Role of reactive oxygen species (ROS) in the diabetes-induced anomalies in rat embryos in vitro: reduction in antioxidant enzymes and low-molecular-weight antioxidants (LMWA) may be the causative factor for increased anomalies

A disturbed embryonic antioxidant defense mechanism may play a major role in diabetes-induced teratogenesis. We therefore studied the antioxidant capacity of 10.5-day-old rat embryos and their yolk sacs after culture for 28 hr in vitro under diabetic conditions (3 mg/ml glucose, 2 mg/ml beta-hydroxybutyrate (BHOB) and 10 microg/ml of acetoacetate), as compared with control embryos in vitro. We found a high rate of congenital anomalies, decreased growth and protein content, and a decrease in the activity of both superoxide dismutase (SOD) and catalase (CAT) under diabetic conditions, as compared with controls. The reducing power, which reflects the concentration and type of water-soluble and of lipid-soluble low-molecular-weight antioxidants (LMWA), was measured by cyclic voltammetry. Generally, LMWA were reduced in the embryos and yolk sacs under diabetic conditions. In the water-soluble fraction of control embryos and yolk sacs, two peak potentials were found, indicating two major groups of LMWA, while only one peak potential was found under diabetic conditions, indicating that an entire group of LMWA is missing. HPLC studies have demonstrated a decrease in vitamin C (water-soluble fraction) and in vitamin E (lipid-soluble fraction) under diabetic culture conditions, and an increase in uric acid. Generally, the concentration of LMWA was higher in the embryos than in the yolk sac. LMWA concentration, protein content, and antioxidant enzyme activity were lower in the malformed experimental embryos than in experimental embryos without anomalies. The addition of vitamins C and E to the diabetic culture medium abolished the deleterious effects of the diabetic serum on the embryos. The disturbed antioxidant defense mechanism under diabetic conditions may be explained, at least in part, by a direct effect of diabetic metabolic factors on the activity of antioxidant enzymes and on the concentration of reducing equivalents. This, in turn, may be embryotoxic.

Is the biological antioxidant system integrated and regulated?

It is proposed that there is biological regulation of the low molecular weight antioxidant (LMWA) status of the body, and that these different antioxidants work in concert and in homeostasis as a system analogous to the regulation of acid-base status (pH). The theoretical design characteristics for such a system include regulation, inducibility, interactivity and balance with the beneficial properties of reactive oxygen species. Testing the hypothesis requires developing suitable methodologies (such as measurement of the redox state) for assessing the total ratio of antioxidant to oxidant activity of both tissues and biological fluids, since it is not clear whether plasma antioxidant status reflects that of the tissues as well. This concept, if accepted, may help explain the contradictions relating to antioxidant therapy and lead to more rational recommendations for dietary intake. It may also help explain the effects of manipulating (increasing/decreasing) specific antioxidants on the overall antioxidant status in health and disease.

Hemolysis of human red blood cells induced by the combination of diethyldithiocarbamate (DDC) and divalent metals: modulation by anaerobiosis, certain antioxidants and oxidants

The objective of the present communication is to describe the role played by combinations between diethydithiocarbamate (DDC) and divalent metals in hemolysis of human RBC. RBC which had been treated with DDC (10-50 microM) were moderately hemolyzed (about 50%) upon the addition of subtoxic amounts of Cu2+ (50 microM). However, a much stronger and a faster hemolysis occurred either if mixtures of RBC-DDC were immediately treated either by Co2+ (50 microM) or by a premixture of Cu2+ and Co2+ (Cu:Co) (50 microM). While Fe2+ and Ni2+, at 50 microM, initiated 30-50% hemolysis when combined with DDC (50 microM), on a molar basis, Cd2+ was at least 50 fold more efficient than any of the other metals in the initiation of hemolysis by DDC. On the other hand, neither Mn2+ nor Zn2+, had any hemolysis-initiating effects. Co2+ was the only metal which totally blocked hemolysis if added to DDC prior to the addition of the other metals. Hemolysis by mixtures of DDC + (Cu:Co) was strongly inhibited by anaerobiosis (flushing with nitrogen gas), by the reducing agents glutathione, N-acetyl cysteine, mercaptosuccinate, ascorbate, TEMPO, and alpha-tocopherol, by the PLA2 inhibitorbromophenacylbromide (BrPACBr), by tetracycline as well as by phosphatidyl choline, cholesterol and by trypan blue. However, TEMPO, BrPACBr and PC were the only agents which inhibited hemolysis induced by DDC: Cd2+ complexes. On the other hand, none of the classical scavengers of reactive oxygen species (ROS) employed e.g dimethylthiourea, catalase, histidine, mannitol, sodium benzoate, nor the metal chelators desferal and phenanthroline, had any appreciable inhibitory effects on hemolysis induced by DDC + (Cu:Co). DDC oxidized by H2O2 lost its capacity to act in concert either with Cu2+ or with Cd2+ to hemolyze RBC. While either heating RBC to temperatures greater than 37 degrees C or exposure of the cells to glucose-oxidase-generated peroxide diminished their susceptibility to hemolysis, exposure to the peroxyl radical from AAPH, enhanced hemolysis by DDC + (Cu:Co). The cyclovoltammetry patterns of DDC were drastically changed either by Cu2+, Co2+ or by Cd2+ suggesting a strong interaction of the metals with DDC. Also, while the absorbance spectrum of DDC at 280 nm was decreased by 50% either by Co2+, Cd2+ or by H2O2, a 90% reduction in absorbance occurred if DDC + H2O2 mixtures were treated either by Cu2+ or by Co2+, but not by Cd2+. Taken together, it is suggested that DDC-metal chelates can induce hemolysis by affecting the stability and the integrity of the RBC membrane, and possibly also of the cytoskeleton and the role played by reducing agents as inhibitors might be related to their ability to deplete oxygen which is also supported by the inhibitory effects of anaeobiosis.

Skin antioxidants: their role in aging and in oxidative stress--new approaches for their evaluation

Skin is a highly metabolic tissue which possesses the largest surface area in the body and serves as the protective layer for internal organs [1]. Skin is also a major candidate and target of oxidative stress. It is designed to give both physical and biochemical protection, and is equipped with a large number of defense mechanisms. The skin tissue is exposed to a variety of damaging species which originate in the outer environment, in the skin itself, and in various endogenous sources [2, 3]. The structure of skin is quite complex being composed of several layers, each of which plays a specific role and carries out different functions [4]. Each layer is equipped with its own arsenal of defense molecules, and the various systems differ from each other based on the layer's susceptibility to oxidative stress and its function. It is generally agreed that one of the major and important contributions to skin aging, skin disorders and skin diseases results from reactive oxygen species (ROS) [1, 5]. Due to the high occurrence of potential biological targets for oxidative damage, skin is very susceptible to such reactions. For example, skin is rich in lipids, proteins, and DNA, all of which are extremely sensitive to the oxidation process [6-8]. Elucidation of the mechanisms involved in skin oxidation and the examination of the defense systems may contribute to the understanding of skin aging and of the mechanisms involved in the various pathological processes of skin. This review addresses the antioxidant defense mechanism of the skin, the role it plays during the aging process, and the role skin has following exposure to oxidative stresses.

Closed head injury in the rat induces whole body oxidative stress: overall reducing antioxidant profile

Traumatic injury to the brain triggers the accumulation of harmful mediators, including highly toxic reactive oxygen species (ROS). Endogenous defense mechanism against ROS is provided by low molecular weight antioxidants (LMWA), reflected in the reducing power of the tissue, which can be measured by cyclic voltammetry (CV). CV records biological peak potential (type of scavenger), and anodic current intensity (scavenger concentration). The effect of closed head injury (CHI) on the reducing power of various organs was studied. Water and lipid soluble extracts were prepared from the brain, heart, lung, kidney, intestine, skin, and liver of control and traumatized rats (1 and 24 h after injury) and total LMWA was determined. Ascorbic acid, uric acid, alpha-tocopherol, carotene and ubiquinol-10 were also identified by HPLC. The dynamic changes in LMWA levels indicate that the whole body responds to CHI. For example, transient reduction in LMWA (p<0.01) in the heart, kidney, lung and liver at 1 h suggests their consumption, probably due to interaction with locally produced ROS. However, in some tissues (e.g., skin) there was an increase (p<0.01), arguing for recruitment of higher than normal levels of LMWA to neutralize the ROS. alpha-Tocopherol levels in the brain, liver, lung, skin, and kidney were significantly reduced (p<0.01) even up to 24 h. We conclude that although the injury was delivered over the left cerebral hemisphere, the whole body appeared to be under oxidative stress, within 24 h after brain injury.

Differences in the reducing power along the rat GI tract: lower antioxidant capacity of the colon

The ability of the gastrointestinal (GI) tract, as well as other tissues, to cope with reactive oxygen species (ROS) efflux in pathological events is determined partly by epithelial antioxidant levels. These levels are comprised of tissue antioxidant enzymes and low molecular weight antioxidants (LMWA). While glutathione levels and the activity of enzymatic antioxidants along the GI tract have been studied, the contribution of the overall LMWA to the total antioxidant capacity has not yet been determined. In this study the overall antioxidant activity in the mucosa/submucosa and muscularis/serosa of various sections along the small intestine and colon of the rat was evaluated by determining the reducing power, which reflects the total antioxidant activity derived from LMWA, using cyclic voltammetry. The activity of the antioxidant enzymes superoxide dismutase (SOD) and catalase was also measured. The reducing power (total antioxidant activity) was higher in the mucosa/submucosa of the small intestine as compared to the mucosa/submucosa of the colon. Similarly, catalase and SOD activity in the mucosa/submucosa of the small intestine was significantly higher than in the mucosa/submucosa of the colon. Differences were also observed in the reducing power and SOD activity in the muscularis/serosa of the rat small intestine as compared to the colon. The low antioxidant capacity in the colon may facilitate reactive oxygen species (ROS)-mediated injury and lead to inflammatory diseases such as ulcerative colitis, specifically in the colon.

Antioxidant mechanisms in apolipoprotein E deficient mice prior to and following closed head injury

Apolipoprotein E deficient mice have distinct memory deficits and neurochemical derangements and their recovery from closed head injury is impaired. In the present study, we examined the possibility that the neuronal derangements of apolipoprotein E deficient mice are associated with oxidative stress, which in turn affects their ability to recover from close head injury. It was found that brain phospholipid levels in apolipoprotein E deficient mice are lower than those of the controls (55+/-15% of control, P<0. 01), that the cholesterol levels of the two mice groups are similar and that the levels of conjugated dienes of the apolipoprotein E deficient mice are higher than those of control mice (132+/-15% of P<0.01). Brains of apolipoprotein E deficient mice had higher Mn-superoxide dismutase (134+/-7%), catalase (122+/-8%) and glutathione reductase (167+/-7%) activities than control (P<0.01), whereas glutathione peroxidase activity and the levels of reduced glutathione and ascorbic acid were similar in the two mouse groups. Closed head injury increased catalase and glutathione peroxidase activities in both mouse groups, whereas glutathione reductase increased only in control mice. The superoxide dismutase activity was unaffected in both groups. These findings suggest that the antioxidative metabolism of apolipoprotein E deficient mice is altered both prior to and following head injury and that antioxidative mechanisms may play a role in mediating the neuronal maintenance and repair derangements of the apolipoprotein E deficient mice.

Noninvasive in vivo evaluation of skin antioxidant activity and oxidation status

The method described here allows noninvasive quantification of reducing LMWA or the lipid hydroperoxide present on the surface of the skin. Quantification of reducing antioxidants can be achieved because they are secreted from the skin surface into a well containing an extraction solution. Analysis of the reducing equivalents released indicates the presence of uric acid and ascorbic acid. Other LMWA released from the skin are as yet unidentified. The secretion of the LMWA reaches a plateau following 20-30 min of incubation. Therefore, a 30-min incubation period was chosen as the optimal time for the extraction solution to be present in the well and in contact with the skin. This extraction procedure can be repeated after 24 hr. This period of time is needed for regeneration of the LMWA to their initial levels. Direct measurement allows continuous determination of the release of LMWA and their interaction with the iron chelate. The reaction is completed after 25-35 min, at which time the final potential can be recorded. When organic peroxides on the surface of the skin are determined, it is important that the glassy carbon electrode be in close contact with the skin, since the reaction occurs on the surface of the electrode and the bound peroxide on the outer layer of the skin. Furthermore, close contact is needed to avoid interference of reducing equivalents secreted from the skin into the well.

Low-density lipoprotein oxidation and its prevention by amidothionophosphate antioxidants

Amidothionophosphates (AMTPs) are a novel group of antioxidants that are lacking in pro-oxidant activity. In this paper, we compare two different amidothionophosphates: 2-hydroxy-ethyl amido, diethyl thionophosphate (AMTP-B), which contains a single primary amido group, and N,N',N-tripropylamidothionophosphate (AMTP-3A), which contains three primary amido groups. The lipoprotein/medium partition coefficients of AMTP-3A and AMTP-B are 74 and 38, respectively. Both protected isolated human low density lipoprotein (LDL) against oxidative damage induced by copper sulfate. Oxidative damage to polyunsaturated acyl chains was determined by gas chromatography (GC), and oxidation kinetics were monitored by following the accumulation of conjugated dienes spectrophotometrically at 234 nm. The AMTP antioxidants significantly protected the LDL against Cu2+-induced oxidation. However, if the LDLs were already partially oxidized, protection against oxidation by the AMTPs was reduced. AMTP-3A was more effective in protecting LDL than was AMTP-B. The difference in antioxidant activity was attributed to the 15-fold higher reactivity of AMTP-3A toward peroxides. Oxidizability of plasma lipoproteins from guinea pigs injected with AMTPs was strongly reduced.

Gs protein-coupled serotonin receptors: receptor isoforms and functional differences

Three distinct mammalian Gs coupled serotonin receptor genes have been identified, 5-HT4, 5-ht6, and 5-HT7, which produce at least seven different functional receptors through alternative splicing. One of the chief questions facing workers in this area mirrors that confronting the serotonin receptor field as a whole: why so many subtypes? The answer to this question is made more elusive at present by two further considerations. First, there may well be additional Gs coupled receptor subtypes yet to be described. Secondly, although the various isoforms of 5-HT4 and 5-HT7 have been shown to be functional in in vitro assays, it remains to be shown that all isoforms have biological significance. This paper will summarize some of the differences at the molecular and cellular level that are becoming apparent among the 5-HT4, 5-ht6 and 5-HT7 receptor subtypes and their various isoforms. As an example, it will focus on the 5-HT7 system, and describe recent developments in ascribing particular functions to differences due to alternative splicing.

Function and distribution of three rat 5-hydroxytryptamine7 (5-HT7) receptor isoforms produced by alternative splicing

Serotonin (5-HT7) receptor pre-mRNA is alternatively spliced in rat tissue to produce three isoforms, 5-HT(7a), 5-HT(7b) and 5-HT(7c), which differ in the amino acid sequences of their carboxyl terminal tails. Substantial species differences in structure and expression patterns exist for 5-HT7 isoforms. We have now compared some of the functional characteristics and level of expression for the three rat 5-HT7 receptor isoforms. Recombinant receptor isoforms were expressed in COS-7 cells for examination of [3H]5-HT binding characteristics and in JEG-3 cells to ascertain their ability to stimulate cAMP production. These studies showed that all three isoforms are functionally active and have similar agonist binding characteristics. Distribution of expression of the three rat receptor isoforms were examined in several brain regions and peripheral tissues using RT-PCR and in situ hybridization. The relative proportions of total 5-HT7 receptor message lent by each isoform varied little between these areas. In contrast to what has been observed in human tissue, the 5-HT(7a) isoform predominated in all regions examined, while the 5-HT(7c) isoform revealed a low level of expression (3% of total transcript). In situ hybridization was used to determine if the overall low level of expression of the 5-HT(7c) isoform by RT-PCR could be attributed to a small localized subpopulation of cells expressing high levels 5-HT(7c) message. In situ hybridization results indicate a generalized low level of expression of the 5-HT(7c) isoform throughout the CNS. These data suggest that while all three known 5-HT7 receptor isoforms in the rat are functionally competent, any functionally important differences between the three isoforms are not likely to involve differences in ligand binding or gross differences in adenylate cyclase coupling. However, differences in receptor phosphorylation, regulation or coupling to other effectors or cell trafficking could still exist.

Neuroprotection against oxidative stress by serum from heat acclimated rats

Exposure of PC12 cells, to 1% serum derived from normothermic (CON) rats resulted in 79% cell death. Sister cultures treated with 1% serum derived from heat acclimated (ACC) rats, were neuroprotected and expressed a significant reduction in cell death. In PC12 cells exposed to a free radical generator causing an oxidative stress, 90% cell death was measured in CON serum treated cultures, while ACC serum treated cultures were neuroprotected. Xanthine oxidase activity and uric acid (UA) levels were lower in ACC serum compared to CON. Addition of UA to both sera abolished the difference in cell viability, and toxicity of ACC serum reached that of CON. These findings suggest a causal relationship between the lower levels of UA in ACC and the neuroprotective effect observed. The present study proposes heat acclimation as an experimental and/or clinical tool for the achievement of neuroprotection.

Skin surface proteolytic activity. Partial characterization and identification

Skin surface proteolytic activity in the living animal was determined by a sensitive, non-invasive methodology developed in our laboratory. A non-leaky well was constructed on the shaved back of an anesthetized guinea pig. The well contained the reaction mixture including the substrate 125I-S-carboxymethylated insulin B-chain (ICMI). The proteolytic activity was shown to be time-dependent. The activity was strongly inhibited by pepstatin A, indicating the involvement of aspartic proteinase(s) such as cathepsin D and/or E. Pretreatment of the skin with propylene glycol blocked the proteolytic activity. The present study demonstrates the presence of proteolytic activity located on skin surface using a unique, non-invasive method for in situ proteinase determination in the living animal.

Noninvasive procedure for in situ determination of skin surface aspartic proteinase activity in animals; implications for human skin

The present study demonstrated a noninvasive procedure for in situ determination of stratum corneum aspartic proteinase in the living animal. A non-leaky well, containing [125I]S-carboxymethylated insulin B-chain (ICMI) as a substrate, was constructed on the shaved back of anesthetized guinea pigs and rats. The enzymatic activity was determined by measuring the radiolabeled trichloroacetic acid soluble material. We demonstrated pepstatin-sensitive proteinase activity bound to the skin surface indicating the involvement of aspartic proteinase(s) such as cathepsin D and/or E. Aged rats had about six fold lower activity than young animals. The proteinase activity was inhibited by the alkylating agent mechlorethamine and by the cosmetic propylene glycol. A similar procedure was carried out with intact human skin pieces obtained during plastic surgery. The activity was inhibited by antihuman cathepsin D antibodies. Cathepsin D was immunohistochemically localized in the corneal and granular layers of the epidermis. Skin surface aspartic proteinase/cathepsin D activity may serve as a marker for skin aging or for certain skin disorders leading to a new approach to their medical treatments.

Oxidative stress in closed-head injury: brain antioxidant capacity as an indicator of functional outcome

It has been suggested that reactive oxygen species (ROS) play a role in the pathophysiology of brain damage. A number of therapeutic approaches, based on scavenging these radicals, have been attempted both in experimental models and in the clinical setting. In an experimental rat and mouse model of closed-head injury (CHI), we have studied the total tissue nonenzymatic antioxidant capacity to combat ROS. A major mechanism for neutralizing ROS uses endogenous low-molecular weight antioxidants (LMWA). This review deals with the source and nature of ROS in the brain, along with the endogenous defense mechanisms that fight ROS. Special emphasis is placed on LMWA such as ascorbate, urate, tocopherol, lipoic acid, and histidine-related compounds. A novel electrochemical method, using cyclic voltammetry for the determination of total tissue LMWA, is described. The temporal changes in brain LMWA after CHI, as part of the response of the tissue to high ROS levels, and the correlation between the ability of the brain to elevate LMWA and clinical outcome are addressed. We relate to the beneficial effects observed in heat-acclimated rats and the detrimental effects of injury found in apolipoprotein E-deficient mice. Finally, we summarize the effects of cerebroprotective pharmacological agents including the iron chelator desferal, superoxide dismutase, a stable radical from the nitroxide family, and HU-211, a nonpsychotoropic cannabinoid with antioxidant properties. We conclude that ROS play a key role in the pathophysiology of brain injury, and that their neutralization by endogenous or exogenous antioxidants has a protective effect. It is suggested, therefore, that the brain responds to ROS by increasing LMWA, and that the degree of this response is correlated with clinical recovery. The greater the response, the more favorable the outcome.

Oxidative stress effect on the integrity of lipid bilayers is modulated by cholesterol level of bilayers

Large unilamellar vesicles (120-160 nm) composed of egg phosphatidylcholine (egg PC) containing approximately 22 wt% of polyunsaturated fatty acids (PUFA) and various mol% (0, 10, 22, or 45) of cholesterol were exposed to oxidative stress. The hydrophilic azo compound 2,2'-azobis-(2-amidinopropane)2HCl (AAPH) which was thermally decomposed to produce a constant flux of peroxy radicals was the source of the oxidative stress (< or = 48 h incubation at 37 degrees C). Cholesterol loss following the oxidation was up to 33%, while PUFA were more extensively damaged; loss was up to 52, 88, and 100% for C-18:2, C-20:4, and C-22:6, respectively. (ii) Oxidizability of cholesterol when quantified in absolute amount was three-fold higher when its level was 45 mol%. The interrelationship between bilayer structure, especially its lateral organization and free volume, and lipid peroxidation are discussed. Differential scanning calorimetry of oxidized multilamellar vesicles lacking cholesterol revealed that a high level of oxidative damage to egg phosphatidylcholine PUFA resulted in the loss of the gel to liquid-crystalline phase transition of egg PC (broad peak at around -8 degrees C).

Four 5-hydroxytryptamine7 (5-HT7) receptor isoforms in human and rat produced by alternative splicing: species differences due to altered intron-exon organization

The serotonin (5-HT) 5-HT7 receptor subtype is thought to mediate a number of physiological effects in mammalian brain and periphery. Previous studies suggested that alternative splicing might contribute to 5-HT7 receptor diversity as well. We now report that alternative splicing in human and rat tissues produces four 5-HT7 receptor isoforms that differ in their predicted C-terminal intracellular tails. Human and rat partial 5-HT7 cDNAs and intronic sequences were identified and compared. In rat tissues, three 5-HT7 isoforms, here called 5-HT7(a), 5-HT7(b), and 5-HT7(c), are found. Rat 5-HT7(a) [448-amino acid (aa)] and 5-HT7(b) (435-aa) forms arise from alternative splice donor sites. A third new isoform found in rat, 5-HT7(c) (470-aa), results from a retained exon cassette. Three 5-HT7 mRNA isoforms were also identified in human tissues, where only one isoform was previously described. Two human isoforms represent 5-HT7(a) and 5-HT7(b) forms (445- and 432-aa), but the third form does not correspond to 5-HT7(c). Instead, it constitutes a distinct isoform, 5-HT7(d) (479-aa), resulting from retention of a separate exon cassette. 5-HT7(d) transcripts are not present in rat because the 5-HT7(d)-specifying exon is absent from the rat 5-HT7 gene. A frame-shifting homologue of the rat 5-HT7(c)-Specifying exon is present in the human gene but is not used in the human tissues examined. Tissue-specific splicing differences are present in human between brain and spleen. These studies suggest that alternative splicing may contribute to diversity of 5-HT7 receptor action and that the human and rat repertoires of 5-HT7 splice variants are substantially different.

Reduced levels of antioxidants in brains of apolipoprotein E-deficient mice following closed head injury

Recent animal model studies using apolipoprotein E (apoE)-deficient (knockout) mice revealed that these mice have memory deficits and neurochemical derangements and that they recover from closed head injury less adequately than control mice. In the present study, we examined the possibility that the diminished recovery of apoE-deficient mice from head injury is related to a reduction in their ability to counteract oxidative damage. Measurements of reducing agents by cyclic voltammetry revealed that cortical homogenates of apoE-deficient and control mice contain similar levels of these compounds whose oxidation potentials for the two groups of mice are at 400 +/- 40 mV and 900 +/- 50 mV. The responses of the apoE-deficient and control groups to closed head injury were both biphasic and were composed of initial reductions followed by subsequent increases in the levels of reducing antioxidant equivalents. However, the two groups differed markedly in the magnitude of their response. This difference was most pronounced with the 400-mV reducing compounds, such that at 4 h after injury their levels in injured control mice increased over twofold relative to the noninjured control mice, whereas the corresponding anodic current of the apoE-deficient mice recovered only to its original level and did not increase further even by 24 h after injury. In vitro studies using recombinant apoE allele E3 and beta very low density lipoprotein revealed that this lipoprotein can delay Cu(2+)-induced lipid peroxidation. This suggests that the inability of the apoE-deficient mice to respond to brain injury by a surge in brain reducing compounds may be related, at least in part to direct antioxidant activity of apoE.