1.  MAMMARY TRANSPLANTS

      A powerful way to investigate the mammary gland is to genetically manipulate the epithelium by reconstructing the mammary gland, in vivo, from transplanted epithelial cells. Mammary epithelial cells are isolated from one mouse mammary gland and put into a mammary fat pad in another mouse, from which the natural epithelium has been removed. Remarkably, transplanted epithelial cells grow into the fat pad and form a mammary epithelium almost exactly like a naturally-occurring one, except that it is not connected to a nipple. To introduce a gene into the mammary epithelium, the mammary epithelial cells to be transplanted can be infected with nonreplicating   retrovirus   vector   between   isolation   from   one   mouse   and transplantation into the host (Fig. 1).

      This approach has a number of advantages over the more familiar approach of creating a transgenic mouse with an exogenous gene driven by a mammary-specific promoter. The first advantage is that the gene is entirely confined to mammary epithelial cells, so any promoter can be used -typically actin, SV40 or retrovirus LTR promoters have been used. In contrast, transgenics have to be made with mammary-specific promoters, the MMTV promoter, the WAP (Whey acidic protein) promoter or the beta-lactoglobulin promoter. All three are much more active in lactation than in virgin animals and the MMTV promoter at least is active in tissues other than mammary gland. A major difference between the approaches is that in the transgenics all mammary epithelial cells contain the introduced gene, and all are capable of expressing it, though in practice expression can be patchy. In the transplants infected with retrovirus, only a few percent of the transplanted cells express the gene, unless those cells have a selective advantage after transplantation. However, while this is a weakness of transplantation where the whole gland is to be altered, it is a unique strength of the approach for modelling cancer development, as individual cells can be genetically altered among an excess of normal cells, a true model of the cancer situation.

      Examples of the use of this approach include the expression of mutant forms of the EGF-receptor family of receptors, such as neu, a point-mutated form of rat erbB2, and v-erbB, the original retroviral truncated EGF-receptor mutant. These gave focal abnormalities of growth pattern, as expected if clones or small groups of cells were expressing the introduced gene. There were striking differences in the patterns of abnormal growth. neu gave a spectrum of lesions, including DCIS and occasional tumours, that resembled histologically a range of lesions found in human breasts. v-erbB on the other hand gave enlarged and distorted ducts with loose cells in the lumens. Thus two closely-related receptors gave different alterations to growth pattern. Whether the difference was between erbB2 and erbB, or was due to the type of activating mutation, or even to the species of origin of the receptor (rat versus chicken) remains to be determined. A very different result was obtained when the genes myc, Wnt-1 and Wnt-4 were expressed this way. In all three cases, large parts or even the whole of a transplanted mammary epithelium were hyperplastic, suggesting that cells expressing the exogenous gene had a selective advantage over their normal neighbours during outgrowth of the transplant. This may be important as a model of cancer development. Also, the pattern of growth in the virgin animal induced by expression of Wnt-1 and Wnt-4 was remarkably similar to the pattern induced early in pregnancy, suggesting that the normal change of pattern in pregnancy may be signalled locally by a member of the Wnt family, perhaps Wnt-4 itself, which is expressed naturally in early pregnancy. Other genes expressed this way include ras and the marker gene beta-galactosidase.

        Transplantation can be used not only to introduce genes into epithelium, but also to manipulate mammary epithelium lacking a gene, and here the method solves different problems of the transgenic approach. In transgenic knockout mice where mammary epithelium is altered, it is not in general possible to tell whether the effect on the epithelium is due to alteration in the behaviour of the epithelial cells themselves, or is an indirect effect of a systemic change, or changes to the environment in which the epithelium grows. And, of course, in some cases no knockout mammary epithelium can be obtained because the mutation is lethal before the mice reach maturity. Both problems can often be solved by transplanting fragments of mammary epithelium between mice.

       For example, mice with null mutations in the cyclinD1 gene develop more or less normally, but the mammary epithelium fails to respond properly to pregnancy. This could be because the epithelial cells were unable to respond to signals, but it was at least as likely that they were not receiving signals. In pregnancy the mammary epithelium proliferates in response to systemic signals coming from, for example, the pituitary and the placenta, and these signals may well act on the stroma of the mammary fat pad to provoke local signals to the epithelium, as for many epithelial-mesenchymal signals. To distinguish these possibilities we transplanted mammary epithelium from knockout mice into normal, histocompatible (an important issue as discussed below) recipients. The resulting mice had knockout epithelium in a normal fat pad and mouse, but they continued to show the same attenuated response to pregnancy, showing that most or all of the reduction in response to pregnancy was a failure within the epithelial cells themselves. In principle, we could also have transplanted normal epithelium into knockout recipients to see if their mammary epithelium had a normal response, but in the mixed-background transgenic line we used, this combination would not have been histocompatible. Transplantation of, or into, transgenic mice is often complicated by the fact that transgenic lines are frequently made by crossing two strains of mice. The resulting transgenics individually have an unknown and variable combination of histocompatibility alleles from the two backgrounds. Such tissue can be transplanted into an F1 mouse from a cross between the two strains, as it has all the alleles, but nothing can be transplanted into these transgenics as they will almost invariably lack at least one allele from any donor chosen. The solution to this problem is for the transgenic knockout line to be made on a pure 129 background.

       To solve the problem of lethality of mutations, it may often be possible to rescue mammary epithelium from embryonic or neonatal mice and transplant it into adult normal mice, where it grows to form epithelium apparently equivalent to mature adult epithelium. Transplantation can be done from about 12 or 13 days of gestation, and indeed the developing mammary glands, or rather the nipples, are easiest to find at or soon after this point, before the presence of hair follicles and folds in the skin make them more difficult to recognise.In the future it will probably be possible to create clones of knockout epithelial cells in a normal fat pad, to model clonal loss of a tumour suppressor gene. This could be done by transplanting epithelium from conditional knockout mice - in which the gene of interest is flanked by lox sites and inducing knockout of the genes in clones by infecting them with a cre -expressing retrovirus. (Cre protein catalyses recombination between lox sites).