Abstract
Oxidative stress, an overproduction of free radicals or a diminution of free radical scavenging ability relative to those of cognitively aged-matched controls, is widely recognized as a critical component of the pathogenesis and progression of Alzheimer’s disease (AD). This recognition arose in significant part from the work in the author’s laboratory, complemented by research from others’ laboratories. The Butterfield laboratory discovered the oxidative stress associated with oligomeric amyloid-β peptide manifested primarily as elevated oxidative modification of proteins and peroxidation of lipids. Such oxidative damage caused neuronal death, which undoubtedly underlies the progressive loss of cognition in AD. Identification of specific oxidatively modified brain proteins in subjects with AD or amnestic mild cognitive impairment was achieved by the methods of redox proteomics, pioneered in the author’s laboratory. The importance and significance of the research emanating from the Butterfield laboratory rest on the paradigm shift of thinking regarding the roles of oxidative stress and resulting damage to key proteins and biochemical pathways in the pathogenesis and progression of AD. Predictions of future research directions also are presented. Given the enormous financial and personal burden placed upon citizens (and governments) of the US from AD, and the surety that the number of AD patients will greatly increase over the next 20–30 years, greater understanding of the molecular basis of pathogenesis and progression of AD is essential. Our laboratory is privileged to have contributed to better understanding of AD and provided rationales to identify effective therapeutic targets for this devastating dementing disorder.
Keywords
PERSPECTIVES ON THE IMPLICATIONS AND IMPORTANCE OF ALZHEIMER’S DISEASE RESEARCH PUBLISHED FROM OUR LABORATORY
The Butterfield laboratory is probably best known in the field of Alzheimer’s disease (AD) research for two major discoveries: 1) The role of oxidative and nitrosative stress in brain and resulting paradigm shift in thinking about in the pathogenesis and progression of AD [1–5]; and 2) The pioneering of redox proteomics methods with which identification of oxidatively and nitrosatively modified, and consequently dysfunctional, brain proteins was achieved in specimens from subjects with AD and amnestic mild cognitive impairment (MCI) and the resulting new insights gained into key molecular pathways affected in these disorders [6–9].
Oxidative and nitrosative stress
Oxidative stress results when the production of oxygen-containing free radicals or molecules from which free radicals could be formed exceeds the rate of scavenging of such moieties by antioxidant enzymes or endogenous small molecule antioxidants [10, 11]. Oxidative stress is indexed in protein oxidation by formation of protein carbonyls and in lipid peroxidation, by among other molecules, 4-hydroxynonenal (HNE) [8, 11]. By Michael addition, a covalent bond is formed when HNE binds to Lys, His, and Cys amino acids in proteins, changing their conformation and function [12]. Nitrosative stress involves actions of oxygen-containing moieties that also contain nitrogen. The most important nitrogen source in this context is nitric oxide (NO), a free radical formed from the action of nitric oxide synthase. Because radical-radical recombination reactions are among the fastest known chemical reactions, NO reacts with incredibly fast kinetics with superoxide free radical anion to form peroxynitrite (ONOO-), a non-radical moiety [13]. This agent in the presence of carbon dioxide leads to nitrogen dioxide (NO2) [13, 14], a free radical that, because the -OH group on an aromatic ring is ortho/para-directing, binds to the 3-position of tyrosine to form 3-nitrotyrosine (3-NT) [10, 14]. Protein carbonyl- and 3-NT covalent modifications on proteins are formal oxidations from a chemical point-of-view. Therefore, from henceforward in this paper, I will use the term “protein oxidation” in discussing both of these covalent modifications of proteins. As noted above, HNE formation is a reactive product of and marker for lipid peroxidation, that when covalently bound to proteins changes their structure and modifies their function [5, 15].
Oxidative stress in AD and MCI brain
Prior to the discovery from our laboratory and that of others that amyloid-β peptide (Aβ) oligomers led to oxidative modification of neuronal proteins and lipids [16–25], it was difficult to rationalize why there were so many reports of altered proteins, enzymes, lipids, and biochemical pathways in AD brain. Why was it not the case that only one major change occurs in AD brain that could account for the progressive pathology of AD and increasing cognitive decline throughout the stages of this dementing disorder?
The discovery of Aβ42-mediated oxidative damage in neuronal cultures, synaptosomes, and in brains from both animal models of AD and subjects with AD and MCI [5, 19–33] opened the possibility that damage following oxidative stress could help explain the following observation: wherever in the brain Aβ42 was abundant, oxidative stress occurred, and in contrast, wherever Aβ42 levels were absent or low (i.e., cerebellum), excess oxidative stress in AD and MCI over that of aged-matched control cerebellum did not occur [28]. Different proteins and lipids, modified by the actions of Aβ42 or from free radicals from other sources, have decreased function [34, 35]. Such a notion in AD, first proffered from the Butterfield laboratory, was a paradigm shift in thinking about the pathogenesis and progression of AD; that is, oxidative stress is a key factor in the pathogenesis and progression of AD [36]. This “radical” idea (please pardon the pun) is now generally accepted dogma about the pathogenesis and progression of AD [37–45].
However, a troubling question about the importance of oxidative stress in AD is: since protein oxidation and lipid peroxidation decreased protein activity after covalent modification or led to their decreased abundance, why have clinical trials employing antioxidants been such failures in AD [46, 47]? Several reasons may address this question. 1) Such clinical trials often occurred late in the disease when neuronal loss is rampant and therefore AD would not be susceptible to potentially positive interventions. Given that AD neuropathology is thought to occur approximately 20 years prior to the onset of clinical symptoms of this disorder [48], the ideal time for antioxidant therapy likely would be at the start or prior to initiation of AD neuropathology. Such a time frame would require some reliable set of biomarkers unique to AD in order to know who should be treated. This notion is discussed further below in the Future Research Directions section of this current paper. 2) Antioxidants in many clinical trials may not have been applied in the most effective manner or had poor penetration of the blood-brain barrier (BBB). For example, vitamin E, which traverses the BBB via a specific transporter [49], requires vitamin C or other agent to reduce the oxidized vitamin E back to vitamin E, i.e., so vitamin E can act as a continuous scavenger of free radicals and not as a saturable sponge-like molecule. This approach often was not the case in clinical trials of antioxidants in AD, likely due in part with the underappreciation of free radical chemistry. 3) The cellular redox state of individuals involved in the clinical trials usually was not considered. Accordingly, a person with a more reductive cellular redox state would not benefit from antioxidants. Consequently, the mean of change from control in a population of subjects with varying cellular redox states likely would not be large, but the standard deviation would be large, leading researchers to conclude antioxidants were not effective in AD.
Redox proteomics studies in AD and its early stages
Redox proteomics, which was pioneered in our laboratory [6, 50–56], leads to identification of excessively oxidized or HNE-bound proteins compared to those proteins in aged matched control brains. Redox proteomics methodology is based on protein separation, selection of oxidatively or nitrosatively modified proteins via sophisticated image analyses, trypsin digestion of these selected proteins, peptide clean-up, application of tandem mass spectrometry (MS/MS) methods to determine the peptide amino acid sequence, followed by interrogation of appropriate databases to identify the specific proteins since each protein has a unique amino acid sequence [6, 50]. Application of redox proteomics to specimens from subjects with AD or MCI led to the identification of many brain proteins altered at different stages of AD progression [6, 51–70]. The reader is directed to the papers cited above and recent reviews [6, 61] from our laboratory for experimental details and list of oxidized brain proteins identified. When classified into pathways, these oxidized proteins were in functional classes, who loss of functions were consistent with the clinical presentation, pathology, and biochemical alterations of AD, MCI, and preclinical AD (Table 1). For example, glucose utilization via glycolysis, the TCA cycle, and the mitochondrial electron transport chain was predicted to be compromised in AD based on redox proteomics-determined oxidative damage to key protein components of each major pathway involved in glucose utilization for ATP production. Our results and predictions are consistent with 18F-glucose PET studies showing progressively decreased glucose utilization with increased stage of AD. In AD and MCI brain, diminution of ATP production following oxidative modification of these proteins is consistent with and likely contributes to: 1) loss of phospholipid asymmetry in cell membranes (that affects both membrane lipid integrity and function of transmembrane proteins and is a marker for apoptosis) [71]; 2) loss of synaptic remodeling associated with decreased neurotransmission and consequent decreased learning and memory [61, 72]; 3) decreased rate of neuronal mitochondrial anterograde and retrograde transport to and from energy-starved pre-synaptic terminals [73–77]; 4) decreased neuritic length (which would decrease efficiency of neuronal communication, clearly important in a disease associated with decreased cognition and memory [78]); and 5) elevated neuronal intracellular Ca2 + (that would both compromise glutamate neurotransmission processes and induce activity of several intracellular destructive enzymes, such as phospholipases, endonucleases, proteases, etc., thereby inducing both apoptotic and necrotic destruction of neurons) [79]; and many other aspects of AD associated with the clinical presentation, pathology, and biochemical alterations known in each stage of AD.
Brain protein and/or pathway dysfunction as a consequence of oxidative damage in Alzheimer’s disease or amnestic mild cognitive impairment revealed by redox proteomics*
*See text for more details.
Some specific proteins are uniquely modified throughout all stages of AD, and I opine that this small subset of oxidized proteins may contribute to the progression of this disorder. Some of these specific proteins are involved in pathways for ATP production (enolase; ATP-synthase) or proteostasis (ubiquitin carboxyl-terminal hydrolyase L1 in the Ubuiquitin-Proteasome System; cathepsin D and V0-ATPase for autophagolysosomal function in autophagy). Defects in glucose metabolism throughout the progression of AD were mentioned above. Autophagy is known to be decreased in AD [80, 81]. This loss of autophagic function throughout the progression of AD, which would lead to accumulation of cellular detritus and therefore cell death, also may be related to activation in the mammalian target of rapamycin (mTOR) pathway throughout the stages of AD [81], an activation that can be initiated by Aβ42, among other means [82]. Our findings that oxidative dysfunction of such a small number of key proteins and pathways related to glucose utilization and removal of aggregated, damaged proteins occurs from the early stages of AD to late-stage AD are consistent with the notions that these proteins and pathways are critical to AD progression and are potentially therapeutically targetable to slow this progression.
Recent studies from our laboratory reported results of investigations of brain from individuals with Down syndrome (DS) obtained as a function of age [81–92]. Though DS individuals exhibit intellectual disability from birth, often AD neuropathology and dementia occur in DS persons at approximately 40–50 years of age [93]. Oxidative stress and redox proteomics studies of brain from DS persons demonstrate changes similar to what is observed in AD brain [81–92]. In addition, activation of the mTOR pathway (with consequent dysfunction of autophagy and insulin signaling in brain), coupled with diminution of glucose metabolism and alterations in the proteostasis network of DS persons who have AD neuropathology and present with dementia, mirrors these characteristics of AD brain [81, 92]. The notion that insights into AD may be gained from study of DS persons is discussed in the section on Future Research Directions below.
Importance and implications of our research on AD
Throughout the above discussion, the importance and implications of our research on AD have been mentioned. Summarizing these discussion points regarding research from our laboratory: Oxidative stress is now considered by most AD researchers and clinicians to be a critical component of this dementing disorder and its earlier stages and a contributor to progression of AD. Neurotoxic oligomers of Aβ42 were shown to be strongly associated with oxidative stress and correlated to protein oxidation and lipid peroxidation in AD and MCI brain and in in vitro and in vivo models thereof. Redox proteomics approaches led to the identification of oxidatively dysfunctional proteins and biochemical pathways, whose dysfunctions are consistent with the clinical presentation, pathology, and biochemical alterations of AD and MCI. A small subset of these redox proteomics-identified, oxidatively dysfunctional proteins and pathways are present from early stages to late-stage AD, suggesting their importance for the progression of this dementing disorder and potential therapeutic targets to slow or retard progression of AD.
PREDICTIONS FROM OUR LABORATORY OF FUTURE RESEARCH DIRECTIONS IN THE FIELD OF ALZHEIMER’S DISEASE
It is my opinion that future AD research will coalesce around several key processes/functions, the protection of which conceivably could slow, or ideally stop, progression of this devastating disorder. Below are some areas of future research in AD that I predict will be among these coalesced areas of research.
Plasma as a biomarker source
As noted above, neuropathology of AD is present approximately two decades prior to the onset of symptoms. Consequently, given the need to diagnose prodromal AD prior to the onset of symptoms, one area of predicted future research is the eventual identification of a reliable and unique set of biomarkers for the unequivocal diagnosis of AD from easily obtainable specimens. This notion is supported by our studies of beagles, in collaboration with the laboratories of Carl Cotman and Elizabeth Head [94]. Brain was isolated from 15-year-old beagle dogs (who have Aβ42 deposition of the same amino acid sequence as humans), who for the preceding 3 years had been on a high antioxidant diet, exposed to a behaviorally enriched environment (to learn new tasks and thereby make new synapses), and given exercise. The oxidative stress levels in brain of such treated beagles were much lower and similar to those of much younger dogs and in marked contrast to unstimulated dogs fed dog chow. Moreover, the treated beagles had lower levels of Aβ42 and performed in behavioral tests like younger dogs [94]. Assuming that these promising results are transferrable to humans, individuals with incipient AD identified by reliable and specific biomarkers could be placed on regimens of high antioxidant diets, exercise (as appropriate and able), and intellectual stimulation (i.e., crossword puzzles, learning a new language, taking up a new musical instrument, etc.). My opinion is that identification of biomarkers for prodromal AD well before onset of symptoms will be a critically important future research effort.
Along this line of thinking, our laboratory, working with that of Patrizia Mecocci, used mitochondria isolated from peripheral lymphocytes to demonstrate elevated oxidative stress and proteomics identification of key proteins of differential levels in both MCI and AD individuals [95, 96]. The elevation of oxidative stress in specific individuals was inversely correlated with cognitive function assessed by the Mini-Mental State Examination and inversely correlated with levels of small soluble antioxidants [95, 96]. In addition, plasma, and to a lesser extent cerebrospinal fluid (CSF) (the hesitation due mostly due to the more invasive method of obtaining CSF), are fluids that may one day be a source of reliable and unique biomarkers. However, plasma has proteins at levels that span a range of 1014, with a small number of proteins comprising 85–90% of total plasma proteins. Thus, in order to reliably measure proteins of much lower concentration, the major proteins (for example, albumin, IgGs, certain glycoproteins) need to be removed and separately analyzed. Using this approach in both fluids, in collaboration with Marzia Perluigi’s laboratory, we reported significant changes in specific proteins using proteomics in AD and MCI [97–101]. At this stage of investigation, such studies are not sensitive enough nor specific enough for unequivocal biomarker-based diagnosis of MCI and AD. Continued improvement in separation technology will lead increased use of these methods for soluble protein-based biomarkers for AD and MCI in my opinion. The major plasma proteins, particularly albumin, also may serve as a biomarker, so, as noted above, the removed proteins also need to be investigated in my opinion.
Interestingly, plasma also has extracellular vesicles (EV) that emanate from neurons, and proteomics and other means of identifying proteins have shown early changes in neuronal protein composition in EVs from persons who would go on to develop AD [102, 103] or who have DS [104]. A good prediction of future research in the AD field is that EV-related research for biomarkers of neuronal origin will greatly expand.
Investigations of pathways identified as important in AD and MCI
Given that type 2 diabetes mellitus is a major risk factor for AD [105], studies of insulin signaling, which is inhibited following activation of the mTORC1 pathway [81], coupled with the role of mTORC1 activation in AD and MCI on inhibition of autophagy [80, 81], I predict that greater investigation of the mTOR pathway activation by Aβ42 and other factors in AD and MCI brain and in patients will occur in the future. Currently, FDA-approved drugs, including rapamycin and metformin (among others), that inhibit mTOR are known. More studies to ensure no harm to patients arises, long-term use of these agents from likely will be investigated.
In the same vein, noting that oxidative stress and redox proteomics studies in AD and MCI brain and mitochondria from peripheral lymphocytes identified enolase, which is a pleitropic enzyme [106], and ATP synthase as oxidatively modified and dysfunctional proteins [96], I predict that research on finding ways to increase glucose metabolism in patients with early stages of AD will be pursued.
Oxidative stress
As discussed above, clinical trials in AD with small antioxidant compounds have been disappointing. However, as also discussed above, there may be key methodological and pedagogical reasons for these failures. Given that in aged beagle dogs use of high antioxidant diets coupled with intellectual stimulation and exercise led to improved cognition and dramatically reduced oxidative stress [94], I predict that research (and clinical practice) in AD in the future will emphasize this multi-pronged approach that was successful in reducing loss of cognition and significantly decreasing brain levels of Aβ42. One particular aspect of this approach that likely will be increasingly emphasized in the future, I predict, will be research on the use of food rich in components that themselves induce a stress, to which cells respond to produce beneficial effects, so called cellular stress response [107] or hormesis [108]. One such cellular response is upregulation of Nrf-2-mediated phase 2 enzymes, such as heme oxygenase, gamma-glutaminylcysteine ligase, and Mn-superoxide dismutase (MnSOD) [107]. The notion of this prediction is that, in an analogous way that cancer therapy is evolving to stimulate the patient’s own immune response to destroy cancer cells, hormetic approaches to cause the AD patient to upregulate her/his own antioxidant or other protective responses to the disease will be beneficial to the patient.
Studies of specific proteins in AD
Redox proteomics and western blot approaches from our laboratory have identified oxidative modification of many proteins, the functions of which are shown in Table 1 above. In the interests of space, I consider two proteins of particular relevance to future AD research (in addition to those mentioned elsewhere in this paper). One protein is peptidyl-prolyl cis-trans isomerase (Pin 1) [53, 109–111]. Pin 1 is a regulatory protein that controls the activity of target proteins by binding to their phosphorylated Ser/Thr-Pro domains and converting the conformation of the proline residue from trans to cis and vice versa [109, 112]. This conformational change produces a very large structural change in the target protein, thereby regulating its activity [109]. Two such target proteins are AβPP, from which Aβ42 is derived, and phosphorylated tau. In addition, protein phosphatase 2A, which removes phosphate moieties from hyperphosphorylated tau, also is a Pin 1-regulated protein. Consequently, oxidative dysfunction of Pin 1 would be related to two principal pathological hallmarks of AD: senile plaques and neurofibrillary tangles. Therefore, additional research on the role of Pin 1 in AD is predicted.
Biliverdin reductase-A (BVR-A), following action of heme oxygenase-1 (HO-1), converts bilverdin to bilirubin as part of the processes needed to rid the brain (and other organs) of toxic heme [113]. Moreover, at low levels, bilirubin is reportedly a powerful antioxidant and de-nitrifying enzyme [114]. However, BVR-A is a pleiotropic enzyme, having both reductive and kinase properties depending on which specific sites on BVR-A are phosphorylated by various kinases [113]. Among other enzymes, insulin receptor substrate-1, needed for transducing insulin signaling, is phosphorylated by BVR-A and may be critical to modulation of insulin signaling [115, 116], which is known to be defective in AD and MCI brain [81]. Our laboratory demonstrated that the HO-1/BVR-A system in brain is highly oxidatively modified and dysfunctional in AD and MCI [145, 146]. I predict that future research on AD will examine further the role of BVR-A in AD and MCI.
Statins
Inhibition of cholesterol synthesis by use of HMG-CoA reductase inhibitors (statins) has been suggested to reduce incidence of AD [119]. However, lowering of cholesterol levels in brain is not the reason for this proposed benefit, since many statins, including atorvastatin, do not cross the BBB [120, 121]. Future research into potential mechanisms by which statins may reduce incidence of AD might provide insights into underlying molecular processes in the disease itself. Consistent with this notion, our research using atorvastatin in the aged beagle dog model system for AD showed this statin greatly reduced oxidative stress and protected BVR-A in brain against oxidative and nitrosative modification, even though atorvastatin does not cross the BBB [120–125]. Atrovastatin also led to lower levels of Aβ and modulated various pathways to provide cognitive benefit [93]. I predict future research in AD will include studies on better understanding the molecular processes involved with non-BBB penetrable statins and potentially reduced incidence of AD.
Inflammation
Notwithstanding that there is ample evidence for inflammation in brain in AD [126], determining if this is causative or a result of AD remains the subject of investigation. As noted above, traditional small molecule anti-inflammatory compounds were not effective in clinical trials in AD. However, newer brain-accessible anti-inflammatory compounds likely will be the subject of future research in AD. Even if such compounds do not get at the primary cause of AD, decreasing neuroinflammation has its own benefits for AD patients and would justify this predicted future research.
Studies in Down syndrome
As discussed above, DS persons often develop AD-like neuropathology and dementia after the fourth decade of life [93]. Insights into the major treatable characteristics of trisomy of chromosome 21 are predicted to result from future research in DS, and, simultaneously, studies of the age-dependent changes in brain from DS individuals are predicted to give new insights into molecular processes and pathways of direct importance to AD, from which new therapies are predicted to result.
CONCLUSION
Given the rapidly aging Baby Boomer cohort in the United States numbering more than 70 million people, coupled with aging being the single most important risk factor for AD, a public health crisis is facing the US in terms of the enormous number of new AD patients predicted to arise over the next 20–30 years. The cost involved in caring for these people, including lost wages, is enormous, and might not be sustainable for the nearly two decade-long population bubble of Baby Boomers. Clearly, some intervention has to emerge to slow the onset of AD. Since the average lifespan from diagnosis to death is about eight years, delaying the onset of AD by five years immediately would cut by more than 50% the number of persons with AD. Such an outcome might be a realistic first goal of future AD research. Consequently, a renewed dedication by government, the private sector, scientists, and physicians to achieving this goal is needed. For ordinary Americans, a greater commitment to lifestyle changes to minimize risk factors for developing AD will be required. All these efforts will necessitate more basic research into the causes and consequences of AD and development of disease-modifying agents and modalities.
This current paper gives the perspectives on the importance and implications of AD-related research done in the author’s laboratory and his predictions of future directions of research into this ominous and challenging disorder. It has, and continues to be, our laboratory’s great privilege to contribute to better understanding of some of the molecular processes involved in AD from which potential disease-modifying therapeutic strategies have emerged.
Footnotes
ACKNOWLEDGMENTS
I thank all my past and current graduate students, postdoctoral scholars, and visiting scientists who trained in my laboratory for their dedication to and success in AD-related research, and to my collaborators mentioned in this paper. I also thank the faculty of the Sanders-Brown Center on Aging for providing well-characterized AD and MCI specimens obtained at short PMI and to the NIH for support of much of the research performed in my laboratory over the years.
