In vivo imaging of prodromal hippocampus CA1 subfield oxidative stress in models of Alzheimer disease and Angelman syndrome

Abstract

Hippocampus oxidative stress is considered pathogenic in neurodegenerative diseases, such as Alzheimer disease (AD), and in neurodevelopmental disorders, such as Angelman syndrome (AS). Yet clinical benefits of antioxidant treatment for these diseases remain unclear because conventional imaging methods are unable to guide management of therapies in specific hippocampus subfields in vivo that underlie abnormal behavior. Excessive production of paramagnetic free radicals in nonhippocampus brain tissue can be measured in vivo as a greater-than-normal 1/T₁ that is quenchable with antioxidant as measured by quench-assisted (Quest) MRI. Here, we further test this approach in phantoms, and we present proof-of-concept data in models of AD-like and AS hippocampus oxidative stress that also exhibit impaired spatial learning and memory. AD-like models showed an abnormal gradient along the CA1 dorsal–ventral axis of excessive free radical production as measured by Quest MRI, and redox-sensitive calcium dysregulation as measured by manganese-enhanced MRI and electrophysiology. In the AS model, abnormally high free radical levels were observed in dorsal and ventral CA1. Quest MRI is a promising in vivo paradigm for bridging brain subfield oxidative stress and behavior in animal models and in human patients to better manage antioxidant therapy in devastating neurodegenerative and neurodevelopmental diseases.—Berkowitz, B. A., Lenning, J., Khetarpal, N., Tran, C., Wu, J. Y., Berri, A. M., Dernay, K., Haacke, E. M., Shafie-Khorassani, F., Podolsky, R. H., Gant, J. C., Maimaiti, S., Thibault, O., Murphy, G. G., Bennett, B. M., Roberts, R. In vivo imaging of prodromal hippocampus CA1 subfield oxidative stress in models of Alzheimer disease and Angelman syndrome.

In neurodegenerative diseases, such as Alzheimer disease (AD), and in neurodevelopmental disorders, such as Angelman syndrome (AS), patients demonstrate impairment in goal location based on surrounding landmarks (i.e., allocentric spatial disorientation) as one of many debilitating morbidities (1, 2). An essential brain region involved in encoding spatial memories is the CA1 subfield of the mid- and posterior hippocampus (human) or dorsal CA1 (CA1d) in rodents (3, 4). A leading hypothesis in the field based on ex vivo measurements is that hippocampus oxidative stress plays a pathogenic role in spatial disorientation (2, 5–16). This hypothesis has developed on the basis of studies in postmortem tissue of oxidative damage biomarkers, such as buildup of the polyunsaturated fatty acid oxidation product 4-hydroxynonenal (HNE) (17, 18). At present, it is not possible to measure excessive production of free radicals from hippocampus CA1 subfields in vivo. Thus, translation of basic science findings into personalized management of antioxidant treatments in patients has been limited and a major problem for clinically testing and improving therapeutic efficacy.

Quench-assisted (Quest) MRI has recently been suggested as a robust and sensitive tool for mapping excessive free radical production in different layers of the retina in vivo without a contrast agent (15, 19). The Quest MRI index of abnormally high production of paramagnetic free radicals is a greater-than-normal spin-lattice relaxation rate R₁ (1/T₁) that can be returned to baseline after acute antioxidant administration (15, 19, 20). Our preliminary studies find agreement between Quest MRI in vivo and reference standard free radical measurements in several animal models (15, 19). For example, in a mouse model of diabetic retinopathy, and in pharmaceutical and genetic models of outer retinal degeneration, Quest MRI, histochemical stains such as dichlorofluorescein and dihydroethidine, and a bioluminescence method such as lucigenin, all identify the outer retina as the first location to show evidence of oxidative stress before overt histopathology is apparent (15, 19, 21, 22). However, more work is needed in phantoms to directly confirm that sustained free radical production per se is what Quest MRI detects. Also, it is unclear whether Quest MRI can be applied to study nonretinal brain tissue, such as hippocampal CA1 subfields.

In this study, we tested the feasibility of using Quest MRI to provide evidence of localized oxidative stress in vivo in brain subfields identified from previous ex vivo studies in animal models that exhibit spatial disorientation. We evaluated Quest MRI during sustained superoxide production from the xanthine/xanthine oxidase reaction (23, 24). We provide proof-of-concept data that Quest MRI can be applied to evaluate hippocampus subfields in vivo in 3 disease models: aldehyde dehydrogenase 2-knockout mice, 5xFAD mice, and UBE3A-knockout mice (25–29). Aldehyde dehydrogenase 2 mice are also called HNE mice because they exhibit a greater-than-normal hippocampus concentration of the peroxidation product HNE (the earliest evidence for oxidative stress in patients with prodromal AD (30), together with spatial learning and memory impairments and age-dependent spontaneous AD-like histopathology (17). 5xFAD mice are a common β-amyloidogenic model of familial AD that demonstrates hippocampus oxidative stress (including HNE accumulation) and spatial cognitive defects (26, 28). UBE3A-knockout mice are a model of AS, a genetic neurodevelopmental disorder characterized by hippocampus oxidative stress and impaired goal location based on external cues and other endophenotypes consistent with that in patients with AS (2, 31). Because oxidative stress impairs l-type calcium channel (LTCC) function and afterhyperpolarization (AHP; due to reduced calcium entry) (32–36), we also examined 5xFAD mice for downstream functional oxidative damage produced by excessive free radical production. LTCC function was measured in vivo along the dorsal–ventral CA1 axis using manganese-enhanced MRI (MEMRI) (37); calcium-dependent AHP in the dorsal hippocampus CA1 subfield was also evaluated (35, 36).

MATERIALS AND METHODS

All animals were treated in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health (Bethesda, MD, USA), the Statement for the Use of Animals in Ophthalmic and Vision Research (Association for Research in Vision and Ophthalmology, Rockville, MD, USA), and the Institutional Animal and Care Use Committee authorization (Wayne State University, University of Michigan, and the University of Kentucky). Animals were housed and maintained in 12:12-h light–dark cycle laboratory lighting, unless otherwise noted.

Experiments

In vitro study

Xanthine/xanthine oxidase is a commonly used system to continuously generate superoxide free radicals in vitro (23, 38). Phantoms used for this study consisted of a 0.7-ml microcentrifuge tube containing either xanthine oxidase (0.135 U/ml, 150 μl) or xanthine oxidase (0.135 U/ml, 150 μl) with either superoxide dismutase (SOD; 125 U/ml, 60 μl) or additional PBS (60 μl). After collecting baseline scans, xanthine (2.9 mM, 300 μl) was carefully added to the reaction vessel inside the magnet via PE20 tubing, and 1/T₁ measurements were obtained. Preliminary experiments were performed to confirm continuous production of free radicals over the examination period based on absorbance spectroscopy (data not shown). All reagents were purchased from Sigma-Aldrich (St. Louis, MO, USA), and all stock solutions were prepared in PBS.

Animal models

Three-month-old HNE mice (17), 2-mo-old 5xFAD mice, and 4-mo-old AS mice [all on C57BL/6 (B6) backgrounds] were studied; additional age-matched B6 mice were purchased as needed (The Jackson Laboratory, Bar Harbor, ME, USA). In all cases, a baseline T₁ data set was collected and mice were allowed to recover from the isoflurane anesthesia. Thirty minutes after recovery, mice were treated with 1 mg/kg methylene blue (MB, i.p., dissolved in saline); MB is an alternate electron transporter that effectively suppresses superoxide generation from a variety of sources (39). The next day, ∼1 h before the second MRI examination, each MB-treated mouse was treated with 50 mg/kg α-lipoic acid (ALA, i.p., dissolved in saline and pH adjusted to ∼7.4). ALA is a potent free radical neutralizer (15, 19, 40). Mice were then reanesthetized with isoflurane, and another T₁ data set was obtained from the same brain slice. All mice were humanely killed after the second MRI examination. Some of the same 5xFAD mice were allowed to recover and were then studied by electrophysiology. A different set of 6-mo-old 5xFAD mice and age-matched controls were separately studied using MEMRI.

MRI

The general mouse preparation for high resolution 2-dimensional (2-D) MRI is well established in our laboratory (15, 19, 41). HNE, 5xFAD, and AS mice were anesthetized with isoflurane (3% induction, 1.2% maintenance) to allow for recovery. 5xFAD mice studied using MEMRI were anesthetized with urethane (36% solution i.p.; 0.083 ml/20 g animal weight, prepared fresh daily; Sigma-Aldrich). MEMRI examination of 5xFAD mice required a prior bolus injection of MnCl₂ (66 mg MnCl₂·4H₂O/kg, i.p.) on the right side of awake mice (42). Four hours after this injection, mice were anesthetized and T₁ data obtained. In terminal studies, mice were humanely euthanized as detailed in our Division of Laboratory Animal Resources-approved protocol.

High-resolution T₁ 2-D data sets were acquired on a 7 T system (Bruker ClinScan; Bruker Daltonics, Bremen, Germany). For the brain studies, a receive-only 4-element phased array coil was used. In all cases, several spin-echo images were collected (single transverse slice, echo time (T_E) 11 ms, 12 × 12 mm², matrix size 192 × 192, slice thickness 400 μm, turbofactor 4). For the phantom studies, a receive-only surface coil (1.0 cm diameter) was used, and 3 baseline T₁ data sets were first collected, followed by 5 additional T₁ data sets after in-magnet addition of xanthine starting at 2 min 40 s after injection, and then at 4 min 30 s, 7 min 20 s, 10 min 20 s, and 15 min 20 s after injection.

Because the Quest MRI experiment is looking for differences in 1/T₁ between baseline and antioxidant treatment, generating a precise measure of 1/T₁ is the major requirement. Thus, any sequence that produces precise 1/T₁ measurements can be used for Quest MRI. Here, we performed a commonly used reference standard saturation recovery approach in which multiple repetition times (T_Rs) are used to precisely measure 1/T₁ over a large range of signal-to-noise conditions, a procedure routinely performed in our laboratory (42–46). Images with different T_Rs were acquired in the following order (number of averages in parentheses): T_R 0.15 s (6), 3.50 s (1), 1.00 s (2), 1.90 s (1), 0.35 s (4), 2.70 s (1), 0.25 s (5), and 0.50 s (3). To compensate for reduced signal-to-noise ratios at shorter T_Rs, progressively more images were collected as the T_R decreased. The time investment is about 45 min per mouse: ∼15 min to initially anesthetize the mouse, gently position it in the cradle, and center it in the magnet; 10 min to autoshim and autocoil calibrate; and ∼15 min to acquire the full T₁ data set. Removing the mouse from the magnet, humanely killing the animal, and cleanup takes an additional ∼5 min.

Hippocampal electrophysiology (AHP)

5xFAD mice that were examined by Quest MRI were then used for electrophysiologic characterization of the CA1d AHP. Hippocampal slices from mice were prepared as previously described in detail (47, 48). Rostral portions of freshly dissected brain hemispheres (including the hippocampus) were cut (350 μm) sequentially using a 3000 Vibratome (Thermo Fisher Scientific, Waltham, MA, USA) in oxygenated low-Ca²⁺ artificial cerebrospinal fluid (ACSF, 0.1 mM Ca²⁺). Slices containing dorsal hippocampus were then placed in an interface-type chamber and maintained in oxygenated regular ACSF at 32°C for at least 2 h. Regular ACSF contains the following (mM): NaCl (128), KCl (3), CaCl (2), MgCl (2), glucose (10), KH₂PO₄ (1.25), and NaHCO₃ (26). Individual slices were transferred one at a time to a microscope stage (E600 FN; Nikon, Tokyo, Japan) equipped with a recording chamber continuously perfused (2 ml/min) with oxygenated ACSF maintained at ∼32°C (TC2Bip; Cell Micro Controls, Norfolk, VA, USA). Sharp electrodes (70–90 MΩ) were used to monitor membrane voltages recorded from CA1 pyramidal neurons using an Axoclamp 2B amplifier (Molecular Devices, Sunnyvale, CA, USA). Electrodes were filled with 2 M potassium methyl sulfate (KMeSO₄) and 10 mM HEPES, pH 7.4, and were pulled from borosilicate glass capillaries (World Precision Instruments, Sarasota, FL, USA) on a P80 pipette puller (Sutter Instrument, Novato, CA, USA). Responses were acquired in current clamp mode and data digitized (5 kHz) and stored to PCs using pClamp 8 (Molecular Devices). To generate intracellular AHPs, neurons were held at −65 mV, and a 100 ms current depolarization pulse was delivered through the electrode to generate 4 Na⁺ action potentials. AHP amplitudes were measured at the negative peak immediately following the depolarization pulse, during the medium AHP, which typically lasts several 100 ms in hippocampal pyramidal neurons, and at 800 ms after the end of the step, during the slow AHP. For each cell recorded, 4–10 depolarizations were averaged to quantify the AHP. Only neurons with input resistances >35 MΩ and overshooting action potentials (>0 mV) were included in the analysis.

MRI data analysis

Within each T₁ data set of 23 images, images acquired with the same T_R were first registered (rigid body) and then averaged to generate a stack of 8 images. These averaged images were then registered across T_Rs. The 1/T₁ maps were then calculated via fitting to a 3-parameter T₁ equation {y = a + b × [exp(−c × T_R)], where a, b, and c are fitted parameters} on a pixel-by-pixel basis using R v.2.9.0 software (R Foundation for Statistical Computing, Vienna, Austria; http://www.r-project.org/) and scripts developed in house.

Initial reports of Quest MRI normalized experimental data to same-day control mice data to minimize day-to-day variations. While effective, the approach is clearly impractical in a clinical setting and in certain experimental models. Here, we introduce new analysis procedure to eliminate the need for normalizing to same-day control mice data. Because of slice bias in the 2-D data and low signal-to-noise ratio, the T₁ estimate is highly dependent on the signal intensity of the T_R 150-ms image and thus imprecise (44). By normalizing to the shorter T_R, some of the bias can be removed giving a more accurate estimate for T₁. We normalized within and between groups for signal intensity differences by first applying 3 × 3 gaussian smoothing, performed 3 times, only on the T_R 150-ms image to minimize noise and emphasize signal. The smoothed T_R 150-ms image was then divided into the rest of the images in that T₁ data set. Preliminary experiments (data not shown) found that this procedure minimized the day-to-day variation in the 1/T₁ profile previously noted, abrogating the need for normalization to a same-day group of control mice (15, 19).

In the brain studies, line regions of interest over presumptive CA1d spanning ∼12.5 pixels (∼780 μm) along the pyramidal region starting just ventral of the subiculum as assessed from the anatomic image, and a region ventral to CA1d called intermediate hippocampus CA1 (CA1i) spanning ∼15.5 pixels (∼969 μm) approximately midway down the pyramidal region) were identified. Average 1/T₁ MRI data from left and right CA1d and CA1i were analyzed (15, 19).

Statistical analysis

In the phantom studies, we used a generalized estimating equation approach to examine differences between groups while appropriately accounting for correlations over time. We used a first-order autoregressive correlation structure, with similar results obtained using an unstructured correlation matrix.

In the brain studies using Quest MRI, we only examined CA1d and CA1i, conducting analyses separately for each region. By limiting the number of brain regions analyzed, we substantially improved the power of hypothesis testing using the methods presented here. While the data for each experimental strain (i.e., HNE, AS, and 5xFAD) were obtained on different dates with a different set of B6 controls, we used a linear mixed model to jointly analyze all data to improve power to detect differences. The dependent variable for these analyses was 1/T₁, and we included the following independent variables: animal strain (B6, HNE, AS, 5xFAD), MB-ALA antioxidant treatment, and the interaction between strain and MB-ALA antioxidant treatment. A random effect of mouse within strain was included to account for measuring mice before and after MB-ALA antioxidant treatment. A random effect of batch was evaluated, but this effect was not significant using a likelihood ratio test, and both Akaike’s information criterion and the Schwarz Bayesian information criterion indicated that the batch random effect did not improve the fit of the models. As such, we did not include this random effect in our analyses. We initially evaluated the statistical significance of the interaction between strain and MB-ALA antioxidant treatment, and we only examined simple effects if the interaction was significant. For the test of the interaction, we used a type I error rate of 0.1 because power is often much lower for interactions. We used the resulting mixed effects model to estimate the MB-ALA antioxidant treatment effect within each strain, the strain differences at baseline, and the strain differences under MB-ALA antioxidant treatment. Two-tailed unpaired Student t tests were performed on MEMRI data in each region (CA1d or CA1i), and to look for AHP differences across transgenic and wild-type (WT) animals. Data are presented as means ± sem.

RESULTS

Detection of sustained production of free radicals by Quest MRI in vitro

Superoxide free radicals inside the magnet were generated using a commonly used xanthine/xanthine oxidase enzymatic system (24). This resulted in a steady ∼6% increase over baseline 1/T₁ values (Fig. 1). SOD added to the reaction vessel prevented the increase in 1/T₁ (24). These results directly confirmed that continuous production of biologically relevant free radicals are readily detectable using Quest MRI.

Figure 1.

Detection of sustained production of free radicals by Quest MRI in vitro. Time course of 1/T₁ from phantoms containing xanthine oxidase (black circles, n = 3) before (left of arrow, baseline) and after (right of arrow) in-magnet addition of xanthine (arrow), or containing xanthine oxidase and SOD (open squares, n = 3). Horizontal bar indicates significant difference range; error bars ± sem.

Control mice demonstrated no evidence for changes in hippocampus CA1 1/T₁ in vivo with or without antioxidants

Comparing all control B6 mice with and without MB-ALA antioxidant treatment revealed no differences in 1/T₁ for both CA1d (95% confidence interval, baseline: 0.65–0.68 s⁻¹ and with antioxidants: 0.65–0.69 s⁻¹) and CA1i (95% confidence interval, baseline: 0.61–0.65 s⁻¹ and with antioxidants: 0.62–0.66 s⁻¹).

Hippocampus CA1 excessive free radical production measured in vivo in HNE, AD-like, and AS models

At baseline, HNE, 5xFAD, and AS mice all had CA1d 1/T₁ that was greater than their respective age-matched WT B6 mice (Fig. 2). After MB-ALA antioxidant administration, 1/T₁ CA1d in all experimental groups were significantly reduced to WT levels. The 1/T₁ of CA1i region in HNE or 5xFAD mice was not different from controls, whereas in the AS mice CA1i 1/T₁ was supernormal and quenchable (Fig. 2).

Figure 2.

Hippocampus CA1 subfields demonstrates disease-dependent excessive production of free radicals in vivo. Representative anatomic image illustrating position of CA1d (A) and CA1i (C) regions (white lines), and corresponding bar graph summaries (B, D). Summary of normalized 1/T₁ at baseline (B) and after treatment with antioxidants (AO) in B6 WT, HNE, 5xFAD, and AS mice. Numbers above bars indicate numbers of mice; error bars ± sem; horizontal bar represents significant difference.

Hippocampus CA1 calcium dysregulation identified in vivo in models of AD

In separate group of 5xFAD mice, a lower-than-normal uptake of manganese was noted in CA1d but not CA1i as measured by MEMRI (Fig. 3) (37, 49–51). We also found subnormal CA1d calcium-dependent AHP (Fig. 3).

Figure 3.

Hippocampus CA1d but not CA1i exhibits impaired LTCC function in vivo, and CA1d shows AHP reductions in 5xFAD mice compared to WT counterpart. A, B) Summary of normalized MEMRI 1/T₁ (see Materials and Methods) of CA1i (A) and CA1d (B) in B6 WT and 5xFAD mice. Numbers above bars indicate numbers of mice; error bars = sem. C) Bar graph of mean medium AHP amplitudes from CA1d neurons of WT (n = 10 cells from 4 mice) and 5xFAD (n = 12 cells from 5 mice). Horizontal bar denotes significant difference.

DISCUSSION

The results of the present xanthine/xanthine oxidase experiment validate a key assumption of Quest MRI: sustained production of free radicals has a net detectable impact as an effective paramagnetic 1/T₁ contrast mechanism in the absence of a spin trap contrast agent and potentially confounding variables that are present in vivo (15, 19, 20, 52, 53). It is commonly thought that because free radicals have very short lifetimes and low concentrations in vivo, they would have a negligible impact on water 1/T₁ (20, 52, 54–58). However, oxidative stress is a condition that is defined as a sustained, net production of free radicals, and the available data clearly demonstrate robust detection via Quest MRI under these conditions (15, 19, 20, 53). Because all species of free radicals are paramagnetic, Quest MRI likely detects the total excessive free radical burden of the tissue (15, 19, 52, 53). We estimate that the xanthine oxidase generated an approximate steady state concentration of superoxide free radicals of ∼60 μM. This concentration estimate is somewhat uncertain because of, for example, potential differences in viscosity between water and the intracellular environment in vivo, which is expected to increase free radical relaxivity (59) and the variable level of superoxide produced that depends in part on the oxygen content of the solution and enzymatic activity of a particular batch of xanthine oxidase (23, 38, 60). Unfortunately, conversion of 1/T₁ to a superoxide concentration is also problematic because the relaxivity of superoxide radicals is not known. Intriguingly, isolated neutrophils produce an estimated total concentration of superoxide of ∼76 μM, although whether this is representative of the concentration of free radicals under oxidative stress conditions in vivo is, to our knowledge, unknown (61). Nonetheless, the above considerations suggest that Quest MRI has sufficient sensitivity to detect an oxidative stress condition in vivo that produces a sustained free radical level of ∼60 μM. On the basis of the data in Fig. 1, it seems likely that lower concentration levels can be detected and more work is needed to define the detection sensitivity of Quest MRI in vivo.

In this study, the use of isoflurane is a potentially confounding factor because it is thought to have antioxidant effects in and of itself (62–67). This claim is somewhat controversial, however, with other studies showing a prooxidant effect of isoflurane (62–67). The experiments performed herein were not designed to resolve this conflict. Nonetheless, we note that if isoflurane exerted a major anti- or pro-oxidant action, we would expect baseline values between control and experimental CA1d and CA1i to be similar. However, this was not observed. Instead, we see some groups in which baseline CA1i values are similar in control and experimental mice (AD-like mice) or elevated compared to controls (AS mice). Thus, we reason that non-isoflurane-based endogenous differences in free radical production in CA1 subregions between controls and experimental mice are the dominant source of sustained free radical production observed.

CA1d and CA1i were examined herein in 3 models that have previously been shown to have hippocampus oxidative damage based on conventional ex vivo methods. Two spatially close hippocampal regions, CA1d and CA1i, were examined. These two regions are functionally distinct (4, 68). For example, they encode for different types of memories [i.e., spatial (CA1d) vs. emotional (CA1i)] (3, 4, 69); they respond differently to ischemia–reperfusion (i.e., CA1d exhibits 9-fold more ischemic necrosis than CA1i)(70); they have ∼10% different total vascular areas (i.e., CA1d < CA1i) (71); and they experience different levels of excitability (i.e., CA1d < CA1i) (72). Thus, the results herein are first-time evidence supporting Quest MRI for detecting in vivo disease-dependent production of abnormally high levels of free radicals in CA1 subfields.

Supernormal production of free radicals is a necessary but not sufficient condition of oxidative stress (15, 19). In addition, it is helpful to provide evidence not just of oxidative stress but also of downstream oxidative damage to related physiology, such as neuronal calcium homeostasis or excitability (32–34, 48). To this end, 5xFAD mice were examined using MEMRI, the imaging modality of choice for measuring neuronal LTCC function in vivo (37, 49–51, 73), and by electrophysiology (35, 36). LTCCs are a cellular redox sensor (74). The combined findings of a net supernormal free radical production in CA1d compared to CA1i (Quest MRI, Fig. 2), impaired LTCC function in CA1d relative to CA1i (Fig. 3), and reduced calcium-dependent AHP in CA1d (Fig. 3) supports the presence of oxidative stress in CA1d relative to CA1i in the 5xFAD mice. The results of this study are also in line with our previous results that found agreement between outer retinal oxidative stress as measured by conventional methods and Quest MRI (15, 19).

Quest MRI offers advantages compared to other imaging methods that report on oxidative stress in vivo (52, 55, 57, 75–79). These other methods involve exogenously administered contrast agents that bind to free radicals (spin traps) and/or require highly specialized equipment (e.g., ref. 52). Thus, their potential clinical application is likely limited because, for example, of a combination of lack of U.S. Food and Drug Administration approval for the contrast agent, difficulty of the contrast agent crossing blood–brain/retina barriers, concerns about how disease may alter contrast agent pharmacokinetics, and possible modifications to endogenous neuronal redox signaling due to binding of free radicals (55, 57, 75–80). However, Quest MRI does not require injection of a contrast agent or require specialized equipment, and Food and Drug Administration–approved antioxidants can be used for the confirmatory quench. Thus, Quest MRI clearly has high potential for translating promising experimental treatments into clinical practice to help manage the pathogenic oxidative stress aspect of devastating neurodegenerative diseases in individual patients.

CONCLUSIONS

Localized overabundance of free radicals leading to regional oxidative stress is one of the earliest abnormalities in neurodegenerative diseases, including AD, and in neurodevelopmental disorders such as AS (2, 5–16, 21, 81–84). However, conventional imaging biomarkers cannot noninvasively measure region excess in free radical production in vivo. For the first time, we show in 3 different models proof of concept that Quest MRI is a promising method to address this major long-standing problem. Quest MRI has high translational potential and the potential for personalized detection over time of supernormal production of free radicals/antioxidant treatment efficacy in localized regions of the brain in vivo. In addition, Quest MRI appears to be useful for measuring in vivo brain region-specific biomarkers of the earliest relevant dysfunction underlying abnormal behaviors.

Glossary

ACSF

artificial cerebrospinal fluid

AD

Alzheimer disease

AHP

afterhyperpolarization

ALA

α-lipoic acid

AS

Angelman syndrome

CA1d

dorsal hippocampus CA1

CA1i

intermediate hippocampus CA1

HNE

4-hydroxynonenal

LTCC

l-type calcium channel

MB

methylene blue

MEMRI

manganese-enhanced MRI

Quest MRI

quench-assisted MRI

SOD

superoxide dismutase

T_R

repetition time

2-D

2-dimensional

WT

wild type

AUTHOR CONTRIBUTIONS

B. A. Berkowitz, O. Thibault, G. G. Murphy, and B. M. Bennett designed research; R. Roberts, J. Lenning, N. Khetarpal, C. Tran, A. M. Berri, K. Dernay, J. C. Gant, and S. Maimaiti performed research; B. A. Berkowitz and R. H. Podolsky, O. Thibault analyzed data; E. M. Haacke, F. Shafie-Khorassani, and R. H. Podolsky contributed new statistical analysis tools; B. A. Berkowitz, O. Thibault, G. G. Murphy, B. M. Bennett, R. Roberts, J. Lenning, N. Khetarpal, C. Tran, A. M. Berri, and R. H. Podolsky wrote the article; and J. Y. Wu developed software necessary to analyze the data.