APOEε4 Impacts Up-Regulation of Brain-Derived Neurotrophic Factor After a Six-Month Stretch and Aerobic Exercise Intervention in Mild Cognitively Impaired Elderly African Americans: A Pilot Study

Abstract

Possession of the Apolipoprotein E (APOE) gene ε4 allele is the most prevalent genetic risk factor for late onset Alzheimer’s disease (AD). Recent evidence suggests that APOE genotype differentially affects the expression of brain-derived neurotrophic factor (BDNF). Notably, aerobic exercise-induced upregulation of BDNF is well documented; and exercise has been shown to improve cognitive function. As BDNF is known for its role in neuroplasticity and survival, its upregulation is a proposed mechanism for the neuroprotective effects of physical exercise. In this pilot study designed to analyze exercise-induced BDNF upregulation in an understudied population, we examined the effects of ApoEε4 (ε4) carrier status on changes in BDNF expression after a standardized exercise program. African Americans, age 55 years and older, diagnosed with mild cognitive impairment participated in a six-month, supervised program of either stretch (control treatment) or aerobic (experimental treatment) exercise. An exercise-induced increase in VO₂Max was detected only in male participants. BDNF levels in serum were measured using ELISA. Age, screening MMSE scores and baseline measures of BMI, VO₂Max, and BDNF did not differ between ε4 carriers and non-ε4 carriers. A significant association between ε4 status and serum BDNF levels was detected. Non-ε4 carriers showed a significant increase in BDNF levels at the 6 month time point while ε4 carriers did not. We believe we have identified a relationship between the ε4 allele and BDNF response to physiologic adaptation which likely impacts the extent of neuroprotective benefit gained from engagement in physical exercise. Replication of our results with inclusion of diverse racial cohorts, and a no-exercise control group will be necessary to determine the scope of this association in the general population.

1. Introduction

Alzheimer’s disease (AD) is associated with many risk factors, advanced age being the strongest, with a doubling of risk every 5 years past the age of 65. African Americans (AA)s are also at higher risk for AD with estimated prevalence ranging from 14% to 100% higher than Caucasians [1–4]. Despite the disproportionately high incidence of AD within the AA population, this group remains severely underrepresented in clinical trials on AD [5]. Importantly, AAs also suffer from higher burdens of several diseases that increase the risk for AD including type-2 diabetes (T2DM) and cardiovascular disease (CVD).

Along-side investigations into putative treatments for AD, there is growing interest in the mechanisms underlying the health benefits of lifestyle adaptation, especially in light of its demonstrated impact on T2DM and CVD [6, 7]. In particular, aerobic exercise has been shown to improve cognitive function, in both young and elderly populations [8–10]. While exercise-induced upregulation of brain-derived neurotrophic factor (BDNF) is well documented [11–16], there are inconsistencies, with a few studies reporting either no change or lowered serum BDNF levels in exercised subjects [17–19]. These inconsistencies are poorly understood. Regardless, BDNF is critically important for neuronal differentiation, synaptic plasticity and neuron survival [20, 21]. In humans, increased BDNF levels have been linked to increased hippocampal volume and spatial memory performance [22], whereas decreased BDNF is linked to AD and other psychiatric disorders [22–27]. Thus, BDNF upregulation is a supported, proposed mechanism for the cognitive-enhancement triggered by lifestyle adaptation from physical exercise.

Recent evidence suggests that BDNF expression is differentially affected by variants of the apolipoprotein E (APOE) gene [28–30]. The APOE gene is polymorphic with three major alleles: ε2, ε3 and ε4. These gene variants produce proteins with differing physiologic properties and variable associations with health risk [31]. APOE is responsible for trafficking of lipoproteins, fat-soluble vitamins and cholesterol[32]. The ε4 allele has been associated with the development of atherosclerosis [33] and cardiovascular disease [34], both of which increase AD risk. In fact, hetero- and homozygosity for the ε4 allele incurs a 3 fold and 12 fold increased risk for AD, respectively, compared to those homozygous for the ε3 allele [35]. Though AAs are at higher risk of AD compared to Caucasian Americans, whether the APOE gene differentially influence exercise training-induced changes in BDNF levels in older AA, mild cognitively impaired (MCI) subjects has not be examined.

The aim of this pilot study was to examine the effects of a 6-month supervised and standardized exercise program, on serum levels of BDNF, in mild cognitively impaired elderly AAs. We hypothesized that an increase in aerobic capacity would result in a parallel increase in BDNF levels. Although, there was some expectation of baseline differences in serum BDNF levels based on APOE genotype, we anticipated an exercise-induced increase in serum BDNF across all APOE genotypes. Through our analyses, we identified an unexpected association between APOEε 4 status and 6-month changes in serum BDNF levels.

2. Methods and Materials

The protocols used in this investigation were approved by the Howard University Institutional Review Board. As required for studies involving human subjects, all participants completed a signed informed consent form prior to enrollment in the study. Participants were recruited mostly through newspaper advertisement, direct mailing, health fairs and hospital clinics.

2.1 Initial Eligibility Screening

Eighty-nine of the volunteers that were screened for eligibility met initial criteria and were tested for MCI and cardiovascular disease. All eligible participants fulfilled the following inclusion criteria: age >55 years; ability to exercise vigorously without causing harm to self (as determined by the cardiovascular disease screening, treadmill test and history of cardiovascular disease); diagnostic designation as MCI according to Petersen criteria[36] using education adjusted scores; have a committed study partner; be in good general health; and ability to undergo required medical and study related assessments. Based on the initial evaluation of medical history, volunteers with a history of the following conditions were excluded: head trauma, uncontrolled diabetes mellitus and hypertension; current chronic renal, liver, respiratory, musculoskeletal, or neurologic disorders; recent myocardial infarction (within the previous 6 months), unstable angina, and chronic alcohol or drug abuse. Persons using hormone replacement therapy (HRT) or medications that may affect memory (e.g., anticholinergics, sedative hypnotics, narcotics, and antiparkinsonian agents), and those starting new medications within 6 weeks of enrollment, were also excluded from the study.

2.2 MCI Diagnosis

Diagnosis of MCI was made using the following criteria: memory complaints, performance scores on the Wechsler Memory Scale Logical Memory II (adjusted for education), Clinical Dementia Rating Scale (CDR) rating of ≤ 0.5, Modified Hachinski Ischemic Score of ≤ 4, Geriatrics Depression Scale (GDS) rating of < 6, and education adjusted Mini-Mental State Examination (MMSE) scores (adjusted MMSE= raw MMSE−(0.471 × [education −12]) + (0.131 × [age−70]) of between 24–30 [37]. Subjects with probable dementia according to National Institute of Neurological and Communicative Disorders and Stroke/Alzheimer's disease and Related Disorders Association (NINCDS/ADRDA) criteria; and those having memory loss from medical, neurological or psychiatric conditions, medication effects, or delirium were excluded from the study.

2.3 CVD Screening

Qualified volunteers underwent a maximal treadmill exercise test using the Bruce protocol [38] to screen for cardiovascular disease (CVD). Blood pressure, heart rate, and ECG were recorded before the test, at the end of every exercise stage, and every 2 min for 6 min after discontinuing the test. Testing was discontinued when the subject unable to continue the required movement, or CVD signs and symptoms occurred. Signs of CVD included >2-mV ST-segment depression, significant ST segment elevation, extra systole, chest pain, arrhythmias, hypotension or dizziness [39–41]. Those showing signs of CVD were not continued in the study.

2.4 Blood Collection and Processing

Participants were instructed to fast (consume only water), refrain from smoking, and avoid use of any anti-inflammatory medications during the 24-hour period prior to the blood draw for baseline and 6-month testing. Additionally, subjects were instructed to refrain from exercise for 72 hours prior to testing, and to confirm that they had no infection in the week prior to testing. All blood samples were drawn using sterile techniques by personnel trained in phlebotomy. Blood samples were taken between 9:30am and 11:00am in order to minimize any possible circadian variations. Blood was drawn from the median cubital or cephalic vein.

For serum collection, blood samples were incubated at room temperature and allowed to coagulate for 45 minutes. After centrifugation at1000 × g for 10 minutes, in a refrigerated centrifuge, the overlying serum was collected, aliquoted and stored at −80°C until used for BDNF assays. For plasma and buffy coat collection, blood samples were collected in tubes containing anticoagulant. Samples were then centrifuged at 200 × g for 20 minutes (without brake). Plasma and buffy coat layers were then collected into cryotubes and stored at −80°C.

2.5 Dietary Guidelines and Randomization

Study partners and subjects were instructed that subjects should maintain their usual caloric intake during the study period. Participants were requested to maintain an American Heart Association Step 1 diet: consisting of less than 30% of energy from fat, approximately 55% from carbohydrate, approximately 15% from protein, and a cholesterol intake of less than 300 mg/day. Twenty-nine subjects were randomized into stretch (n=12) and aerobic (n=17) exercise groups prior to baseline measurements and genotyping.

2.6 VO₂Max Testing

Maximal oxygen consumption (VO₂Max) was determined at baseline and repeated after subjects completed the 6-month exercise program. VO₂Max and endurance capacity were determined using a modified Bruce protocol and measured by a validated, customized telemetric gas analysis system (K4b² by Cosmed, Chicago Illinois). Discontinuation criteria for this test were similar to those used for the CVD screening test.

2.7 Aerobic Exercise-Training protocol

Each subject's maximum heart rate was inferred from baseline VO₂Max tests. Both aerobic exercise and stretch exercise group subjects participated supervised training 3 days/week. The aerobic exercise protocol complied with the American College of Sports Medicine Guidelines (ACSM) [40], and included treadmill walking or jogging, stair-stepping, and elliptical training. Subjects underwent a warm up period, followed by intensity targeted-training and an appropriate cool-down period. Initial training sessions lasted 20 min at an intensity targeted at 50% VO₂Max. Recordings of exercise heart rate and duration were used to monitor and ensure protocol compliance. Training duration increased by 5 min. each week until subjects attained 40 min. of exercise at 50% VO₂Max. Subsequently, training intensity was increased by 5% VO₂Max weekly until achieving 70% VO₂Max. Aerobic group participants were asked to include an unsupervised 45–60 min. lower intensity walk during the weekend after the first 4–6 weeks of exercise in order to ensure maintenance of acquired fitness levels as well as to maintain motivation and interest. Exercise-training lasted 6 months and took place at the Howard University Hospital, Clinical Research Unit, under the supervision of trained personnel.

2.8 Stretch Exercise-Training Protocol

Stretch activity consisted of static stretch of joints designed to increase flexibility and joint range. Stretch exercise positions produced a slight pull on the muscle without triggering pain and were maintained for 15–30 seconds. Using different positions for a total of about 40 min, each stretch was directed at muscles that are prone to tightness (e.g., hamstrings, hip flexors, calves and chest). Stretches were repeated slowly, 3–5 times on each body side, 3 days/week [40]. To more closely match the increasing intensity and duration (minutes) of aerobic-exercise, stretch training targeted an increasing number of muscles each week up to week 4 and was maintained thereafter. Training began with hamstrings during week-1; hamstring and hip flexors during week-2; hamstring, hip flexors and calves during week-3; and hamstring, hip flexors, calves and chest muscles during week-4 and continued until the end of the 6 month training period.

2.9 Measurement of Serum Levels of BDNF

Serum levels of BDNF were quantitatively determined by ELISA using the human BDNF ELISA kit (Abcam, Cambridge, MA USA). No significant cross-reactivity or interference has been observed using this assay. To minimize assay variance, baseline and 6-month serum samples from a particular participant were measured on the same ELISA plate. A second, separate assay of all baseline samples was also performed on the same ELISA plate, for baseline only comparisons. Protocols were performed according to the manufacturer’s instructions. Each serum sample was tested in duplicate and results were averaged. Intra-assay coefficient of variability (CV) was 8.79, according to the formula [SD/Mean]*100. Serum samples were diluted 1:20 and each ELISA plate contained no-serum controls as well as calibration controls for generation of standard curves. Measurements were expressed in ng/ml after correcting for sample dilution. The optical density of each well was measured using an automated microplate reader (BioTek, Winooski, VT USA).

2.10 APOE Genotyping

Subjects were genotyped for the APOE single nucleotide polymorphisms (SNPs) rs429358 and rs7412 (which define the ε2, ε3 and ε4 alleles). Genomic DNA was extracted from buffy coat samples using standard procedures and quantified using Nanodrop 8000 (ThermoScientific, Wilmington, DE USA). TaqMan (Applied Biosystems) methods were used for genotyping. Manufacturer's protocols and recommendations were followed. Whole genome amplified DNAs were added to SNP reaction mix in 96-well optical reaction plates. Each 96-well plate contained 2 control wells without DNA in order to detect possible contaminations. Endpoint fluorescence reading of TaqMan SNP assays was performed using the ABI Prism 7900HT Sequence Detection System.

2.11 Statistical Analyses

All data were analyzed using SPSS (SPSS Inc., Chicago, IL) and SigmaStat 4.0 (Systat software Inc.,) statistical software. The two-tailed hypothesis was tested with significance set at p ≤ 0.05. The Shapiro-Wilk Test was used to check the assumption of normality and Brown-Forsythe test was used to determine equality of variance. Standard t-tests and Two-Way ANOVAs were performed for analysis of parametric data. Non-parametric group comparisons were analyzed using the Mann-Whitney U Statistical test. Since Generalized Linear Regression Analysis (GLM) allows for the inclusion of additional factors and is most appropriate and often used for unbalanced designs [42, 43], we used GLM to analyze baseline versus 6 month levels of serum BDNF by exercise group, ApoE4 status and gender, while controlling for age. Regression analysis was also performed to determine associations between serum BDNF levels and change in VO₂Max, as well as Baseline BMI and change in serum BDNF.

3. Results

Baseline Characteristics of Study Participants

The detailed demographic characteristics and baseline measurements of participants by exercise groups and ε4 carrier status are shown in Tables 1 and 2, respectively. Twenty-nine participants were randomized into exercise groups and 22 completed six months of exercise training. There were no significant differences in characteristics of age, gender, or education between Aerobic and Stretch participants or between non-ε4 carriers and ε4 carriers. Whereas, baseline BMI was significantly lower in the Aerobic- compared to the Stretch-group, baseline VO₂Max and serum BDNF levels were similar between exercise groups and also ε4 status.

Table 1.

Baseline Characteristics of Participants by Exercise Group.

Characteristics	Participants n=22	Stretch n=9	Aerobic n=13	P value
Age in years	72.0 (7.2)	70.41 (6.3)	73.1 (7.8)	0.393
Female (%)	15 (68.2)	6(66.7)	9 (69.2)	0.900
Body Mass Index, Kg/m2	28.9(5.5)	32.8 (5.8)	26.2(3.4)	0.003
Years of Education	16.1(3.4)	15.0 (3.1)	16.8 (3.5)	0.215
BL VO2Max, mL/min/Kg	23.9 (5.5)	24.8 (8.1)	23.2 (2.9)	0.520
BL BDNF, ng/mL	72.9 (27.9)	68.1 (28.3)	76.3 (28.3)	0.513
APOEε4 Carrier	9(40.9)	2 (22.2)	7 (53.8)	0.367

All data are presented as mean (standard deviation) unless otherwise specified. A total 22 participants completed the 6-month exercise program. BL= baseline; VO₂Max = maximum rate of oxygen uptake. P value indicates significance level of differences between aerobic and stretch groups.

Table 2.

Baseline Characteristics of Randomized Participants by APOEε4 Status

Characteristics	Genotyped Participants n=21	Non- ε4 Carriers n=12	ε 4 Carriers n=9	P value
Age in years	72.5(7.0)	72.9(7.2)	72.0 (7.1)	0.76
Female (%)	14 (66.7)	8 (66.7)	6 (66.7)	1.000
Body Mass Index, Kg/m2	28.3 (4.9)	28.7 (5.6)	27.9 (4.1)	0.716
Years of Education	16.3 (3.3)	16.0 (2.6)	16.7 (4.3)	0.662
BL VO2Max, mL/min/Kg	23.9 (5.7)	23.6 (7.3)	24.2 (2.5)	0.794
BL BDNF, ng/mL	73.3 (28.5)	65.5 (22.4)	83.6 (33.7)	0.154

All data are presented as mean (standard deviation) unless otherwise specified. Twenty-one of the 22 participants, who completed the 6-month exercise program, were successfully genotyped. BL= baseline; VO₂Max = maximum rate of oxygen uptake. P value indicates significance level of differences between ε4 carriers and non- ε4 carriers.

3.1 Distribution of APOE Genotypes

Twenty-one of the 22 participants who completed the 6 month exercise program were successfully genotyped at the APOE locus for the 3 major alleles. Those carrying two copies of the APOEε3 allele constituted the majority (11) of participants, followed by ε4/ε3 heterozygotes (6). There were two ε4/ε4 participants, one ε2/ε4, and one ε2/ε3. One participant’s APOE genotype remained undetermined. Within the stretch exercise group there were six non-ε4 carriers, two ε4 carriers and one undetermined. Within the aerobic exercise group there were six non-ε4 carriers and seven ε4 carriers.

3.2 Changes in VO₂Max

Compared to baseline, Stretch and Aerobic groups showed increases in VO₂Max by 5.62% (P = 0.44) and 11.34% (P = 0.21), respectively, at the 6-month follow-up (Figure 1A.); but the differences were not statistically significant. Similarly, the percent change in VO₂Max between Stretch and Aerobic groups was not significantly different (P= 0.667, t = 0.44) (Figure 1A). Since changes in VO₂Max did not differ between Stretch and Aerobic groups, data from the two groups were combined for assessment of change from baseline by sex and ε4 carrier status. A significant difference in percent change in VO₂Max between men and women (P= 0.024, t = 2.49) became evident (Figure 1B), and remained significant (P= 0.027, t=−2.43) after adjustment for age. Though both men and women were similar with respect to baseline VO₂Max (P= 0.23, t = −0.75), men showed improvements (37.1%) in VO₂Max, compared to a 4.7% decrease in women. In order to determine if the gender effect on exercise-induced changes in VO₂Max was accounted for by changes in BMI, we also evaluated percent change in absolute values of VO₂Max (VO₂MaxAbs) which is not normalized to bodyweight. VO₂MaxAbs also differed significantly between the genders (P=0.01, t = −2.48) but not by exercise groups (P=0.38, t=0.19,). Also, APOEε4 carrier status after adjustment for age showed no effect on change in VO₂Max (relative P=0.857, t= −0.18, (Figure 1C); absolute P=0.32, t= 1.02).

Figure 1.

Box plot of percent (%) change in VO₂Max from baseline values, after 6 months of exercise training (a) between exercise groups, (b) between genders, and (c) between non-ε4 and ε4 carriers. Middle line in box represents the median; lower box bounds the first quartile; upper box bounds the 3^rd quartile. Whiskers represent the 95% confidence interval of the mean. Open circles are outliers from 95% confidence interval. *Significant difference between groups (P=0.024). Non-ε4 = non carriers of the APOE ε 4 allele, ε4 = carrier of the APOEε4 allele.

3.3 Changes in Serum BDNF

While both Stretch and Aerobic groups showed mean increases in serum BDNF (stretch=46.29%; and aerobic=15.12%) (Figure 2A), these increases were not significantly different from baseline values (stretch P=0.24; aerobic P=0.82), and median percent change in serum BDNF were similar for the two groups (stretch= 6.82; aerobic= 0.415; P= 0.950, U = 51). There were no gender effects on changes in serum BDNF levels (P=1.0, U=52.0) (Figure 2B). Linear Regression Analysis also revealed that there was no significant correlation between changes in BDNF and changes VO₂Max (R=0.292, P=0.20). Baseline BMI and change in BMI also did not correlate with change in BDNF levels (R=0.028, P=0.900 and R=0.114, P= 0.623 respectively). Importantly, percent change in serum BDNF levels was significantly different between non-ε4 carriers and ε4 carriers (P= 0.012, U=18.0) (Figure 2C). With adjustment for age, these differences remained statistically significant (Non- ε4 carriers=27.2% versus ε4 carriers =−8.6%; P=0.019). Addition analyses using Generalized Linear Regression analysis was also performed on combined aerobic and stretch groups (Table 3) as well as the aerobics groups only (Table 4). Analyses of only aerobic exercise participants, showed a maintained, significant effect of ε4 carrier status on change in serum BDNF levels after accounting for gender, exercise group and age (P=0.021)

Figure 2.

Box plot of percent (%) change in serum levels of BDNF from baseline values, after 6 months of exercise training (a) between exercise groups, (b) between genders, (c) between non-ε4 and ε4 carriers and (d) between non-ε4 and ε4 carriers only within the aerobic exercise group. Middle line in box represents the median; lower box bounds the first quartile; upper box bounds the 3^rd quartile. Whiskers represent the 95% confidence interval of the mean. Open circles are outliers from 95% confidence interval. *Significant difference between groups (P=0.016 and P=0.021 for C and D respectively). Non-ε4 = non carriers of the APOE ε 4 allele, ε4 = carrier of the APOEε4 allele.

Table 3.

Association of Change in Serum BDNF and ApoEε4 Carrier Status (Combined Stretch and Aerobic Participants)

Parameter	Estimate	Standard Error	t Value	P-Value
Intercept	144.86	101.80	1.42	0.174
ApoE4 Status (C vs. N)	−47.04	17.38	−2.71	0.016
Age	−1.55	1.33	−1.17	0.261
Group (A vs. S)	1.15	17.92	0.06	0.950
Gender (F vs. M)	−6.85	18.97	−0.36	0.723

Generalized Linear Regression analysis (6 months vs. Baseline); ApoE4 Status: C = Carriers, N = Non-carriers; Group: A = Aerobic, S = Stretch; Gender: F = Female, M = Male

Table 4.

Association of Change in Serum BDNF and ApoEε4 Carrier Status (Aerobic Participants Only)

Parameter	Estimate	Standard Error	t Value	P-Value
Intercept	198.06	133.58	1.48	0.172
ApoEε 4 Status (C vs. N)	−59.44	21.37	−2.78	0.021
Age	−2.21	1.63	−1.36	0.208
Gender (F vs. M)	−2.37	24.83	−0.10	0.926

Generalized Linear Regression analysis (6 months vs. Baseline); ApoE4 Status: C = Carriers, N = Non-carriers; Gender: F = Female, M = Male

4. Discussion

Our major finding suggests that APOE genotype was a crucial modulatory factor for BDNF upregulation in our aged cohort of 55–89 year-old AAs with MCI. In addition, changes in serum levels of BDNF did not parallel changes in VO₂Max, suggesting that they occur through distinct physiologic pathways.

VO₂Max, a measure of the maximum rate of oxygen consumption, is regarded as a standard measure of cardiorespiratory fitness. Physiologically, it is determined by several factors including red blood cell concentration, skeletal and cardiac muscle function, and oxygen transport. In our female participants, VO₂Max response to stretch and aerobic exercise did not differ; however, a gender-based difference was noted. Exercise did not induce statistically significant increases in VO₂Max of female subjects. In contrast, male participants showed an increased VO₂Max. This result is a good indication that in older men, stretch exercise regimens may also be advantageous in elevating fitness levels, especially given the propensity for long term adherence. The significant gender-based difference in exercise-induced alteration in VO₂Max denotes a role for sex distinct physiological characteristics in the regulation of this effect. A diminished VO₂Max response to exercise training in women has been previously shown [44]. One suspected candidate for this gender effect is body composition. Previous studies have demonstrated that training-induced changes in VO₂Max is at least partially explained by differences in body composition [45, 46], with increased lean mass positively affecting exercise-induced increase in fitness levels.

In our sample, we did not observe ε4-based differences in VO₂Max or exercise-induced adaptations in VO₂Max from baseline. This observation is not in concordance with that of Thompson and colleagues [47] who demonstrated a significant effect of APOE genotype on exercise-induced increases in VO₂Max, with an enhanced effect in those carrying one ε3 and one ε4 allele compared to those carrying two ε3 alleles. A larger cohort may be necessary to reveal such an effect in our study.

Recently, a few studies have reported that ε4 carriers have lower serum levels of BDNF [28, 29, 48]. This is in accord with our observations of significant APOE gene-based and training-related increases in BDNF levels (P=0.012). At baseline, the ε4 carriers in our study had 21% lower mean level of serum BDNF compared to non-ε4 carriers, though the difference was not statistically significant (p= 0.15). Conversely, a lack of association has been reported by others. [49, 50]. Given its highly protective and reparative role in the brain, it is reasonable to suspect a critical role for BDNF in cognitive decline and neurodegenerative disease. Unfortunately, steady state levels of BDNF can vary drastically, obscuring potential differences. This study supports the use of an inducer of BDNF for revealing differences in BDNF regulation between groups.

BDNF is mainly produced in the brain with the highest concentrations found in the hippocampus [51, 52]. Correlation between brain and serum levels of BDNF has not been established; however, several studies suggest that serum levels of BDNF are positively correlative with levels found in brain [53–56]. Notably, exercise has been established as an up-regulator of BDNF in both animal and human subjects [14, 57–63], as well as in various states of cognitive decline including Alzheimer’s disease [11] for which lowered levels of BDNF have been reported [64–67]. Few studies have demonstrated direct positive correlations between exercise-induced BDNF upregulation and improved cognitive function in human subjects (Kimhy D 2015); however, the neuroprotective impact of BDNF upregulation has been well documented [68–70]. Together, these studies suggest that blood levels of BDNF are a likely reflection of brain levels; and the after-training levels may reflect training effects. In addition, BDNF upregulation through exercise or other means would be expected to have positive effects on brain health, particularly neuroplasticity and neuron survival.

Our data suggests that APOE genotype is an important factor for exercise-induced BDNF upregulation in elderly AA MCI subjects. Replication of our results in larger multi-racial cohorts will be necessary to determine the scope of this association within the general population. Weather this effect is also relevant for younger and cognitively normal populations should also be examined.

Our study is the first to offer evidence in support of a differential effect of APOEε4 carrier status, on exercise training-induced changes in BDNF levels. This result suggests that the ε4 allele of the APOE gene may influence the neuroprotective benefit gained from physical activity. Conversely, one group that examined the effects of exercise-induced BDNF production in ε4 compared to ε3 transgenic APOE-humanized mice [71], reported that the ε4 carrier mice had robust up-regulation of BDNF similarly to the ε3 mice, though the ε4 mice had lower baseline levels of BDNF. However, it remains possible that BDNF response may be under an alternate system of regulation in rodents compared to humans [72].

A few recent studies provide some insight into the mechanisms that may potentially explain our observations in this pilot study. BDNF expression is regulated by nine functional promoters [73]. It is now realized that exercise modulates BDNF expression through epigenetic mechanisms by causing demethylation of at least one BDNF promoter region[74]. For example, one study demonstrated that APOE functions as a transcription factor, and binds to various gene promoter regions including those regulating neurotropic factors [75]. Another study showed that BDNF promoter methylation was significantly higher in AD cases [76], and that APOE variants may differentially bind to and regulate BDNF promoter regions. Furthermore, it has been demonstrated that the APOEε4 variant increases translocation of histone deacetylases in human neurons, resulting in decreased BDNF expression [30]. Whereas, the amount of physiologic adaptation needed to improve BDNF levels is not immediately apparent in this study, collectively, these array of other evidence suggest that APOE may mechanistically explain differences in training-induced changes in BDNF.

4.1 Limitations

This paper presents analyses from a pilot study that was designed to primarily examine the effects of a standardized 6-month aerobic exercise regime on the mechanisms underlying changes in biomarkers of fitness and the enhancement of neuroprotection. Although there was some expectation of possible baseline differences in serum BDNF levels based on APOE genotype, we anticipated an exercise-induced increase in serum BDNF across all APOE genotypes. Therefore, APOE genotype was not incorporated into the study design, and was not factored into the randomization scheme. Consequently, the stretch group consisted of a larger proportion of non-ε4 carriers than the aerobic group. Conceivably, this may have accounted for the higher mean change in serum BDNF in our stretch group compared to the aerobic group. Since this was a highly supervised exercise intervention, and stretch participants not required to maintain exercise diaries or wear activity monitoring devices. As such, there may be some possibility that participates engaged in additional, unapproved and unrecorded exercise activity that may have contributed to our findings of no differences between stretch and aerobic groups. It may also be possible that males engaged in more outside exercise activities compared to women, which could have potentially contributed to gender differences found in our study.

Recruiting and retaining African American research participants is a highly recognized challenge within the scientific community [77, 78], which also impacted our sample size. Only 29 of the 89 volunteer that met initial criteria began the study, and 22 completed our six month exercise intervention. Our relatively small sample size is likely to have contributed to the lack of differences found between stretch and aerobics groups. Still, we believe that this study demonstrates an effect of APOEε4 on exercise-induced BDNF regulation, and that this effect may not be specific to AAs. Given the lack of significant training-induced differences in VO₂Max or BDNF levels between exercise groups, and the absence of a “no-exercise control”, we cannot be certain that the increase in BDNF levels in non-ε4 carriers is definitively an exercise-induced response. However, we note with interest that a lack of direct correlation between changes in VO₂Max and BDNF has been previously reported [79]. Importantly, our study is yet the most rigorous, supervised, and randomized controlled and mechanistic exercise study in mild cognitively impaired older AAs.

4.2 Conclusions

We believe we have identified a relationship between the APOEε4 allele and BDNF response to physical activity in a cohort of AAs with MCI. This observation provides a deeper understanding of the impact of the APOE gene variants on brain health, and the efficacy of exercise as a strategy for the upregulation of neuroprotective biomolecules. Potentially, this finding may have implications for the use of exercise-induced serum BDNF response as a diagnostic measurement for predicting disease progression in preclinical stages of Alzheimer’s disease. Our expectation that aerobic exercise would increase VO₂Max and BDNF levels, in parallel, was unmet. Our subsequent, post hoc analyses led to unexpected results and thus, the generation of a new hypothesis: that the extent of exercise-induced up-regulation of serum levels of BDNF in humans is dependent on APOE genotype. This hypothesis should be tested in new studies specifically designed to examine APOEε4 effects on exercise-induced BDNF up-regulation.

Highlights.

• Older adult men but not women demonstrate exercise-induced increases in VO₂Max.

• Exercise-induced up-regulation of BDNF varies by APOE genotype.

• Muted BDNF response to physiological adaptations may explain ε4-related AD risk.