Reproduction and development.

The currently developed and validated in vitro tests for female and male fertility and also for developmental toxicity are described and evaluated according to their potential use as screening or replacement alternatives to the established in vivo tests in reproductive and developmental toxicology. Alternative methods today can only be used to evaluate a few specific components of the integrated reproductive functions in both females and males. However, in the field of developmental toxicity testing there is a strong theoretical and empirical basis for the predictive power of in vitro screens using mammalian embryos as well as embryonic cells and tissues. Several of these assays have been validated or are currently undergoing validation in several laboratories and are > 80% concordant with in vivo results. Failure to achieve 100% accuracy reflects the inherent limitations of these systems, which are manageable, as the concordance rates are still good. The level of concordance suggests that these assays are adequate for screening purposes to complement traditional in vivo testing. The use of these assays as screens will save valuable in vivo testing resources for those compounds most likely to enter the market and to which people will be exposed.


Introduction
Reproduction is a continuous cycle. For the conducting mechanistic reproductive purposes of toxicity testing, however, it is toxicity studies. Additionally, in vitro methbroadly divided into pregnancy in females, ods already play a valuable role in so-called including prenatal or postnatal developsecondary testing, i.e., in the screening of mental toxicity, and the remainder of the series of structurally related chemicals when cycle in both males and females during at least one of the chemicals is of known which fertility may be impaired.
reproductive toxicity in vivo. During the past 20

years research in
The majority of research into the develreproductive toxicology has focused on opment of alternative tests has concentrated the use of alternatives to living mammals on teratogenicity, which is only one manifor testing the potential reproductive toxi-festation of adverse effects on development cities of chemical and physical agents. and does not cover fertility, which includes Recent reviews include an Organisation for sexual behavior, spermatogenesis, oogenesis, Economic Co-operation and Development fertilization and the development of the workshop in Ottawa, Canada, in 1992 (1) zygote up to term, postnatal development, and a European Centre for the Validation of and hormonal activity.
Alternative Methods (ECVAM) workshop InVivoTesting for in Ispra, Italy, in 1994 (2). International IngVivo Purpor experts concluded at both conferences Regulatory Purposes that the use of in vitro methods is well Currently, reproduction/development established and that they are invaluable for screening tests (3,4) or multigeneration This paper was prepared as background for the 13th Meeting of the Scientific Group on Methodologies for the Safety Evaluation of Chemicals (SGOMSEC): Alternative Testing Methodologies held 26-31 January 1997 in Ispra, Italy. Manuscript received at EHP9 May 1997; accepted 29 May 1997. studies must be conducted to provide information on the effects of industrial chemicals on all aspects of the highly complex reproductive cycle (5,6). For chemicals used as drugs, segment studies must be conducted covering three important phases of preand postnatal development and fertility (7). Because of the complexity of the reproductive cycle, from gamete maturation through implantation of the early embryo into the uterus, and because of the lack of validated alternative tests for most steps in the cycle, testing in living animals is the only option currently available for assessing the possible effects of chemicals on reproduction. Moreover, because of the complexity of functions that are only found in living animals, in vitro screening may never be able to cover all of the aspects of fertility. Thus, the key question is whether sufficient information can be derived from alternative tests to be able to classify and label chemicals as toxic to the reproductive system.

Fertility: In Vitro Approaches
Female Fertility Some aspects of female reproductive function can be modeled in vitro, and several cellular components of the female reproductive organs can be maintained in culture (8). Although none of the organs have been used or validated as toxicity screens, several may be useful for specialized toxicologic studies. In the future, a battery of such systems may be able to cover a large proportion of the female reproductive cycle.
In females, the proliferation of primordial germ cells and the initial steps of meiosis occur long before birth. From puberty onward, a small number of the primary oocytes complete oogenesis and are released from the ovary. Ovarian somatic cells (granulosa, thecal, and stromal cells) can be maintained in culture (8) and any adverse effects can be assessed by examining cell morphology, viability, and hormonal responsiveness.
Techniques for in vitro fertilization (including functional maturation of spermatozoa) are routine, both clinically and in farm and laboratory animals. The methods used have not been validated for testing purposes, although some toxicologic studies have been performed (9). For example, the mammalian preimplantation period can be investigated by culturing embryos from the first cleavage divisions up to implantation, Environmental Health Perspectives * Vol 106, Supplement 2 -April 1998 and toxicologic investigations at the chromosomal level have been published (10). Thus, there are no alternative tests available in the near future that would enable the screening of chemicals for female reproductive toxicity with the predictivity required for the safety assessment of chemicals. In the long term, a complex battery of in vitro tests may be devised.

Male Fertility
There is particular interest in the development of alternative approaches for assessing male reproductive toxicity. The human male has a relatively low sperm count; the number of sperm per ejaculate is typically only between 2-and 4-fold higher than that at which fertility is significantly impaired. In contrast, the number of sperm in a rat or rabbit ejaculate is many times (up to 1000-fold) that which will produce maximum fertility. Epididymal sperm count can be reduced by as much as 90% in the rat without significantly affecting fertility. Consequently, animal models may be insensitive indicators of human reproductive hazards. Studies on male reproductive toxicity are aided by the ready availability of human target cells. The production of spermatozoa from stem cells is a complex process that takes about 5 weeks in mice and 11 weeks in humans. Chemicals can disturb normal spermatogenesis by direct interaction with targets within the testis itself, or indirectly by interfering with hormonal stimulation or alterations in blood supply.
It is not possible to mimic the whole of the male reproductive cycle in vitro, but several components can be studied individually. Although they have been used extensively in toxicologic studies, they still do not represent a viable alternative to in vivo tests. Male germ cells are produced from stem cells throughout mature life. Thus, it should be easier to devise culture systems that are able to support the whole of spermatogenesis.
Several testicular cell types can be maintained in culture, either alone or in combination; these include Sertoli-germ cell cocultures, Sertoli cell-enriched cultures, germ cell-enriched cultures, Leydig cell cultures, and Leydig-Sertoli cell cocultures. All of these systems have been used successfully to study specific features of testicular toxicity (11). Primary cultures of testicular cells retain many of the in vivo characteristics but only have a limited life span. The ability to study individual cell populations from a heterogeneous organ such as the testis is a powerful tool for probing mechanisms of toxicity. However, the loss of interactions with other cell types is a serious limitation to their use for screening purposes.
Additional measurements may be incorporated into current in vivo testing protocols. High quality histopathology of the testes and epididymis would enable effects on specific cell populations to be evaluated. Alternatively, recently developed flow cytometric techniques could be used. One of the flow cytometric methods developed for detecting alterations in spermatogenesis allows simultaneous measurement of cellular DNA content, RNA content, and stainability (12). This procedure provides a rapid assessment of up to eight different testicular cell populations.

Semen Analyses
There are several techniques available for monitoring sperm motility, morphology, and various other aspects of semen composition, including the fertilizing ability of sperm. Such approaches could be used for both human and animal semen analyses. The direct addition of test chemicals to semen samples in vitro may be a valuable approach, given the availability of human material. Fertilizing capacity declines with increasing proportions of abnormal sperm head morphology, and there is a high correlation between chemical-induced sperm head abnormalities and altered sperm chromatin structure in the mouse (13).

Developmental Toxicity: In Vitro Tests
Over the past 20 years, more than 30 different culture systems have been proposed as tests for developmental toxicity. The majority of these tests have each been used by only one laboratory. The culture systems fall into four categories: established cell lines, primary cell cultures, nonmammalian embryos, and mammalian embryos or primordia.

Cell Lines and Embryonic
Stem CelLs Established cell lines that have been used for developmental toxicity screening include human embryonic palate mesenchymal (HEPM) cells (14), mouse ovarian tumor (MOT) cells (15), and neuroblastoma cells (16). The results of a blind trial with a dual HEPM/MOT approach showed an unacceptably high level of false positives (> 50%) (17).
In a more recently developed approach, blastocyst-derived totipotent embryonic stem (ES) cell lines of the mouse were used for in vitro embryotoxicity testing. ES cells can be maintained in an undifferentiated state in the presence of feeder layers and/or purified leukemia inhibitory factor (LIF). When the LIF is removed, ES cells differentiate into a variety of cell types depending on the culture conditions. For example, in the mouse, muscle cell differentiation from ES cells reflects myogenesis in vivo (18), and the development of hemopoietic cells parallels hematopoiesis in the developing embryo (19).
ES cells offer several new approaches with respect to screening for embryotoxicity in vitro, enabling the use of differentiating permanent embryonic cells. Cytotoxicity (20) and effects on differentiation (19,21) have been used as end points in embryotoxicity tests with ES cells; inhibition of the differentiation of ES cells in a micromass culture has proved a particularly promising assay under routine testing conditions (21). Determination of the two essential features of embryotoxic agents (i.e., inhibition of differentiation combined with a higher sensitivity of embryonic cells than adult tissues to cytotoxic damage) in a single assay procedure with ES cells was recently attempted (22). The predictive value of this embryonic stem cell test (EST) was as good as the results obtained with in vitro embryotoxicity tests using either rodent whole embryos or embryonic tissues.
ES cells are routinely used in the production of transgenic mice, and methods to introduce targeted mutations and reporter constructs are well established. Transgenic markers could be devised to simplify the end points used in a particular toxicity test and to enable the automation of such assays. ECVAM supports the continued development ofthese approaches.

Aggregate and Micromass Cultures
Aggregates of primary cultures of chick embryo neural retina cells (CERC) provided encouraging results as a screen for developmental toxicity (23). The CERC assay may not have gained wider acceptance because it is based on the differentiation of nonmammalian cells from an organ that is not the primary target of embryotoxic agents in the human embryo.
When cells from the undifferentiated mesenchyme of early chick embryo limbs were cultured in small volumes at high density, they formed numerous small foci of differentiating chondrocytes within a background of apparently undifferentiated cells. Cell adhesion, movement, communication, Environmental Health Perspectives * Vol 106, Supplement 2 a April 1998 division, and differentiation all occur in micromass cultures (24). In principle, the micromass test is based on detecting the ability of a particular chemical to inhibit the formation of foci. Embryonic limb or central nervous system (CNS) cells (usually midbrain, which form foci of neurons) from chick, mouse, or rat can be used (25,26). The technique has subsequently been modified for use with 96-well microtiter plates (27). Cells can be exposed either directly in culture or transplacentally prior to culture (28).
Several structure-activity studies have shown that the micromass test can distinguish between teratogens and nonteratogens within a particular chemical class, e.g., retinoids (28) and triazole antifungal drugs (29). In some cases, organ-, species-, and strain-specific toxicity have been modeled in micromass cultures. For example, ethylenethiourea is more toxic to midbrain than to limb cultures (30) and is more toxic to the rat CNS than to the mouse CNS (31).
Differentiating cells in both midbrain and limb cultures express cytochrome P450 isozymes (32) and, at least in the case of limb cells, these are able to metabolize chemicals such as phenytoin and cyclophosphamide to their toxic metabolites. Validation studies using chemicals from a variety of classes indicate that the percentage of teratogens detected with micromass cultures may vary between approximately 60 to 90%, and that the percentage of nonteratogens identified correctly may vary between 89 and 100% (26,27,33). It is possible that much of this variation is accounted for by differences in the exact methodology used, and none of these studies are considered definitive.
Thus, micromass cultures represent robust test systems for studying potential teratogens. It was recommended that the micromass method be included in a comparative trial to determine its applicability relative to several other available in vitro systems (2).

Embryos of Lower Order Spedes
Numerous tests that use embryos of submammalian vertebrate and invertebrate species for detecting the teratogenic potentials of chemicals have been described. The organisms that have been used include hydra, fish, frogs, crickets, Drosophila, brine shrimp, and slime mold. Several of these are currently being used extensively as models for investigating mechanisms of development. Because any stage or component of development is a potential target for toxicants, the existence of species differences is a strong argument in favor of using vertebrate models for predictive screening. However, subvertebrate systems may have applications in ecotoxicologic monitoring.
Of the nonavian vertebrate systems available, only the frog embryo teratogenesis assay Xenopus (FETAX) has undergone limited validation using about 40 different substances (34). The overall accuracy in predicting teratogenic potential has been claimed to be 79 to 83% (34). FETAX is low cost and rapid and uses a species commonly maintained under laboratory conditions. The assay is limited by the aqueous solubility of test substances, the relative lack of validation, and the small number of laboratories that have used the system. Nevertheless, it has been recommended that FETAX be induded in a comparative trial of alternative tests for developmental toxicity (1,2).

Avian Embryos
Although avian embryos are widely used as models in developmental biology, they have rarely been used for embryotoxicity testing. The chick embryotoxicity screening test (CHEST) was devised by Jelinek and co-workers (35) and has been used extensively in their laboratory, but not elsewhere. Intraamniotic injection eliminates the problem of continuous exposure of the embryo because the test substance is readily distributed to the extraembryonic compartments. Growth retardation, malformation, and death as well as dose-response and stage-response relationships and malformation spectra are easily determined. The results obtained from testing over 130 compounds have been published (35,36). One general problem with CHEST has been the inability to distinguish general toxicity from specific developmental effects.

Mammalian Whole Embryo Culture
Mammalian embryos can be maintained in culture for short periods throughout the phase from fertilization to the end of organogenesis (10). For toxicity testing, the period from the end of gastrulation to midorganogenesis has been investigated extensively. Screening systems using mouse (37) and rat (38) embryos have been proposed, and the culture of rabbit embryos has recently been optimized (39).
Head fold or early somite stage embryos are dissected free from maternal tissue, parietal yolk sac, and Reichert's membrane, leaving the visceral yolk sac and ectoplacental cone intact. The conceptus is cultured in medium under defined gassing conditions for 24 to 48 hr, usually in a roller bottle system. A variety of media have been used, all of which contain a high proportion of serum. Rat serum is most common (40), but mouse, rabbit, cow, monkey, and human sera (41) have been used. The test compound can be added to the cultures for defined periods or for the entire culture period. Metabolic activation systems can be incorporated, including the addition of S9 or microsomal fractions of liver from different species, coculture with hepatocytes, sequential hepatocyte/whole embryo culture, and the addition of serum from treated animals or humans (42).
At the end of the culture period a number of end points can be measured, including effects on the development of the visceral yolk sac vascularization and circulation; effects on hematopoiesis, embryonic growth (e.g., size and protein and DNA contents); and differentiation (number of somites, morphologic score); and dysmorphogenic effects (37,38,43). Interpretation of the results takes into account adverse effects on yolk sac development and embryonic growth and differentiation as well as adverse effects specifically on dysmorphogenesis. Validation studies have been carried out (43) and an interlaboratory validation study has been conducted (44). In a validation study on different culture systems, six pairs of coded compounds were tested in chick and rat embryo cultures and in brain cell aggregate cultures (45). Bechter et al. (46) reported an excellent agreement between in vivo and in vitro data for a series of retinoids.
Mammalian whole embryo culture systems are well developed in vitro tests for the detection of potentially teratogenic compounds and for the elucidation of mechanisms of teratogenicity. This in vitro system has been used in many academic and industrial laboratories and has proved to be a valuable tool. It allows the detection of dysmorphogenesis in many organs and the comparison of specific dysmorphogenic effects with general adverse effects on growth and differentiation. In addition, it enables the potencies of structurally related compounds to be ranked. Concentrations of test compounds and metabolites can easily be monitored in the culture medium and embryonic tissues. However, the system has clear limitations: it is relatively complex, covers only a part of organogenesis, and requires high technical skills. The test can be costly and it uses mammalian tissue and serum. Whether this is justified with respect to its use as a screen may be evaluated by including it in a comparative trial with other simpler in vitro systems.

Toxicokinetics and Metabolism
The production of a direct effect on the developing organism depends on the concentration/time relationship of the chemical and/or its active metabolite(s) in the target cells. Therefore, toxicokinetic and metabolism studies are of crucial importance for the design and interpretation of developmental toxicity studies with both in vitro and in vivo methods (47,48). In vivo target concentrations are dependent on maternal absorption of the compound, its distribution, metabolism, and excretion, and its placental transfer and distribution in the embryo. Toxicokinetic studies are also important in vitro. The presence of the compound and its stability in the culture medium must be verified, along with an assessment of its transport to, and uptake by, the tissues and cells in culture, its metabolic activation, and its cellular distribution. Toxicokinetic studies are essential for interpreting results obtained in vitro and for extrapolating these to the in vivo situation. Activities of added metabolizing systems, such as liver homogenate fractions, isolated enzymes, and hepatocytes, can be assessed by analytical techniques. Measurement of the compound in the cultured tissues and cells is critical, so the target concentration needed to yield an effect can be determined. Such measurements are especially important if little or no activity is observed in vitro so that false negatives can be excluded.
Toxicokinetic parameters often differ drastically between in vivo and in vitro situations. For example, in vivo drug levels can fluctuate markedly between doses because of the short half-lives of many chemicals, and high concentration peaks alternate with low or negligible drug levels. In contrast, the chemical is added in vitro to the culture medium and may persist for extended periods of time unless it is degraded by hydrolysis or enzymes present in the culture medium.

Discussion
The current status of in vitro tests for reproductive and developmental toxicity testing is summarized in Table 1. It is obvious that for reproductive toxicity, alternative methods can only be used to evaluate a few components of integrated reproductive functions both in the female and the male. Table 1 also shows that the situation is more promising in the field of in vitro embryotoxicity testing, as many tests can be used for screening purposes and a few have undergone interlaboratory validation with coded chemicals. Some of the tests using mammalian cells and embryos have provided promising results and may be used as screens to set priorities for in vivo testing for regulatory purposes, e.g., the ESTs, micromass cultures, and whole embryo cultures. Therefore, ECVAM is currently funding a combined prevalidation and validation trial of the three latter tests according to ECVAM's schemes for the prevalidation and validation of toxicity tests (49,50). One important limitation of the in vitro screening tests in developmental toxicity shown in Table 1 is that they have been selected to detect structural alterations but not any other potential manifestations of developmental toxicity. Therefore, one would not expect these tests to predict decreased fetal weight or mortality. Moreover, none of the tests would predict functional abnormalities induced by chemicals.
Therefore, even the best in vitro assays are limited in their biology because of the following facts: a) none of the tests represent the entire spectrum of developmental events and embryotoxic mechanisms; b) the metabolizing capacity of embryonic cells, tissues, and organs is limited; and c) the pharmacokinetic parameters in vitro are not identical to the situation in vivo.
However, the limitations of in vitro screens for embryotoxicity can be managed by test selection and design. In vitro systems can include specific metabolizing systems or be manipulated to approximate the pharmacokinetic behavior of the test agent in the species of interest. In fact, the ease with which such test systems can be manipulated may in some cases mean that limitations can be turned into advantages.
The primary reason for the general bias of toxicologists against in vitro screens is the unrealistic standards that they are Environmental Health Perspectives a Vol 106, Supplement 2 * April 1998 expected to meet. Despite protests to the contrary, it appears that the in vitro tests are expected to be as predictive as, or more predictive than, traditional in vivo screens. In other words, they are expected to meet the standards of replacement. For example, it is usually anticipated that in vitro tests are able to identify thalidomide as a teratogen despite the fact that thalidomide cannot be identified when tested in vivo in the most common rodent species. Today in vitro assays for developmental toxicants can be used for a number of product development and industrial purposes. The three major applications are as follows (51): a) in the earliest stage of product development, to select from among a group of candidate compounds for a particular indication those compounds that are the least likely to cause developmental toxicity; b) to compare the developmental toxicity potential of a new chemical that is only a slight modification of an existing chemical that has already been tested in vivo; and c) to evaluate compounds for which testing is not routinely performed, usually because the anticipated exposure is very low.
Eventually in vivo screening will only have to be conducted on compounds that passed the initial in vitro screen. In the latter two instances it may not be necessary to carry out any additional testing in vivo.
There are other applications of in vitro screens that should be considered seriously. The most significant potential application would be to set priorities for definitive testing of compounds that have been on the market for many years and for which no developmental toxicity data exist to date. Among these untested existing chemicals, in vitro screens would permit identifying compounds that may have developmental toxic potential. Using them will save time and money, but they will not replace in vivo testing for compounds to which people may substantially be exposed. They will, however, provide information on developmental toxicity for entire classes of compounds for which these data are never routinely obtained. Therefore, using validated in vitro tests will increase flexibility in product development and testing without compromising safety. In vivo tests will still be conducted on all new pharmaceutical agents and commercial chemicals for which exposure is significant. However, the in vitro tests will obviate the need to test in vivo materials when there is a high likelihood that their developmental toxicity potential would prevent their introduction into the market.