Retroviruses and the study of cell lineage

In this review, I want to discuss a new way of tackling an old problem. The problem is how to mark a cell such that its developmental capacity can be assayed. The solution I want to consider is gene transfer using retroviruses. There are many ways of marking cells, but a genetic marker has a number of obvious advantages. It is indelible, heritable and need not damage a cell or distort its development. However, it is not always easy to introduce a genetic marker into a cell, especially if the number of marked cells as well as the precise time in development at which the marker is introduced need to be controlled.For several years now, molecular biologists have been using a variety of gene transfer techniques (see Gordon & Ruddle, 1985, for review), the most familiar of which to embryologists is probably the microinjection of DNA into the mouse pronucleus as a means of generating transgenic mice (Gordon et al. 1980; Wagner, Stewart & Mintz, 1981; Harbers, Jahner & Jaenisch, 1981; Brinster et al. 1985). However. it is not obvious how the methods of molecular biology, which are primarily in vitro techniques, can be applied to problems of gene transfer in vivo. Retroviruses may be a way out of this predicament.The work I shall discuss in this review represents the initial studies from a small number of laboratories, in which retroviruses have been applied to the study of cell lineage. In concentrating on this aspect of retroviral technology, I am ignoring several other interesting applications of retroviruses (such as in-sertional mutagenesis and gene therapy) and also excluding much of the molecular biology of the structure and function of retroviruses. (Good reviews of this area exist in any case: see Varmus, 1982; Coffin, 1985; Bernstein, Berger, Huszar & Dick, 1985.)Retroviruses have a number of properties that make them versatile and powerful tools in the study of development. Their principal advantages stem from the fact that they are a naturally evolved system for transferring genes into cells of a host animal (Fig. 1). As a consequence, this transfer is highly efficient (unlike most artificial means) and also highly accurate in that a faithful copy of the retroviral genome is integrated into the host cell chromosome. Furthermore, the retrovirus itself carries the nucleic acid sequences required to direct the host cell in the transcription of the retroviral genes. This means that in many situations the virus can be relied upon to express itself without recourse to further genetic manipulation. On the other hand, viruses can be engineered to provide other promoter elements should they be required (Wagner, Vanek & Vennstrom, 1985; Stewart, Vanek & Wagner, 1985; Emerman & Temin, 1984, 1986).The structure of the retroviral genome has evolved with the retroviral genes located in the middle of the genome, flanked by sequences called the long terminal repeats (LTRs). The significance of the LTRs is that they contain all the sequences required in cis for the integration and expression of the retrovirus. This organization is principally a reflection of the mechanism of replication that retroviruses have evolved (Varmus, 1982), but it is fortunate from the point of view of the molecular biologist because it means that, in principle, any gene could be introduced in place of the retroviral genes, and the virus could still infect a cell, integrate and transcribe that cloned gene. (The genetic engineering of retroviruses is essentially the same as cloning in plasmids and mostly uses the techniques of conventional molecular biology.) Such expression of cloned DNA in retroviruses has now been shown to work in a number of instances, as will be described below.Of equal importance has been the development of methods for the packaging of recombinant viruses into infectious particles (Fig. 2). This is done by transfecting a plasmid that contains the engineered virus into what is termed a packaging cell line (Mann, Mulligan & Baltimore, 1983; Watanabe & Temin, 1983; Cone & Mulligan, 1984; Sorge, Wright, Erd-man & Cutting, 1984). This permanent cell line makes the retroviral gene products but makes no viral genomic RNA that is capable of forming a retroviral particle. In other words, it has all the ingredients necessary for making a virus with the exception of packageable retroviral RNA. The engineered virus, however, once introduced into the cell will form such an RNA, which consequently is packaged and released by the transfected packaging cells. In this manner, the tissue culture supernatant from these cells constitutes a permanent supply of high titre virus. Viruses of this type, that encode foreign genes but not the endogenous retroviral genes, are usually termed retroviral vectors. They are infective in the same manner as wild-type viruses, but, on infecting a cell, they cannot complete the wild-type life cycle shown in Fig. 1, because they do not contain the endogenous retroviral genes required to package an RNA.A variety of studies has now been done using retroviral vectors of this type to infect cells in culture. In the earliest experiments, the cloned genes used were predominantly those that encode selectable marker genes such as the Herpes simplex thymidine kinase (tk) gene or the Tn5 neo gene, which infers G418 resistance on eukaryotic cells (see Bernstein et al. 1985 for references). But more ambitious constructs have also begun to appear and the expression of preproparathyroid hormone (Hellerman et al. 1984), granulocyte-macrophage-colony-stimulating factor (Lang et al. 1985), the polymeric immunoglobulin receptor (Deitcher, Neutra & Mostov, 1986) and fibronectin polypeptides (Schwarzbauer, Mulligan & Hynes, 1987) have all now been reported.Once a retrovirus has infected a cell, it integrates into the host cell genome, so that the pro virus is inherited by all the progeny of that cell. Hence, the clone of cells derived from the infected cell is genetically marked. Moreover, when the provirus integrates, it does so randomly, so that each integration site is unique. Consequently, on infection each cell (and subsequent clone) is given a genetic label. So if the host cell DNA is cut with a restriction enzyme, and the DNA fragments separated electrophoretically and hybridized with a probe recognizing the viral sequences, the fragment of DNA from each clone that contains the provirus will be unique and of a characteristic size, which will distinguish it from any other such fragment.This manner of marking clones with a retrovirus provides one way in which they can be used to study cell lineage. One of the advantages of this approach is that expression of the viral genome is not required, the presence of the integrated provirus being sufficient to recognize the clone. However, a potential disadvantage of the approach is that in situations where there is retroviral expression, infected cells can generate new retroviral particles and so spread the virus in a horizontal fashion, thereby obscuring any clonal analysis.The most elegant way of avoiding this problem is to use the retroviral vectors described above, which are replication defective and, therefore, cannot spread horizontally to other cells. I will return to this approach to lineage later in this review. However, another solution to the problem is to study systems in which virus does not express. For example, the Muloney murine leukaemia virus (MoMLV) does not express in cells of the preimplantation mouse embryo (Jaenisch et al. 1975). It appears that the proviral DNA becomes methylated (Stewart, Stuhlman, Jahner & Jaenisch, 1982), although it is not clear that this is the primary reason for the lack of expression (Gautsch & Wilson, 1983; Niwa, Yokata, Ishida & Sugahara, 1983). Consequently, studies of lineage in the early mouse embryo are possible using wild-type MoMLV retrovirus, and these experiments are described below.This manner of recognizing clones - studying band sizes on Southern blots - has been productive in two principal areas of research into cell lineage. These are haematopoiesis and the early development of the mouse embryo.The development of blood cells was in many ways an obvious place to begin applying gene transfer techniques to the study of cell lineage. It is one of a few vertebrate systems where a considerable amount was already known about cell lineage relationships, largely as a result of experiments using nonretroviral chromosomal markers (see Quesenberry & Levitt, 1979, and Till & McCulloch, 1980 for reviews). Also, techniques already existed for the removal and culture of mouse bone-marrow cells and their subsequent introduction into an irradiated syngeneic host. Probably the main drive to research in this area is the hope that the haematopoietic system might be amenable to gene therapy.The lineage of the haematopoietic system is possibly better understood than that of any other mammalian system. It has been clear for some time that there is a stem cell (CFU-S) that has the ability to generate the entire myeloid lineage (Till & McCulloch, 1961). This stem cell is selfreplicating (Simno-vitch, McCulloch & Till, 1963) and is believed to generate committed progenitor cells, which can give rise to particular differentiated cell types (Lewis & Trobaugh, 1964; Curry & Trentin, 1967).Several workers have shown that bone marrow cells can be infected in vitro with retroviral vectors such that a proportion of the stem cell population becomes marked. On reintroduction into an irradiated host, these cells can myeloid and cell & Bernstein, 1983; et al. 1984; et al. 1984; & in the earliest studies a small proportion of the stem cells was techniques have more been so that now it is possible to to of the stem cells & 1985; et al. 1985; & Mulligan, This is done by the bone marrow cells with high titre viral and by the predominantly stem cells into with or by the animal with to the This cells and so stem cells into to for the Consequently, there is a proportion of stem cells in the bone marrow The of the stem cells is because cells to integrate virus. The proportion of stem cells is because these cells are a of the population and as such are to of some was the retroviral could be used to the stem cell that rise to myeloid and It is now clear from a number of studies that such cells can be with this et al. 1984; et al. 1985; & Wagner, 1985; & Mulligan, The importance of this not so much in the of a haematopoietic stem cell has been in this for many years Till, & & but in the with which it is possible to such cells and their development. these better marking techniques, more ambitious experiments have possible and the initial are For example, and were to how the of marked stem cells to cell of in the same animal et al. they that each cell was of the of few but that a of the as clones from the differentiated cell now This is of a as it is from and bone marrow these that a number of stem cells to the haematopoietic system at any one The system in other words, and, with stem cells to to the of differentiated cells and are was also possible in the same study to the that marked clones make to and in mice et al. Consequently, the stem cells that a potential in of the cell types to which they could give to be in the they clone could for but not cells, or bone marrow but not (The do however, which of cell types and could be derived and which could These of be with because a stem cell does not that its was and that it could not have other this the that haematopoietic cells can in of the that they as well as in the types of blood cell to which they give this to be it will be interesting to see how it conventional of haematopoietic cell studies of haematopoiesis are still at an early and they I to more than on lineage. obvious is to introduce genes into haematopoietic stem cells with a view to the this has on their development. and all to in this as do the genes that, it is may for gene for example, & & Mulligan, et al. to that described for haematopoiesis has also been applied to the study of lineage in the preimplantation mouse can be infected at early introduced into for This approach has been used by a number of workers & & & Jaenisch, Stewart, Vanek & Wagner, this is to other methods of generating but it is no of DNA or of cells et al. or stem cells or can be infected in culture & 1985; & 1986) and used to form using In this manner, the retroviral sequences are introduced into a of the techniques have been used for a number of which will not be with see & Jaenisch, 1983; et al. 1985; et al. 1985; et al. For the study of this approach has been by & as a means of early in the They infected at the to by them for with MoMLV cells. the into a they development to and were to in which the progeny of marked For example, they clones were and or and and the clones were in all or to a of some interesting They for a lineage to and embryo at this of proviral integration at were in the in the and in In that were to to the were to to what a marked that to one the for example, also to They that in the of a that to any one of the or that they also to the were to this of a Southern of DNA from each the could the of any given proviral band to DNA from of that were for the provirus all cells of the tissue one copy of the In other words, if the cells were to have a of 1, one copy cell in the entire a that of the cells of that tissue a copy of that particular provirus (and were derived from the that been This the want to an of the proportion of the cells in any tissue that are derived from an infected They that for any given proviral integration this was the same in all the of marked a equal proportion of all This was for all the out of These in other words, This as the that considerable stem cell and to the of tissue mice and the of in the the were to not that to the but also that some of these to to This that at the time of proviral some were in the cell to of these with those from by the of preimplantation 1984). For example, it has been that at such can to the lineage but not to the cell The that the can to the line but not the lineage is more as it that the line is at an than has been other techniques have to these as as or & 1983; et al. above, and have their Southern by the by clones to a number of all these were the that the that was they that they could have as as a of is one as a this means that if a clone to a it at an of the cells of that The this to that cells are to make the entire I two problems with this one some that this of has some the fact that this is the in this particular cannot be to that it is without some that it is the this is a problem and should be as as has already been by one with this at is that it is not clear when these cells be problem with the and point is that although infection place it by with tissue culture studies that the provirus can in the cell for some time it Stewart et al. have that with MoMLV of cells, this can be to This means that although the infection place at the to in and experiments, integration (and may have place as as the So if the is one which cells are being and when are they from the of their the studies of haematopoiesis and lineage described have the of retroviruses as lineage Furthermore, I it is to that they have been retroviruses to a of their the lineage studies so have a the of the form of a Southern the is to of cells. This has two small clones cannot be it is not known which cells a population to the marked so that, if the population contains more than one cell type, it is not possible to the marked cells one or many cell these a number of the more interesting types of lineage study that might be For example, in the it be interesting to and cells are derived from the same of progenitor cells, or and are early in development. The to of this type is a considerable in developmental biology, not in the study of development. It is in to from a to a of developmental The reason for this is clear from the given the of and the in which are are it is to which of the potential and are to be much of the in lineage in from the with which those studying upon the in the The of in is well known now and no But it is that the of lineage in has also been in developmental where lineage a in One of this is the development of the where it has been shown that lineage no in cell type & & This has on the later that are the in how cells form an 1985; & & there are other ways in which a retrovirus can be to study lineage. It is possible to use a that not integrates and genetically a but also an gene. the of this gene can be a cell this can be This the of being to study small clones and the ability to which cells a structure are of a clone. There are so two of this approach being In one of and at the have used a retroviral expression system to at in the mouse embryo & The other was work in which I have been in with and of have lineage in the system of the & & two are In retroviral vectors on MoMLV were that the gene, which for the However, the two viruses were In virus, the gene the of the endogenous retroviral promoter and et al. used an early promoter to drive the has a number of obvious advantages as a It has a number of which make the of expression there are a variety of and the there is a to cells that are the This uses the and can be used on cells or tissue or tissue to cells to these it that cells could express high of without it with development. that is to for such might be in ways that However, in other high of expression have with no & 1983; & et al. 1987) and, in the studies there was no of (see the study by et al. virus was into the mouse embryo in and to this of it is not possible to see the embryo the because of the The hope is to the virus into the or and so the embryo to virus. There is also a to the amount of virus that can be into such a small structure and it is that these workers could sufficient virus into the embryo to infect any cells at this they were to study the lineage of a number of In they a of clones in the and in the they virus at the clones they upon subsequent of the cells from the and In other words, all the cell types in the that are derived were in of the were not virus was at the of clones cells and one clone cells and it no other cell They also one clone The these to that at there is a progenitor which rise to the and however, it that exist that give rise to and cells, or to This that a progenitor that all cell types rise by to two with a more the the is their the clones of cells and, in out of these clones also cells of the is the of cells but not these clones could contain cells of the and of the the structure that into a the clones from were out of clones but not and the but not The therefore, that at a progenitor which rise to cells that by can make or but not In the and therefore, and were to a restriction in the cell types to which cells could give They were also to to some when these restriction studies are for a number of they have some out some of the lineage of the they have But this study the of retroviral lineage studies at this of development. to be is the of can be One might it to be more to virus into some of the or as this infection of cells embryo It to be or not this is retroviral approach that has been to lineage of the system is to that of et al. that on the later of development in an to the of cell et al. & the study of et al. has that early in the in the system have that some types of progenitor cells exist that can give rise to an of cell types in was shown most in the cell types of the are but most and some cells, cells and are the virus is into the of and their in for clones that express one clones all these cell types et al. & It can also be shown that the number of clones is to the amount of virus and that, if the viral titre is in this fashion, it from the in vitro titre on cells by a factor of & The clones were small or cells on as be so in and by the of clones as might be given that this is the main cell type being at this interesting is what are the lineage of these cell and have of clones and not they clones more than one cell type, but they also clones possible of cells, cells and that no clones all cell However, given the of clones and the of of some of the cell types such cell clones be to be of these is that each progenitor in the can give rise to all cell is that there are types of progenitor cell, each of which can give rise to a of the cell In it that a progenitor cell can give rise to two cell types to its is of the in the to but to what is to in the mammalian system where and are to be early (see Levitt, & 1981; 1983). This view has been in years by the work of and which has shown the in the of a progenitor which can give rise to one type of and but not to other types of or any cells & 1983). This the in the at of that have the potential to make than the of cell of these are of the developmental that the animal in the might be from that used in the It is now necessary to the retroviral marker to other of the and that is being et al. some clones in the although no clonal was have infected cells in culture and, primarily by clones of cells and types of et al. and It however, early to these with all of the I have one possible problem with the In the early mouse embryo and haematopoietic was by the fact that a proviral integration marked a unique site in the In the studies using the this was not areas of cells were as being There however, the that the was the result of two cells being infected with virus. In the of two clones being is in most the use of viral that an infection is a For example, and used of virus that one clones two cells an entire in experiments with cells in vitro et al. 1987) a culture with about cells have or marked an result should be and, in all the experiments most types of clones were several The were some of the clones by et al. in their clones were In the studies on the each clone could be as such by its of in so that clones could be So is not a problem one that there is some for infective to and there is no for is that the studies described in this review are the of to The potential of retroviruses is to be there are many more retroviral constructs than those described Consequently, it should now be possible to the lineage of many cells in the vertebrate using some of the techniques This is an to and all of were to and on this