The Hsp90 cochaperone p23 is essential for perinatal survival

Abstract

The functions of molecular chaperones have been extensively investigated biochemically in vitro and genetically in bacteria and yeast. We have embarked on a functional genomic analysis of the Hsp90 chaperone machine in the mouse by disrupting the p23 gene using a gene trap approach. p23 is an Hsp90 cochaperone that is thought to stabilize Hsp90-substrate complexes and, independently, to act as the cytosolic prostaglandin E2 synthase. Gene deletions in budding and fission yeasts and knock-down experiments with the worm have not revealed any clear in vivo requirements for p23. We find that p23 is not essential for overall prenatal development and morphogenesis of the mouse, which parallels the observation that it is dispensable for proliferation in yeast. In contrast, p23 is absolutely necessary for perinatal survival. Apart from an incompletely formed skin barrier, the lungs of p23 null embryos display underdeveloped airspaces and substantially reduced expression of surfactant genes. Correlating with the known function of glucocorticoids in promoting lung maturation and the role of p23 in the assembly of a hormone-responsive glucocorticoid receptor-Hsp90 complex, p23 null fibroblast cells have a defective glucocorticoid response. Thus, p23 contributes a nonredundant, temporally restricted, and tissue-specific function during mouse development.

FIG. 1.

Functional disruption of the mouse p23 gene. (A) Schematic representation of the p23 gene disruptions. In both lines the gene trap vectors integrated within the first intron. Numbers indicate the integration site (in bp) relative to the ATG of the open reading frame, which begins with AT in the first exon and continues with the G of the initiation codon in the second exon. (B) Immunoblot of liver extracts from p23 null (−/−), wild-type, and heterozygous (+/−) embryos of line A. Hsp90 was revealed as a loading control. (C) Two litters of 18.5-dpc p23 null, heterozygous, and WT embryos of line A.

FIG. 2.

p23 null embryos develop but do not survive. The graph shows the genotypes of the offspring of p23+/− parents. Crosses of p23+/− animals gave no live p23 null mice in either of the two gene trap lines. In contrast, the genotypes of sacrificed 18.5-dpc embryos are compatible with a Mendelian distribution. Total numbers of animals that were considered for this analysis are as follows: for line A, 125 adults and 79 embryos; and for line B, 135 adults and 65 embryos.

FIG. 3.

Lung with underdeveloped airspaces in p23 null mice. (A) The lungs of the p23 null newborn are not aerated and sink when immersed in water. The floating lung of a newborn p23+/− animal and the sunken lung of a WT embryo sacrificed before birth (dead WT) are shown as controls. (B) Peripheral lung sections of control (WT) and p23 null (−/−) 18.5-dpc embryos of line A stained with hematoxylin and eosin. Magnification, ×10. (C) Morphometric analysis of histological sections of the p23 null and WT lungs (line A). The quantitation was done by measuring five different fields each in comparable peripheral sections from three wild-type and two p23 null lungs. The bars represent the percentages of section area occupied by airspace in the 18.5-dpc lungs; errors bars show standard deviations from the average. (D) Bronchi and bronchioles of p23 null embryos develop normally. C and P, central and peripheral lung sections of 18.5-dpc embryos. Magnification, ×40.

FIG. 4.

Defective maturation of the lung. (A) Peripheral lung ultrastructure in 18.5-dpc p23 null embryos (line A). In the WT lung, type II pneumocytes contain lamellar bodies (white arrows) and apical microvilli, but these features are absent from the p23 null littermate. Surfactants released into airspaces in the p23 WT lung (black arrow). Magnification, ×7,100. (B) Expression levels (n-fold) of surfactant protein SP-A, SP-B, SP-C, and SP-D genes over that of a housekeeping gene as assessed by quantitative real-time PCR at 18.5 dpc. Three WT and three −/− lung extracts of line A embryos were analyzed. Error bars ± standard deviations are shown.

FIG. 5.

Defective skin in p23 null mice. (A) Skin sections of 18.5-dpc p23 null (−/−) embryos (line A), stained with hematoxylin and eosin, have fewer SC layers (arrows) than WT littermates. Magnification, ×10. (B) Electron micrographs of the boundary between stratum granulosum and stratum corneum of the epidermis from p23−/− embryos and p23+/− littermates (line A). Arrows indicate lamellar bodies fusing with a membrane of the uppermost cell of the stratum granulosum. K, keratohyalin granules; SG, stratum granulosum; SC I, first layer of stratum corneum. Magnification, ×21,000. (C) Compromised skin barrier in p23 null embryos. The 18.5-dpc p23 null embryos and their littermates (WT) of line B were stained with toluidine blue solution.

FIG. 6.

GR function is defective in p23 null MEFs. (A) Levels of endogenous GR vary between MEF cultures but do not correlate with genotype. Immunoblots of GAPDH and p23 serve as the loading control and the confirmation of genotype, respectively. (B) Hormone binding of endogenous GR is severely impaired in p23 null cells. Specific dexamethasone binding at about 2 nM was determined with three p23 null (−/−) and three WT MEF cultures. The graph shows the averages with standard deviations. The difference in bound hormone levels between wild-type and null cells is 5.2-fold. (C) Transcriptional response of GR displays decreased dexamethasone potency in p23 null MEFs. Relative GR reporter gene activity was measured as a function of dexamethasone concentration. The plot represents data collected from three p23 null and three p23 WT MEF cultures. Concentrations from 0.1 to 1,000 nM are plotted in a log scale, and error bars indicate standard errors of the mean.