Flavin homeostasis in the mouse retina during aging and degeneration

Involvement of flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN) in cellular homeostasis has been well established for tissues other than the retina. Here, we present an optimized method to effectively extract and quantify FAD and FMN from a single neural retina and its corresponding retinal pigment epithelium (RPE). Optimizations led to detection efficiency of 0.1 pmol for FAD and FMN while 0.01 pmol for riboflavin. Interestingly, levels of FAD and FMN in the RPE were found to be 1.7- and 12.5-fold higher than their levels in the retina, respectively. Both FAD and FMN levels in the RPE and retina gradually decline with age and preceded the age-dependent drop in the functional competence of the retina as measured by electroretinography. Further, quantifications of retinal levels of FAD and FMN in different mouse models of retinal degeneration revealed differential metabolic requirements of these two factors in relation to the rate and degree of photoreceptor degeneration. We also found twofold reductions in retinal levels of FAD and FMN in two mouse models of diabetic retinopathy. Altogether, our results suggest that retinal levels of FAD and FMN can be used as potential markers to determine state of health of the retina in general and more specifically the photoreceptors.     

Frontal gel chromatographic analysis of the interaction of a protein with self-associating ligands: aberrant saturation in the binding of flavins to bovine serum albumin

Frontal gel chromatography is an accurate method to obtain the total free ligand concentration of a protein-ligand mixture in which ligands self-associate. The average number of bound ligands per protein molecule is obtained as a function of the total free ligand concentration. The method was applied to the interaction of bovine serum albumin with self-associating flavins. The binding curves for FMN and FAD leveled off at about 0.7 and 0.5, respectively. These data were simulated well by a binding model where flavins undergo isodesmic indefinite self-association and the monomer alone binds to a single binding site of albumin. The isodesmic association constants of FMN and FAD were (1.7 +/- 0.1) x 10(2) and (2.2 +/- 0.3) x 10(2) M(-1), respectively. The binding constants of the monomer of FMN and FAD were (7.6 +/- 0.2) x 10(2) and (3.5 +/- 0.2) x 10(2) M(-1), respectively. FMN competitively inhibited the binding of FAD to albumin. The affinity to flavins was in the following order at pH 5.8: lumiflavin, FMN, riboflavin, and FAD. The SH modification and the binding of palmitate did not affect the FMN binding to bovine albumin at pH 5.8. As pH increased from 5.8 to 9.0, the affinity to FMN of bovine albumin decreased 3-fold, whereas that of human albumin increased about 80-fold. The present study clearly showed how isodesmic self-association of a ligand can cause apparent saturation of the interaction of a protein with the ligand at levels lower than 1.     

31P NMR spectroscopic studies on purified, native and cloned, expressed forms of NADPH-cytochrome P450 reductase.

31P NMR spectroscopy has been utilized in conjunction with site-directed mutagenesis and phospholipid analysis to determine structural aspects of the prosthetic flavins, FAD and FMN, of NADPH-cytochrome P450 reductase. Comparisons are made among detergent-solubilized and protease (steapsin)-solubilized preparations of porcine liver reductases, showing unequivocally that the 31P NMR signals at approximately 0.0 ppm in the detergent-solubilized, hydrophobic form are attributable to phospholipids. By extraction and TLC analysis, the phospholipid contents of detergent-solubilized rat liver reductase, both tissue-purified and Escherichia coli-expressed, have been determined to reflect the membranes from which the enzyme was extracted. In addition, the cloned, wild-type NADPH-cytochrome P450 reductase exhibits an additional pair of signals downfield of the normal FAD pyrophosphate resonances reported by Otvos et al. [(1986) Biochemistry 25, 7220-7228], but these signals are not observed with tissue-purified or mutant enzyme preparations. The Tyr140----Asp140 mutant, which exhibits only 20% of wild-type activity, displays no gross changes in 31P NMR spectra. However, the Tyr178----Asp178 mutant, which has no catalytic activity and does not bind FMN, exhibits no FMN 31P NMR signal and a normal, but low intensity, pair of signals for FAD. The latter experiments, taking advantage of mutations in residues putatively on either side of the FMN isoalloxazine ring, suggest subtle to severe changes in the binding of the flavin prosthetic groups and, perhaps, cooperative interactions of flavin binding to NADPH-cytochrome P450 reductase.     

Flavin-sensitized photoreduction of thymidine glycol

Photochemical reactivity of flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN) toward thymidine glycol (dTg) has been investigated. Fluorescence intensity of FAD was enhanced as increasing the concentration of dTg, suggesting that adenosine moiety of FAD interacts with dTg. However, photoreduction of dTg using reduced form of FAD gave repaired thymidine in almost the same yield as when reduced FMN was used alternatively, and thus such interaction seems to have no effect on the reduction. Oligodeoxynucleotides containing dTg were also photochemically repaired by reduced form of flavins in different yields depending on the sequence, which could be related to electron affinity of the nucleobases in DNA.     

Thermodynamic basis of electron transfer in dihydroorotate dehydrogenase B from Lactococcus lactis: analysis by potentiometry, EPR spectroscopy, and ENDOR spectroscopy

Dihydroorotate dehydrogenase B (DHODB) is a complex iron-sulfur flavoprotein that catalyzes the conversion of dihydroorotate to orotate and the reduction of NAD(+). The enzyme is a dimer of heterodimers containing an FMN, an FAD, and a 2Fe-2S center. UV-visible, EPR, and ENDOR spectroscopies have been used to determine the reduction potentials of the flavins and the 2Fe-2S center and to characterize radicals and their interactions. Reductive titration using dithionite indicates a five-electron capacity for DHODB. The midpoint reduction potential of the 2Fe-2S center (-212 +/- 3 mV) was determined from analysis of absorption data at 540 nm, where absorption contributions from the two flavins are small. The midpoint reduction potentials of the oxidized/semiquinone (E(1)) and semiquinone/hydroquinone (E(2)) couples for the FMN (E(1) = -301 +/- 6 mV; E(2) = -252 +/- 8 mV) and FAD (E(1) = -312 +/- 6 mV; E(2) = -297 +/- 5 mV) were determined from analysis of spectral changes at 630 nm. Corresponding values for the midpoint reduction potentials for FMN (E(1) = -298 +/- 4 mV; E(2) = -259 +/- 5 mV) in the isolated catalytic subunit (subunit D, which lacks the 2Fe-2S center and FAD) are consistent with the values determined for the FMN couples in DHODB. During reductive titration of DHODB, small amounts of the neutral blue semiquinone are observed at approximately 630 nm, consistent with the measured midpoint reduction potentials of the flavins. An ENDOR spectrum of substrate-reduced DHODB identifies hyperfine couplings to proton nuclei similar to those recorded for the blue semiquinone of free flavins in aqueous solution, thus confirming the presence of this species in DHODB. Spectral features observed during EPR spectroscopy of dithionite-reduced DHODB are consistent with the midpoint reduction potentials determined using UV-visible spectroscopy and further identify an unusual EPR signal with very small rhombic anisotropy and g values of 2.02, 1.99, and 1.96. This unusual signal is assigned to the formation of a spin interacting state between the FMN semiquinone species and the reduced 2Fe-2S center. Reduction of DHODB using an excess of NADH or dihydroorotate produces EPR spectra that are distinct from those produced by dithionite. From potentiometric studies, the reduction of the 2Fe-2S center and the reduction of the FMN occur concomitantly. The study provides a detailed thermodynamic framework for electron transfer in this complex iron-sulfur flavoprotein.     

Inhibition of muscle glycogen phosphorylase b by vitamin B2 and its coenzyme forms

Kinetic studies have demonstrated that vitamin B2 and its coenzyme forms FMN and FAD are potent inhibitors of glycogen phosphorylase b from rabbit skeletal muscle. The inhibition of the enzyme by flavins has a co-operative character (Hill coefficients exceed unity). Glycogen phosphorylase b bound to FMN or FAD does not reveal catalytic activity, whereas the enzyme bound to riboflavin retains about 16% of the initial catalytic activity.     

Entamoeba histolytica: flavins in axenic organisms.

The individual flavin species of axenic Entamoeba histolytica were assayed: separated riboflavin was assayed by the lumiflavin method; flavin-adenine dinucleotide (FAD), by an enzymatic method; flavin mononucleotide (FMN) was calculated from the difference, total flavin minus FAD and riboflavin. The amount of flavin in micrograms per grams fresh cells follows: total flavin, 7.6 ± 0.9 calculated as riboflavin; riboflavin, 1.6 ± 0.7; FMN, 6.6 ± 0.5; and FAD, 1.2 ± 0.1. Recalculated to nanomoles per milligrams total amebal protein these values were: total flavin, 0.21; riboflavin, 0.04; FMN, 0.15; and FAD, 0.02. The identity of each flavin was confirmed by a paper chromatographic method. Analyses on Panmede, the main source of flavins in the TP-S-1 medium, indicate that it contains all three forms of flavin. Its contribution to growth medium in micrograms per milliliters: riboflavin, 2.1 ± 0.3; FMN, 0.6 ± 0.1; and FAD, 0.4 ± 0.1. The in vivo biosynthesis of FMN and FAD from riboflavin by E. histolytica is demonstrated. A new and convenient method was found to separate riboflavin from flavin nucleotides in tissue extracts.     

Singlet oxygen generation by UVA light exposure of endogenous photosensitizers

UVA light (320-400 nm) has been shown to produce deleterious biological effects in tissue due to the generation of singlet oxygen by substances like flavins or urocanic acid. Riboflavin, flavin mononucleotide (FMN), flavin adenine dinucleotide (FAD), beta-nicotinamide adenine dinucleotide (NAD), and beta-nicotinamide adenine dinucleotide phosphate (NADP), urocanic acid, or cholesterol in solution were excited at 355 nm. Singlet oxygen was directly detected by time-resolved measurement of its luminescence at 1270 nm. NAD, NADP, and cholesterol showed no luminescence signal possibly due to the very low absorption coefficient at 355 nm. Singlet oxygen luminescence of urocanic acid was clearly detected but the signal was too weak to quantify a quantum yield. The quantum yield of singlet oxygen was precisely determined for riboflavin (PhiDelta = 0.54 +/- 0.07), FMN (PhiDelta = 0.51 +/- 0.07), and FAD (PhiDelta = 0.07 +/- 0.02). In aerated solution, riboflavin and FMN generate more singlet oxygen than exogenous photosensitizers such as Photofrin, which are applied in photodynamic therapy to kill cancer cells. With decreasing oxygen concentration, the quantum yield of singlet oxygen generation decreased, which must be considered when assessing the role of singlet oxygen at low oxygen concentrations (inside tissue).     

Location of functional centers in the microsomal cytochrome P450 system.

Fluorescence quenching and energy-transfer studies have been carried out to determine the position of FAD and FMN groups of NADPH-cytochrome P450 reductase and of the heme and substrate groups of cytochrome P450 with respect to the lipid/water interphase. Quenching by iodine of the fluorescence of the flavins of the reductase shows a biphasic pattern, due to the different accessibility of FAD and FMN to the solvent with Stern-Volmer constants of 7.9 x 10(-4) and 2.7 x 10(-3) mM-1, respectively. Both prosthetic groups appear to be buried within the three-dimensional structure of the native reductase, FAD more deeply embedded than FMN and with a relative contribution to the total fluorescence of flavins of 84% (FAD) and 16% (FMN). The lack of significant energy transfer (less than 5%) from FAD+FMN to the rhodamine group of the N-labeled phosphatidylethanolamine incorporated in membranes reconstituted with NADPH-cytochrome P450 reductase and phosphatidylcholine points out that both groups are located at a distance greater than 5 nm from the lipid/water interphase. Steady-state fluorescence intensity and anisotropy data obtained with native and FMN-depleted NADPH-cytochrome P450 reductase show that energy transfer between both prosthetic groups occurs in the native reductase with an efficiency of ca. 31%, consistent with a separation between these groups of 2 nm as suggested earlier by Bastiaens, P. I. H., Bonants, P. J. M., Müller, F., & Visser, A. J. W. G. [(1989) Biochemistry 28, 8416-8425] from time-resolved fluorescence anisotropy measurements.(ABSTRACT TRUNCATED AT 250 WORDS)     

High pressure refolding, purification, and crystallization of flavin reductase from Sulfolobus tokodaii strain 7

Flavin reductase HpaC(St) catalyzes the reduction of free flavins using NADH or NADPH. High hydrostatic pressure was used for the solubilization and refolding of HpaC(St), which was expressed as inclusion bodies in Escherichia coli to achieve high yield in a flavin-free form. The refolded HpaC(St) was purified using Ni-affinity chromatography followed by a heat treatment, which gave a single band on SDS-PAGE. The purified refolded HpaC(St) did not contain FMN, unlike the same enzyme expressed as a soluble protein. After the addition of FMN to the protein solution, the refolded enzyme showed a higher activity than the enzyme expressed as the soluble protein. Crystals of the refolded enzyme were obtained by adding FMN, FAD, or riboflavin to the protein solution and without the addition of flavin compound.     

Reduction of exogenous flavins and mobilization of iron from ferritin by isolated mitochondria

The release mechanism for ferritin iron and the nature of the compound(s) which donate iron to the mitochondria are two important problems of intracellular iron metabolism which still await their solution. We have previously shown that isolated mitochondria reduce exogenously added flavins in a ubiquinol-flavin oxidoreductase reaction at the C-side of the inner membrane and that the resulting dihydroflavins function as reductants in mitochondrial mobilization of iron from ferritin (Ulvik, R. J., and Romslo, I. (1981). Biochim. Biophys. Acta 635, 457-469). In the present study it is shown that the rate at which iron is removed from ferritin depends on the capability of the flavins to penetrate (1) the mitochondrial outer membrane and (2) the intersubunit channels of the ferritin protein shell. Intact mitochondria reduce flavins at rates which decrease in the following order: riboflavin > FAD > FMN. The ferritin iron mobilization rates decrease in the order of riboflavin > FMN > FAD. The results are further support for the operation of a flavin-dependent mitochondrial ferrireductase, and strengthen the suggested role for ferritin as a donor of iron to the mitochondria.     

Picomolar analysis of flavins in biological samples by dynamic pH junction-sweeping capillary electrophoresis with laser-induced fluorescence detection

Sensitive capillary electrophoresis (CE) methods are required for emerging areas of biochemical research such as the metabolome. In this report, dynamic pH junction-sweeping CE with laser-induced fluorescence (LIF) detection is applied as a robust single method to analyze trace amounts of three flavin derivatives, riboflavin, flavin mononucleotide (FMN), and flavin adenine dinucleotide (FAD), from several types of samples including bacterial cell extracts, recombinant protein, and biological fluids. Submicromolar amounts of flavin coenzymes were measured directly from formic acid cell extracts of Bacillus subtilis. Significant differences in flavin concentration were measured in cell extracts derived from either glucose or malate as the carbon source in the culture media. Quantitative assessment of FAD and FMN content from selected flavoenzymes was demonstrated after heat denaturation to release noncovalently bound coenzymes and deproteinization. This method was also applied to the analysis of free flavins in pooled human plasma and urine without the need for laborious off-line sample preconcentration. Picomolar detectability of flavins by CE-LIF detection was realized with on-line preconcentration (up to 15% capillary length used for injection) by dynamic pH junction-sweeping, resulting in a limit of detection (S/N = 3) of about 4.0 pM for FAD and FMN. This represents over a 60-fold improvement in concentration sensitivity compared to those of previous techniques using conventional injections. The method was validated in terms of reproducibility, sensitivity, linearity, and specificity. Flavin analysis by dynamic pH junction-sweeping CE-LIF offers a simple, yet sensitive way to analyze trace levels of flavin metabolites from complex biological samples.     

Photoreception inEuglena gracilis: Fluence-rate and time dependence of photoaccumulation and photodispersal

The fluence-rate and time dependence for photoaccumulation and photodispersal ofEuglena gracilis was measured for the wild-type strain and three white mutants. For wavelengths of 453 or 463 nm the threshold for photoaccumulation was close to 6×10⁻²Wm⁻². Photoaccumulation increased steadily with time and reached a maximum after about 4 hr. Red light elicited substantial photoaccumulation in the wild type and photodispersal in the white, non-photosynthetic mutant 1224-5/9f. The chromophore mediating the red-light response needs to be a non-photosynthetic pigment which remains presently unidentified. A whiteEuglena mutant, FB, which had retained a reduced stigma and a paraflagellar body, showed weak photoaccumulation. Two white mutants, 1224-5/1f and 1224-5/9f, both of which lacked the stigma and positive phototaxis, displayed during the first 90 min of irradiation photodispersal; after longer irradiations they showed instead photoaccumulation. These results contradict a widely held belief that the presence of a stigma represents a stringent requirement for photoaccumulation. Our results imply that phototaxis is not a prerequisite for photoaccumulation. Exogenous flavins and 5,10-methenyl-tetrahydrofolate (MTHF) influenced in a wavelength-dependent manner photoaccumulation and photodispersal. In the wild type FAD and riboflavin (RB) caused at 453 nm an increase of the responsiveness for photoaccumulation. The photoaccumulation of the white mutant FB, was sensitized by FMN and FAD. In the white mutant 1224-5/9f exogenous flavins lowered the threshold for photodispersal. FMN, which absorbs only blue light, altered also the responsiveness to red light: in the wild type FMN reduced photoaccumulation and in the white mutant 1224-5/9f it reduced photodispersal.     

Identification and properties of 8-hydroxyflavin--adenine dinucleotide in electron-transferring flavoprotein from Peptostreptococcus elsdenii.

1. A new flavin prosthetic group has been isolated in pure form from the electron-transferring flavoprotein of Peptostreptococcus elsdenni. Its structure has been established as the FAD derivative of 7-methyl-8-hydroxyisoalloxazine: (see article). Proof of this structure has been obtained by chemical syntehsis of 7-methyl-8-hydroxyisoalloxazine models, and by stepwise degradation of the native compound to 7-methy-8-hydroxyalloxazine. The orange chromophore is characterized by a strong absorption band with a maximum at 472 nm (xi = 41 000 M-1 CM-1) and a pK at 4.8 due to the ionisation of the C(8)-OH group. 2. The properties of a series of functionally substituted derivatives of 8-hydroxy flavins and lumichromes have been investigated to provide a basis for interpreting the effects of pH on the spectroscopic properties of the 8-hydroxy derivatives of FAD and FMN. 3. The 8-hydroxy derivative of FAD is bound by apo-D-amino acid oxidase; the complex shows no catalytic activity. The 8-hydroxy derivative of FMN is bound by apoflavodoxin to give a complex which has catalytic activity similar to that of native flavodoxin. The complex is reversibly reduced by dithionite, first to a relatively stable semiquinone and further to the dihydroflavin form.     

Blue-light reactivation of spinach nitrate reductase inactivated by acetylene or cyanide. Effects of flavins and oxygen

Nitrate reductase (NADH: nitrate oxidoreductase, EC 1.6.6.1) of spinach ( Spinacia oleracea L.) leaves, inactivated in vitro by acetylene, was reactivated by irradiation with blue light. Red + infrared, green or white light of the same irradiance were less effective. The dehydrogenase activity of the nitrate reductase complex was not required for pliotoreactivation. Photoreactivation of cyanide-inactivated nitrate reductase was greatly enhanced by the addition of 1 and 20 μ of either FMN or FAD; however, flavins showed a much smaller stimulatory effect on photoreactivation of acetylene-inactivated enzyme. The effect of flavins was higher under anaerobic conditions. This might imply the direct ievolvement of excited flavins in the photoreactivation mechanism. Besides promoting photoreactivation, blue light irradiation led simultaneously to a gradual inactivation of the enzyme especially under air and 20 μ FMN, eventually abolishing the recovered activity of the enzyme.     

Chimeric enzymes of cytochrome P450 oxidoreductase and neuronal nitric-oxide synthase reductase domain reveal structural and functional differences

The nitric-oxide synthases (NOSs) are comprised of an oxygenase domain and a reductase domain bisected by a calmodulin (CaM) binding region. The NOS reductase domains share approximately 60% sequence similarity with the cytochrome P450 oxidoreductase (CYPOR), which transfers electrons to microsomal cytochromes P450. The crystal structure of the neuronal NOS (nNOS) connecting/FAD binding subdomains reveals that the structure of the nNOS-connecting subdomain diverges from that of CYPOR, implying different alignments of the flavins in the two enzymes. We created a series of chimeric enzymes between nNOS and CYPOR in which the FMN binding and the connecting/FAD binding subdomains are swapped. A chimera consisting of the nNOS heme domain and FMN binding subdomain and the CYPOR FAD binding subdomain catalyzed significantly increased rates of cytochrome c reduction in the absence of CaM and of NO synthesis in its presence. Cytochrome c reduction by this chimera was inhibited by CaM. Other chimeras consisting of the nNOS heme domain, the CYPOR FMN binding subdomain, and the nNOS FAD binding subdomain with or without the tail region also catalyzed cytochrome c reduction, were not modulated by CaM, and could not transfer electrons into the heme domain. A chimera consisting of the heme domain of nNOS and the reductase domain of CYPOR reduced cytochrome c and ferricyanide at rates 2-fold higher than that of native CYPOR, suggesting that the presence of the heme domain affected electron transfer through the reductase domain. These data demonstrate that the FMN subdomain of CYPOR cannot effectively substitute for that of nNOS, whereas the FAD subdomains are interchangeable. The differences among these chimeras most likely result from alterations in the alignment of the flavins within each enzyme construct.     

Modulation of rat polymorphonuclear leukocyte 5-lipoxygenase activity by 5-HPETE and NADH-dependent flavin inhibition

The effect of nicotinamide and flavin coenzymes on the 5-lipoxygenase activity has been determined in cell-free extracts from rat polymorphonuclear leukocytes. 5-lipoxygenase was assayed in the presence of 5-hydroperoxyeicosatetraenoic acid (5-HPETE), which caused a 3 to 4-fold stimulation in the maximal conversion of radiolabeled arachidonic acid to 5-hydroxyeicosatetraenoic acid (5-HETE) and 5,12-dihydroxyeicosatetraenoic acid (5,12-di-HETE). Addition of FMN or FAD to the assay mixture had little effect on the 5-lipoxygenase activity and caused inhibition only at high concentrations (IC50 greater than 100 microM). NADH markedly potentiated the inhibition of lipoxygenase by flavins with a 100-fold decrease in the FMN concentration required to inhibit the enzyme (IC50 approximately equal to 2 microM). Similar effects were observed for FAD although this flavin derivative was slightly less potent than FMN (IC50 congruent to 10 microM). NADH could be substituted by NADPH but not by NAD or NADP, indicating that the inhibition was not due to the production of the oxidized forms of these co-factors. These results show that the 5-lipoxygenase activity is stimulated by 5-HPETE and inhibited by flavin-dependent redox transformations.     

Nitric oxide synthase regulatory sites. Phosphorylation by cyclic AMP-dependent protein kinase, protein kinase C, and calcium/calmodulin protein kinase; identification of flavin and calmodulin binding sites.

Nitric oxide (NO) is an important molecular messenger accounting for endothelial-derived relaxing activity in blood vessels, mediating cytotoxic actions of macrophages, and functioning as a neurotransmitter in the brain and periphery. NO synthase (NOS) from brain has been purified to homogeneity and molecularly cloned. We now report that NOS is stoichiometrically phosphorylated by cAMP dependent protein kinase, protein kinase C, and calcium/calmodulin-dependent protein kinase, with each kinase phosphorylating a different serine site on NOS. Activation of PKC in transfected cells reduces NOS enzyme activity by approximately 77% in intact cells and by 50% in protein homogenates from these cells. Utilizing fluorescence spectroscopy we find that purified monomer NOS contains 1 molar equivalent of both FMN and FAD. This stoichiometry is supported by enzymatic digestion of the flavins with phosphodiesterase, and titration of the FMN with a specific FMN binding protein. We demonstrate that purified NOS is labeled by a photoaffinity derivative of calmodulin. These recognition sites on NOS provide multiple means for regulation of NO levels and "cross-talk" between second messenger systems.     

Analytical and preparative high-performance liquid chromatography separation of flavin and flavin analog coenzymes

Reversed-phase high-performance liquid chromatography has been used to separate a number of flavin and flavin analogs at the riboflavin, FMN, and FAD coenzyme level. Analytical methods were developed which enable the facile determination of a particular flavin or mixture of flavins present. These methods also allowed the separation of oxidized from reduced forms of oxygen-stable flavin analogs. Past investigations have utilized enzymatically synthesized FAD analogs with the problem of potential contamination by other levels of the coenzyme or ATP a cosubstrate in the flavokinase/FAD synthetase reaction. Preparative methods show that all the potential reaction products may be separated from one another thereby allowing the rapid purification of these redox coenzyme analogs. To demonstrate the utility of this method, radiolabeled FAD and 1-deazaFAD were prepared and purified.