In the absence of detergent, the transfer of mannose from GDP-mannose to rat liver microsomal vesicles was highly stimulated by exogenous retinyl phosphate in incubations containing bovine serum albumin, as measured in a filter binding assay. Under these conditions 65% of mannose 6-phosphatase activity was latent. The transfer process was linear with time up to 5min and with protein concentration up to 1.5mg/0.2ml. It was also temperature-dependent. The microsomal uptake of mannose was highly dependent on retinyl phosphate and was saturable against increasing amounts of retinyl phosphate, a concentration of 15mum giving half-maximal transfer. The uptake system was also saturated by increasing concentrations of GDP-mannose, with an apparent K(m) of 18mum. Neither exogenous dolichyl phosphate nor non-phosphorylated retinoids were active in this process in the absence of detergent. Phosphatidylethanolamine and synthetic dipalmitoylglycerophosphocholine were also without activity. Several water-soluble organic phosphates (1.5mm), such as phenyl phosphate, 4-nitrophenyl phosphate, phosphoserine and phosphocholine, did not inhibit the retinyl phosphate-stimulated mannosyl transfer to microsomes. This mannosyl-transfer activity was highest in microsomes and marginal in mitochondria, plasma and nuclear membranes. It was specific for mannose residues from GDP-mannose and did not occur with UDP-[(3)H]galactose, UDP- or GDP-[(14)C]glucose, UDP-N-acetyl[(14)C]-glucosamine and UDP-N-acetyl[(14)C]galactosamine, all at 24mum. The mannosyl transfer was inhibited 85% by 3mm-EDTA and 93% by 0.8mm-amphomycin. At 2min, 90% of the radioactivity retained on the filter could be extracted with chloroform/methanol (2:1, v/v) and mainly co-migrated with retinyl phosphate mannose by t.l.c. This mannolipid was shown to bind to immunoglobulin G fraction of anti-(vitamin A) serum and was displaced by a large excess of retinoic acid, thus confirming the presence of the beta-ionone ring in the mannolipid. The amount of retinyl phosphate mannose formed in the bovine serum albumin/retinyl phosphate incubation is about 100-fold greater than in incubations containing 0.5% Triton X-100. In contrast with the lack of activity as a mannosyl acceptor for exogenous dolichyl phosphate in the present assay system, endogenous dolichyl phosphate clearly functions as an acceptor. Moreover in the same incubations a mannolipid with chromatographic properties of retinyl phosphate mannose was also synthesized from endogenous lipid acceptor. The biosynthesis of this mannolipid (retinyl phosphate mannose) was optimal at MnCl(2) concentrations between 5 and 10mm and could not be detected below 0.6mm-MnCl(2), when synthesis of dolichyl phosphate mannose from endogenous dolichyl phosphate was about 80% of optimal synthesis. Under optimal conditions (5mm-MnCl(2)) endogenous retinyl phosphate mannose represented about 20% of dolichyl phosphate mannose at 15min of incubation at 37 degrees C.