Human Reproduction vol 7 no.10 pp.1339-1341, 1992
OPINION
Co-culture for embryo development: is it really necessary?
Barry D.Bavister Department of Animal Health and Biomedical Sciences, and Regional Primate Research Center, University of Wisconsin, Madison, WI 53706, USA Numerous studies have shown, in several species including man, that co-culture with somatic cells or with conditioned medium can enhance in-vitro development of embryos. There are a variety of ways in which this improvement is manifested: by overcoming blocks to development, amplifying the numbers of embryos able to reach the blastocyst stage, increasing embryo cell numbers, and/or improving viability of embryos following transfer. Although co-culture can improve embryo development, however, these studies raise two major questions: how does co-culture exert its effect on embryo development, and is co-culture really necessary? In at least some applications, the need for somatic cell contributions can be eliminated by careful preparation of culture media. Nevertheless, examination of the plethora of studies on co-culture indicates that this technology is being accepted uncritically by some investigators as showing that somatic cells have 'embryotrophic' properties. This attitude sometimes leads to an a priori approach to the design of studies, which is inimical to good scientific principles. I will argue that, while co-culture can be useful under some conditions, in general, there is no evidence as yet for any embryotrophic activity, and the need for co-culture may reflect the use of inappropriate culture media or problems in their preparation. At worst, the apparent beneficial effect of co-culture is sometimes used to justify reporting of inconclusive data in the guise of research. The rationale for using oviductal cell co-culture in human in-vitro fertilization—embryo culture (TVF—EC) is logical: cleavage stages of development normally reside within the oviduct, and using oviductal epithelial cells in vitro may mimic the natural environment more closely. Conversely, the absence of oviductal influences in vitro may compromise the viability of IVF embryos. Reports of improved quality and/or development of embryos with co-culture fall into two categories: (i) empirical use of co-culture to improve embryo development and/or viability, and (ii) attempts to improve the co-culture technology itself, and perhaps to attribute the 'embryotrophic' effect to some secretory property of (e.g.) oviduct cells. These efforts have by no means been confined to human IVF —EC: a large number of studies, especially with cattle IVF embryos, have been conducted during the past 5 years or so, to the point that co-culture is frequently used almost unquestioningly as the method of choice for sustaining embryo development in vitro. The somewhat different goals reported above need to be kept © Oxford University Press
in mind when evaluating the utility of co-culture studies. With respect to (i), if co-culture with somatic cells enhances embryo development, then the benefits may outweigh the disadvantages. It may be easier in some locations to use co-culture, regardless of how it works, than to troubleshoot culture media problems to eliminate the need for co-culture. Even so, it would be wise to realize that the co-culture effect may be simply masking an underlying culture media problem, and co-culture aficionados might want to consider using a different formulation and/or being more critical about media preparation. Co-culture is not without a price, since it represents an additional layer of technology, with some risk of introducing disease organisms into the embryo culture environment, especially when animal cells are used for co-culture with human embryos (e.g. Weimer et al., 1989). Moreover, there is a danger of accommodating the culture needs of the somatic cells at the expense of the embryos (see 4 below). In the second approach (ii), attempting to 'prove' that co-culture is beneficial, and to characterize the effective agent(s), is fraught with problems. In the majority of studies in which an effect of co-culture has been claimed, although good supporting data are lacking, authors seem to prefer the explanation that somatic cells contribute some embryotrophic factor to the medium (we call this 'positive conditioning'). Yet removal of detrimental components from the culture environment ('negative conditioning' of the medium), such as the well-known case of hypoxanthine (Loutradis et al., 1987), is at least an equally valid explanation for the effects of co-culture on embryo development, and one for which there is good, albeit indirect, evidence. Problems with the 'positive conditioning' hypothesis are as follows. (1) First and foremost, the beneficial effects of co-culture can disappear when different culture media are used, or when the preparation of media is done more rigorously. For example, using 'standard' medium (TCM-199) containing serum, which is reported by several laboratories to cause a block to development of IVF cow embryos at the 8- to 16-cell stage, we found no such block: embryos developed to the morula/blastocyst stages with equal frequency in the presence or absence of oviduct cells (Pinyopummintr and Bavister, 1991). We ascribed this either to our very rigorous quality control of water used in culture media preparation, or to careful selection of the supplier of serum, or to the use of unheated serum in our experiments. (Incidentally, the necessity to heat serum used in cell culture is influenced by the age and species of the serum donor; and heating serum may adversely alter its cell growth supporting properties or even render it toxic.) Sheep zygotes also developed in vitro to blastocysts (85 %) in a simple medium supplemented with serum (Walker et al., 1992). This is in contrast to a study showing that 1339
B.D.Bavtster
sheep embryos were blocked in vitro unless co-cultured with oviduct cells (Gandolfi and Moor, 1987); this work, more than any other, stimulated the resurgence of interest in co-culture. Porcine embryos, too, can develop into blastocysts in vitro without co-culture (Reed et al., 1992; and see Table 1 of Walker et al., 1992). These contradictory data, arguing for and against a need for co-culture, surely arise from the selection and preparation of the culture media for embryo development. One way that somatic cells in co-culture might benefit embryo development is by stabilizing or altering physico-chemical conditions, such as the pH of the culture medium or oxygen/carbon dioxide concentrations, as a consequence of their metabolic activities. Adjustments in the gas levels used in culture can have striking effects on embryo development (Carney and Bavister, 1987; McKiernan and Bavister, 1990). The need for co-culture in particular locations might thus be eliminated by placing very strict quality controls on the environment actually being provided by the culture incubator, which is notoriously variable. (2) Reported beneficial effects of co-cultured cells on embryo development are neither species- nor tissue-specific. This detracts from attempts to demonstrate that oviduct cell secretory products, such as 'oestrus-associated' proteins, are embryotrophic. Neither is there evidence for cycle stage-specific effects of oviduct cells on embryo development, thus further arguing against a specific embryotrophic factor. Although oviductal proteins are reported to associate with cultured embryos (discussed by Bongso et al., 1990), there is as yet no evidence that they are responsible for improvements in embryo development observed in other studies. The crucial evidence is lacking, i.e. a demonstration that secreted proteins isolated from oviductal cell cultures improve embryo development. Without this information, data on oviductal protein secretion profiles should not be assumed to be relevant to embryogenesis, tempting though this may be. Other tissues shown to date to enhance embryo development, in a variety of species, include cumulus and granulosa cells, trophoblastic vesicles, uterine endometrial cells, monkey kidney cells and buffalo rat liver cells (the lack of co-culture tissue specificity is discussed by Bongso et al., 1990). It would be remarkable if all of these different cell types were producing the same specific embryotrophic factor, although this remains a possibility. On the contrary, the fact that such widely different cell types are effective when used in co-culture for embryo development could well indicate support for the 'negative conditioning' concept. (3) Many studies, particularly (but understandably) those with human embryos, claim beneficial effects of co-culture without reporting any control data, i.e. results of embryo culture using the same medium in the absence of somatic cells (there are a few notable exceptions (e.g., Weimer et al., 1989). Without control data, claims of enhanced embryo development using co-cultured cells cannot be evaluated. In a study with FVF cow embryos, using a factorial experimental design, the same medium (TCM-199) was tested with and without oviduct cell conditioning, so that any enhancement of development could be statistically analysed; none was found, and embryos developed equally well to morulae/blastocysts in both situations (Pinyopummintr and Bavister, 1991). In contrast, a study of human embryo development showed a reduction in blastomere fragmentation, together with a doubling of the incidence of pregnancy, with (bovine) 1340
uterine fibroblast co-culture (Weimer et al., 1989). Here again, perhaps a high incidence of fragmentation and loss of embryo viability with time in vitro is caused by unsuitable culture conditions, and somatic cell co-culture simply ameliorates this condition by rehabilitating the medium. It should also be noted that the gross appearance of embryos under a microscope can be misleading about their quality, especially if examined around the time of cleavage when major cytoplasmic rearrangements take place. Furthermore, analysis of FVF results for comparison of embryo treatments is constrained by the need for large numbers of transfers to provide statistically valid data; some studies may have failed to meet this requirement. (4) The emphasis on co-culture is sustaining a belief by some that improving the culture conditions for somatic cells will lead to increased embryo development. Several studies have already been published in which the primary focus is on the growth of somatic cells rather than development of embryos. This seems to be a prime case of 'putting the cart before the horse'. From an embryologist's viewpoint, the problem with this approach is that the metabolism and nutritional needs of preimplantation embryos and of most somatic cells are quite different. For example, glucose cannot be utilized or may be downright inhibitory to development in cleavage stage embryos, whereas it is suitable as the sole exogenous energy substrate for growth of most somatic cells. The explanation for this difference is that early embryos have incomplete glycolytic metabolism and/or use alternative pathways to generate energy for development (Leese, 1991). The logical extension of this experimental focus, which has actually been expressed by some investigators, is to develop media that are optimal for the culture of both somatic cells and preimplantation embryos. However, to my mind this approach will only lead to an unhappy compromise, i.e. a medium that is not optimal for either cell type, because of their conflicting requirements. I fear that undue emphasis on the needs of somatic cells for the purpose of improving co-culture will only distract us from our goal of understanding the regulation of embryo development. (5) The experimental consequence with the most serious potential for error is a disturbing trend in the scientific literature to 'prove' the embryotrophic benefits of co-culture by using mouse embryos as a test system. This approach is frequently coupled with, and justified by, the statement that development of human IVF embryos is suboptimal, so improved culture conditions need to be found. While this is perfectly true, there is no good rationale for using mouse embryos as models for the human. In fact, the whole question of what are the most appropriate models for human embryo development needs to be addressed, but elsewhere. The most alarming aspect of these studies with mouse embryos is that commonly, development from 1- or 2-cell stages in vitro to the blastocyst stage is extraordinarily low, sometimes only 20-30% or even lower. This contrasts with 'standard' results from many laboratories showing blastocyst development of 80—90% from mouse pronucleate embryos, with the majority of embryos hatching, in simple culture media developed specifically for mouse embryos, such as BMOC, BWW, CZB or SOM. In co-culture studies, sub-standard development responses in the control treatments immediately suggest that either the medium selected for mouse embryo culture is unsuitable for
Is co-culture really necessary?
this purpose (e.g. Ham's F-10 or F-12), and/or quality control during preparation of the medium is inadequate. The authors of such reports seem unaware of the contradiction that blastocyst development frequencies
OPINION
Co-culture for embryo development: is it really necessary?
Barry D.Bavister Department of Animal Health and Biomedical Sciences, and Regional Primate Research Center, University of Wisconsin, Madison, WI 53706, USA Numerous studies have shown, in several species including man, that co-culture with somatic cells or with conditioned medium can enhance in-vitro development of embryos. There are a variety of ways in which this improvement is manifested: by overcoming blocks to development, amplifying the numbers of embryos able to reach the blastocyst stage, increasing embryo cell numbers, and/or improving viability of embryos following transfer. Although co-culture can improve embryo development, however, these studies raise two major questions: how does co-culture exert its effect on embryo development, and is co-culture really necessary? In at least some applications, the need for somatic cell contributions can be eliminated by careful preparation of culture media. Nevertheless, examination of the plethora of studies on co-culture indicates that this technology is being accepted uncritically by some investigators as showing that somatic cells have 'embryotrophic' properties. This attitude sometimes leads to an a priori approach to the design of studies, which is inimical to good scientific principles. I will argue that, while co-culture can be useful under some conditions, in general, there is no evidence as yet for any embryotrophic activity, and the need for co-culture may reflect the use of inappropriate culture media or problems in their preparation. At worst, the apparent beneficial effect of co-culture is sometimes used to justify reporting of inconclusive data in the guise of research. The rationale for using oviductal cell co-culture in human in-vitro fertilization—embryo culture (TVF—EC) is logical: cleavage stages of development normally reside within the oviduct, and using oviductal epithelial cells in vitro may mimic the natural environment more closely. Conversely, the absence of oviductal influences in vitro may compromise the viability of IVF embryos. Reports of improved quality and/or development of embryos with co-culture fall into two categories: (i) empirical use of co-culture to improve embryo development and/or viability, and (ii) attempts to improve the co-culture technology itself, and perhaps to attribute the 'embryotrophic' effect to some secretory property of (e.g.) oviduct cells. These efforts have by no means been confined to human IVF —EC: a large number of studies, especially with cattle IVF embryos, have been conducted during the past 5 years or so, to the point that co-culture is frequently used almost unquestioningly as the method of choice for sustaining embryo development in vitro. The somewhat different goals reported above need to be kept © Oxford University Press
in mind when evaluating the utility of co-culture studies. With respect to (i), if co-culture with somatic cells enhances embryo development, then the benefits may outweigh the disadvantages. It may be easier in some locations to use co-culture, regardless of how it works, than to troubleshoot culture media problems to eliminate the need for co-culture. Even so, it would be wise to realize that the co-culture effect may be simply masking an underlying culture media problem, and co-culture aficionados might want to consider using a different formulation and/or being more critical about media preparation. Co-culture is not without a price, since it represents an additional layer of technology, with some risk of introducing disease organisms into the embryo culture environment, especially when animal cells are used for co-culture with human embryos (e.g. Weimer et al., 1989). Moreover, there is a danger of accommodating the culture needs of the somatic cells at the expense of the embryos (see 4 below). In the second approach (ii), attempting to 'prove' that co-culture is beneficial, and to characterize the effective agent(s), is fraught with problems. In the majority of studies in which an effect of co-culture has been claimed, although good supporting data are lacking, authors seem to prefer the explanation that somatic cells contribute some embryotrophic factor to the medium (we call this 'positive conditioning'). Yet removal of detrimental components from the culture environment ('negative conditioning' of the medium), such as the well-known case of hypoxanthine (Loutradis et al., 1987), is at least an equally valid explanation for the effects of co-culture on embryo development, and one for which there is good, albeit indirect, evidence. Problems with the 'positive conditioning' hypothesis are as follows. (1) First and foremost, the beneficial effects of co-culture can disappear when different culture media are used, or when the preparation of media is done more rigorously. For example, using 'standard' medium (TCM-199) containing serum, which is reported by several laboratories to cause a block to development of IVF cow embryos at the 8- to 16-cell stage, we found no such block: embryos developed to the morula/blastocyst stages with equal frequency in the presence or absence of oviduct cells (Pinyopummintr and Bavister, 1991). We ascribed this either to our very rigorous quality control of water used in culture media preparation, or to careful selection of the supplier of serum, or to the use of unheated serum in our experiments. (Incidentally, the necessity to heat serum used in cell culture is influenced by the age and species of the serum donor; and heating serum may adversely alter its cell growth supporting properties or even render it toxic.) Sheep zygotes also developed in vitro to blastocysts (85 %) in a simple medium supplemented with serum (Walker et al., 1992). This is in contrast to a study showing that 1339
B.D.Bavtster
sheep embryos were blocked in vitro unless co-cultured with oviduct cells (Gandolfi and Moor, 1987); this work, more than any other, stimulated the resurgence of interest in co-culture. Porcine embryos, too, can develop into blastocysts in vitro without co-culture (Reed et al., 1992; and see Table 1 of Walker et al., 1992). These contradictory data, arguing for and against a need for co-culture, surely arise from the selection and preparation of the culture media for embryo development. One way that somatic cells in co-culture might benefit embryo development is by stabilizing or altering physico-chemical conditions, such as the pH of the culture medium or oxygen/carbon dioxide concentrations, as a consequence of their metabolic activities. Adjustments in the gas levels used in culture can have striking effects on embryo development (Carney and Bavister, 1987; McKiernan and Bavister, 1990). The need for co-culture in particular locations might thus be eliminated by placing very strict quality controls on the environment actually being provided by the culture incubator, which is notoriously variable. (2) Reported beneficial effects of co-cultured cells on embryo development are neither species- nor tissue-specific. This detracts from attempts to demonstrate that oviduct cell secretory products, such as 'oestrus-associated' proteins, are embryotrophic. Neither is there evidence for cycle stage-specific effects of oviduct cells on embryo development, thus further arguing against a specific embryotrophic factor. Although oviductal proteins are reported to associate with cultured embryos (discussed by Bongso et al., 1990), there is as yet no evidence that they are responsible for improvements in embryo development observed in other studies. The crucial evidence is lacking, i.e. a demonstration that secreted proteins isolated from oviductal cell cultures improve embryo development. Without this information, data on oviductal protein secretion profiles should not be assumed to be relevant to embryogenesis, tempting though this may be. Other tissues shown to date to enhance embryo development, in a variety of species, include cumulus and granulosa cells, trophoblastic vesicles, uterine endometrial cells, monkey kidney cells and buffalo rat liver cells (the lack of co-culture tissue specificity is discussed by Bongso et al., 1990). It would be remarkable if all of these different cell types were producing the same specific embryotrophic factor, although this remains a possibility. On the contrary, the fact that such widely different cell types are effective when used in co-culture for embryo development could well indicate support for the 'negative conditioning' concept. (3) Many studies, particularly (but understandably) those with human embryos, claim beneficial effects of co-culture without reporting any control data, i.e. results of embryo culture using the same medium in the absence of somatic cells (there are a few notable exceptions (e.g., Weimer et al., 1989). Without control data, claims of enhanced embryo development using co-cultured cells cannot be evaluated. In a study with FVF cow embryos, using a factorial experimental design, the same medium (TCM-199) was tested with and without oviduct cell conditioning, so that any enhancement of development could be statistically analysed; none was found, and embryos developed equally well to morulae/blastocysts in both situations (Pinyopummintr and Bavister, 1991). In contrast, a study of human embryo development showed a reduction in blastomere fragmentation, together with a doubling of the incidence of pregnancy, with (bovine) 1340
uterine fibroblast co-culture (Weimer et al., 1989). Here again, perhaps a high incidence of fragmentation and loss of embryo viability with time in vitro is caused by unsuitable culture conditions, and somatic cell co-culture simply ameliorates this condition by rehabilitating the medium. It should also be noted that the gross appearance of embryos under a microscope can be misleading about their quality, especially if examined around the time of cleavage when major cytoplasmic rearrangements take place. Furthermore, analysis of FVF results for comparison of embryo treatments is constrained by the need for large numbers of transfers to provide statistically valid data; some studies may have failed to meet this requirement. (4) The emphasis on co-culture is sustaining a belief by some that improving the culture conditions for somatic cells will lead to increased embryo development. Several studies have already been published in which the primary focus is on the growth of somatic cells rather than development of embryos. This seems to be a prime case of 'putting the cart before the horse'. From an embryologist's viewpoint, the problem with this approach is that the metabolism and nutritional needs of preimplantation embryos and of most somatic cells are quite different. For example, glucose cannot be utilized or may be downright inhibitory to development in cleavage stage embryos, whereas it is suitable as the sole exogenous energy substrate for growth of most somatic cells. The explanation for this difference is that early embryos have incomplete glycolytic metabolism and/or use alternative pathways to generate energy for development (Leese, 1991). The logical extension of this experimental focus, which has actually been expressed by some investigators, is to develop media that are optimal for the culture of both somatic cells and preimplantation embryos. However, to my mind this approach will only lead to an unhappy compromise, i.e. a medium that is not optimal for either cell type, because of their conflicting requirements. I fear that undue emphasis on the needs of somatic cells for the purpose of improving co-culture will only distract us from our goal of understanding the regulation of embryo development. (5) The experimental consequence with the most serious potential for error is a disturbing trend in the scientific literature to 'prove' the embryotrophic benefits of co-culture by using mouse embryos as a test system. This approach is frequently coupled with, and justified by, the statement that development of human IVF embryos is suboptimal, so improved culture conditions need to be found. While this is perfectly true, there is no good rationale for using mouse embryos as models for the human. In fact, the whole question of what are the most appropriate models for human embryo development needs to be addressed, but elsewhere. The most alarming aspect of these studies with mouse embryos is that commonly, development from 1- or 2-cell stages in vitro to the blastocyst stage is extraordinarily low, sometimes only 20-30% or even lower. This contrasts with 'standard' results from many laboratories showing blastocyst development of 80—90% from mouse pronucleate embryos, with the majority of embryos hatching, in simple culture media developed specifically for mouse embryos, such as BMOC, BWW, CZB or SOM. In co-culture studies, sub-standard development responses in the control treatments immediately suggest that either the medium selected for mouse embryo culture is unsuitable for
Is co-culture really necessary?
this purpose (e.g. Ham's F-10 or F-12), and/or quality control during preparation of the medium is inadequate. The authors of such reports seem unaware of the contradiction that blastocyst development frequencies
