*For offprint requests: tel: (609)785-0074 ext.127, fax: (609)785-1544, email: sallen@hsrl.rutgers.edu, xguo@hsrl.rutgers.edu

SUMMARY

In all the attention given to risk assessment in genetically modified organisms, practically none has addressed marine or estuarine organisms. This is not surprising considering the tremendous difficulty with the notion of containing the dispersive reproductive strategies of marine organisms in an environment where barriers are hard to define. On the other hand, the rapid development of the aquaculture industry in the US and around the world has evoked considerable effort for the development and use of aquatic GMOs, and the need for biological containment and risk assessment is urgent. One of the only feasible modes of containment for marine GMOs or non-native species is reproductive sterility, and at present there is really only one method to achieve sterility -- induced triploidy. However, recent data collected by our lab indicates a high proportion of heteroploid mosaics among triploid Pacific oyster, Crassostrea gigas, suggesting that at least some triploids are unstable, i.e., that they revert from triploidy to mosaics. This poses a real problem for use of triploids in population control.

The goal of this ongoing research is to clearly establish the origin and nature of heteroploid mosaics among triploid populations. This task is essential for evaluating the risks associated with using triploids for biological containment of GMOs or for population control of non-native species. Our objectives, using the Pacific oyster, are as follows:

To determine the origin of heteroploid mosaics by comparing the frequency of mosaics in two types of triploids: those produced by standard induction procedures and those produced from matings of tetraploids with diploids.

To determine the stability of chromosome constitution of triploid populations.

To determine the fate of mosaics through time.

To assess the productive potential of mosaics.

The problem of mosaics is important. More and more, triploids are being offered as a means for population control. This research could be viewed as crucial for population control of genetically modified or non-native marine animals generally. If reversion occurs, there are dire consequences for the use of triploids for this purpose.

Key words: Triploidy, tetraploidy, Pacific oyster, population control, heteroploid mosaics, reversion, chromosome

INTRODUCTION

In all the attention given to risk assessment in genetically modified organisms, practically none has addressed marine or estuarine organisms. This is not surprising considering the tremendous difficulty with the notion of containing the dispersive reproductive strategies of marine organisms in an environment where barriers are hard to define. On the other hand, the rapid development of the aquaculture industry in the US and around the world has evoked considerable effort for the development and use of aquatic GMOs, and the need for biological containment and risk assessment is urgent. One of the only feasible modes of containment for marine GMOs or non-native species is reproductive sterility, and at present there is really only one method to achieve sterility -- induced triploidy.

All applications of triploidy are based on their sterility. For this proposal we are concerned with sterility in the sense of capacity to reproduce versus market quality or growth rate. In terms of reproductive capacity, our lab has been working on two important aspects of this in the oyster Crassostrea gigas: the reproductive potential of gametes produced by triploid C. gigas and the problem that populations of triploid are never 100% triploid.

We have studied reproductive potential of triploid C. gigas and C. virginica and found them incapable of reproducing themselves. We used this data to convince authorities to allow field tests on C. gigas, a non-indigenous species in the mid-Atlantic region and a worse case scenario than GMOs. This is the only case of the use of biological containment for a marine species. We have also addressed the second problem of creating 100% triploids by producing and testing tetraploids. Crosses of tetraploids and diploids produced 100% triploids.

It would seem, then, that there is great potential in the use of triploids for biological containment. However, recently, we were surprised by data acquired from a two populations of "certified" (see below) triploids used in field tests. After a season of disease challenge in the field, about 15 and 20% of supposed triploids were, in fact, heteroploid mosaics in which individuals contained both diploid and triploid cells. This and other evidence suggests that heteroploid mosaics may have arisen as a consequence of reversion of triploids to a mosaic state, through chromosome set loss. If this is true, it is a serious problem for the use of triploids for biological control in invertebrates and perhaps vertebrates as well. In this proposal, we plan to study the risk of heteroploid mosaics associated with the use of triploidy for biological containment. This paper describes our approach to the problem. As a project in progress, results are preliminary.

MATERIALS AND METHODS

Our overall goal is to establish the origin and nature of heteroploid mosaics among triploid populations. For the origin of mosaics, we aim to discriminate between our two main hypotheses concerning the origin of mosaics: that they are induced at the time of treatment or that they revert from the triploid state afterward. For the nature of mosaics, we aim to isolate numerous mosaic individuals and follow the frequency of diploid cells and determine whether the mosaic condition can give rise to reproductively capable gonads. Our design is basically a brute force approach using FCM as our major tool and using cytogenetics to confirm the cytometry. It is an effort that can not be routinely accomplished during the course of other studies, such as, triploid grow out trials, and it is an effort for which our lab is uniquely suited.

We will use Crassostrea gigas as our test species. Our lab maintains several closed populations of C. gigas for experimental and breeding purposes. In fact, most of the principal accomplishments by us in the last several years has been on this species. We think that it is appropriate to use C. gigas for a number of reasons. First, it is in this species that we have the most preliminary data. Second, C. gigas is by far the most important commercial oyster species and, as such, the most likely candidate for genetic modification. Third, we have demonstrated that C. gigas harbors resistance to parasites infecting the native species and it is likely that there may be need for future field tests of C. gigas, of C. gigas gene constructs, or hybrids and their back crosses. Settling the issue of the origin of mosaics is key to approval of future tests. Fourth, it is far easier to manipulate the gametes of C. gigas than other species we work with, including the native C. virginica.

1. To determine the origin of heteroploid mosaics by comparing the frequency of mosaics in two types of triploids: those produced by standard induction procedures and those produced from matings of tetraploids with diploids

Hypothesis. Heteroploid mosaics are produced as a artifact of the treatment process using cytokinetic inhibitors.

Task. We will create triploid C. gigas by two fundamentally different methods. The standard method is to induce triploidy by treatment of newly fertilized eggs with CB to prevent extrusion of PB2 (Allen et al., 1989). The alternative method, possible because of our development of tetraploid oysters, is by mating tetraploid and diploids (Guo et al., 1996). The triploids created by mating will serve as a control for the CB induction. Since producing triploids by mating involves no disruption of early cell development (that might promote formation of mosaics), we expect to find no mosaics in the progeny groups.

Our crosses will be based on the matings depicted in Table 1. Female ploidy is always listed first. One replicate will consist of the two types of triploids. We consider the reciprocal crosses of [female tetraploid x male diploid] and [female diploid x male tetraploid] equivalent for this experiment. The crossing scheme below produces two progeny groups: induced and mated triploids All four crosses will be done simultaneously forming "half"-sib groups (although relatedness is complicated by the chromosome set manipulation). Our lab does this type of simultaneous, diallele experiment frequently (Allen et al. 1993; Allen and Gaffney, 1993; Scarpa and Allen, 1992; Guo and Allen. 1994b). A brief explanation of the progeny groups follows:

Ploidy analyses will be done principally by flow cytometry (FCM) (e.g. Allen, 1983). FCM has been established as the fastest and most effective method of ploidy determination, and is quite useful to detect mosaics. The advantage of FCM is the rapid assessment of large numbers of cells according to distributions in ploidy. It is not as sensitive as karyology. The advantage of karyology is the identification of fine differences in the karyotype (e.g., chromosome fragments, aneuploidy or trisomy). But it is difficult to categorize large numbers of cells. The two methods combined is a powerful tool.

For analysis of mosaics in induced and mated populations of triploids, we will sample about 130 individuals from each treatment (2) x replicate (3) combination, a sample size that will yield a standard deviation of about 1% for the initial level of mosaics expected (about 2%). FCM analysis on all groups will be accomplished three times: as embryos using karyology (next paragraph), and as larvae >250m (Chaiton and Allen, 1985) and newly settled spat (<3 mos old) using FCM. The ability to assay individual larvae is a powerful tool for the analysis of initial proportions of mosaics because it gives us a window into ploidy composition of individuals before selective mortalities associated with metamorphosis.

2. Determine the stability of chromosome constitution of triploid populations

Hypothesis. Chromosome constitution of triploids is stable over time, without chromosome (chromosome set) loss.

Task. Mark (label) individual oysters from each of the treatment x replicate combinations, establish a ploidy profile for each, and monitor them by repeatedly sampling them every ~6 months.

From the progeny groups sampled above, we have a precise baseline assessment of the proportion of mosaics. However, this is data from the populations as a whole. We already have data on populations (preliminary data cited below) and, although suggestive, it is indirect evidence of reversion. To disprove the hypothesis above (i.e., to confirm reversion) unequivocally, it is necessary to "catch reversion in the act," that is, to show transition from the triploid state to a mosaic one. This can only be done by monitoring triploid individuals over time, and since we do not know a priori the frequency at which reversion occurs, a "brute force" approach is proposed. Using the triploid progeny groups created by induction and mating allows us to discriminate between reversion that occurs as a result of the effects of CB treatment versus reversion that is a consequence of triploidy per se. In the former case, incidence of mosaics should increase in CB induced oysters only; in the latter case, incidence of mosaics should increase in both groups.

For determination of baseline ploidy profile of every individual, we will label and biopsy about 130 oysters (8-9 months old) from each treatment x replicate combination. Determination of ploidy will be accomplished by bioassay of the gill tissue. Individually labeled oysters will be placed in 5% solution of MgCl₂ overnight to induce relaxation and gaping of the shell. A small piece of gill will be snipped away with dissecting scissors and prepared for FCM using standard technique. Oysters then will be allowed to recuperate and subsequently will be held in our holding tanks. Every six months, all individuals will be re-biopsied and DNA profiles compared with the baseline for signs of changes in cell DNA content.

3. Determine the fate of mosaics through time

Hypothesis. In a mosaic individual, the frequency of diploid cells (in the biopsied tissue) remains stable with time.

Task. From the screening above, individually label mosaics and hold them separately. Re-biopsy them about every three months for about two years to monitor the percent of diploid cells.

The screening from objective (2) above should uncover an initial population of mosaics. Based on our preliminary data, we might expect at least 1-2% of animals will be mosaic, so we should be able to identify at least 9-18 of them, and perhaps more with time. Fortunately, this is not our only source of mosaics. We are actively engaged in projects in the Pacific northwest of the US and in France where several large populations of triploids are available for screening. Moreover, we are also actively collaborating on a project in Louisiana where our preliminary estimate of mosaics is 6.4% (7 mosaics identified among 110 sample so far) in a population of triploid C. virginica. We therefore anticipate no difficulty in isolating 20-25 mosaic individuals for objective (3) above.

Mosaics will be isolated from the general populations from objective (1) or from other triploid populations as needed. Between 20-25 mosaics will be monitored by frequent (every ~3 months) biopsies of gill tissue. The proportion of diploid cells in each individual will be calculated from the curve fitting program Modfit. Correlations of the proportion of diploid cells with time, with individuals as replicates, will be the principal data analysis. Other changes in the ploidy profile, e.g., broadening of the CV or changes in the mean DNA content, will be evident.

We will also sacrifice 5-10 mosaic individuals to better characterize the ploidy composition of various tissues. At issue is the similarity among tissues in proportions of diploid and triploid cells, but especially important is the consistency of these proportions within the same tissue. Our preliminary data indicate that there tends to be differences in ploidy composition among tissues of the same oyster, but that a particular tissue is consistent. For example, proportions of diploid cells in gill, mantle, adductor and hemolymph might be 10, 20, 30, and 40% in a particular individual, but all the gill tissue is 10% diploid. This latter fact is important to establish since objectives (2) and (3) will rely on repeated samplings of only one tissue type. We realize that our ability to sample only one tissue type for the biopsies is a potential drawback. We will carefully "dissect" a number of mosaics to satisfy this question. We will also experiment with taking biopsies of more than one tissue type. However, the gill is a good candidate tissue for biopsies because it is normally a rapidly dividing one with high regenerative capability.

4. Assess reproductive potential of mosaics

Hypothesis. Mosaics produce haploid gametes.

Task. Sample gonad tissues via flow cytometry for evidence of production of haploid cells (products of meiosis from diploid gonia).

From mosaics isolated from west coast populations (in year one of this study) and those isolated from our own spawnings ( in year two), we will obtain biopsies of the gonadal tissue during sexual maturation. Because of the nature of meiosis during gametogenesis, the gonadal tissue is normally mosaic in constitution (Allen et al., 1986; Guo and Allen, 1994a). Diploid and triploid males and females all have a distinctive signature pattern of ploidy composition. Those patterns and the expected ones derived from mosaics are summarized in Table 2.

EXPECTED AND PRELIMINARY RESULTS

1. To determine the origin of heteroploid mosaics by comparing the frequency of mosaics in two types of triploids: those produced by standard induction procedures and those produced from matings of tetraploids with diploids

We expect to be able to prove or disprove the null hypothesis that mosaics are produced as an artifact of the treatment process using CB, and by extension, other cytokinetic inhibitors. This result establishes only whether or not baseline proportions of mosaics are present in induced populations, and not whether triploids have a stable chromosome constitution.

Our preliminary data show no difference in the frequency of mosaics in embryos between induced and mated triploids. Table 3 contains data from the original certification of triploids from the 6 groups (3 CBs and 3 DTs).

2. Determine the stability of chromosome constitution of triploid populations

This is a very simple experiment designed simply to observe changes in individual ploidy profiles that might indicate reversion. This data could take the form of secondary peaks, whether diploid or aneuploid, or changes in the nature (mean or CV) of the baseline data. Because we are sampling so many individuals, we will be able to generate a frequency distribution histogram of mean DNA content among triploids to establish the variability in DNA content of the population as a whole. Then, if reversion of individual oysters occurs, we will be able to see whether the reverts are those individuals in the tails of the population (of DNA content) or randomly distributed. Sampling in progress.

3. Determine the fate of mosaics through time

We expect that if the occurrence of mosaics is a dynamic process, there should be a progression in the proportion of diploid cells within an individual. Static proportions of diploid cells over time supports the null hypothesis. Sampling in progress.

4. Assess reproductive potential of mosaics

This objective is a preliminary probe into the question of whether mosaics are reproductively competent. Several scenarios can be envisaged. If reversion is absent and mosaics are stable and no haploid cells are found, then they probably pose no problem for population control. The converse of this is that reversion occurs, is dynamic and results in the production of haploid cells. This is a severe problem. Intermediate scenarios are also possible, and we intend to illustrate these risk factors. However, a detailed analysis of gamete production in mosaics is outside the scope of this proposal.

Using FCM, we have examined 36 mosaics, looking at ploidy in about 4-5 tissue types, including gonad. With a few exceptions, mosaics have had 2n and 3n cells in all tissues examined. In the exceptions, one tissue type would have either diploid or triploid cells. There has been no evidence of haploid gamete production in any of them, so far.

DISCUSSION

The problem of risk management of marine GMO's. The culture or mere testing of marine GMOs poses unique problems for risk management. First, marine culture most frequently employs some sort of impoundment in open systems, either because the biology of the species requires water flow or because of economic considerations. Second, while it might be possible to physically contain large numbers of adults, it will not be possible to contain their gametes. Gene flow, e.g., between GMOs and native populations, will be a function of interpopulational exchange of planktonic larvae that persist in the water column anywhere from minutes to months, depending on the species. Indeed, even small scale tests of transgenic marine species would be the equivalent of large scale agricultural plots which are only now under consideration (Stone, 1994).

Despite these unresolved difficulties over containment, work continues on the production of transgenic marine species. These efforts include a number of fish (Chen and Powers, 1990) and shellfish species, e.g., abalone (D. Powers, Hopkins Marine Lab, personal communication), clams and oysters (T. Chen, COMB, U. Md., personal communication). Ultimately, the commercial feasibility of genetically modified marine species will rest on the demonstration that they will not interbreed with native species. Many investigators working on transgenic marine species have embraced the use of triploids or monosex triploids for this purpose.

Induced triploidy is a mature technology and widely available for population control. That triploid aquatic species has become a common tool and could be used for population control is evident by the maturity of the field of chromosome set manipulation. At the 1994 meeting of Genetics in Aquaculture held in Halifax, Nova Scotia, Canada, at least a third of the papers presented were on ploidy manipulation, either induced triploids or gynogenesis, and represented contributions from all over the world, including so-called third world nations. The popularity of this topic is an attestation to the practicability of the technology and the potential value of chromosome set manipulation for aquaculture and population control. For aquaculture, triploids are used in several industries because of their reduced sexual maturation. The west coast oyster industry produces approximately 15 billion eyed triploid larvae per year (Allen, 1988). Triploid oysters are commercially valuable because they produce significantly less gonad than normal diploids (Allen and Downing, 1986) and, consequently, are more marketable during certain seasons. Triploid rainbow trout are cultured, especially monosex populations, in commercial facilities in Great Britain.

The best example of the use of triploids for population control is grass carp in the US. An almost universal characteristic of triploid fish is the failure of triploid females to produce any eggs (Thorgaard and Allen, 1986), although significant sperm production is possible in males. Nevertheless, it is generally accepted that 100% triploid populations of fish are sterile and therefore will not reproduce. Data from cytogenetic analyses of sperm from triploid male grass carp (Allen, 1986) and rainbow trout (Benfey et al., 1986) confirm this. Recently New Brunswick (Canada) passed an ordinance that pen reared Atlantic salmon must be triploid, to prevent their interbreeding with native populations (T.J. Benfey, University of New Brunswick, personal communication), and the Alaska Department of Fish and Game is considering a similar requirement for rainbow trout stocked in the Yukon river drainage (J. Seeb, AK Dept. Fish and Game, personal communication).

Triploids can not reproduce themselves. To date, there is but a single example of the use of triploids for population control in the marine environment. Our lab was interested in field tests of the non-native species, Crassostrea gigas, on the east coast to determine whether or not C. gigas was resistant to the oyster diseases ravaging the native species. Estimation of reproductive likelihood in triploid oysters was not quite as simple as the case for fish. Triploids do in fact make significant numbers of eggs and sperm (Allen and Downing, 1990).

Our analysis of reproductive potential of triploid Pacific oyster revealed that although gametes from triploids were fully capable of fertilization, aneuploid progeny resulted (Guo and Allen, 1994a). When triploids were crossed with themselves, the ploidy of resulting embryos was 2.88n on average, that is, hypo-triploid. Survival of fertilized eggs to metamorphosis and settlement was only about 0.0085%. More recent data obtained this summer show that triploid males are about 1000 fold less fecund than diploid males; triploid females about 20 times less fecund. So, although triploid oysters are not sterile in terms of gamete production, their reproductive potential is extremely low, by all practical measures, 0.

Our risk analysis on the reproductive potential of triploid oysters for the sake of population control enabled us to obtain permission from state authorities in New Jersey and Virginia to plant several hundreds of triploid C. gigas in Delaware and Chesapeake Bays for disease challenges. The principal requirement of this work was that the oysters must be certified as triploid before placement in either estuary. Certification required pre-screening over 1300 oysters individually by taking a biopsy and examining DNA content by flow cytometry (FCM). This tedious requirement illustrates another impediment to the use of triploids for population control of either non-native species or GMOs: the need for 100% triploids.

100% triploidy is feasible. Until very recently, the production of spawns of 100% triploids seemed all but impossible. This is because the state of the art for making triploids involved an induction procedure in which the newly fertilized egg is poisoned with an antibiotic, usually cytochalasin B (CB), to cause the failure of cytokinesis during the elimination of the second polar body (PB2) (Allen et al., 1989). The chromosome contained in the polar body contributes the third chromosome set to the embryo. Because the treatment (whether CB or anything else) has to be coordinated with the elimination of the second polar body and because PB2 elimination in a population of newly fertilized eggs is subject to variability inherent in any biological system, some eggs invariably escape treatment and remain diploid. This imprecision gives rise to populations of oysters with varying proportions of triploids. For perfect biological containment, pure triploid populations are necessary.

Our lab has also been working on this problem. In summer of 1993, we were successful in creating the first viable tetraploid bivalves, specifically C. gigas. Tetraploids theoretically hold the potential to create all triploid populations because they should produce diploid gametes. Tetraploid x diploid crosses should yield triploids.

We have begun to evaluate the reproductive potential of tetraploids and the early data are encouraging. In studies in August 1994, we found that fecundity of tetraploid females was relatively high, only slightly lower than diploids.

We have also learned that the breeding potential of tetraploids for creating triploids is high. In experiments conducted to examine survival of 2 x 2 crosses between diploids and tetraploids, survival of [4n x 2n] and the reciprocal were at least as high as the diploid controls, and two orders of magnitude higher than triploids produced by standard induction procedures (Guo et al., 1996). Finally, we also examined the ploidy of progeny derived from 2 x 2 crosses of diploids and tetraploids in a number of replicates. The proportion of triploids produced from 4n x 2n crosses was, compared to standard induction techniques, almost perfect (exception: 1 tetraploid progeny produced) (Guo et al., 1996). It is feasible to create 100% triploids using a tetraploid breeding population.

A careful look at triploids reveals mosaics. Up to this point in the discussion, several things are clear. Practically the only form of containment for GMOs of marine species is biological containment, and biological containment is possible via triploidy. In C. gigas, the only marine species in which risk analysis of biological containment has been performed, triploids are suitable, 100% triploids are feasible, and field tests with C. gigas in a non-native setting have been allowed. Extension of this work to genetically modified oysters and perhaps marine species in general seems possible.

It was therefore with surprise that we obtained data that suggests triploids may not be stable over time. Routinely, our lab routinely assays triploid populations to assess differential survival and to separate the triploids and diploids in mixed populations. In the course of this work, we sometimes observe heteroploid mosaics, that is, individual oysters that have populations of both diploid and triploid cells. The occurrence of mosaics themselves is not particularly surprising since the triploid induction process affects cytokinesis in newly dividing embryos. Abnormal progeny, such as mosaic individuals with two cell types, might be expected as a matter of course.

The surprising result was that the frequency of mosaics in several triploid populations increased over time, suggesting that some triploids may have a tendency to lose chromosome sets. We have three data sets with such evidence. Data set 1: In 1993, our lab examined a large population of induced triploids to separate the triploids from the diploids. Of 1299 individuals, 487 were diploid, 787 triploid, and 25 mosaic (1.9%). Because the sample size is so large, we have a very accurate estimate of the proportion of mosaics that is present in the entire population of triploids. Again, mosaics might be expected at some baseline proportion if they happen as a matter of course during induction. We then removed all diploids and mosaics from the population, leaving only triploids that were used for a disease exposure field experiment. In five consecutive sampling periods, the percentage of mosaics (total sample size) was 0% (n = 20), 11% (n = 28), 15% (n = 26), 4% (n = 24), 20% (n = 83). A replicate site had 10% mosaics by the last sampling period. Data set 2: Another example of this same phenomenon was observed in an triploid experiment run over the course of 2½ years. Three triploid populations were sampled periodically for percent triploidy (n = 100 from each cross at each sampling period). In the first 1½ years, no mosaics were discovered during sampling. In the last three sampling periods, mosaics were discovered in all three populations in frequencies as high as 20%. Data set 3: Finally, the results of ploidy analysis on another population of triploids that we followed for almost two years are shown in Table 4 below. What these data demonstrate is an increase in mosaics over time with a corresponding decrease in the percentage of triploids, suggesting that the mosaics are coming at the expense of the triploids.

Individually, any one of the examples demonstrating an increase in the percentage of mosaics over time is not compelling, principally because mosaicism is most often a low frequency event and statistical differences among low frequency events are difficult to show. However, taken together, there is a clear suggestion that there may be more to mosaics than can be explained by abnormalities caused to embryos during CB treatment.

What we know and don't know about mosaics. The problem of mosaicism is a vexing one. The classic definition of mosaicism is the presence of two or more cell lines of different chromosomal constitution, both cell lines derived from the same zygote. In our case, it is the presence of triploid and some other cell type(s) within the same oyster. This other cell type is generally diploid, although (i) whether or not the "diploid" cells are euploid is unknown; (ii) there can be more than one other cell type, as has been recently found in our lab among tetraploid oysters; and (iii) some mosaic conditions, like that found in the gonad of triploids, is natural because of the process of meiosis. The presence of mosaics among triploid populations is generally unappreciated for two major reasons. First, it requires some level of sophistication in ploidy analysis, for example, FCM, to find mosaics. With FCM, the frequency distribution histograms of mosaics appear as distinct ploidy types, usually triploid and something else (Figure 1). Techniques using simple procedures like nuclear size or whose methods involve the analysis of relatively few numbers of cells (100 - 200) probably would fail to detect mosaics. Even with FCM, where thousands of cells can be analyzed quickly, there is a question of whether a sample with two or more distributions of DNA content is the result of an artifact of sampling, an accidental mixing of cells, or a true mosaic. The second reason mosaics have gone unnoticed is that they generally occur in very low frequency (e.g., <2%), although if sample size is large enough they always seem to be found. For example, on sabbatical, Allen recently examined eight population of triploids and found mosaics in six, ranging in percent mosaics from 0.6 to 7.5%. However, in general, very few labs re-examine large numbers of animals from triploid populations.

We have ruled out artifact as an explanation for mosaics. We have found and re-confirmed about 30 mosaics by taking repeated biopsies. In no case were findings of mosaicism disputed. In addition, we have verified that mosaicism occurs throughout the body by examining five distinct tissues (gill, adductor muscle, mantle, labial palps, hemolymph) from about 25 individuals. Also, we never find mosaics among the diploids, which would be expected if mosaics were an artifact of sample preparation or technique.

There are two hypotheses for the origin of mosaics. This proposal is designed to distinguish between the two. The first hypothesis is that mosaics arise during the process of triploid induction, whereby the polar body is inhibited and incorporated into only one of two, or one of four, blastomeres, giving rise to two stem cell populations. Cytological evidence exists that shows the retained polar body may be incorporated after the first mitosis (Komaru et al., 1990; Guo et al., 1992). Under such a scenario, different tissues would have varying compositions of diploid or triploid cells, depending upon the fate of the blastomeres during epiboly and development. Another expectation of this hypothesis is that the frequency of mosaics is randomly distributed with respect to different triploid populations, that is, the occurrence of mosaics is a stochastic event and has no relationship to, for example, environmental variables or age of the oyster. A third expectation is that the relative proportion of diploid and triploid cells in the oyster should remain stable with time, that is, there should be stability in frequency of both cell types within an individual.

The second hypothesis to explain our observations is that reversion occurs in triploid cells back to the diploid state. Reversion may be a feature of triploidy per se or it may be a consequence of creating triploids by induction methods. Hypothetically, triploids would begin to lose chromosomes or chromosome sets through some sort of mitotic miscue, particular only to triploids. A resulting diploid cell, or cells, then might act as a stem cell line and, possibly, out multiply triploid cells, increasing in frequency relative to triploid cells. Under this scenario, we might expect oysters to also vary in ploidy composition among different tissue types, with those tissues that are most rapidly dividing showing the first or most evidence of reversion. A second expectation would be that there might be some correlation between frequency of mosaics and age (as we have seen in the preliminary data) or an environmental correlate where certain environmental conditions might promote reversion. And a third expectation under this hypothesis would be that the relative proportion of diploid and triploid cells in mosaics will change with time, whereby the diploid population increased in frequency.

Revealing DNA content(s) through FCM. Our principal means of analyzing mosaics has been FCM. There are several levels of analysis of flow cytometric data. A discussion of them will make the experimental design more clear. The unit of measure in the FCM data is the fluorescence emitted from a nucleus that has been stained with the fluorescent dye, in our lab, 4',6-diamidino-2-phenylindole (DAPI). The fluorescence emitted is dependent upon a number of things, some real and some artifact.

Real variables that affect the fluorescence emitted from a nucleus are the amount of DNA (hence the amount of stain attached to it) and the way in which the DNA is packaged. DNA packaging is especially relevant with some stains, such as, propidium iodide, a stain that intercalates in the major grooves of DNA. Chromosomes tightly wound or packed with histones, e.g., in sperm, will not bind as much dye as naked DNA. This is not a problem with DAPI which binds covalently.

Artifacts that affect the apparent staining intensity of the nuclei include variation in the concentration of the stain, improper storage of samples, ionic composition of the staining solution, and a myriad of variables that are generally held constant by the flow cytometer itself. Because of these factors, standardization of the protocol among samples is essential. Our standardization procedures are rigorous.

The fluorescence emitted from the stained nucleus is one of many such measurements taken from an individual, or more correctly, from the cells of the tissue sampled from the individual. With FCM, thousands or even tens of thousands of cells can be measured in minutes. The frequency distribution generated from this population of measurements yields a frequency distribution histogram (e.g., Figure 1) can be described by a mean and a coefficient of variation (CV). The mean and CV so obtained is that of a single individual. The population as a whole can also be described by a mean and CV of the individual means. Sampling 100,000 nuclei versus 1000 does not necessarily yield more accurate statistics, but measuring large numbers of nuclei is useful in detecting secondary peaks, such as those found in mosaics.

REFERENCES

Allen, S. K., Jr. 1983. Flow cytometry: assaying experimental polyploid fish and shellfish. Aquaculture 33: 317-328.

Allen, Jr., S.K. 1988. Triploid oysters ensure year-round supply. Oceanus 31: 58-63.

Allen, S. K., Jr. and S. L. Downing. 1986. Performance of triploid Pacific oysters, Crassostrea gigas (Thunberg). I. Survival, growth, glycogen content, and sexual maturation in yearlings. J. Exp. Mar. Biol. Ecol. 102: 197-208.

Allen, S.K., Jr. and S.L. Downing. 1990. Performance of triploid Pacific oysters, Crassostrea gigas. Gametogenesis. Can. J. Fish. Aquat. Sci. 47: 1213 - 1222.

Allen, Jr., S.K. and P.M. Gaffney. 1993. Genetic confirmation of hybridization between Crassostrea gigas and C. rivularis. Aquaculture 113: 269-289.

Allen, Jr., S.K., Downing, S.L., and Chew, K.K. 1989. Hatchery Manual for Producing Triploid Oysters. University of Washington Press, 27 pp.

Allen, Jr., S.K., R.G. Thiery, and N.T. Hagstrom. 1986. Cytological evaluation of the likelihood that triploid grass carp will reproduce. Trans. Am. Fish. Soc. 115: 841-848.

Allen, Jr., S.K., P.M. Gaffney, J. Scarpa, and D. Bushek. 1993. Inviable hybrids of Crassostrea virginica (Gmelin) with C. rivularis (Gould) and C. gigas (Thunberg). Aquaculture 113: 291-300.

Benfey, T.J., I.I. Solar, G. de Jong, and E.M. Donaldson. 1986. Flow-cytometric confirmation of aneuploidy in sperm from triploid rainbow trout. Trans. Am. Fish. Soc. 115: 838-840.

Chaiton, J.A. and S.K. Allen, Jr. 1985. Early detection of triploidy in the larvae of Pacific oysters, Crassostrea gigas, by flow cytometry. Aquaculture 48: 35-43.

Chen, T.T. and D.A. Powers. 1990. Transgenic fish. Trends in Biotechnol. 8: 209-215.

Guo, X. and S.K. Allen, Jr. 1994a. Reproductive potential and genetics of triploid Pacific oysters, Crassostrea gigas. Biol. Bull. 187: 309-318.

Guo, X. and S.K. Allen, Jr. 1994b. Viable tetraploid in the Pacific oyster (Crassostrea gigas Thunberg) produced by inhibiting PB i in eggs from triploids. Mol. Mar. Biol. Biotechnol. 3: 42-50.

Guo, X., Cooper, K., Hershberger, W.K., and Chew, K.K. 1992. Genetic consequences of blocking PB I with cytochalasin B in fertilized eggs of the Pacific oyster, Crassostrea gigas: II. Segregation of chromosomes. Biol Bull 183:387-393.

Guo, X, G.A. DeBrosse and S.K. Allen, Jr. 1996. All-triploid Pacific oysters (Crassostrea gigas Thunberg) produced by mating tetraploids and diploids. Aquaculture 142: 149-161.

Komaru, A, H. Matsuda, T. Yamakawa, and K.T. Wada. 1990. Chromosome behavior of meiosis inhibited eggs with cytochalasin B in Japanese pearl oyster. Nippon Suisan Gakkaishi 569: 1419-1422.

Scarpa, J. and S.K. Allen, Jr. 1992. Comparative kinetics of meiosis in hybrid crosses of Crassostrea gigas and C. rivularis with C. virginica. J. Exp. Zool. 263: 316-322.

Stone, R. 1994. Large plots are next test for transgenic crop safety. Science, 266: 1472-1473.

Thorgaard, G.H. and S.K. Allen, Jr. 1986. Chromosome manipulation and markers in fishery management. In N. Ryman and F.M. Utter, eds. Population Genetics and Fishery Management. University of Washington Press, Seattle. pp. 319-331.

Table 1. Schematic of diallele crosses between diploid and tetraploid C. gigas. Shown are PLOIDY of the female and male parent and the DNA content of the gametes from them; the ploidy of the progeny groups resulting from these matings are shown in the boxes and are explained below in the text.

FEMALE

Ploidy

2n

4n

	Ploidy	DNA	1c	+PB2 = 2c	2c
MALE	2n	1c		INDUCED* 3npb2	MATED** 3n(4x2)
	4n	2c	MATED 3n(2x4)

*Induced: 3n_pb2 -- Induced triploids are produced by inhibition of the PB2 with CB in a diploid x diploid cross, a standard procedure now used commercially (Allen et al., 1989).
**Mated: 3n_(4x2) -- A mated triploid results from the combination of a fully reduced (but 2c) egg of a tetraploid with a normal 1c sperm of a diploid. There is no effect from CB treatment. 3n_(2x4) -- A mated triploid also results from the combination of a fully reduced (1c) egg of a diploid with the sperm (2c) of a tetraploid. There is also no effect from CB treatment.

Table 2. Observed and expected mosaic constitutions of cell types in gonads from diploid, triploid, and diploid/ triploid mosaic oysters. We have observed mosaic constitutions in diploid and triploid gonads, and this condition is normal in and in accordance to meiotic events in the gonad (pre-meiotic doubling followed by two maturation divisions in the male; only pre-meiotic doublings in the female). Constitution of gonadal tissues in heteroploid mosaics is unknown. -- indicates that cells of this ploidy are expected.

Sex

Ploidy of oyster

Cell types within gonad 1n 1.5n 2n 3n 4n 6n

MALE	Diploid
	Triploid
	2/3n Mosaic1
FEMALE	Diploid
	Triploid
	2/3n Mosaic

¹Expected distributions based on assumption that both diploid and triploid cells are contributing gonia to the maturing gland

Table 3. Results of original certification of 9 month-old triploids from CB and DT groups, before field deployment. Only triploids were deployed to field.

		2n	3n	2/3n	TOTAL	% Triploid	% Mosaic
CB	1	11	356	3	370	96.2	0.8
	2	43	367	6	416	88.2	1.4
	3	84	319	6	409	78.0	1.5
	Total	138	1042	15	1195
DT	1	0	405	0	405	100.0	0.0
	2	0	357	0	357	100.0	0.0
	3	0	394	0	394	100.0	0.0
	Total	0	1156	0	1156

Table 4. Ploidy of oysters subsampled over a 20 month period showing % triploids and % mosaics. Data demonstrate the covariance between these two measures.

Age (months)	n	A: % triploid	B: % mosaic	TOTAL: A + B
0.1	larvae	98	---	98
2	50	96	0	96
7	100	97	0	97
11	98	96	2	98
15	99	95	11	97
20	509	91	7	98

Figure 1. Example of a mosaic individual Crassostrea gigas, age 2+. Cells from the gill were dissociated, nuclei stained with DAPI, and examined on flow cytometer. About half of the cells are something other than triploid (mean channel number, i.e. relative DNA content (128). In this case, there are diploid cells (mean channel number, 86) and aneuploid 1.5n cells (mean channel number, 62). Aneuploid cells result from meiosis in the gonads of triploid, which reduces the chromosome complement by half, and is a normal process in triploids. The diploid cells, however, are of unknown origin.