Contents
-Home
-Welcome Letter
-Purpose of Workshop
-Program
-Presentations

Appendix
-Participants
-Steering Committee
-Program Committee

Workshop Report
-Preface
-Final Report

Abstract: Recent studies on the state of marine and freshwater fisheries demonstrate a need to develop new lines of fish to meet demands of increasing world population. To meet these needs we have examined new methods for identifying the functional activities of genes and developed new methods that employ transposons and border elements to genetically engineer vertebrate animals. Although many phenotypes of interest to aquaculture and mariculture probably already exist or have been produced naturally, they have not survived selection in nature. Consequently, genetic engineering of fish offers a way of achieving human needs. There are several technical difficulties that will have to be overcome to allow effective and efficient genetic engineering of fish stocks. These include identification of the functions of genes as well as enhanced techniques for their transfer and reliable expression. The functions of genes can be deduced from knockdown phenocopies augmented by analyses of the effects of overexpression and misexpression of the genes. The Sleeping Beauty (SB) Transposon System is a genetic tool for introducing single copies of specific genes into animal chromosomes to confer new and desirable traits with more reliable patterns of expression. These tools can be adapted to many species of fish. Short-term goals are the identification of function of the 30,000-40,000 genes in vertebrate (fish) genomes. Intermediate and long-term goals include the further development of transgenesis for strain improvement or adaptation to particular aquacultural environments. The bottom line is: This is the time to test transgenic methods for modifying animals - before they are needed so that they can be applied when needed.

Key words: fish, GMOs, gene function, genetic engineering, risk assessment, transposons

INTRODUCTION
Over-harvesting of fish is rapidly depleting aquatic resources worldwide (Roughgarden and Smith, 1996; Cook et al., 1997; Malakoff, 1997). As the world's population grows, its fisheries are being depleted at an increasing rate. One potential solution is aquaculture. But, aquaculture often is inefficient and ecologically harmful. As an example, in the decade 1986-1997 the annual production of farmed salmon increased almost 6.5-fold to about 644,000 metric tons valued at about $2 billion. However, about 1.8 million metric tons of fish were required to produce these fish for human consumption (Naylor et al., 1998). Moreover, farmed fish are often grown to high densities, which can lead to environmental pollution. Calls for long-term (e.g., 20 years) moratoria on fishing in some regions is an economically and politically difficult choice and they can at best only restore fisheries to levels that were unable to satisfy consumer demand. Another way to increase yields of fish is through aquaculture of genetically superior stocks of fish that can be farmed at a faster rate for lower cost. This will require genetic engineering (Hackett and Alvarez, 1999).

Genetic Engineering - What are we Fearing?

Genetic engineering of animals, plants and other life forms to produce genetically modified organisms (GMOs) has received harsh responses. The most common reason is that GMOs will introduce change into the environment. The rationale for this objection is an unstated assumption that "This is the best of all possible worlds." (Voltaire, 1759). It is certainly true that GMOs will alter their environments. But, change is neither bad nor harmful by definition. The important question is whether alterations of local ecosystems after stabilization would be bad or harmful. These problems are hard to address in scientific terms as they involve extremely emotional issues. However, there are a number of potential problems regarding genetic engineering that have been raised and that can be addressed scientifically.

1) Insertion of a transgene into a resident gene will inactivate that gene.

2) The expression of transgenes cannot be reliably controlled.

3) The transgene may influence the expression of neighboring genes.

4) We know too little about the functions and subtle effects of transgenes to be able to predict the overall consequences on the GMO.

5) We do not know the functions of most genes, some of which may be more appropriatefor a specific goal.

6) Transgenic animals may escape into the wild.

OBSERVATIONS AND EXPERIMENTAL RESULTS

Answers and Solutions to Perceived Problems of Genetic Engineering

The goal of this paper is to address these specific concerns. Here political and/or emotional objections to genetic engineering are not discussed. Essentially, genetics has made rapid advances since the first genetically engineered organisms and has addressed some of these points and new technologies apply to others. The first concern is what differentiates a GMO from selection of natural mutants for specific phenotypes. Currently, many of the types of phenotypes that are of interest to aquaculture/mariculture and that are based on altered expression of endogenous genes are likely to exist or have occurred naturally. Mutation rates have been most intensively studied in humans, where mutations of every sort have been catalogued (Table I -Crow, 1997; International Human Genome Sequencing Consortium, 2001; various textbooks on human genetics). This means that for populations of billions of fish, there may be more than 1000 different mutants for every gene in the total population of the species. Moreover, single base-pair changes in regulatory regions can have profound effects leading to overexpression as well as underexpression. The transcriptional regulatory regions of vertebrate -globin genes have been intensively studied. The levels of these genes can vary more than 30-fold as a result of single base mutations (Meyers et al., 1986). In the case of the globin genes, where the ratio of alpha to beta polypeptides is essential for healthy growth and survival, fish with unbalanced levels of gene expression will be negatively selected, unless carefully maintained. This case is an example of "genetic buffering", which applies to most genes through their interactions with other gene products (Hartman et al., 2001). The result is the activities of most transgenes will be modified by epistatic mechanisms.

TABLE I: GENETIC VARIATION IN HUMANS

DNA polymerase error rate per base pair ca. 10^-9

Observed mutation rate per gene ca. 10^-5

Single Nucleotide Polymorphism rate per base pair ca. 10^-3

Chromosome duplication rate (trisomies 21, 13, 18) 10^-3 - 10^-4

Chromosomal translocations per live birth ca. 2x10^-5

The data in Table I demonstrate that there are natural gene duplications, deletions and rearrangements that occur constantly at lower rates. We humans are acutely aware of this - except in the case of identical twins, none of us are identical and we cherish the differences that result from the myriad of genetic differences between us. Moreover, we vigorously protect subspecies of fish and other organisms - which means that we actually appreciate diversity in germlines of animals. Hence, in reality, highly targeted genetic engineering that introduces single new alleles into a genome is more closely associated with background mutations than it is with either the introduction of exotic species into a naïve environment or even crosses between sub-species wherein hundreds to thousands of altered alleles are mixed in one event. Indeed, selection can produce naturally the same phenotypes as genetic engineering in fish, where addition of extra growth hormone genes to domesticated salmonids has a lesser effect than to wild fish (Devlin et al., 2001). The bottom line is that duplication and loss of genes is not uncommon in fish. With this in mind, let us consider the specific objections listed above regarding genetic engineering of fish and other animals.

1) Insertion of a transgene into a resident gene will inactivate that gene. The polyploid genomes of higher animals introduce significant redundancy (e.g., zebrafish like others are partially tetraploid and even the basic unit has many genes with redundant function). This redundancy coupled with the observation that haploinsufficiency rarely causes a phenotypic effect, means that phenotypes are quite resilient to most small changes in genotype. For this reason, the thousands of experiments done in mice with insertional mutagens such as retroviruses have not uncovered evidence of deleterious effects. A second reason why mutant phenotypes are very rarely seen is that only about 30% of the genomes of higher animals encode functional genes and of that most of the sequences are in introns where they may not elicit a phenotypic effect. The forgoing does not apply, of course, to inbred animals that may receive mutations in both alleles. In this sense, transgenic animals will fall into the same category as those animals that inherit any set of mutated genes from both parents. For this reason, it will be important to have lines of fish with genes inserted at multiple sites in order to find those that will cause minimal to no undesirable side effects.

2) The expression of transgenes cannot be reliably controlled. This problem can be overcome by the inclusion of DNA sequences that shield genes from effects on each other, termed position variegation. These sequences are called "border" or "insulator" elements (Bell et al., 2001). They have been associated with DNA sequences called nuclear matrix attachment sites and/or scaffold attachment regions. Border elements appear to protect certain stretches of DNA from being transcriptionally silenced. In experiments using transgenic zebrafish, we have shown that transgenes lacking border elements gave non-uniform expression whereas either of two different types of border elements conferred position-independent expression that was maintained through three generations of fish (Fig.1). Moreover, within a factor of two in zebrafish, the

Figure 1: Effects of Border/Insulator Elements in Fish

expression per gene per cell at one, two and five days of development was nearly constant for transgenes flanked with either of the two types of border elements (Caldovic and Hackett, 1995; Caldovic et al., 1999). These data suggest that the border elements can confer reliable levels of expression of a transgene in fish and protect the transgenes from being switched off as they are passed from generation to generation.

3) The transgene may influence the expression of neighboring genes. This problem also can be overcome by the inclusion of border elements in transgenic constructs. That is their function. The shielding of effects by border elements works bidirectionally by blocking enhancer effects from one region on another (reviewed in Hackett and Alvarez, 2000).

4) We know too little about the functions and subtle effects of transgenes to be able to predict their overall consequences on the GMO. This is true. Genetic engineering efforts have concentrated on elucidating the effects of only a handful of genes whose effects are substantial, and in many organisms confer unwanted side effects. Transgenic growth hormone is one example of where it has been shown to behave as advertised (e.g. Palmiter et al., 1982; Devlin et al., 1994). However, some side effects such as fertility declines have been noted (e.g., Pursel et al., 1989; Juskevich and Guyer, 1990; Nagai et al., 1991; Devlin et al., 1995). Such deficiencies will be overcome as we learn more about the molecular genetics of fish through genome sequencing (Duyk and Schmitt, 2001). QTL mapping in fish (e.g., Liu and Dunham, 1998) as well as high density maps for RAPD and AFLP marker-assisted selection of genes that direct useful phenotypes (e.g. in catfish Liu et al., 1998a, 1998b, 1999). More exciting are recent developments of methods for inactivating gene expression in fish using morpholino-modified oligonucleotides that are resistant to enzymatic degradation and function through inhibition of translation (Nasevicius and Ekker, 2000; Fig. 2). Morpholino-based screening of gene function could be done at an exceptionally high throughput rate of about 10,000 genes per year, which is nearly an order of magnitude higher than that achievable in mice using standard "gene knock-out" technology.

Fig. 2: Morpholino Knock-down Analysis in Fish

Furthermore, more effective gene delivery systems have been developed for assessing the effects of overexpression of genes and misexpression of genes. In particular, the Sleeping Beauty transposon system developed in our lab raises the integration frequencies of transgenic DNA more than 100-fold higher than by other means in some cells (Ivics et al., 1997). This transposon was developed from an inactive Tc1/mariner-type salmonid transposon (Ivics et al., 1996) and can be used for a variety of purposes including mapping (Ivics et al., 1999). The bottom line is that these developments will allow sophisticated investigation of gene function in fish.

5) We do not know the functions of most genes, some of which may be more appropriate for a specific goal. This is certainly true. The same resources described above in (4) apply to this consideration. Functional genomics is a hot area throughout biology as genomes are sequenced. The use of a variety of gene traps, enhancer traps, and poly(A) traps, built into Sleeping Beauty transposons will permit detailed understanding of gene functions Fig. 3). It is

Fig. 3: Sleeping Beauty-based traps for elucidating gene function

only a matter of time and effort before appropriate genes are discovered that will permit more sophisticated gene transfer to confer desirable traits. However, for some types of genetic engineering, such as for disease resistance (e.g., Anderson et al., 1996), the functions of the transgene are sufficiently clear that they can be used to achieve very precise goals.

6) Transgenic animals may escape into the wild. There is no doubt whatsoever that it will be impossible to keep transgenic fish from entering into natural ecosystems. Likewise, containing transgenic constructs is virtually impossible over the long run. The questions really are how fast will transgenes and GMOs spread and what will be the consequences. Part of the argument against genetic engineering has been that this process can transcend normal species and even kingdom barriers. Genomics has now demonstrated that nature, as usual, is far ahead of the engineers. For instance, the genome of the small mustard plant Arabidopsis contains transposons of insect origin (The Arabidopsis Genome Initiative, 2000) and human chromosomes contain bacterial genes (The Human Genome Sequencing Consortium, 2001; Venter, et al., 2001). Transposons are notorious for spreading DNA between all types of organisms over hundreds of millions of years of evolution. Spreading of genes far and wide in terms of genomes as well as geographically is responsible for the wonders of nature that we treasure so deeply. By not acknowledging extremely low-risk potentials and appropriately informing the public, companies using genetic engineering face huge obstacles in marketing products that should be safe based on all that we know about allergy and other potential side-effects. The concern about the escape of Aventis' Star-Link corn is an excellent example of how the public has become concerned about what appears to be a non-issue. The result is the spending of fantastic amounts of money (ca. $100 million) by both the company and the USDA - without much hope of success. This is money that could be used to ameliorate a myriad of problems that really do affect the public. However, because near term rates of spread and escape can be controlled, regulations on release of GMOs are not unreasonable.

CONCLUSIONS AND RECOMMENDATIONS

The results of work in our lab, and others, to improve transgenesis in fish, suggest that within a few years far more will be known about how to efficiently obtain transgenic animals that will be able to meet human nutritional needs. In particular, border elements flanking transgenes can be used to stabilize the expression of the exogenous genes such that their expression is consistent from line to line and is maintained during vertical transmission through the germ line. Recent methods of elucidating the bottom line functions of genes have now been developed for one model fish species. That should provide a roadmap to allow fine precision genetic engineering of fish in the near future. To achieve this goal, I suggest USDA should undertake the following actions with respect to aquaculture of commercially important fish:

Short-term (1-3 years):

The first four recommendations are required for any type of directed improvement of fish stocks. The fifth recommendation is to begin testing of enhanced methods of gene transfer for their efficacy in larger fish. The sixth is long overdue.

At the rate that molecular, cellular, and developmental biology information is being accumulated in many species, the first five perceived problems raised in introduction should be largely addressed. That should permit a more rationale discussion of the merits of widespread use of genetically engineered fish.

Acknowledgments

I thank my many colleagues in and out of the lab, especially John Liu, for stimulating discussions. The work described here was supported by grants from NIH (RO1-RR06625) and SeaGrant (USDOC/NA46RG O101-04).

References