INTRODUCTION

TOP ABSTRACT INTRODUCTION DESCRIPTIONS OF BASIC AND... ADVANTAGES OF TRANSGENIC ANIMAL... POTENTIAL PITFALLS AND... SUMMARY AND CONCLUSIONS REFERENCES

THE COMBINED APPLICATION OF sophisticated methods for assessment of cardiovascular, respiratory, behavioral, and metabolic physiology in intact animals with techniques to generate defined genetic modifications in model organisms creates an almost unlimited horizon of opportunity for integrative biologists. Here, we review the methods used to generate transgenic animal models, the important advantages provided by this class of experimental approaches, and potential problems faced by those using them.

	DESCRIPTIONS OF BASIC AND ADVANCED TRANSGENIC METHODS

TOP ABSTRACT INTRODUCTION DESCRIPTIONS OF BASIC AND... ADVANTAGES OF TRANSGENIC ANIMAL... POTENTIAL PITFALLS AND... SUMMARY AND CONCLUSIONS REFERENCES

At the present time, the mouse is the model organism most suitable for the application of transgenic methods to the study of problems in integrative biology. This is because of the well-developed state of techniques for manipulating the genome of this species, the rapid breeding time, the lower maintenance costs of mice compared with larger animals, and our general knowledge of mouse genetics. In addition, the mouse has been selected recently as a model organism in which every gene will be sequenced. Although the small size of these animals leads to challenges in the design and application of instrumentation for physiological measurements, this disadvantage is usually counterbalanced by the many positive features of this species for transgenic experiments.

It is possible currently to introduce transgenes into rats and larger mammals, and physiological assessment of mutant strains of genetically modified zebrafish also holds promise for the future. Model experimental organisms larger than mice offer the obvious advantage that physiological assessments are easier and also provide alternatives when manipulation of the mouse genome does not produce the phenotype one wishes to investigate. Examples of this situation are provided by the hypertensive response of rats but not mice to forced expression of the REN-2 gene (12) and the more severe spondyloarthropathy produced by B27 and 2-microglobulin transgenes in rats (5). For most experimental questions that are addressed by transgenic methods, however, the mouse is likely to be the best choice, and this review will focus primarily on this species.

There are two basic approaches to manipulate the mouse genome: random chromosomal integration and homologous recombination of foreign DNA (Fig. 1). The first method is based on integration of DNA into unspecified locations of chromosomes following microinjection into one-cell embryos (fertilized oocytes). The microinjected oocytes are implanted into pseudopregnant females, and the resulting offspring are screened to identify those animals in which the transgene has inserted stably into a host chromosome. The second approach, targeted modification of a specific chromosomal locus by homologous recombination, is more complicated. This procedure requires introduction of the foreign DNA sequence into cultured embryonic stem (ES) cells, followed by identification of individual ES cell clones that have the correct mutation, and injection of these ES cells into early mouse embryos at the blastocyst stage. If all goes well, the genetically modified cells contribute to the germ line (sperm or egg cells) of mice that are born following this procedure, such that these animals can pass the modified allele to their offspring. Homologous recombination of foreign DNA, as opposed to random chromosomal integration, allows the investigator to create very precise changes in the genome.

View larger version (1425K): [in this window] [in a new window]   Fig. 1.   Schematic illustration of basic transgenic procedures. A: generation of germ line transgenic mice by random integration of foreign DNA into host chromosomes following microinjection into one-cell embryos and implantation into pseudopregnant females. B: generation of mice bearing specific mutations in endogenous genes by homologous recombination of foreign DNA into specific chromosomal locations. Technique requires manipulation of embryonic stem (ES) cells in culture, followed by microinjection of cloned ES cells carrying the properly targeted allele into early mouse embryos at the blastocyst stage. Mice resulting from this procedure are chimeric, meaning that they include cells derived from the microinjected ES cells as well as the host embryo. If the targeted ES cells have contributed to the germ line (sperm and egg cells) of the chimera, then this animal can pass the targeted allele to its offspring. In the example shown here, the goal was a gene knockout, creating a null allele. Other applications of this procedure are discussed in the text. Initial offspring of a chimeric mouse carry only a single copy of the targeted allele (+/), but animals homozygous for the null allele (/) are generated by mating of 2 heterozygotes. Unless the product of the targeted gene is essential for embryonic and fetal development, the frequency of each possible genotype in the progeny of matings between heterozygous males and females is predicted by Mendelian genetics.

Random integration of a transgene is employed most commonly for one of two purposes: the forced expression of a recombinant protein to alter the physiology and/or morphology of the animal or the analysis of transcriptional control mechanisms involved in regulatory pathways. For most applications, there is no intent to modify endogenous genes. To express a recombinant protein in intact animals, the foreign DNA that is injected consists of at least two modules: a transcriptional regulatory region selected to direct expression of the foreign gene to a specific cell type or tissue and the region that encodes the foreign protein to be expressed.

To alter the phenotype of an intact animal, the transgene may encode a native protein, so that the experiment addresses the consequences of producing an abnormal quantity of this protein in a cell that normally expresses it ("overexpression"). In other experiments, the design is to express a protein in a cell type that normally does not produce that gene product ("ectopic expression"). Alternatively, the transgene is designed to encode a mutated protein that has been modified for a special purpose: to produce a constitutively active ("gain-of-function mutant") or dominant-negative ("loss-of-function mutant") form of a specific protein or to mimic a mutation observed in a human genetic disease.

Another common application of random chromosomal integration of transgenes is to identify transcriptional control elements (promoters, enhancers, and locus control regions) that respond to developmental cues or physiological stimuli. In this application, the coding region of a so-called "reporter gene" is linked to the segment of DNA thought to contain the regulatory elements of interest (binding sites for transcription factors). Reporter genes, by definition, encode biologically innocuous proteins that are easily detected by convenient histological or biochemical assays. The goal is to avoid perturbation of the biology of cells in which the reporter gene is expressed while assessing the function of the transcriptional regulatory sequences that are attached to the reporter gene. Commonly used reporter genes encode -galactosidase, which provides a brilliant blue stain in histological sections; green fluorescent protein, which lights up cells expressing it as green when examined by fluorescence microscopy; and chloramphenicol acetyltransferase or luciferase, which is reliably quantified in cell or tissue extracts using sensitive and simple biochemical assays.

The more complex procedure of homologous recombination of a transgene is used most commonly to produce a deficiency of a specific protein by disrupting transcribed regions (exons) of a specific endogenous gene. This application is called a "gene knockout." The mammalian genome includes two copies of each chromosome (except for the X and Y chromosomes in males), but only one copy of the gene of interest (the "targeted allele") is disrupted in the mice produced initially. This permits analysis of heterozygous animals bearing a single targeted (null) allele, which may or may not produce an abnormal phenotype. "Haploinsufficiency" is a term that is used when abnormalities result from mutation of only one of the two copies of a given gene. If heterozygous null animals are viable and fertile, they are mated. The mating of two heterozygote null animals generates litters in which, on average, 25% have two copies of the modified allele (homozygous null, /) and therefore are completely deficient in the protein encoded by this gene. On average, one-half of such litters will be heterozygous (+/) for the null allele (like the parents) and the remaining 25% will be normal (+/+ or "wild type"). The actual percentages of genotypes represented in the offspring of such a mating will differ from those predicted by Mendelian genetics if the targeted gene has an essential function during embryonic or fetal life. Many different outcomes are possible, ranging from 100% survival of homozygous null animals to complete embryonic lethality of genetically modified animals. Prenatal or perinatal death of heterozygous or homozygous null animals may be an exciting result for the developmental biologist but a disappointing outcome for the integrative biologist interested in the function of the targeted gene in adult animals. For example, this problem has limited studies of homeostatic responses to hypoxia in adult animals mediated by the hypoxia-inducible transcription factor HIF-1 (7) or the angiogenic growth factor VEGF (vascular endothelial growth factor) (3).

A solution to this problem can be provided by "conditional knockouts" (Fig. 2). A method successful for this purpose is based on the ability of an enzyme called Cre recombinase, a 30-kDa enzyme native to bacteriophage P1, to remove segments of DNA that lie between two copies of a unique 34-bp sequence termed a LoxP site (9, 15). To apply this "Cre-Lox" strategy, the gene of interest is first modified by homologous recombination in ES cells in a manner analogous to a gene knockout but with the important difference that the targeted allele is modified by the insertion of two LoxP elements without disrupting the normal function of the gene. This usually means their insertion into introns at sites that do not influence RNA splicing but that flank coding regions (exons) essential to produce a functional gene product. After a number of technical steps we will not discuss here, transgenic mice are then produced that carry two copies of the targeted allele in all cells. These animals are phenotypically normal, since the function of the targeted gene has not yet been altered. There are then two ways to create functionally null alleles at the targeted locus in the tissue of interest. One can introduce Cre recombinase via a viral vector, such that the targeted gene is knocked out only in infected cells. Alternatively, one can breed the mice carrying the LoxP-tagged gene to another line of transgenic mice in which Cre recombinase is expressed from a tissue-specific or drug-regulated promoter. By either method, Cre recognizes LoxP sequences, excises the segment of DNA between the LoxP sites, and recombines the remainder of the gene, thereby rendering it nonfunctional. Such a procedure will inactivate the gene but only in those cells in which Cre has been expressed. The alteration of the genome produced by Cre is permanent and will be passed to all daughter cells arising by mitosis of the cell originally expressing Cre, even if Cre is no longer present.

View larger version (27K): [in this window] [in a new window]   Fig. 2.   A conditional gene knockout strategy using Cre recombinase. This sophisticated technique is based on the ability of Cre recombinase to recognize a unique nucleotide sequence (LoxP site), to remove the intervening DNA, and to recombine the ends. The usual procedure requires generation of 2 independent varieties of transgenic mice. The first is generated by homologous recombination (see Fig. 1B) to insert LoxP sites into the gene of interest in a manner in which function of the gene is not impaired (see text). A second line of transgenic mice is generated by random chromosomal insertion (see Fig. 1A) of a transgene expressing Cre recombinase under the control of a tissue-specific promoter. Double transgenic animals resulting from matings of these original transgenic lines will carry null alleles of the gene of interest only in those cells that have expressed Cre. It also is possible to introduce Cre by somatic cell gene transfer using viral vectors without a requirement for establishing germ line transmission of Cre.

An example of how a conditional knockout strategy can be applied to a problem in cardiovascular physiology is provided by recent studies with respect to the role of endothelins in cardiac hypertrophy. A standard knockout of the endothelin-1 (ET-1) gene is lethal at birth in homozygous null animals because of craniofacial abnormalities that interfere with normal respiration (8). In contrast, animals bearing a conditional knockout of the ET-1 gene in cardiomyocytes (based on expression of Cre recombinase under the control of the cardiac-specific -myosin heavy chain promoter) are viable and avoid the extra cardiac manifestations of ET-1 deficiency. When these animals are subjected to hypertrophic stimuli, however, their altered responses reveal autocrine or paracrine functions of ET-1 synthesized in cardiomyocytes in the control of hypertrophic growth of the heart (R. Shohet and M. Yanagisawa, personal communication).

These Cre-Lox strategies add additional layers of complexity and expense to the experiment. Nevertheless, these powerful techniques are likely to become a standard methodology used to approach problems in integrative biology, as applied to many genes of physiological importance. In principle, conditional knockouts permit the investigator to control the timing at which cells experience a deficiency in a given protein, thereby circumventing both embryonic lethality and confounding effects of complex adaptive responses that can occur when the physiological observations follow the gene knockout event by days or weeks.

Another important variation on the theme of using homologous recombination to produce a null (knockout) allele is the related approach of introducing a new gene into the same locus (termed a "knock-in"). This strategy may be used to insert a reporter gene (e.g., green fluorescent protein), the expression of which is then subjected to the identical regulatory controls that were placed on the gene that was replaced. This is useful for tagging particular cell types or to facilitate studies of gene regulation. Alternatively, gene replacement may be used to assess the degree of functional redundancy among two related proteins or to examine the phenotype produced by replacing a normal protein with a mutated form.

The technique of drug-regulated transgene expression was mentioned previously with respect to controlling the timing at which Cre recombinase is expressed in a conditional knockout experiment. This strategy also can be applied productively to other types of experiments in which the investigator wishes to control the timing of transgene expression. Such control becomes important when acute rather than chronic responses to the transgene are the issue of greatest biological interest or when chronic transgene expression is lethal. A variety of systems have been developed for this purpose, the principles of which are illustrated in Fig. 3. The fundamental requirements include the expression of a transcription factor (either a repressor or an inducer of transcription) that becomes active in the presence of a drug that can be administered systemically to the animal. Methods based on binding of tetracycline to the inducible transcription factor have been most widely used (23). The transcriptional regulatory region of the transgene is designed to include a binding site for this drug-regulated transcription factor. Like conditional gene knockouts, this strategy requires two separate genetic modifications: the first to express the drug-responsive transcription factor in the correct cell type and the second to insert the ultimate transgene of interest with its binding site for the drug-responsive transcription factor. This approach has great appeal, since an ability to regulate transgene expression effectively circumvents the difficulties posed by embryonic lethality or chronic adaptive responses. In practice, however, the current inducible systems appear often to suffer from either excessive transgene expression in the "off" state or insufficient transgene expression in the "on" state, and successful applications of drug-regulated transgene expression to questions of integrative biology have been uncommon. It is reasonable to anticipate that further technical improvements will bring this general approach into widespread use.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   A system for drug-regulated transgene expression. Investigators can control the expression of a transgene by engineering animals to express a transcription factor that is regulated by binding of small molecules. One such system is based on the ability of the bacterial Tet repressor (Tet-R) protein to inhibit transcription of genes that contain a unique nucleotide recognition sequence to which Tet-R binds (binding site for Tet-R). Transcriptional repression by Tet-R (small arrow) is relieved in the presence of tetracycline (drug) and the transgene becomes activated (large arrow).

A number of other advanced transgenic technologies may be useful in special circumstances. It is possible, for example, to use homologous recombination to insert a single copy of a transgene construction into a specific chromosomal location that has been selected not to modify an endogenous gene but to provide a genomic site that is insulated from effects of flanking DNA that otherwise would modify transgene expression. This strategy markedly reduces the variations in transgene expression that plague experiments based on random integration of transgenes. Another maneuver that sometimes proves advantageous is the use of bicistronic transcriptional units in the transgene construction. This means placing cDNA segments encoding two different proteins downstream of a single promoter, separated by an internal ribosomal entry site element. The transgene is transcribed to generate a single mRNA species that is translated to make two different proteins at the same time. Finally, it is sometimes advantageous to study genetic chimeras, animals that include cells of two different genotypes. The generation of chimeric mice is a necessary step in the creation of gene knockout models, since genetically modified ES cells are admixed with wild-type cells at the blastula stage, and cells of both genotypes contribute to the adult animal produced by this procedure. In a typical knockout project, the chimeras are usually discarded after breeding. However, if one develops ES cells that are homozygous for a targeted allele and uses these to produce chimeric mice, it is then possible to compare the behavior of cells lacking the protein produced from the targeted gene with wild-type cells directly in the same animal.

	ADVANTAGES OF TRANSGENIC ANIMAL EXPERIMENTS TO ADDRESS QUESTIONS IN INTEGRATIVE BIOLOGY

TOP ABSTRACT INTRODUCTION DESCRIPTIONS OF BASIC AND... ADVANTAGES OF TRANSGENIC ANIMAL... POTENTIAL PITFALLS AND... SUMMARY AND CONCLUSIONS REFERENCES

A properly designed transgenic experiment can be a thing of exquisite beauty in that the results support absolutely unambiguous conclusions regarding the function of a given gene or protein within the authentic biological context of an intact animal. A transgenic experiment may provide the most rigorous test possible of a mechanistic hypothesis that was generated by previous observational studies. A successful transgenic experiment can cut through layers of uncertainty that cloud the interpretation of the results produced by other experimental designs. For example, drugs employed to inhibit the function of a given protein in an intact animal invariably produce effects other than the specific perturbation that is the focus of the investigation. Experiments that rely on gene transfer into tissues of intact animals (as opposed to the germ line transmission mode) suffer from the often inconsistent or transient nature of transgene expression and from artifacts relating to the gene transfer vector (e.g., immune responses to adenoviral proteins). However, genes may serve different functions during embryonic and fetal development as opposed to postnatal life, and unexpected consequences of genomic modifications are frequent. The initial results of a transgenic experiment not infrequently mark the beginning, and not the end, of a mechanistic journey.

Genomic interventions may reveal the biological importance of a gene product when other experimental strategies fail. A good example of this principle is that of HIF-1 in the hypoxically challenged lung (22). The observation that normal mice exposed to 10% O₂ do not elevate HIF-1 protein suggested that HIF-1 is not involved in the pulmonary response to hypoxia. However, when HIF-1 (+/) animals were exposed to this level of hypoxia for 1-6 wk, there was substantial attentuation of the pulmonary vascular remodeling, pulmonary hypertension, and right ventricular hypertrophy seen in wild-type (HIF-1 +/+) mice. Although these results cannot define the molecular mechanism(s) by which HIF-1 is acting in this situation, they reveal unequivocally a functional involvement for this hypoxic response protein that could not be discerned by observational studies only. The explanation for how HIF-1 is functioning without being detected by immunoblotting techiques may lie in the very rapid decay of this protein after hypoxia is relieved (19).

Genetically modified strains of mice also can provide convenient and authentic models of human diseases. Disease phenotypes produced by a given transgene are often highly penetrant within the inbred strains of mice that are commonly employed for such models, yielding reliable and consistent results. Such consistency of a disease phenotype contrasts favorably with disease models created by dietary, pharmacological, or surgical manipulations, and transgenic models can be generated quite rapidly when all goes well. For laboratories in which transgenic technology is well established, recombinant DNA is the only additional reagent that is required to produce transgenic mice by random chromosomal insertion, and the biochemical techniques required for assembly of the various components of the transgene construction are simple and rapid.

The application of transgenic technology to problems in integrative biology has several other advantageous features. Transgenic lines generated with one transgene can be crossed with other lines bearing a different transgene to create double mutants. It is particularly useful to identify measures capable of correcting ("rescuing") a disease phenotype that was modeled initially by a single transgenic approach. For example, two groups have recently described the genetic rescue of cardiomyopathic mice that were generated intially by a knockout of the gene encoding muscle LIM protein (MLP). MLP-null (/) mice develop a dilated cardiomyopathy in the first few weeks of postnatal life. Rockman et al. (16) prevented cardiomyopathy in mice deficient in MLP by forced overexpression in the heart of a peptide inhibitor of the -adrenergic receptor kinase. Minamisawa et al. (11) showed that mice deficient in expression of phospholamban because of disruption of this gene also are protected against the cardiomyopathy that otherwise would result from a deficiency of MLP. These studies have important implications both for our understanding of the pathobiology of heart failure and for the design of novel therapeutic approaches. These studies also demonstrate how genetically modified mice can provide a source of stably modified cells for experiments in tissue culture. In this manner, reductionist approaches in a more controlled environment can be used to complement analyses performed in intact animals.

Finally, the attractiveness of transgenic experiments in mice has been enhanced markedly by the ingenuity of integrative and systems biologists who have developed techniques for sophisticated physiological assessments of these tiny animals. For example, the heart of an adult mouse is smaller than a raisin, but a number of laboratories have achieved remarkable proficiency in the analysis of ventricular performance, contractile physiology, and myocardial metabolism within the intact animal or in isolated heart preparations. It also has been possible to assess respiratory control mechanisms in genetically modified mice, even at a relatively young age (1).

	POTENTIAL PITFALLS AND LIMITATIONS OF TRANSGENIC ANIMAL EXPERIMENTS

TOP ABSTRACT INTRODUCTION DESCRIPTIONS OF BASIC AND... ADVANTAGES OF TRANSGENIC ANIMAL... POTENTIAL PITFALLS AND... SUMMARY AND CONCLUSIONS REFERENCES

All current transgenic procedures require specialized equipment and skilled operators. The best results are obtained in settings where the responsible individuals are dedicated to these activities. Even in settings where core labs provide this technology as a service to other investigators, the time and expense required to generate and analyze informative transgenic models should not be underestimated. Animal husbandry becomes a major component of the experiment, and it is easy to become mired down in the effort to produce and maintain a sufficient number of animals of the correct genotypes for the desired analyses. Even a modestly sized transgenic colony can consume $5,000-10,000 a month or more in animal housing costs and occupy the efforts of several personnel.

Initially, positive and exciting results may be only a beginning. For example, Huang et al. (6) deleted the gene encoding endothelial nitric oxide synthase (eNOS) and found that mean arterial blood pressure was 20 mmHg higher than in wild-type controls. The appealingly simple explanation that peripheral vasodilation is impaired because of reduced synthesis of nitric oxide (NO) may not, however, be sufficient to account for this result. Perhaps eNOS-generated NO is important in regulating other molecules with vasoactive properties, in controlling cardiac output as well as peripheral vasodilatation as a determinant of blood pressure, or in modifying baroreceptor function or vasomotor controls in the central nervous system. Our point is to emphasize that a genomic manipulation followed by an initial description of a phenotype often opens up additional questions, and investigators should plan accordingly.

Negative results of an initial transgenic experiment, too, may represent only the beginning of an interesting and important scientific journey. If trivial or artifactual explanations for the lack of phenotype (i.e., no measurable structural or functional consequences) following a given genetic modification can be excluded, then a number of possible explanations should be considered, demanding additional experiments. A common trivial problem in gain-of-function or dominant-negative research designs is a failure to express the transgene to physiologically relevant levels in the temporal and spatial pattern expected. When the transgene integrates randomly into host chromosomes, several variables that can exert profound influences on expression of the transgene are uncontrolled, sometimes leading to negative or artifactual results. The foreign DNA usually integrates as linear arrays, comprising variable numbers of copies of the transgene construction. Although one might expect intuitively that a greater number of copies of the transgene would result in higher levels of expression, this is often not the case. In addition, the desired function of transcriptional regulatory elements included with the transgene (e.g., muscle-specific expression) can be influenced profoundly by the chromosomal location in which the transgene integrates. Random transgene insertion occasionally may alter endogenous genes (insertional mutagenesis), thereby confounding the interpretation of the phenotype. Finally, more exotic problems, such as genetic imprinting (transcriptional silencing of a gene based on transmission from parent to offspring of repressive nucleosomal structures), may arise and produce confusing results. Variability in transgene expression based on differences in gene dosage and in DNA sequences flanking the insertion site are a universal feature of all experiments using this approach, whereas the latter problems arise only occasionally.

The conventional practice to deal with this problem is to establish and analyze multiple lines of transgenic mice bearing any specific transgene, each of which represents a different chromosomal insertion event. It is mandatory for most purposes to assess at least two independent lines. Whenever possible, it is advantageous to assess dose-response relationships between transgene expression and a given phenotype by analyzing separate lines of transgenic mice that express the transgene product at each of several levels of abundance. For experiments designed to assess the function of transcriptional control elements contained within the transgene, it may be necessary to assess 5-10 independent transgenic lines to be confident of a correct interpretation.

Targeted modifications of endogenous genes by homologous recombination circumvents the uncertainties associated with random chromosomal insertion of transgenes. However, surprises may occur even with this more sophisticated and demanding technology. For example, in creating a null allele in one gene, it is possible unwittingly to destroy transcriptional control elements that govern expression of a neighboring gene. Such events seem to explain occasions when two different labs knock out the same gene but observe different phenotypes based on subtle differences in the specific design of the targeting vector (13). It is advisable to have as much information as possible about the genomic organization of the targeted region to avoid or at least to be aware of potential difficulties of this nature.

We have already discussed how embryonic lethality may preclude testing of the original hypothesis, a disappointing initial result for some experiments. Transgenic experiments designed to test hypotheses that relate to physiological regulation or pathophysiology of adult animals become infeasible if the specific genetic modification impairs an essential developmental function and transgenic animals die in utero. An essential function for a given gene during development may be unrelated to its functions during adult life. For experiments in which overexpression or ectopic expression of a transgene or gene deletion is the goal, the problem of embryonic lethality can be avoided by the use of transcriptional control regions that are inactive (or nearly so) during embryonic and fetal life but highly active in the adult tissues of interest. The -myosin heavy chain promoter has this property with respect to cardiac-specific expression (20) and has been widely and successfully employed to drive expression of transgenes in the adult or neonatal heart that would be likely to produce lethal effects in embryos. Unfortunately, correspondingly timed promoters are not available for transgene expression in many other cell types.

Investigators using transgenic animals also must be cognizant of potential differences in phenotypes observed when an apparently identical genetic modification is examined in different inbred strains of mice or in outbred animals. The same overexpressed transgene or gene knockout may produce a phenotype that is severe in one strain and mild in another. The hypoxia-responsive transcription factor EPAS-1 is essential for survival in certain strains of mice (21) but EPAS-1 / animals survive when the null allele is crossed into different genetic backgrounds (R. Hammer, personal communication). The basis for such differences lies in so-called "modifier genes," allelic variations that influence the responses to a transgene. It is a good practice to assess the effects of transgenes or knockouts in more than one mouse strain.

Adaptive responses to a genetic modification may confound the interpretation of phenotypes. It is important to remember that the phenotype observed in any transgenic experiment is a function both of the planned genetic modification and of secondary responses of the organism to that perturbation. A dramatic example of this principle was provided recently by our own studies of mice in which the myoglobin gene was disrupted. Previous experiments using pharmacological inhibitors of oxymyoglobin formation demonstrated that myoglobin was essential to maintain energy metabolism and contractile function of the myocardium. Surprisingly, however, we found that normal cardiac function could be maintained in mice completely devoid of myoglobin (4). Although this result could be interpreted to indicate that myoglobin is unimportant for oxygen transfer in the myocardium, this is probably not correct. We have subsequently determined that survival in the absence of myoglobin is possible only because of powerful adaptive responses that compensate for the absence of myoglobin. The examination of a large number of offspring from heterozygote crosses (i.e., mating of two myoglobin +/ animals) revealed that more than half of embryos without myoglobin (/) die in utero between embryonic day 9.5 and 11.5, a period of rapid growth (and presumably increasing energetic demands) of the embryonic heart. The myoglobin / animals that survive demonstrate increased expression of hypoxia-responsive genes, increased vascularity of the heart, increased coronary blood flow, increased Hb concentrations, and reprogramming of myocardial gene expression with respect to many other genes (D. Garry and R. S. Williams, unpublished observations). Apparently, myoglobin deficiency is fatal unless these pleiotropic adaptive responses are sufficiently robust to compensate for the defect in oxygen transfer. Further exploration of the molecular basis for survival in the absence of myoglobin using this transgenic model should extend our understanding of the repertoire of defense mechanisms used by mammalian organisms to maintain function when oxygen transport is limited.

The HIF1- gene deletion studies of Yu et al. (22) further illustrate the care that must be taken in studying the effects of a gene deletion. When HIF-1 +/ (heterozygous null) mice are exposed to 10% O₂ for 6 wk, two of the major physiological adaptations to hypoxia, polycythemia and right ventricular hypertrophy, are similar to those observed in wild-type (HIF-1 +/+) littermates. If the animals had been examined at only this single time point, one would conclude that haploinsufficiency of HIF-1 is unimportant for these adaptations. However, measurements made after 1, 2, 3, 4, and 5 wk of hypoxia showed a delay in both erythrocytosis and right ventricular hypertrophy in the early weeks, demonstrating a role for HIF1- in these processes.

Many mammalian proteins are present as multiple isoforms, closely related proteins derived from different genes. The planning and interpretation of gene knockout experiments, in particular, must take into account the potential for overlapping or redundant functions of such proteins. It often may be necessary to generate animals bearing null alleles in two or more proteins of multigene families to gain an understanding of their function. A prominent example of such functional redundancy among individual members of multigene families involves members of the MyoD family of basic helix-loop-helix proteins that we now know exert overlapping but distinctive functions during development of skeletal muscles and in muscle regeneration following injury. Skeletal muscle development is nearly normal in mice that lack either MyoD or the closely related Myf5 protein (2, 17); however, the double knockout (MyoD /:Myf5 /) has a severe phenotype, and skeletal muscles fail to form (18). Although these two genes are not entirely redundant (14), each is capable of compensating for a deficiency of the other with respect to myogenic differentiation during embryonic life. Interestingly, however, the degree of functional redundancy of these two proteins is less complete during muscle repair following injury to adult muscles, and animals lacking only MyoD have a severe deficit in muscle regeneration (10).

Finally, it is important to use caution in interpreting the results of transgenic experiments that involve overexpression of a given protein in transgenic mice, as produced by linkage of a highly active promoter to the protein coding region of the gene of interest. It should be remembered that exaggerated physiological effects produced in such an experiment demonstrate only that the gene/protein in question is capable of the observed function. The results of an overexpression experiment cannot, in the absence of other data, establish the normal physiological role of that gene or protein.

	SUMMARY AND CONCLUSIONS

TOP ABSTRACT INTRODUCTION DESCRIPTIONS OF BASIC AND... ADVANTAGES OF TRANSGENIC ANIMAL... POTENTIAL PITFALLS AND... SUMMARY AND CONCLUSIONS REFERENCES

This review has been intended as an introduction to transgenic technology for investigators skilled in physiological disciplines who are moving to incorporate transgenic technology into their research. The analysis of genetically modified animals can provide otherwise unattainable opportunities to advance our understanding of homeostatic mechanisms and pathophysiological principles. Integrative biologists have already leaped forward aggressively to embrace this technology, sometimes with dramatically pleasing results. Investigators who are just beginning to incorporate transgenic approaches into their experimental strategies are well advised to seek counsel from laboratories with longstanding experience in this methodology. Nuances of the design of transgenic experiments, in addition to those discussed here, may have an important influence on the outcome and interpretation of research. Overall, perhaps the best advice to investigators new to this mode of experimentation is "Expect the unexpected!"