Biochimica et Biophy,~ica Acta. I 12(} (1992) 183- 186 ~0 1992 Elsevier Science Publishers B.V. All rights reserved 1)167-4838/92/$05.00

183

BBAPRO 34154

Microsomal retinal synthesis: retinol vs. holo-CRBP as substrate and evaluation of NADP, N A D and N A D P H as cofactors

Joseph L. Napoli, Katalin C. Posch and Robbin D. Burns Departnlent of Biochenli.~t~'. School of Medicine attd Bionu'dical Sciences. The State Unh'crxity of New York at BufJakJ.

Buffalo. NY (USA)

( Received ! 2 August 1991 )

Key words: Retinol: Retinal: Relinoic acid: Microsome: Cellular relinol binding protein: Dehydrogenasc

Hoh)-CRBP (cellular retinol binding protein) is recognized specifically by an NADP-dcpcndcnt micro~mal retinol dchydro- gcna~ and protects retinol from conversion into retinal by NAD and NADPH dependent dehydrogcnascs. The synthesis of retinal from frcc rctinol is catalyzed by both NADP- and NAD-dcpcndcnt pathways, with the former being the preferred one (Km of 4 vs. 22 p.M for rctinol, and I"m~,/K,~ of 33 vs. 9, respectively). NADPH does not support quantitatively significant retinal synthesis from physiological concentrations of rctinol or holo-CRBP, if an NADPH regenerating system is used to prevent NADP formation.

Introduction

Retinoic acid is an activated retinol mctabolitc that supports vitamin A-dependent differentiation begin- ning during embryonic development and continuing throughout the life of animals [1-3]. Its synthesis, therefore, is critical [4]. The first and rate-limiting step in the biogenesis of retinoic acid from retinol in cul- tured cells is reversible dehydrogenation into retinal by enzyme(s) distinct from the alcohol dehydrogenases that catalyze ethanol metabolism [5,6]. Isolating the specific enzyme(s) that catalyze the dehydrogenation of retinol has been a problem because multiple dehydro- genases, including alcohol dehydrogenases, convert free rctinol into retinal in both cytosol and microsomes, usually with NAD as cofactor [7-11]. A microsomal NADP-dependent retinol dehydrogenase has been identified recently for which holo-CRBP 3 (cellular retinol binding protein) is substrate [ 12]. Because CRBP and other retinoid binding proteins exceed the concen- trations of unesterified retinol in vivo, resulting in low steady-state free retinoi concentrations, reactions of retinol metabolism that recognize holo-CRBP as sub-

Abbreviations: CRBP. cellular retinol binding protein; D'IT, dithio- threitol: Hepes. 4-(2-hydroxyethyl)-I-piperazineethanesulfonic acid: HPLC, high-performance liquid chromatography.

Correspondence: J.L. Napoli. Department of Biochemistry, 140 Far- bet Hall. Main St. Campus, SUNY-Buffalo, Buffalo, NY 14214, USA.

strate, such as retinol dehydrogenation [12] and esteri- fication [13,14], arc likely to bc physiologically signifi- cant in cells that express CRBP. The present work compares kinetic data of microsomal retinal synthesis with NAD, NADP and NADPH as cofactors, and free retinol and holo-CRBP as substrates.

Materials and Methods

Retbzoids Radioinert retinoids wcrc purchased from Eastman

Kodak and were purified by normal-phase HPLC and used within two weeks.

C R B P CRBP was expressed in E. coil with the vector

pMONCRBP, a gift from Dr. Marc Lcvin [15]. The CRBP was purified by pelleting the E. coil from 4.8 I of medium at 5 0 0 0 × g for 15 rain and resuspending the pellet in ~ 50 ml of 50 mM Tris-Hcl, 10% sucrose and 0.5 mM PMSF, pH 7.9. The suspension was pro- cessed through a French press at 1100 psi. The mate- rial recovered was spun at 100000 x g for h and was divided into two portions, one to provide hoio-CRBP (after saturation with retinol) and the other to provide apo-CRBP. Each was purified through a Sephadcx G-50 column as described [12,15].

Rat ricer microsomes Male Sprague-Dawley rats (Harlan, Indianapolis,

IN) were starved overnight and were decapitatcd. The

184

livers were removed and rinsed in ice-cold saline and were homogenized (1 g /4 ml) in 111 mM Hepes, 25(I mM sucrose, 1 mM EDTA, 2 mM DTT, pH 7.5. Micro~mes were prepared by differential centrifuga- tion as reported [8].

Retinal synthesis Enzyme assays were done in duplicate, unless stated

otherwise, at 37°C in buffer A (2 mM DTI', 150 mM KCI, 1 mM EDTA and 20 mM Hepes) at pH 8 with the cofactor indicated and with 150-2110 ttg protein in a final volume of 2(10 /zi for 30 rain when holo-CRBP was substrate and 51)0 #l for 20 min when unbound retinol was substrate. In some cases an NADPH gener- ating system was used in conjunction with the NADPH, which consisted of 1 unit of glucose-6-phosphate dehy- drogenase and 3 mM glucose-6-phosphate. Generally, duplicates were within I(1% of their averages. Un- bound retinol was added in 2 #l of dimethylsulfoxide, whereas CRBP was added in buffer A. Controls con- sisted of incubations with native protein in the absence of cofactors, or boiled protein, as specified in the figure legends. Reactions were quenched by the addi- tion of sufficient 0.025 N KOH/ethanol to raise the pH to at least 12 [16]. Buffer A was added to increase the volume to 500 /~l where appropriate, and the neutral retinoids were extracted with 2.5 ml of hexane. Retinal was quantified by elution (in -- 5 rain) from a normal-phase HPLC column (DuPont Zorbax-Sil Re- liance Cartridge, 0 .4 .4 cm) at a flow rate of 2 ml/min with an 8 min linear gradient of tetrahydrofur- an/hexane ( 1/99) to tetrahydrofuran/hexane (15/85). Sensitivity was at least 1 pmol (0.4 cm peak height at 370 nm).

Data analyses Kinetic data were fit to Michaelis-Menten plots with

the non-linear regression analysis program "Enzfitter" [17].

20

z E

1 0

E t~

a:

G

0 O ,

, , - o o o o o -

, • - - . i . , , , . . . . .

0 % 10 1 5

n o l o C R B P ( u M )

Fig. I. Cofactor effects o n microsomal retinal synthesis from ho l (~

CRBP. The cofactors used were 2 mM NADP (filled circles]. 2 mM NAD (open circles). I mM NADPH (trianglesk or no co-factor (squares). NADPtt was used in the pre~nce of an NADPH generat- ing system. Each concentration of holo.CRBP was obtained with a constant 2-molar excess of total CRBP/totai retinol, yielding a i / I molar ratio of holo-CRBP/aI~-CRBP at all holo-CRBP concentra- tions. As a result, the concentration of free retinol was constant at 16 nM over the range of htflo-CRBP concentrati~ms. This would result in a constant rate of retinal production of 11.5 p m o l / m i n / m g protein with NADP if only free retinol were substrate, calculated from the kinetic constants in Table I. With NAD the rate from free retinol

would have been 0.15 pmol /min /mg protein.

nal formation from holo-CRBP was driven prefer- entially by NADP vs. NAD over the entire range of substrate concentrations, such that the rate of retinal synthesis from NAD with holo-CRBP as substrate was minimal. This shows that CRBP protects retinol from both artifactual oxidation and NAD-dependent dehy- drogenation, but allows it to be metabolized by an NADP-supportcd pathway.

The K,, with holo-CRBP and NADP was lower than the Km with free retinol using either NADP or NAD (Table 1), suggesting preferential binding of holo-CRBP to the dehydrogenase relative to free retinol. The dehydrogenation of free retinol also was driven preferentially by NADP. Retinol had 5-fold

Results and Discuss ion

We reported recently that holo-CRBP is a substrate for NADP-supported microsomal retinal synthesis and showed that the rate of retinal synthesis from 5 p M retinol was similar with either NADP or NAD as cofactor, but with 5 /.tM holo-CRBP, NADP was the preferred cofactor [12]. To expand these data and to determine the precise relationship between cofactor preference and retinal production from retinol and holo-CRBP in microsomes. Michaelis-Menten curves were obtained with either NAD or NADP over con- centration ranges of free retinol or holo-CRBP. With holo-CRBP, retinal was not formed without cofactor except at 15 pM, and then the retinal was only about 3% of the amount obtained with NADP (Fig. 1). Reti-

TABLE I

Kinetic constants of microsomal retinal synthesis as a fimction of cofactor atul substrate "

Substrate Cofactor N b Km i/ma,,

/,t M p m o l / m i n / m g

Retinol NAD 4 22 _+8 205+-81 Retinol NADP 3 4 +0.7 131 + 15 Retinol NADPH 2 > ll)[I NS t. Holo-CRBP NAD 2 NA d I Holo-CRBP NADP 5 1.9+_11.6 23+ 9 Holo-CRBP NADPH 2 NA d ND ~

~' Data are the mean_+S.D. Two mM NAD or NADP or I mM NADPH with an NADPH generating system were used.

h Number of independent determinations. " NS, not saturated al a retinol concentration of I(10 p.M. d NA. not applicable.

~' ND. no retinal detected above control.

185

higher affinity (lower K.,,) for an NADP-dependent reaction relative to a NAD-dependent process. The maximum velocities of retinal synthesis were both co- factor and substrate dependent. With holo-CRBP, NADP supported a velocity at least an order of magni- tude higher than NAD, but with free retinol, NAD supported a maximum velocity 2-fold higher than NADP. The ratios of V,,~x/K,, for NADP and free retinol, NADP and holo-CRBP, and NAD and free retinol were 33, 12 and 9, respectively. This indicates that regardless of substrate, the NADP-supported pathway is the kinetically favored one. When the physiologically available substrate is considered, then NADP and hoIo-CRBP are the major source of retinal. Concentrations of CRBP and the serum retinol binding protein are equal to or greater than the retinol concen- trations in liver: therefore concentrations of unbound retinol are likely to be less than 100 nM, making holo-CRBP the quantitatively major substrate available physiologically []4].

A retinol dehydrogenase with a K m of 120/zM for free retinol that prefers NAD as cofactor has been reported as a major route of retinal synthesis in micro- somes [9]. The difference between this result and the current one may stem in part from the calculation of net retinal. A substantial amount of retinal is produced art]factually from free retinol regardless of the type of control used. An example of this problem with a no- eofactor control is shown in Fig. 2, which depicts one of the four experiments summarized in Table l. In these experiments, for example, retinal observed in the absence of NAD ranged from 25 to 44% of that found with l mM NAD at 25 # M retinol. This background was just" as evident with boiled protein or no protein controls. In fact, the data were erratic using controls other than no-eofactor controls with free retinol, sug-

3 0 0

/ / f O 2 0 0

c_

1 0 0 • 0 ~ ~

r , o . ~ 0 , , i , i I i i i i i i , , P • . • 1

O 2 0 4 0 6 0 8 0 1 ( . ) 0

R e t t n o ~ { I J M )

Fig. 2. Microsomal retinal synthesis from unbound retinol. Retinal generated in the control reactions run in the absence of cofactors (open circles) was subtracted from the retinal observed in the pres- ence of 2 mM NAD (filled circles) to provide the cofaclor-dependent rate of retinal production (triangles). The kinetic values for this specific experiment, as determined by non-linear regression analyses (Enzfitter) were: for total retinal (filled circles) K m = 76 # M and Vr,~ = 475 p m o l / m i n / r a g protein: for net retinal (triangles) K m = 20

/zM and Vm~ = 124 p m o l / m i n / m g protein.

2 z~ / j / . . / / ~ / / / / /

1 5 0

o 0 2 5 5 0 7 5 1 0 0

R e t i n o l ( p M )

Fig. 3. Effects of NADPH on micmsomal retinal synthesis from untn)und retinol. Retinal was measured with a control reaction done in the absence of cofactor ((}pen circles), with I mM NADPH (filled circles), or with I mM NADPH plus an NADPII generating system

(triangles).

gesting that thc native protcins solubilized and /o r stabilized retinoi. If total retinal measured from free retinol and NAD had been used to calculate kinetic constants from our data, rather than the net retinal (total retinal minus the retinal in the control), then the K m would have been 76 + 14 p.M and the Vm~,x would have been 475 + 102 pmol re t inal /min/mg protein.

Recently, an observation was reported of an NADPH driven rat liver microsomal retinol dehydrogenase, i.e. a mixed function retinol oxidase [18]. in an attempt to confirm this observation and to determine whether holo-CRBP could be substrate for such an oxidase, the effect of NADPH on the conversion of CRBP-bound retinol and free retinol was examined. NADPH does not support retinal synthesis from holo-CRBP when NADPH is used in conjunction with an NADPH re- generating system to prevent accumulation of NADP (Fig. l, Table i). On the contrary, rather than stimulat- ing retinal formation, NADPH inhibits it. With 5 # M holo-CRBP and 2 mM NADP, adding 0.4 mM NADPH caused 50% inhibition of retinal synthesis (51-± 8 vs. 25 + 3 pmol/0.2 /~g protein/30 rain; mean + S.D., n = 4) and 2 mM NADPH reduced retinal to non-detec- table levels.

Relative to a no-cofactor control, NADPH de- pressed the retinal measured with retinol as substrate (Fig. 3). NADPH plus an NADPH generating system suppressed retinal much more severely, perhaps through reduction of retinal formed art]factually in the absence of cofactor. Notably, net retinal formation with NADPH in the presence of a regenerating system occurred only at high and non-physiological retinol concentrations and the rate was low ( ~ 6 pmol / rain/rag protein at 100/zM retinol). This differs from the recent report of NADPH-supported rat liver mi- crosomal retinal formation with a K m of 285 /~M, apparently calculated from rates that had not reached saturation and therefore likely to be an underestimate, and a Vm,, x of 300 pmol /min /mg protein, obtained without an NADPH regenerating system [ 18]. As shown

186

here, NADPH provides different results than NADPH plus a regenerating system, probably because the for- mer allows production of sufficient NADP to support retinal synthesis. This is probable because the K~, for NADP is 10 p.M with frec retiaol as substra,¢ [12].

In summary, this report shows that holo-CRBP is recognized specifically by an NADP-dependent micro- somal retinol dchydrogenase and protects retinol from oxidation by NAD and NADPH dependent dehydro- genases. Free retinol is dehydrogenated by both NAD and NADP dependent enzymes, with the NADP-de- pendent process being the preferred one (higher Vm~x/Km). NADPH does not support quantitatively significant retinal synthesis from physiological concen- trations of free retinol. Retinol in liver is essentially completely bound to either the serum retinol binding protein or CRBP, with holo-CRBP concentrations about 5 p.M. Therefore the concentrations of free retinol are likely to be less than 0.1 pM, from retinolfr~ = (Ka • retinolt,,t,,i)/(K d + binding pro- tein,,t~ I -binding protein~.,~). Calculating from the K m values with the two substrates (retinol or holo- CRBP) and the two effective cofactors (NAD and NADP), and considering the physiological concentra- tions of available substrates (-,-5 p.M holo-CRBP vs. ~ 0.1 p.M free retinol), the quantitatively most impor- tant route of retinal synthesis would occur from holo- CRBP and NADP in cells that express CRBP.

Acknowledgement

This work was supported by National Institutes of Health Grant DK36870.

References

I Lotan. R. (1988) Prog. Clin. Biol. Rcs. 259. 261-2."1. 2 Noji, S.. Nohno. T., Koyama, E.. Muto. K.. Ohyama. K.. Aoki, Y..

Tamura. K.. Ohsugi. K.. lde. H,, Taniguchi, S. and Saito, T. (1991) Nature 350, 83-86.

3 Wanek. N., Gardiner, D.M., Muneoka. K. and Bryant, S.V. (1991) Nature 350. 81-83.

4 Napoli. J.L. (1990) in Chemistry and Biology of Synthetic Retinoids (Dawson. M.I. and Okamura. W.H., cds.) pp. 229-249. CRC Press, Boca Raton.

5 Napoli. J.L. (1986)J. Biol. Chem. 261. 13592-13597. 6 Siegenthaler, G.. Saurat. J.H. and Ponce. M. (19~0) Biochem. J.

268. 371-378. 7 Nicolra. C. and Livrca. M.A. (1982) J. Biol. Chem. 257, 11836-

11841. 8 Nal~)li, J.L. and Race. K.R. (1987) Arch. Bi(~hem. Biophys. 255,

95-101. 9 Leo, M.A., Kim, C.-L. and Lieber, C.S. (1987) Arch. Biochem.

Biophys. 259. 241-249. l0 Posch, K.C.. Enright, W.J. and Napoli, J.L. (1989) Arch. Biochem.

Biophys. 274. 171-178. I I Napoli, J.L. and Race. K.R. (1990) Biochim. Biophys. Acta 1034.

228-232. 12 Posch. K.C.. Boerman. M.E.H.. Burns. R.D. and Napoli, J.L.

O991) Biochemistry 30, 6224-6230. 13 Ong. D.E., MacDonald. P.N. and Gubitosi, A.M. (1988) J, Biol.

Chem. 263. 5789-5796. 14 Yost, R.W., Harrison, E.H. and Ross, A.C. (1988)J. Biol. Chem.

263, 18693-18701. 15 Levin. M.S., Li, E. and Gordon, J.l. (1990) Methods En~mol.

189, 506-520. 16 Napoli, J.L. (1990) Methods Enzymol. 189, 470-482. 17 Leatherbarrow, RJ. (1987) Enzfitter: A Non-linear Regression

Data Analysis Program. EI~vier-BIOSOFT. Cambridge. UK. 18 Shih, T.W. and Hill. D.L. (1991) Drug Metabol. Disp. 19, 332-335.