Retinol dehydrogenases (RDH) convert retinol into retinal, the intermediate in the biosynthesis of retinoic acid. All-trans-retinoic acid (atRA) regulates gene transcription and/or translation through retinoic acid receptors (RARs) and PPARδ (1). To test function of Rdh1, an efficient (Vmax/Km) and widely distributed RDH (2), our lab created Rdh1 knockout (KO) mice (3). Initial study of Rdh1-KO mice determined that when fed a low or vitamin A-deficient (VAD) diet, Rdh1-KO mice gain 33% more weight than wild-type (WT). However, when fed a chow diet (22+ IU vitamin A/g diet--i.e. copious), KO mice appeared identical to WT.

Our continued work on the Rdh1-KO mouse reveals additional insight into the role of Rdh1 in both retinoic acid synthesis and energy regulation. Both a 10% increase in length and 33% increase in adiposity contribute to weight gain in VAD-fed KO animals. Based on measurements of adipocyte size, increased adiposity results from hyperplasia of white adipose tissue. Rdh1-KO mice, fed a diet with the recommended amount of vitamin A (4 IU/g), also increase in weight and adiposity, but not length. Interestingly, high fat diet feeding fails to exacerbate weight or adiposity. Increased weight contributes to a decline in Rdh1-KO mouse health. Whereas young animals respond normally to tests of glucose and insulin tolerance, older Rdh1-KO mice become insulin resistant and glucose intolerant, with 2.5-fold higher insulin levels. These changes in Rdh1-KO glucose metabolism cannot be attributed to changes in pancreatic 9-cis-retinoic acid (9cRA) (4). As expected in heavier animals, leptin levels also increase 1.8-fold in Rdh1-KO mice. In addition, loss of Rdh1 reduces circulating thyroid hormones T4 and T3 by 12% and 17%, respectively, though TSH levels remain unchanged. These data demonstrate that atRA, generated by Rdh1, contributes to control of body weight, and does so to a physiologically relevant extent.

Increased weight results from either increased caloric intake or decreased energy expenditure. Overall, Rdh1-KO mice move just as much as WT, and their food intake increases only in proportion to body weight. Normalized to body weight, we observe no change to energy expenditure by indirect calorimetry. Thermogenesis, the generation of heat for body temperature maintenance, also contributes to energy expenditure. Rdh1-KO mice respond normally to cold challenge and β3-adrenergic stimulation, but appear reduced in body temperature, especially when stressed by fasting. Consistent with these observations, the brown adipose tissue (BAT) of Rdh1-KO mice expresses normal levels of the β2- and β3-adrenergic receptors and lower levels of Gbpar1, a bile acid receptor also involved in thermogenesis (5). Ucp1 expression increases when KO mice are fasted and returns to WT levels upon refeeding, suggesting UCP1 compensates for, rather than causes, reduced thermogenesis.

We used stable isotopes to measure de novo lipogenesis and triglyceride turnover in Rdh1-KO animals. Long term studies reveal reduced de novo lipogenesis, with less newly synthesized palmitate accumulating in white adipose tissue. In short term studies, brown adipose accumulated less new palmitate during 5 hours of refeeding. Comparable levels of total palmitate per gram tissue suggest increased storage of dietary lipid. Reduced rates of de novo lipogenesis, however, cannot explain increased adiposity in Rdh1-KO mice.

In samples of liver, white and brown adipose, we measured the relative levels of the most common fatty acids. In animals refed after fasting, the relative contribution of stearate is less in the brown adipose and liver of Rdh1-KO mice. In ad lib fed animals, white and brown adipose have reduced relative stearate levels. These changes to lipid composition are likely due to reduced expression of elongase Elovl6 in Rdh1-KO livers.

Loss of Rdh1 does not affect atRA, retinal, retinol or retinyl ester levels in ad lib fed mice in any tissue studied. However, studies of atRA during fasting and refeeding proved insightful to atRA function in WT and KO animals. In WT animals, atRA levels increase 1.3-1.5-fold in brown adipose tissue when refed 2-4 hours after a fast. In liver, atRA levels decline 30% after 5 hours of refeeding. Furthermore, reductions in hepatic retinol and retinyl ester during refeeding coincide with transient increases to serum levels, suggesting export of retinoid in the transition from fasting. Cold exposure had a similar impact on liver retinoids, but did not affect levels in brown adipose tissue. During refeeding, Rdh1-KO animals fail to maintain normal atRA levels in BAT and appear delayed in the export of retinol and esters from liver to serum.

Gene expression of retinoid synthesis genes also suggests coordination or retinoid metabolism with fasting and refeeding. In brown adipose tissue, Rdh1, Aldh1a1, Rbp1, Rbp4 and Rbp7 expression decrease between fasting and 6.5 hours of refeeding. Increased atRA, despite decreased gene expression, suggests additional regulatory mechanisms of atRA synthesis in BAT. In liver, multiple Rdh (Rdh1, Dhrs9, Rdh10), retinal dehydrogenases (Aldh1a1, Aldh1a2, Aldh1a3), retinol binding proteins (Rbp1, Rbp4) and Cyp2c39 decrease expression during refeeding. Alternately, Cyp26a1 expression increases with refeeding. Hepatic gene expression changes during refeeding are generally consistent with decreased atRA. Additional work on retinoid gene expression suggests some of these expression changes occur within 2.5 hours of refeeding and that high retinol diet disrupts the regulation of some retinoid metabolism genes.

To better understand the molecular mechanisms underlying the Rdh1-KO phenotype, we performed global gene expression analysis in liver, epididymal white adipose tissue (EWAT), testes and brown adipose tissue. Despite little to no Rdh1 expression in EWAT, we found nearly 3,000 gene changes in Rdh1-KO. We suspect many of these changes are secondary to weight gain. Microarray of liver, EWAT and BAT identified the gene Gbp1 upregulated in Rdh1-KO mice. Using qPCR, we found increases of 3-fold, 40-fold, and over 500-fold, respectively. Many inflammatory cytokines regulate the expression of Gbp1, yet serum levels of TNFα and IL-6 appear normal in KO mice. Based on our microarray study, we identified increased expression of Cap1, Klf2, Rbp4 and Gmpr and decreased expression of Adh1, Car5b and Orm2. Interestingly, increased Rbp4 expression does not lead to increased BAT expression of Pparg or Socs3, gene expression targets of RBP-STRA6 signaling (6). We have yet to determine which gene changes in BAT are direct retinoic acid targets or which lead to the weight and adiposity increases in Rdh1-KO mice.