natural killer cells, macrophages, neutrophils, and T cells), platelets, fibrocytes, and other cells are found in inflammatory reactions of every type, including those observed in xenografts. These cells are thought to participate in the innate immune reactions that accompany ischemia and surgical disruption of tissues and acute and chronic rejection of transplants,^163–166 modifying the functions of endothelial cells, especially in transplants.^167–169 Cellular elements can recognize foreign or injured-autologous cell surfaces and products released from cells (eg, agonists of inflammatory receptors) and initiate the ensuing reactions, or cellular elements can be recruited to inflammatory reactions begun by other recognitive systems (eg, responses to binding of Ab or activation of complement). Cellular elements can also play a tangential role in pathogenesis (eg, severe complement-mediated injury can destroy an organ whether or not inflammatory cells are present). Although effector and regulatory pathways and specific cellular interactions are often identified in cell culture systems, the contribution of cells to pathologic processes must be ascertained in vivo, typically by use of animal models. For example, foreign NK cells rapidly bind to and kill or activate foreign endothelial cells, including xenogeneic cells^{167,170–172}; yet NK cells do not evidently cause early injury of porcine organs transplanted into nonhuman primates but rather might contribute to chronic vascular injury.¹⁷³

Inflammatory reactions associated with xenotransplantation appear to be greater than those typically seen in ischemia-reperfusion injury or in allogeneic transplant reactions. Whether interactions between macrophages, NK cells, and other cells and xenografts initiate or cause that increase or whether these interactions merely reflect greater tissue damage caused by antibodies, complement, etc. is not entirely clear. However, some of the pathways that constrain cellular interactions in autologous systems fail in allogeneic and especially in xenogeneic systems. For example, NK activity is suppressed by interaction of inhibitory receptors with classical or nonclassical MHC class I, but porcine MHC class I (and some allogeneic MHC class I) fail to interact with these inhibitory receptors.^174,175 Macrophage activity is suppressed in autologous systems by interaction of signal regulatory protein (SIRP)-alpha, with CD47, expressed by nearly all cells.¹⁷⁶ Failure of these interactions increases interaction and effector activity of NK cells and macrophages with xenografts.¹³⁰ As potential solutions to the incompatibilities, pigs have been genetically engineered to express transgenes for human HLA-E and human CD47.^177–179

Adaptive Immunity

More than 100 years have elapsed since Ehrlich and Morgenroth showed that foreign cells elicit abundant antibody responses,¹⁸⁰ Nuttall¹⁸¹ showed that those responses recognize multiple determinants, and Fleisher¹⁸² observed a more intense “leukocytic reaction” in xenografts than in allografts. Still, the rapidly destructive impact of innate immunity and biological incompatibilities generated by admixing of cells of disparate species left some uncertainty about the dimensions of the barrier to xenotransplantation posed by adaptive immunity, particularly cell-mediated immunity.¹²⁷ Long-term survival of porcine organs in nonhuman primates is presently pursued by genetic engineering to minimize the impact of innate immunity (ie, use of pigs deficient α1,3-galactosyltransferase and expressing human complement regulatory proteins) and by administration of regimens of immunosuppressive agents more severe than those typically used in clinical transplantation. Because innate and adaptive immunity intersect (eg, complement activation promotes B cell responses) and because both innate and adaptive immunity induce thrombosis, coagulation, and inflammation, advances in survival of experimental xenografts have been realized through use of combinations of intense regimens of immunosuppression (to suppress adaptive immunity) and genetic engineering to efface innate immunity, thrombosis, coagulation, etc. However, as clinical application of xenotransplantation approaches, it will be important to determine whether less intrusive regimens of immunosuppression can be employed. Identifying the model(s) that can reliably and efficiently evaluate immunosuppression for clinical xenotransplantation thus should be a key objective in the field.

Cell-Mediated Immunity

Isolated xenogeneic proteins and intact cells elicit cell-mediated immunity in humans and animals. Because the number of foreign peptides in xenogeneic antigens exceeds the number in allogeneic antigens, one might expect cellular immune reaction to the former would be especially robust. Therefore, it was striking, to say the least, when in vitro analyses of T cell responses to xenogeneic cells revealed profound limitations of cell-mediated responses.^183,184 These limitations mainly resulted from incompatibility of cytokines and co-recognition receptors between species under conditions that minimized the impact of recognition of foreign peptides presented with self-MHC (ie, mixed leukocyte cultures, which preferentially detect responses of naïve T cells to intact foreign MHC but not de novo responses to foreign peptides).¹⁸⁵ The observations raised the possibility that cell-mediated immunity to xenografts might be vulnerable in ways that allogeneic responses are not. Consistent with that possibility, work in mice revealed that antibodies against CD4 suppress T cell responses to and cellular rejection of xenografts.¹⁸⁶ Subsequent investigation revealed that anti-CD4 antibodies, engineered to block co-recognition but not to deplete CD4+ T cells, induce tolerance to xenogeneic protein.¹⁸⁷ Today, most work in xenotransplantation employs agents such as anti-CD154 and anti-CD40 that disrupt co-stimulation more broadly.¹⁸⁸ The toxicity of these broadly active agents will probably encourage a revisiting of less severe approaches or the possibility of inducing tolerance.

For reasons described above, immunosuppressive regimens for xenotransplantation are commonly investigated using nonhuman primates as recipients of porcine xenografts. Although nonhuman primates probably offer a more stringent model of the cellular immune barrier to xenotransplantation than small animals, including “humanized”-mice, nonhuman primates also potentially can mislead. One problem is that nonhuman primates might develop immunity to “humanized” antibodies or to human proteins expressed as transgenes in pig organs. If such immunity blocks the action of immunosuppressive agents, rejection could ensue or higher doses or more severe regimens might be needed to sustain the graft. For this reason, selection of the optimal and least intrusive immunosuppressive regimens for swine-to-human xenotransplantation might prove futile in nonhuman primates and might rather depend on analysis in early clinical trials.

One potential avenue that might be optimally tested first in nonhuman primates is the induction of tolerance. Some have argued that successful application of xenotransplantation might depend on devising approaches for the induction of immune tolerance.^188–191 Depending on the approach and regimen used, tolerance might well be extended to human proteins expressed in a xenograft and to “humanized” agents used for tolerance induction or maintenance.

As xenotransplantation approaches clinical application, the emphasis of research will almost certainly shift from preventing and treating acute rejection to the promoting of long-term function and avoidance of chronic disease of the graft and such comorbidities as cancer and cardiovascular disease. Toward that objective, it will be important to consider whether manipulation of the recipient and/or genetic engineering of swine modify cell-mediated immunity in ways that promote or hinder long-term function and well-being. As only one example, consider the impact of transgenic expression of human CD46 in a swine organ to control activation of complement in the graft.^192,193 CD46 facilitates the proteolytic action of factor I on C3b, generating C3d, among other fragments, which can amplify the effector activity of cell-mediated immunity.¹²² Therefore, if efforts to decrease the intensity of immunosuppression and/or to induce tolerance fail, there might be reason to revisit the approaches used to control complement. For reasons given above, pig-to-nonhuman primate models are best suited if not essential for testing the impact of genetic engineering in xenotransplantation.

Humoral Immunity

There is general appreciation that elicited antibody responses might limit the success of xenotransplantation. In the absence of immunosuppression, all xenografts and nearly all heterologous proteins elicit T cell-dependent B cell responses. The specificity, concentration, and avidity of antibody responses to xenotransplantation in immunosuppressed nonhuman primates have been the subject of a few reports.^194–197 However, the full range of antibody responses in xenotransplant recipients is incompletely understood at best. Given the successes achieved by targeting the α1,3-galactosyltransferase gene, there might be some temptation to catalogue the specificities of elicited antibody responses to xenotransplantation. Hopefully, consideration of the potential diversity of these responses and the variation in specificities likely to be found between recipients will build resistance to such temptations.

Several obstacles hinder investigation of elicited antibody responses to xenotransplantation. Natural antibody responses to Galα1-3Gal and other antigens can increase after transplantation, possibly owing to inflammation, making these responses more difficult to distinguish from de novo responses. Another obstacle, and one that also impairs full understanding of donor-specific antibody responses in allotransplantation, is that a functioning graft, especially an intact organ graft, can absorb enormous amounts of antibody and the antibodies so absorbed will be of the highest affinity and specific for the antigens of greatest density.^198–200 Hence, the antibodies remaining in the serum do not necessarily represent the antibodies of greatest importance.²⁰¹ Another obstacle is that antibodies and complement bound to healthy cells can be taken up and processed, impairing the ability of immunopathology to detect early stages of injury. These problems are potentially overcome by investigation of B cell responses.²⁰¹

Some antibody responses to xenotransplantation may induce antigen-specific pathophysiology. Human kidney allotransplant recipients with the X-linked Alport syndrome (mutant gene encoding collagen type IV alpha 5 chain) sometimes produce antibodies against the wild-type collagen in the transplant, leading to antiglomerular basement membrane nephritis.²⁰² Similarly, human kidney transplant recipients sometimes produce antibodies against angiotensin receptor allotypic variants, evoking malignant hypertension.²⁰³ In principle, all kidney xenograft recipients are at risk for producing antibodies directed against the swine homologues of these or other pathophysiologically significant targets. Thus, antibody-inhibitors of heterologous factor VIII have been described in xenograft recipients.¹⁹⁷ Because every protein in a xenograft is a potential immunogen, identifying the most common pathophysiologically vulnerable targets could prove challenging but clinically significant. Although this problem is likely to be idiosyncratic, it is best pursued in animal models in which the importance of antibody binding to a specific antigen can be distinguished from pathology caused by antibody binding to any target and solutions, such as antigen specific tolerance, can be explored.

Pig-to-nonhuman primate xenotransplantation models offer 2 important advantages for investigation of elicited humoral immune responses to transplantation. One advantage is the high frequency of antibody-mediated rejection (compared to clinical allografts) makes it easier to link B cell and antibody responses to the development of graft pathology. The other advantage is the ready access to blood and tissue samples. The main disadvantage is the relative paucity of information concerning non-uman primate variable region genes and the mixture of species (baboon and monkey) used as recipients.

The Impact of Immunity on Xenografts

Three distinct factors determine the impact of innate and adaptive immunity on xenografts. These factors are: (1) the source(s) of the blood vessels in the graft; (2) the intrinsic and induced resistance of the graft to immune and inflammatory injury (a condition we named accommodation); and (3) the nature and kinetics of immunity directed at the graft. The nature and kinetics of immunity to xenografts were discussed above. But as important as the intensity of immunity may be, the impact of immunity on a graft is to a large extent determined by the origin of blood vessels—whether derived from the recipient by ingrowth or originating with the source (as in organ grafts)—which determines the pathogenic processes invoked by immunity (Figure 4). Intrinsic and induced resistance to injury determines whether the immune-induced pathogenic pathways destroy the graft or allow repair and recovery from immune assault.

Figure 4.

The type of xenograft determines the source of endothelium, which in turn determines the impact of immunity on graft pathology. Organ xenografts (top) contain blood vessels originating from the source of the xenograft. Antibodies and complement directly attack the endothelial lining of organ xenografts, causing hyperacute rejection (HAR) in minutes to a few hours. If HAR is averted, acute vascular rejection (AVR) also called antibody-mediated rejection (AMR) and also caused by antibodies and complement, can ensue over the next days, weeks, or months. Organ grafts are also susceptible to chronic rejection (CR), sometimes caused by antibodies and complement, developing over months or years. Immunity also can induce changes in endothelium that render blood vessels and other cells resistant to injury (accommodation). These resistive changes counter pathogenesis and allow function to persist in the face of immunity to the transplant. Cell or tissue xenografts (bottom) contain blood vessels and endothelium of the recipient. Recipient blood vessels are not directly targeted by xenoreactive antibodies and hence are not susceptible to HAR, AVR, and CR (usually). However, T cells and phagocytes actively penetrate blood vessels, and hence cell and tissue grafts are susceptible to CMR. When cell and tissue xenografts are introduced into blood vessels of the recipient (eg, into the portal vein), antibodies and complement can attack the grafts, causing “instant blood mediated inflammatory reaction” (IBMIR). Injury begun by IBMR can persist after the grafts pass out of the circulation, but susceptibility to de novo IBMR ceases once engraftment outside of blood vessels occurs. Studies in cell culture systems suggest cell and tissue grafts, like organ grafts, may be protected by accommodation.

Source of Blood Vessels in a Graft

We have long emphasized that the origin of blood vessels in a xenograft determines the conditions potentially induced by the immune response of the recipient.^114,130 Organ xenografts, in which blood vessels are mainly of donor origin, are susceptible to conditions generated by the direct action of antibodies, complement, and inflammatory cells on graft endothelium (Figure 4). Cell and tissue xenografts (eg, pancreatic islet and hepatocyte, respectively) in extravascular sites are not susceptible to these conditions (hyperacute and acute antibody mediated rejection) because all IgM and most IgG and complement are retained within blood vessels. Cell and tissue xenografts introduced via the blood (rather than injection into extravascular spaces) are susceptible to injury by antibodies and complement during the period of passage through the blood of the recipient, but once engrafted, this susceptibility wanes.^204,205 All xenografts are susceptible to cellular rejection because stimulated lymphocytes and phagocytes migrate actively through blood vessel walls.^206,207

The pathology of ischemia and rejection of xenografts reflects the distinct assaults by immunity on blood vessels.^30,126 Rapid activation of abundant amounts of complement and assembly of terminal complement complexes cause endothelium to lose functions (especially barrier and vasoregulatory functions), eventuating immediately in the pathology of hyperacute rejection. Activation of smaller amounts of complement or interaction of leukocytes (macrophages, NK cells, T cells) changes the transcriptional program and physiology of endothelial cells, causing the cells to promote coagulation, leukocyte activation, and migration, etc. and eventuating in coagulation and intra- and perivascular inflammation. In allografts, this condition is often called antibody-mediated rejection and that term could be applied to xenografts. However, antibodies do not necessarily mediate this condition, and therefore we have preferred to use the term acute vascular rejection because blood vessels are the target and the instrument of pathological processes.^30,126,208 Migration of activated T cells and macrophages through otherwise undamaged endothelium causes the cellular infiltrates typical of cellular rejection and, perhaps more importantly ,increases interstitial pressure and exposes regional cells to cytokines, proteases etc., the clearance of which is impaired.

Baseline and Induced Resistance to Injury (Accommodation)

Allografts and xenografts exhibit a wide range of responses to assault by ischemia and immunity. In part these differences reflect the intensity of the ischemic insult and of the immune response of the recipient directed at the transplant. Indeed, the intensity of injury to the graft is often used as an index of ischemia or immunity. Yet decades of investigation have established that properties of target can govern the outcome of inflammation and immunity.^209,210 Thus, among pairs of recipients from the same renal transplant donors, up to 60% of early and late outcome can be ascribed to the donor organ.^211–213 Some graft-associated determinants of the outcome of transplants (besides MHC) are inherited. In clinical transplantation, the race of the donor influences early and possibly long-term outcome. Among inherited factors in donors are polymorphisms encoding variants of APOL1, caveolin-1, ABCB1, and eNOS, and donor-recipient pairing of certain alleles beyond MHC²¹⁴ have a discernable impact as well.^215,216

One property of grafts, especially xenografts, that determines outcome is the ability to resist and repair injury. We refer to this ability as “accommodation.”³⁰ Accommodation was first observed in ABO-incompatible kidney transplants and in heterotopic cardiac xenografts that continued to function despite the presence of antibodies against the grafts in the blood of the recipients.³⁰ Accommodation in ABO-incompatible transplants explained why in some circumstances antibodies that can initiate devastating injury sometimes fail to do so and why surveys of anti-graft antibodies in the blood of recipients often revealed little or no relationship to the presence or severity of graft injury in ABO-incompatible transplants.^217,218 Accommodation is also observed in conventional (ABO-compatible) organ allografts, but the frequency is unclear because alloantibodies specific for HLA can be absorbed in and taken up by organ transplants.^199,219

Accommodation develops over a period of days, and establishment of that condition obviously requires sufficient baseline resistance to injury. As a working hypothesis, we suggest that baseline resistance to injury reflects some constitutive properties of cells,^210,220 and heightened expression of the products of “cytoprotective genes”²²¹ allow cells, tissues, and organs to repair initial damage and dispose of waste. More enduring changes increase the efficiency of these processes, restoring function and shifting the level of resistance to cytotoxicity.^137,222,223

Accommodation, sometimes referred to by other terms, has been implicated in the cancer phenotype, responses to infection and physical injury, and autoimmunity.^{200,222,224,225} Efforts to identify genes involved in accommodation in clinical settings have met with limited success,²²⁶ in part because surveys focused on “cytoprotective genes,” which are also expressed in rejection. Genome-wide association studies (GWAS) reveal regions of the genome and heritable traits that confer resistance to injury from infection^227,228 and improved quality of meat.²²⁹ GWAS in models of tissue injury in pigs and transcriptional profiling of human subjects with rejection reveal potential involvement of tissue repair.^230,231 Still, identifying the optimal genetic background and understanding how expression of sets of genes over time drives (or allows) accommodation to occur will require a more incisive analysis, probably both GWAS and dynamic gene expression.^232,233

The more proximal consideration, however, is that pigs, like humans, undoubtedly vary greatly in the levels of baseline and induced resistance to immune and inflammatory injury. Differences in genetic background thus may confound efforts to compare efficacy of therapeutic regimens or genetic manipulations of xenografts from distinct sources. Table 2 shows outcomes of heart and kidney transplants with various genetic manipulations performed in nonhuman primates. Some of the variation in outcome reflects the efficacy of genetic manipulations but some reflects differences in the background of the transplants and some differences in immunity between recipients. There is a tendency to think that introduction of more human genes that counter pathologic changes will improve results,³⁴ but comparison of the outcomes of kidney xenografts listed in Table 2 might suggest otherwise. This problem can be addressed in part by cloning to make the genetic background of the source homogeneous. However, cloning (or inbreeding) potentially fixes in the background gene variants that undermine resistance to injury and restoration of function. A recent report on the outcome of kidney transplants from “multi-transgenic” inbred mini-pigs (a1,3 GT KO, and combinations of human CD55, Hu CD46, Hu CD59, and Hu CD39) in baboons revealed little or no advantage of the transgenes and survival at 3–14 days.²³⁴ These results, disappointing compared to the results shown in Table 2, could reflect various aspects of the regimens or transgenes used, but they could also reflect properties of the genetic background of the source. Nor can one be certain whether off-target effects of genetic manipulation have affected physiology (the heart xenografts are mainly nonfunctional heterotopic grafts, the kidney xenografts are functional). Because the more dramatic barriers to xenotransplantation are overcome and clinical application and long-term function are prized, there will be much potentially to be gained by focusing on function rather than survival and pathology and potentially from optimizing the background of the sources through breeding or engineering or both.

Physiological and Biochemical Barriers to Xenotransplantation (Incompatibilities)

During the first half of the twentieth century, the forebears of transplantation biology and immunology struggled to understand why grafts of foreign tissue fail. One theory, put forward by Leo Loeb, held that proteins produced by genetically-different individuals and especially by individuals of disparate species fail to support healthy and functional interactions between cells from those different individuals.²³⁵ The differences between proteins of different individuals might be assayed by serologic methods, but the failure of grafts and attendant pathology reflected the extent of incompatibility. Although Loeb was recognized widely for his contributions,²³⁶ his theory of individuality obscured recognition of immunity as a cause of allograft failure. However, the development of Loeb’s theory presaged the challenges one inevitably faces in attempting to understand why xenografts fail.

Because human cells can survive enduringly in immunodeficient mice⁹⁹ and immune-competent pigs,⁹³ whatever incompatibilities may exist between disparate species need not preclude survival of xenografts. However, comparison of protein structure and physiology between species could suggest that xenografts would likely fail to meet the physiologic needs of recipients and incompatibilities of biochemical systems could engender distinct toxicities,^237,238 as first suggested by Loeb. Obviously, the preeminent question for clinical application of xenotransplantation is whether and to what extent biochemical and physiologic incompatibility between species diminish the value of xenotransplantation and whether the defects can be overcome without converting the swine genome to human. These questions have been addressed at least in part by demonstration that xenografts of swine lungs, kidneys, hearts, and pancreatic islets can temporarily support the life of nonhuman primates.^239–245 In contrast, orthotopic porcine liver xenografts can engender life-threatening complications (eg, thrombocytopenia), and some believe incompatibility of the swine liver precludes successful xenotransplantation. However, porcine liver xenografts do exhibit measurable function for a period of days,²⁴⁶ and whole porcine livers²⁴⁷ and isolated porcine hepatocytes in liver assist devices²⁴⁸ can augment functions in patients with acute liver failure, suggesting physiology is not limiting. The observations on transplantation of porcine tissues and organs other than liver into nonhuman primates thus argue against Loeb’s idea that xenografts inevitably fail because of incompatibilities generate lethal toxicity.²³⁵ The observations also argue against the proposition that incompatibilities decrease the level or modify the nature of physiologic support to the point where xenografts might not offer an acceptable replacement for a failing tissue or organ.

The acceptable function of porcine xenografts in nonhuman primates for periods of months or even years cannot be taken as evidence of absence of significant incompatibilities between species that would impair clinical utility. Regardless of the extent of species-specific regulation of complement^{30,117,138,139} (or lack thereof¹⁴⁰), experience during the past 20 years provides numerous examples of the benefit for xenograft function and survival conferred by expression of human complement regulators in swine tissues. This benefit indicates that, for whatever reason, xenotransplantation effectuates a functional deficiency of complement regulation. However, functional deficiency or incompatibility are not necessarily apparent immediately but rather might appear months or years after birth (in the case of inherited deficiencies) or transplantation. Some individuals with inherited deficiency (or nonfunction) of complement factor H, factor I, and CD46 (regulators of the alternative complement pathway) develop thrombotic microangiopathy of the kidney (atypical hemolytic uremic syndrome) but do so years after birth or in adulthood.^249,250 Some never develop this condition. A xenotransplantation model functioning for months or even years might very well fail to reveal clinical evidence of some incompatibilities of complement regulation.

Xenotransplantation of swine organs in humans or nonhuman primates would generate incompatibility between the coagulation system of the recipient and coagulation regulators, particularly thrombomodulin, expressed in blood vessels of the transplant.^251,252 Thrombomodulin expressed in porcine blood vessels is appreciably incompatible with human protein C^113,253,254 and that should eventuate in excess generation of thrombin and coagulation and/or inflammation.²⁵⁵ This incompatibility sparked the development of transgenic pigs expressing human thrombomodulin, among other modifications, and testing with favorable results in pig-to-nonhuman primate organ xenograft models.^256,257 These and more recent results encourage the view that generation of pigs expressing multiple transgenes has advanced xenotransplantation toward clinical application.³ Whether correct or not, the successes achieved by expression of multiple transgenes, such as those listed in Table 2, should not be taken as critical proof that the transgenes address key molecular incompatibilities. Using the expression of human thrombomodulin as only one example, the incompatibility of the human protein for swine has been proved but the importance of the incompatibility has not. Thrombotic microangiopathy, as observed in organ xenografts, is characteristic of inherited defects in regulation of the alternative pathway of complement (eg, atypical hemolytic uremic syndrome). In contrast, inherited deficiency of thrombomodulin activity typically causes late-onset large vessel or coronary thrombosis,²⁵⁸ if it causes any disease at all.²⁵⁹ That is not to question the benefit of expressing human thrombomodulin in xenografts. Transgenic expression of human thrombomodulin is certainly more convenient and less toxic than administration of anticoagulant agents (or correction of what we think might be the more fundamental problem with complement regulation). Rather, it argues that molecular incompatibilities may have less impact than vitro experiments suggest. Inherited defects in regulation of complement or coagulation (among other pathways) are not immediately pathogenic because the system adjusts to increased pathway activity. Such adjustment is likely to be found for countless biochemical or structural “incompatibilities.”

Infection Between Species as a Barrier to Xenotransplantation

The possibility that xenotransplantation would convey or heighten the risk of infection has been viewed as a significant barrier to xenotransplantation.^260–262 Accordingly, approaches to prevention and surveillance and standards for microbiological safety have been extensively discussed and recently reviewed.^263–268 Although these approaches will continue to be applied, dimensions of this barrier are now generally viewed as “small”²⁶⁹ and “manageable.”³ Here we shall consider the general nature of the biological barrier infection poses and the extent to which current models can provide useful insights. For reasons we shall mention, this consideration must remain a matter of speculation until xenotransplantation enters clinical practice.

Transplants of every type potentially convey infectious agents, particularly viruses, to the recipient. That risk is greatest when transplants originate from deceased donors because only limited time can be devoted to screening and because screening might fail to detect a recently acquired transmissible infectious agent. When transplants originate from living human donors, more time and resources can be devoted to screening and clinicians can weigh the risks against potential benefits if the donor has a transmissible virus. In xenotransplantation, the potential for screening is greater because multiple generations of source animals can be evaluated and risks can be further decreased by isolation of source animals, breeding, treatment, vaccination, or genetic engineering to eliminate existing agents.²⁷⁰ Therefore, in principle, the risk transmitting an infectious agent from a graft to a recipient should be lower in clinical xenotransplantation than in allotransplantation. This risk is further decreased because some agents capable of infecting pigs are not infectious for humans.

However, xenotransplantation does potentially engender several risks distinct from those experienced in allotransplantation. One such risk is that infectious agents harbored by the graft, whether or not transferred to human cells, might be less effectively controlled by human immunity, particularly T cells and cytokines, and in this setting cause tissue or organ damage. Porcine CMV has been found to be activated in xenotransplants, capable of activating endothelial cells and associated with thrombotic microangiopathy.^271,272 It is possible that immunity to the graft and rejection causes activation of the virus as virus activation was seen in early rejection.²⁷³ But it is also possible that this or some other virus underlies damage or dysfunction observed over time in xenotransplants. On the other hand, if the swine agent is controlled by the immune system of the recipient, it is possible the agent could serve as a source of peptide targeted by cell-mediated rejection.

Another risk of infection pertinent to xenotransplantation is the possibility that innocuous retroelements or an endogenous retrovirus of the pig could undergo activation and/or recombination to generate a novel virus transferable to the human recipient and potentially more broadly in society. The porcine endogenous retrovirus (PERV) has been thought potentially to be such an agent.²⁶⁰ A gammaretrovirus, PERV can be activated and transferred to human cells in culture. Concern about PERV has fueled efforts to eliminate elements from the porcine genome by selection and gene targeting.²⁷⁴ However, humans subjects exposed to pigs in the workplace and subjects whose blood was perfused through porcine livers for treatment of liver failure or through porcine kidneys for kidney failure or recipients of porcine xenografts of skin or other tissues reveal no evidence of PERV transmission to humans.^275–278 Consequently, concern about potential risk of PERV transmission has decreased substantially.³

Yet another “infectious” risk unique to xenotransplantation involves the potential consequences of genetic recombination caused by spontaneous fusion of swine and human cells.^91,279 Although fusion of heterologous cells is probably rare, when it occurs, the potential for recombination is increased by aberrant hybridization and DNA breaks, among other events, and recombination potentially generates novel genes.^280–282 Cell fusion has been considered mainly from the perspective of risks of oncogenesis and tumor progression,^282,283 but the same mechanism potentially can underlie emergence or evolution of viruses, for example, acquisition of a ligand for an existing cell surface receptor.²⁷⁹ Although this mechanism might explain rare emergence of new viruses by evolutionary leaps, we think the likelihood that swine to human xenotransplantation would cause emergence of new viruses by this mechanism is exceedingly low because pigs and humans have lived in proximity for thousands of years and blood is continuously exchanged on farms and other settings (some of those engaged in agriculture could have immunodeficiency or receive immunosuppressive agents). The introduction of human genes in the swine genome at increasing numbers of loci, however, does potentially increase the potential for recombination in hybrids and that might warrant consideration in the future.

On the other hand, because pigs and nonhuman primates do not naturally share habitats and exchange flora, the use of nonhuman primates as recipients of experimental xenografts does potentially generate conditions that could increase the rate of viral evolution. Although accelerated viral evolution is probably not a unique risk of xenotransplantation, for reasons mentioned above, these models potentially offer an opportunity to investigate processes important for public health. As a related consideration, however, experimental transplants of tissues or organs from pigs into nonhuman primates probably offer a poor (and exaggerated) model of the infectious risks of clinical xenotransplantation. Not only are humans better adapted than nonhuman primates to pig flora, but the sophisticated diagnostic tools, range of therapeutic agents, and established regimens and doses of antimicrobials in the clinical setting, among many other factors, probably decrease the risk and improve the outcome of infection in the clinical setting.

Concluding Remarks on Potential Limitations of Current Models of Xenotransplantation for Clinical Application

Advances in experimental xenotransplantation have generated much excitement and the perception that xenotransplantation is rapidly advancing toward clinical application.^1,2,284 To a large extent, this excitement and the perception of progress spring from improvements in the survival, now sometimes exceeding 1 year, of porcine islets, hearts, and kidneys transplanted into nonhuman primates. At this juncture, then, it would seem appropriate to consider how well the preclinical models of xenotransplantation are likely to predict the outcome of clinical xenografts performed for treatment of disease. The questions we think most timely are two. The first question is whether the results in experimental models suggest that in a given condition and circumstance (eg, unavailability of an allogeneic organ), a xenograft could provide a better option than alternative therapies. This question is frequently addressed by practitioners and regulators and hence needs little comment here. The second question, not adequately addressed in the literature, is whether and how pig-to-nonhuman primate models depart systematically from what might be expected of pig-to-human xenografts performed in the clinical setting.

We believe clinical xenografts might well perform better than experimental xenografts discussed above. One reason for this view is that the resources, expertise, fund of knowledge, diagnostics, therapeutics, etc. that can be directed at the recipient of a clinical xenograft vastly exceed what can be directed at recipients in animal models. Another reason for this view is that much of the genetic modification of pigs for xenotransplantation has gained expression of human genes, the products of which (eg, CD46 and thrombomodulin) better regulate complement and coagulation of humans than of nonhuman primates. Even if these proteins have normal function in isolation, the proteins may interact aberrantly in complex networks, the impact of which extends beyond complement and coagulation, potentially influencing expression and function of a broad set of genes,^232,285,286 cellular functions, and signaling pathways,²⁸⁷ components of which can be physiologically discontinuous between species.²⁸⁸ One extended network potentially pertinent here concerns coagulation and complement. Nearly all components of the coagulation system vary greatly in the population, reflecting tuning by regulation,²⁸⁹ and modifying one protein at one anatomic location changes the system in others.²⁹⁰ Although nonhuman primates are used to model humans, individual proteins and complex systems of nonhuman primates likely have some incompatibility with human proteins and systems. Such incompatibilities might explain some of the abnormal function of organs from nonhuman primates transplanted into patients (Table 1). As a related concern, when nonhuman primates are used as recipients of porcine organ xenografts, the nonhuman primates might develop immunity to human proteins expressed as the products of transgenes in pigs. Immune responses of nonhuman primates to human proteins might thus limit the duration or level of action of the human proteins in pig-to-nonhuman primate xenograft models. Immunity to the human proteins would be far less likely to compromise the function porcine xenografts in human recipients.

The potential usefulness of xenotransplantation in clinical settings remains a matter of speculation. Because nonhuman primates do not model the diseases and pathophysiologies that would be addressed by transplantation, it is impossible to accurately compare the potential efficacy of xenotransplantation against the efficacy of other therapies for most conditions. Only clinical trials can be expected to test the efficacy for some potential applications of xenotransplantation.

Two exceptions might be transplantation of islets for treatment of diabetes and transplantation of hepatocytes for treatment of acute liver failure. Both applications are limited at least in part by availability of human tissues and in both settings retransplantation could be performed if the initial transplant failed. Xenotransplantation of hepatocytes for treatment of severe acute liver failure might be especially compelling. Orthotopic liver transplantation is the only life-saving treatment currently available for most severely afflicted individuals. Xenotransplantation might be considered if a human organ (or an effective liver assist device) was not available. Orthotopic liver xenotransplantation seems unlikely to provide a permanent solution, although it might serve as a surgically intrusive bridge to allotransplantation. Hepatocyte xenotransplantation, on the other hand, might avoid removal of the native liver and potentially allow the diseased liver to regenerate.²⁹¹ Therefore, the development of a model for acute liver failure in nonhuman primates²⁹² is a timely advance.

Pig-to-nonhuman primate models for xenotransplantation have proven essential for advancing xenotransplantation. The models established the significance of immune and biochemical barriers to xenotransplantation, especially the significance of Galα1-3Gal as a target natural antibodies and defective control of complement as a mechanism responsible heightened susceptibility of xenografts to complement-mediated injury. Pig-to-nonhuman primate models have been essential to testing physiologic incompatibility of xenogeneic blood vessels with primate coagulation and thromboregulation. Finally, pig-to-nonhuman primate models have proven essential to the testing of genetic engineering as a central approach to addressing those barriers. However, we also believe it is important now to consider the limitations of pig-to-nonhuman primate models, especially as the models are used to test genetic engineering. Little attention has been devoted to incompatibilities between nonhuman primates and humans that might confound efforts to test more subtle genetic manipulations. One consequence could be the introduction of genes to solve problems that would not exist in pig-to-human xenografts. Another might be that the models underestimate the survival and function would be exhibited by pig xenografts in humans. On the other hand, we also suspect that once clinical trials are begun, barriers unappreciated in pig-to-nonhuman primate models will be found and these might well be addressed by introduction of further genetic modifications in pigs and tested in nonhuman primates or perhaps sometimes preferably in “humanized” mice^101,293 to avert limitations of nonhuman primate models.

Having commented extensively on the models used to advance xenotransplantation toward clinical application, we would be remiss not to add that a byproduct of the preclinical investigation of xenotransplantation includes fundamental discoveries. Fundamental discoveries will have value and affect whether xenotransplantation becomes part of clinical practice. These discoveries include the importance of endothelial cells as the engine of changes in tissues targeted by immune responses, accommodation as a response by cellular targets of immunity that subverts injury, rekindled interest in humoral immunity as a determinant of the outcome of organ transplants, including allotransplants, and an impetus for discoveries and applications at the nexus of developmental biology and genetics.