Casein is a major protein component ofmilk and, in conjunction with the other caseins, it is assembled into micelles. The casein miceiles determine many of the physical characteristics of milk, which are important for stability during storage and for milk-processing properties. There is evidence that suggests that j-casein may also possess other, nonnutritional functions. To address the function of (-casein, the mouse P-casein gene was disrupted by gene targeting in embryonic stem cells. Homozygous i-casein mutant mice are viable and fertile; females can lactate and successfully rear young. -Casein was expressed at a reduced level in heterozygotes and was completely absent from the milk of homozygous mutant mice. Despite the deficiency of 1-casein, casein micelles were assembled in heterozygous and homozygous mutants, albeit with reduced diameters. The absence of R-casein expression was reflected in a reduced total protein concentration in milk, although this was partiay compensated for by an increased concentration ofother proteins. The growth ofpups feeding on the milk ofhomozygous mutants was reduced relative to those feeding on the milk of wild-type mice. Various genetic manipulations of caseins have been proposed for the qualitative improvement of cow's milk composition. The results presented here demonstrate that f-casein has no essential function and that the casein micelle is remarkably tolerant of changes in composition. Milk is usually the sole source of nourishment of young mammals, and the milk of domestic animals is an extremely important food source for much of the world's population, both as liquid milk and in a very wide variety of processed forms. Most of the protein in milk is found in the caseins, which are aggregated into large micellar structures that are in colloidal suspension in native milk. Although the detailed structure of casein micelles is not yet established, there are a number of models of micelle structure (for review, see ref. 1). The "calcium-sensitive" caseins (asi-, a52-, and 3-caseins in cow's milk) are generally thought to be located predominantly within the micelles, and K-casein is thought to coat the micelle, serving to stabilize the structure. The calciumsensitive caseins are highly phosphorylated, and calcium phosphate interacts with them via their phosphate groups. In this way the casein micelles carry large amounts of (normally highly insoluble) calcium phosphate into milk and retain it in suspension. A further consequence of the assembly of the caseins into micelles is that the viscosity of milk remains low despite the high protein concentration. Thus, casein micelles are offunctional importance for protein and mineral nutrition of the young and in determining the physical properties of milk. Selective breeding has been very successful in increasing milk yields of cattle, but this approach has been found to have limited potential for the alteration of milk composition. We and others have previously demonstrated the profound alteration of the protein composition of milk by expression of milk protein genes in transgenic mice (2-4), and the expression of human pharmaceutical proteins in the milk of transgenic mice, rabbits, goats, pigs and sheep has been demonstrated (5-13). In some cases very high yields of active human proteins have been obtained (6, 8, 11-13), and this approach is being commercialized. In addition to pharmaceutical applications, genetic manipulation may be of use for the production of milk with altered nutritional, allergenic, or processing properties. The storage and processing properties of milk are, in large measure, determined by the caseins, and these proteins thus represent a potentially important target for genetic manipulation (14, 15). There is some evidence that suggests that caseins may possess additional functions. A number of peptides derived by proteolytic cleavage of bovine /-casein and known as /3-casomorphins have been shown to possess potent opioid activity and have been suggested to be natural agonists for opioid receptors in the gut (16). In addition to expression in the mammary glands, casein expression has been detected at the RNA level in mouse cytotoxic T-lymphocyte cell lines and in thymus (17). The authors suggested that casein micelles may serve to sequester perforin in cytotoxic T lymphocytes in conditions that prevent its polymerization. The sequestration of perforin may protect the cytotoxic T lymphocytes against perforin-mediated lysis, and casein micelles may function to deliver perforin to the target cells. To investigate the function of 3-casein in milk and to determine whether this protein possesses any other essential function, we generated mice deficient for /3-casein by gene targeting in embryonic stem cells. The effects ofthe mutation on gene expression, milk protein content, and casein micelle structure have been characterized. MATERIALS AND METHODS Gene Targeting. The gene targeting vector (pP3MClneo/ TK) includes a 4.7-kb Sca I-EcoRI fragment of the (3-casein gene (18) from a C57BL/6 mouse, with the neomycinresistance gene from pMClneo(C) (19) inserted into the Asp700I site in exon 2 of the gene in the same orientation. A herpes simplex virus thymidine kinase expression cassette was placed at the 3' end of the region of homology, for use in positive-negative selection (20). Embryonic day-14 stem cells (21) were electroporated with p.3MClneo/TK at 31.25, 93.75, or 187.5 pg ml-1. After 24 hr, the cells were selected with Geneticin (0.3 or 0.5 mgml-1), and aftera further4 days, IlTo whom reprint requests should be sent at the address. 6138 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. Proc. Nati. Acad. Sci. USA 91 (1994) 6139 ganciclovir (2 ,uM) selection was applied. The enrichment obtained with ganciclovir was 4.4to 15-fold. DNA and RNA Analysis. Genomic DNA was prepared from embryonic stem cells and digested by the method of Laird et al. (22) and from mouse tails as described (23). RNA was isolated by the method of Chomczynski and Sacchi (24) and redissolved in 100% formamide (25). After electrophoresis, DNA or RNA was blotted onto Hybond-N (Amersham) and hybridized with probes labeled by random priming (26, 27) using the method of Church and Gilbert (28). Production of Chimeras. Chimeras were produced essentially as described (29) by microinjection oftargeted cells into the blastocoel cavity of 3.5-day C57BL/6 x CBA F2 embryos. Microinjected blastocysts were reimplanted into the uteri of 2.5-day pseudopregnant mice. Chimeras were identified on the basis of the presence ofpale patches of fur on an agouti or nonagouti background. Milk Protein Analysis. Collection of milk and electrophoretic analysis of milk proteins were done essentially as described (2). For quantitative protein analysis, pools were made of equal volumes of milk from 18 wild-type and 18 homozygous mutant mice. Total protein and whey protein concentrations were estimated by the micro-Kjeldahl method as described (30), except that protein concentration was obtained by multiplying the nitrogen value by 6.38 (to take account of the amino acid composition and nonprotein nitrogen content of milk). The factor 6.38 was derived from calibrations with cow milk and may not be entirely accurate for mouse milk. By SDS/PAGE, no casein contamination of the whey fractions was detected. The (-casein concentration was determined by measuring A280 of eluates from ionexchange fast-protein liquid chromatography (31); the extinction coefficient of pure (-casein was found to be 1.4 [1% (wt/vol) solution, 1-cm path length]. Casein micelle sizes were estimated by the dynamic light-scattering method (32, 33). Analysis ofPup Growth. Homozygous mutant females were mated with wild-type males, and homozygous mutant males were mated with wild-type females. All the pups were thus heterozygous for the P-casein mutation. Entire litters were weighed three times per week (Monday, Wednesday, and Friday) between birth and 11 days of age. Litter weight was found to be linearly dependent upon age over this period; correlation coefficients between weight and age were greater than or equal to 0.% for all individual litters. The growth rates (slopes) obtained from individual litter weight vs. age regressions were compared across genotypes and litter sizes by using regression analysis. The model included effects for genotype, litter size, and a quadratic effect of litter size with interactions between genotype and litter size terms. Predicted growth rates for each combination of genotype and litter size were then obtained.