Kinetic and thermodynamic studies of the conversion of previtamin D3 to vitamin D3 in human skin.

The thermoisomerization of previtamin D3 to vitamin D3 is the last step in the synthesis of vitamin D3 in human skin. Kinetic and thermodynamic studies of this reaction in human skin and an organic solvent revealed that not only the equilibrium of the reaction was shifted in favor of vitamin D3 formation in human skin (equilibrium constant K at 37 degrees C = 11.44) compared to hexane (K = 6.15), but also the rate of the reaction was increased by more than 10-fold in human skin (T1/2 at 37 degrees C = 2.5 h) when compared to hexane (T1/2 = 30 30 h). This extraordinarily fast reaction rate was also confirmed in vitro in chicken skin and in vivo in human subjects. The enthalpy change for the reaction determined by the van't Hoff plot was delta H degree = -21.58 kJ mol-1 in human skin and delta H degree = -15.60 kJ mol-1 in hexane. Arrhenius plots showed that the activation energies for both the forward and the reverse reactions were lower in human skin (Ea1 = 71.05 kJ mol-1 and Ea2 = 92.63 kJ mol-1) than in hexane (Ea1 = 84.90 kJ mol-1 and Ea2 = 100.5 kJ mol-1). Activation parameters for the reaction in human skin and in hexane were also reported. Subcellular fractionation of human epidermal tissue revealed that most epidermal 7-dehydrocholesterol and previtamin D3 were in the membrane fraction, while only 20% were in the cytosol. The interaction of previtamin D3 with intracellular lipids and/or proteins in skin may be responsible for the increased vitamin D3 formation rate in the skin.

Of all the vitamins, vitamin D3 is unique in that it can be photosynthesized in the skin. When human skin is exposed to sunlight, the high energy ultraviolet (UVB)' (290-315 nm) radiation penetrates the epidermis and photolyzes epidermal 7-dehydrocholesterol (7-DHC) to previtamin D3 (preDJ and other photoisomers (Holick, 1987). Once formed, preD3 immediately begins to isomerize to vitamin D3 by a temperaturedependent process (Velluz et al., 1949;Holick et al., 1980). The reaction rate of this isomerization is enhanced as the temperature increases. Early studies found that the preD3 isomerization rate could not be influenced by acids, bases, catalysts, and inhibitors of radical chain processes * This work was supported by National Institutes of Health Grant R01-AR36963. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The stereochemical mechanism of this [1,7]-sigmatropic hydrogen migration was postulated correctly even before the enunciation of the concepts of orbital symmetry rules. According to the rules, Woodward and Hoffmann (1965) predicted that the thermal [1,7]-sigmatropic hydrogen migration should be an antarafacially allowed and suprafacially forbidden process (Woodward and Hoffmann, 1965). Recently, experimental evidence supporting this mechanism has been obtained from the study of thermal behavior of deuteriumlabeled cis-isotachysterols (Hoeger et al., 1987). Although the skin's ability to generate preD3 and vitamin D3 has been appreciated for decades, the kinetics and thermodynamics of the preD3 s vitamin D3 reaction in human skin following UVB irradiation has not been elucidated. Much of the information about the preD3 thermoisomerization has been obtained in organic solvents (Hanewald et al., 1961) and assumed to be the same in human skin. However, the slow isomerization rate of preD3 in organic solvents at 37 "C could not explain the relatively rapid increases in the serum vitamin D3 concentrations 12 to 24 h after the whole body UVB irradiation (Adams et Clemens et al., 1982). To evaluate this apparent discrepancy, we quantitatively measured the rate constants and equilibrium constants of the preD3 $ vitamin D, reaction in human skin and organic solvents. We also assessed the effects of temperature on the rate constants and the equilibrium constants and calculated the Arrhenius activation energy and the enthalpy change of the reaction as well as other activation parameters in both models.

Photolysis of Samples-The UVB source consisted of eight Medical
Chemicals-7-DHC and vitamin DI were purchased from Sigma. Sunlamps (National Biological Corp., Cleveland, OH). Spectral output for these lamps peaked at about 314 nm with broad output between 270 nm and 380 nm as measured by a Spectroradiometer (Model 742, Optronic Laboratories, Orlando, FL). About 56% of the total output from these lamps was within the UVB range. The amount of UVB that reached the samples was measured with an 11-1700 International Light Research Radiometer (Newburyport, MA) with a UVB probe. The accumulated UVB output during the entire exposure time was 0.5 J cm-2.
Fresh neonatal foreskin, fresh chicken shank skin, and quartz test tubes containing 7-DHC in n-hexane were irradiated on ice. Control samples of skin and 7-DHC solutions were kept in the dark. The human epidermis was separated from dermis as previously described (Holick et al., 1980). For the chicken skin, the subcutaneous fat was removed and the epidermis and dermis were irradiated followed by extraction. Triplicate samples of exposed skin and 7-DHC solutions in sealed ampules that were purged with argon were incubated at various temperatures including 30, 37, 50, 60, 70, and 80 "C for different durations as indicated. Although the human skin is viable for up to 50 h at 30 and 37 "C, it is likely that at higher temperatures the cellular activity of the skin ceased. The epidermis was extracted three times with 8% ethyl acetate in hexane at -20 "C overnight. Skin extracts and solution samples were dried under a stream of nitrogen and reconstituted with 0.45% isopropyl alcohol in hexane.
Experimental Subjects-The study population was comprised of three healthy white subjects (two females ages 42 and 22 with skin types 111 and IV, respectively, and a 29-year-old male with skin type 11) without a history of skin, hepatic, or renal disease. None was taking vitamin D, anticonvulsant medications, or corticosteroids. All subjects gave their informed consent as approved by the Jefferson Medical College Institutional Review Board.
Cutaneous UV Irradiation in Vivo-Subjects were exposed to UVB irradiation in a walk-in radiation chamber (photounit Psoralite UVA/ UVB, Columbia, SC). The radiation delivered by the wavelength in the 260 to 320 nm band is 0.2 milliwatt cm-', determined at a distance 30 cm from the source. The previously determined one minimal erythema dose for this photounit is 33 to 36 mJ cm-' in whites with skin type 111. 1 min, 30 min, and 60 min after the exposure, a 4-mm skin biopsy was taken from the exposed buttock. The biopsy specimens were snap-frozen in liquid nitrogen for 5 min. The epidermis was separated from dermis and extracted with organic solvents as previously described (Holick et al., 1980). Separation and Quantification-Vitamin D3 and the photoisomers of 7-DHC were analyzed at 254 nm by a high performance liquid chromatography system (HPLC) (Model 440, Waters Associates, Inc.) with an Econosphere silica column (250 X 4.6 mm, 5 pm, Alltech Associate Inc., Deerfield, IL). The mobile phase was 0.45% isopropyl alcohol in n-hexane. The amount of vitamin DI and preD3 was determined from the peak areas based on the standard curves as previously described (Holick et al., 1980). Cellular Fractionation of Epidermis-Fresh newborn human foreskin epidermis was separated from dermis by heating the skin at 60 "C for 30 s. Histological analysis of similarly treated skin samples showed that their cellular structure remained intact (Webb et al., 1989). The separated epidermal layer plus basal cells which were scraped from the surface of dermis were combined and homogenized together in 0.05 M sucrose buffer, pH 7.5 at 4 "C, and fractionated by differential centrifugation to obtain membrane, cytosolic, and mitochondria fractions. The total lipid wa$ extracted from each fraction, and 7-DHC was isolated and quantitated by the HPLC method described above.
Determination of the Rate and Equilibrium Constants-The thermoisomerization between preD3 and vitamin D3 is a first order reversible reaction: where kl and kz represented rate constants for the forward and the reverse reaction. The equilibrium constant ( K ) was calculated as follows: The rate constants were obtained by the equation where D, and De were vitamin D3 concentrations at times t and the time to reach the equilibrium, respectively (Alberty et al., 1992). Measurements of Thermodynamic Parameters-Applying the van't Hoff equation dlnK/d(l/T) = -m / R , the enthalpy change for the reaction was determined from the slope of the linear relationship between the natural logarithm of K versus the inverse of the absolute temperature observed from the experimental data. The free energy change (AGO) and the entropy change (AS') at 25 "C were calculated from A@ = -RTlnK and ASo = ( a H 0 -A@)/T.
Determination of the Activation Energy and Other Activation Parameters-The activation energies for the forward and reverse reaction were determined according to the Arrhenius equation: dlnk/dT = E,,/RTZ. Integration of the equation yielded Ink = InA -EJRT, where k is the rate constant for the reaction, E. is the Arrhenius activation energy, R is the molar gas constant, and T is the temperature in degrees Kelvin, while A is a constant, the "pre-exponential factor." A plot of Ink versus 1/T shows straight lines. The value of E.
can be calculated directly from the slope (-EJR) of the line.
According to Eyring's transition-state theory, the rate constant is expressed as where kB is Boltzmann constant, h is Planck's constant, and K is the transmission coefficient. The enthalpy of activation (&) can be calculated from the equation & = E, -RT. Then if it is assumed that the transmission coefficient is unity, both A@ and ASt can be obtained from the above equations.

RESULTS
HPLC Analysis-The HPLC conditions used in this study allowed baseline separation of 7-DHC from its photoisomers including preD3, lumisterol, and tachysterol, as well as the separation of vitamin D3 from tachysterol ( Fig. 1). No interfering peaks were found in the preD3 and vitamin D3 regions in unexposed controls (Fig. 1, A and D). After irradiation and before incubation, the major photolytic product found in human skin and in the organic solvent was preD3 with small amounts of lumisterol and tachysterol. No vitamin D3 was detected ( Fig. 1, B and E ) . The vitamin Da peak was apparent after incubation at 37 "C for 2 h (Fig. 1, C and F ) . uersw time of the preDB + vitamin D3 reaction in human skin and in hexane. The linear plot indicated that the kinetics of the isomerization was a first order reversible reaction. The T1/2 for the conversion in human skin was 2.5 h, which was 12 times faster than that in hexane (Tl/z = 30 h). In hexane at 37 "C, the rate constant for the forward reaction kl was 6.89 X s-', and the rate constant for the reverse reaction k2 was 1.12 X 1O"j s-'. The rate constants measured in human kZ = 6.93 x s". Compared to the reaction in hexane, in human skin the forward reaction rate and the reverse reaction rate were increased by more than 10-and s-fold, respectively. To determine whether the extraordinarily fast reaction rate was unique to human skin, a similar study was conducted in chicken skin. versus time for the preD3 e vitamin D3 reaction at 40 "C (the normal temperature of a chicken) in chicken skin, human skin, and in hexane, and the calculated total rate constants were 1.30 X lo-* s-l, 1.20 X s-l, and 1.09 X loL5 s-', respectively. The reaction rates were similar in chicken skin and in human skin, and both rate constants were more than 10 times higher than the reaction rate in hexane.
Thermal Conversion of preD3 to Vitamin Da in Vivo-The male subject received 36 mJ cm-' ; the females with skin types 111 and IV received 51 and 54 mJ cm-' of UVB, respectively.
After UVB irradiation, skin biopsies were obtained serially, at 1 min, 30 min, and 60 min following UVB irradiation. The relative amounts of vitamin D3 formation (mean f S.D.) at these times were 1.3 k 2%, 9 f 4%, and 13 f 3%, respectively (Fig. 2, inset). Thermodynamics of preD3 Vitamin D3 Reactions-The enthalpy change for the reaction ( H ) was determined by the van't Hoff equation. Plotting I n K against 1/T gave straight lines for the reaction in human skin and in hexane (Fig. 4A). The @ for the reaction in human skin was -21.58 kJ mol", while the AH for the reaction in hexane was -15.60 kJ mol". At 25 "C, the free energy change for the reaction was lower in human skin (A@ = -6.83 kJ mol") than in hexane (A@ = -5.14 kJ mol-'). The AS' was -49.24 J K" mol" in human skin, while it was -35.25 J K" mol" in hexane.
Activation Energy and Activation Parameters- Fig. 4B and 4C were Arrhenius plots for the preD3 e vitamin DS reaction in human skin and in hexane based on the kinetic data summarized in Table I as well as Fig. 5, A and B. The straight lines were obtained in both systems. In human skin, the activation energy for the forward reaction (EDl) was 71.05 kJ mol" and for the reverse reaction (Err2) was 92.63 kJ mol", both of which were lower than the correspondent activation energies in hexane, i.e. Eal = 84.90 kJ mol" and Eo2 = 100.5 kJ mol". The thermodynamic activation parameters for the preD3 + vitamin D3 reaction in human skin and in hexane were calculated from Eyring plots and were summarized in Table 11. The activation parameters for the reaction in hexane agreed very well with the reported values by Yamamoto and Borch (1985) ( Table 11). The data in Table I1 also showed that in human skin the free energies of activation for the forward and reverse reactions were reduced by 6.62 and 5.00 kJ mol-l, respectively, compared to those in hexane.
The rate and equilibrium constants for the preD3 e vitamin D3 reaction at any temperature (T) were calculated from the Arrhenius equations:    where, for the reaction in human skin: lnA1 = 18.15, E,I = 71.05 kJ mol-', lnA, = 24.10, and Ea2 = 92.63 kJ mol", whereas for the reaction in hexane: 1nA1 = 21.05, Ea, = 84.90 kJ mol", lnA, = 25.26, and Ea2 = 100.5 kJ mol". The equilibrium constant ( K ) is temperature-dependent and was equal to the ratio of the kl versus k:! rate constants.

Subcellular Localization of 7-DHC in Human Epidermis-
The human epidermis was homogenized and separated by differential centrifugation into nuclear plus mitochondria, cytosolic, and membrane fractions. The total content of 7-DHC in the soluble fraction of homogenized human epidermis was 618 ng cm-', wherein 65% was found in the membrane fraction (400 5 8.7 ng cm-' of epidermis) while only 15% and 20% were found in the nuclear with mitochondrial (96.3 k 9.0 ng cm-' of epidermis) and cytosolic fractions (122 f 11.7 ng cm-' of epidermis), respectively.

DISCUSSION
The isomerization rate of preDB to vitamin D3 is relatively slow in an isotropic organic solvent such as hexane. At 37 "C, it takes about 6 to 10 days to reach equilibrium (Fig. 2). However, when healthy young adults were subjected to whole body UV radiation, the serum vitamin D3 concentrations reached peak values within 12 to 24 h (Adams et al., Clemens et al., 1982). One hypothesis attributed the relatively rapid increases in serum vitamin D3 concentration to the specific translocation of vitamin D3 from skin into the circulation by the vitamin D binding protein, but not to the cutaneous effects on the reaction rate (Holick et al., 1980). The consequence of this highly specific translocation would result in the removal of vitamin DS from the skin as it was being formed, thereby changing the thermal reaction from a reversible process to an irreversible process. Based on the kinetic data presented in this study, it was calculated that for the reversible reaction in hexane, the T I / , was 30 h, whereas for the irreversible process the Tl12 was 28 h. Since there was little difference between the two TI/, values, and both were longer than 1 day, this theory alone was not adequate to explain the differences between the in vivo and in vitro data. Thus, although the specific translocation of vitamin DS into circulation might thermodynamically favor vitamin D3 synthesis in high yield, this phenomenon would have little effect on the kinetics of preD3 isomerization to vitamin D3 due to the much slower reverse reaction rate at 37 "C. As would be expected, in addition to the selectivity and specificity, the vitamin D3 synthesis in the biological system should also be more efficient in terms of a faster reaction rate and higher equilibrium constant. As reported in this study, the Tl12 for the preD, $ vitamin DS reaction in human skin was only 2.5 h compared to 30 h in hexane, and the equilibrium constant for the vitamin DS formation was 86% greater in human skin compared to hexane. It has been reported that the serum concentrations of vitamin D, of healthy adults who received whole body UV radiation had increased to peak values within 24 h and then gradually declined to basal values (Adams et al., 1982;Clemens et al., 1982). This fact immediately eliminates the possibility that the rate of vitamin D3 formation in vivo was similar to that in organic solvents since the Tllp for the reaction in hexane at 37 "C is 30 h, which is longer than 1 day. On the contrary, as reported in this study, the Tl12 for vitamin D3 formation in human skin is only 2.5 h, and the time to include four T112 intervals is 10 h. This observation explains the relatively rapid rise in the blood levels of vitamin Preuitamin 0 3 Thermoconversion in Human Skin D3 after exposure to UVB radiation (Adams et al., 1982;Clemens et al., 1982). The in vitro observations on the rates of vitamin D3 formation from preDs in human skin were confirmed in vivo in human subjects (Fig. 2) and in vitro in chicken skin (Fig. 3B). Fig. 2 also indicates that the difference between the percent vitamin D3 formation in excised skin and in vivo becomes larger as the post UVB time becomes longer.
A comparison of the amount of vitamin D, formed from preD3 in human skin at 30 and 60 min in vivo and in vitro revealed that there was 11 and 26% less vitamin D3 in the epidermis from the in vivo study. Since there is no evidence that vitamin D3 is rapidly metabolized in the skin to 25-hydroxyvitamin D3 (MacLaughlin et al., 19911, the most likely explanation is that as vitamin D3 is being formed, it is selectively translocated from the epidermis into the circulation (Holick et al., 1980).
The equilibrium as well as the rate of preDB vitamin D3 reaction depends strongly on temperature. The temperature effects are quantitatively given by the van't Hoff equation 1nK = constant -AH/RT and the Arrhenius equation Ink = constant -E,/RT. In general, the equilibrium favors vitamin D3 formation as temperature goes down, and the conversion rate increases as the temperature goes up due to the negative AH and the positive E, in the above equations. For example, in order to achieve the relatively high yield of vitamin DB a t 37 "C in hexane (about 86%), the reaction rate is relatively slow and takes about 7 to 10 days to reach equilibrium. In human skin due to the decreased E, and increased reaction rate, the preD3 isomerization rate was increased more than 10 times than in hexane, while the activation energy was lowered by 13.85 kJ/mol. Thus, it is possible to achieve both a high yield and a rapid equilibrium rate in human skin at a body temperature which would be impossible to achieve for the reaction in hexane. Based on the Arrhenius equation, it was calculated that the temperature required for the preD3 isomerization to vitamin DO in hexane to proceed as rapidly as in human skin at 37 "C was 62 "C. It was apparent that this temperature could not be reached in human skin under physiological conditions. Using the van't Hoff equation, the calculated temperature required for the equilibrium constant in hexane to reach the higher value in human skin a t 37 "C was 9 "C. Thus, human skin profoundly changed the rate constant and equilibrium constant in favor of vitamin D3 formation which would be impossible to achieve by heating or cooling the human body under physiological conditions. The thermal formation of vitamin D3 occurs exclusively from certain classes of conformations of preD,, and the reaction kinetics depends on the properties of the reaction medium, i.e. isotropic organic solvents versus anisotropic media such as highly ordered phospholipids (Cassis and Weiss, 1982;Yamamoto and Borch, 1985). We found that most of the epidermal 7-DHC were localized in the membrane fraction and only about 20% in the cytosol fraction. We propose that the increased reaction rate in human skin is primarily due to the influence of the highly ordered phospholipids environment in which preDa is generated. Molecular mechanic calculation (Dauben and Funhoff, 1988a) as well as 'H NMR evidence (Dauben and Funhoff, 1988b) indicate that there existed two main triene types, corresponding to two cZc conformers and two tZc forms each. During transition state, hydrogen migration might take place from the two cZc conformers only, with a preference for the right-handed helical conformer (+)cZ(+)c (Enas et al., 1991). The (+)cZ(+)c conformer with ring A lying below the plane of the C/D ring permitted the hydrogen to be delivered axially from CI9 to C9, leading to the more stable conformers. The ratio of the right-to left-handed conformer had been determined experimentally to be 2:1 a t the [1,7]-migration in organic solvents (Sheves et al., 1979).
Based on the conformational analysis of preD3, a possible explanation of the faster conversion rate in human skin may result from a change of conformational equilibrium of preD3 in favor of cZc conformers or result from the increased ratio of right-to left-handed cZc conformers in the membrane due to the presence of less fluid, more ordered membrane phospholipids. Thus, the activation energy of the reaction could be lower and the reaction rate would be faster in human skin than those in an isotropic organic solvent. Similar rate enhancement and lowered activation energy had been reported in the liposome models (Yamamoto and Borch, 1985). Yamamoto and Borch reported that at 40 "C, the rate of thermal isomerization of preDB to vitamin D, in egg lecithin liposomes and in dipalmitoylphosphatidylcholine liposomes increased by 8-and 11-fold, respectively, compared to that in hexane (Yamamoto and Borch, 1985). Rate enhancement was also reported by Cassis and Weiss using cholesteric liquid crystalline solvents (Cassis and Weiss, 1982). But, it should be pointed out that unlike in human skin, in the liposome models the equilibrium constant actually decreased compared to that in hexane (Cassis and Weiss, 1982). Thus, the use of the liposome models to study the formation of vitamin D3 from preDB in the biological systems is not ideal.