Curcumin Improves Reversal Learning in Middle Aged Rhesus Monkeys

Abstract

Age-related impairments in cognitive function occur in multiple animal species including humans and non-human primates. Humans and rhesus monkeys exhibit a similar pattern of cognitive decline beginning in middle-age, particularly within the domain of executive function. The prefrontal cortex is the brain region most closely associated with mediating executive function. Previous studies in rhesus monkeys have demonstrated that normal aging leads to an increase in myelin degradation in the prefrontal regions that correlates with cognitive decline. This myelin deterioration is believed to result, at least in part, from the age-related emergence of chronic low levels of inflammation. One therapeutic that may arrest the deleterious effects of neuroinflammation is curcumin (CUR), the primary component of the spice turmeric. CUR has been shown to be a potent anti-inflammatory and anti-oxidant, and has demonstrated the ability to improve performance on tasks for working memory and motor function. In the present study, middle aged monkeys (12–21 years old), were given daily dietary supplementation of 500mg of curcumin or vehicle over a period of 3–4 years. Here we present data from a series of both object and spatial reversal tasks. Compared to vehicle, the CUR group showed enhanced performance on object, but not spatial reversal learning. These findings suggest that curcumin may improve specific aspects of executive function.

Introduction

Age-dependent cognitive decline occurs both in humans and non-human primates (Cahn-Weiner et al., 2000; Fristoe et al., 1997; Lai et al., 1995; Moore et al., 2003, 2006; Moss et al., 1997; Rabbitt and Lowe 2000; Souchay et al., 2000). Declines in the domains of short-term memory, working memory and executive function have been observed as early as middle age (Albert, 1984; Chodosh et al., 2002; Drag and Bieliauskas; 2010; Hara et al., 2012; Kwon et al., 2016; León et al., 2016; Light 1991; Simen et al., 2011; Park and Reuter-Lorenz, 2009; Moore et al., 2003, 2006; Zeamer et al., 2012; Zeamer et al., 2011). Age associated cognitive decline was originally thought to be due to loss of neurons from the cerebral mantle (Brody 1955). Since then, rather than a loss of neurons, significant disruptions of the myelin sheaths have been observed with aging (Peters et al, 1998). These pathologic changes occur predominantly in the frontal lobe and include cytoplasmic ballooning, degeneration of myelin sheaths, as well as impaired and reduced remyelination. Further, this myelin pathology observed in non-human primates is correlated with diminished cognitive ability that is reminiscent of human age-related cognitive decline (Bowley et al., 2010; Makris et al., 2007; Peters and Sethares 2002; Wisco et al., 2008).

Inflammation and oxidative stress also increase with age and may contribute to myelin pathology (De la Fuente and Miquel, 2009; Poliani et al., 2015; Rawji et al 2016; Ruckh et al., 2012). Therapeutics which dampen or mitigate inflammation and oxidative stress may therefore rescue or delay age related cognitive decline. Curcumin (CUR) is a naturally occurring polyphenol, derived from the rhizomes of Curcuma Longa, that has been shown to be a potent antioxidant and anti-inflammatory nutraceutical (Queen & Tollefsbol 2010; Salminen et al., 2008; Sikora et al., 2010a, Sikora et al., 2010b). CUR has been receiving attention as a potential anti-aging therapeutic that may slow or reverse inflammaging and cognitive decline. Numerous studies have demonstrated that administration of CUR improves cognitive and motor performance in rodents and humans alike. (Cox et al., 2015; Nam et al., 2014; Rainey-Smith et al., 2016; Salvioli et al., 2007; Sikora et al., 2010a; Sikora et al., 2010b). Our laboratory has demonstrated beneficial effects of CUR in the middle-aged non-human primate on a task of spatial working memory, but not one of object recognition memory (Moore et al., 2018; 2019). These findings are of particular interest as several studies have demonstrated that the domains of working memory and executive function, are most sensitive to age-related cognitive decline (Arshad et al., 2016; Borella et al., 2014; Funahasi et al, 2017; Johnson et al., 2016; Kwon et al., 2016; McEwen and Morison 2013; Moore et al., 2005; Raz et al., 1997; Toepper et al., 2014; Writ and Hyman 2017). To this end, we hypothesize that CUR would ameliorate deficits in executive function in the aged monkey (Moore et al, 2009; 2017; Bartus et al 1979; Herndon et al., 1997; Moss et al., 1988).

In the present study we assessed the effects of long-term CUR administration on two forms of a reversal learning task - object and spatial reversals. Reversal tasks require subjects to alter their behavioral response following changes in reward contingencies. Studies report that reversal testing can engage both the prefrontal cortex (PFC) and the hippocampus (HPC), dependent on the modality of testing performed (Lai et al., 1995, Jones and Mishkin 1972; Mahut 1971; 1972 Pohl 1971). Age monkeys exhibit impairments in reversal learning, failing to respond to switches in reward contingencies, when compared to young monkeys (Gray et al., 2017; Munger et al., 2017; Lai et al., 1995; Rapp 1990; Bartus et al., 1979). In our longitudinal study of the efficacy of oral administration of CUR we asked the question if CUR would reduce age-related impairments on both object and spatial reversal learning tasks given that both tasks share the feature of set shifting or cognitive flexibility.

Methods

Subjects

Eight middle aged male and female rhesus monkeys (Macaca mulatta) were used in this study (Table 1). The monkeys ranged from 12–21 years of age, equivalent to approximately 36–63 human years (Tigges et al., 1988). All monkeys were obtained from either private vendors or National Primate Research facilities. All monkeys arrived with complete health records and underwent a full medical examination including magnetic resonance imaging to rule out overt brain pathology. All monkeys were healthy at the start of the study. Monkeys were individually housed in cages, within visual and auditory range of the other monkeys of the colony, located in rooms at the Animal Science Center (ASC) at Boston University School of Medicine. Several studies are ongoing within our laboratory, and new cohorts of monkeys are introduced regularly. The decision to individually house monkeys was made to minimize disruption to the consistency and quality of behavioral testing reducing overall study durations. Furthermore, we aimed to reduce the risk of injury that can occur with pair housing of rhesus macaques. Lastly, individualized housing also allowed for greater control of the food intake and enrichment that each monkey received. Monkey rooms were kept at a 12-hour light-dark cycle. The daily diet included Purina Monkey Chow (Purina Mills Inc, St. Louis, MO) with 12–20 biscuits per day (by weight) and fresh fruit and vegetables given after completion of daily testing and following CUR administration. During testing monkeys received small food rewards including fruit and candy. Monkeys had free access to water in their home cage and enrichment in the form of toys and food treats located either directly inside or attached to their cages. Enrichment items were changed regularly. The Boston University ASC is fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC). All procedures were performed following the guidelines set by the National Institutes of Health and the Institute of Laboratory Animal Resources Guide for the Care and Use of Laboratory Animals (2011). This study was approved by the Boston University Institutional Animal Care and Use Committee (IACUC).

Table 1.

Study Subjects

Group	Monkey	Sex	Age
Control (n=4)	AM352	M	12
	AM347c	F	14
	AM350c	F	16
	AM340	F	17
Mean			15
SD			2.4
Treated (n=4)	AM344c	F	16
	AM349c	F	19
	AM312c	M	21
	AM310c	M	21
Mean			19
SD			2.4

Monkeys used in study, listed in order by age

Prior to starting the CUR treatment, all monkeys were behaviorally tested to establish baseline cognitive performance and were then randomly assigned into two groups, receiving either CUR or vehicle control (VEH). It should be noted that while the CUR treatment group was significantly older than the VEH group (t(6)=2.77, p=0.03), no differences were observed in the initial acquisition of any phase of reversal testing. Testers were blind to the treatment condition of the monkeys. The CUR treated group consisted of two males and two females and the VEH treated group consisted of one male and three females. All monkeys received CUR or VEH daily (including weekends) for 3 years prior to testing on the Reversal Learning Tasks and dosing continued during testing. CUR dosage was 500mg per day. The CUR and VEH were provided courtesy of Verdure Sciences (Noblesville, IN). Their formulation is an optimized (solid lipid CUR particles) CUR for enhanced bioavailability, to increase its absorption within the gastrointestinal system and permeability of the blood brain barrier. Both CUR and VEH were mixed with either 150ml of yogurt or Prima-Burger TM (BioServ, Flemington, NJ). Treatments were given immediately following testing and monkeys were observed to verify that the treatment was consumed. If a monkey did not immediately eat the treatment, (e.g. discarding or dropping it into the waste pan of their cage), a second treatment was given at a later time. Monkeys ate the treatments (both VEH and CUR) on more than 98% of the study days.

Cognitive Testing

Apparatus:

All testing was conducted in a Wisconsin General Testing Apparatus, in a darkened room with white noise played on overhead speakers. Monkeys used in this study had already completed a battery of cognitive and motor tests (Moore et al., 2017; 2018) in this apparatus and were thus acclimatized to the testing environment. Briefly, the monkeys and tester were separated by a double set of windows, one fully opaque (between monkey and testing board) the other semi-opaque (between tester and testing board) allowing for one way viewing of the monkey by the tester. A stimulus tray of 3 equally linear spaced wells was located between the two doors. Monkeys had been previously trained to displace grey plaques to uncover the wells and retrieve a food reward (small pieces of candy, fruit or nuts). For reversal testing only the left and right wells were utilized. The opaque door was raised to initiate and terminate trials, while the semi-transparent door remained lowered during each trial allowing the tester to observe the monkey’s behavior.

Reversal Learning Tasks:

Two unique pairs of dissimilar objects were used for object reversals and two identical black plaques were used for spatial reversals (Figure 5). For both tasks, each trial consisted of the simultaneous presentation of the rewarded and unrewarded stimulus over the lateral wells. Thirty trials per day were administered with an inter-trial interval of 15 seconds. Initial learning criterion for the acquisition phase was defined as selection of the rewarded stimulus for 90% (27/30 trials) of trials within a single session of testing. For each task, initial acquisition of the rewarded stimulus was established and then four reversals were administered. The object Reversal Learning Task was administered twice followed by the spatial Reversal Learning Task which was only administered once due to budgetary constraints.

Figure 5. Representative Sample of the Plaques and Testing Board Used for the Object Reversal Learning Task.

For both object and spatial reversals, the day after the initial learning criterion was achieved, testing was conducted with the same rewarded stimulus as the day prior. If the monkey selected the rewarded stimulus for 90% of the first 20 trials (18/20 trials), meeting the reversal criterion, the rewarded and unrewarded stimuli were switched on the 21^st trial, with testing continuing for an additional 20 trials, for a total of 40 trials on the reversal day. If the monkeys did not select the rewarded stimulus for 90% of the first 20 trials, testing continued for 30 trials per day until the reversal criterion was met. The next day, following a reversal, testing continued with the ‘new’ rewarded stimulus for 30 trials/day until the reversal criterion was met (90% correct of the first 20 trials in a single day of testing) triggering another switch between the rewarded and unrewarded stimuli. This pattern of 40 trial reversal days, and 30 trial testing days was performed for a total of 4 reversals of the rewarded and unrewarded stimuli.

Object Reversal Learning Task:

For object reversals, the location of the rewarded object was pseudo-randomized and balanced for both the right and left wells to prevent side-based selection bias. One object from a pair of objects was selected at random to be the rewarded stimulus, this was randomized for each monkey. Testing began with an initial acquisition phase and once criterion was reached on the initial acquisition phase (< 3 errors in 30 trials), the 1^st reversal was administered the following day. Testing continued until reversal criterion was reached and then the 2^nd reversal was administered (the initially rewarded object was once again rewarded). This pattern continued for 2 additional reversals. Once criterion was reached on the last reversal, the entire procedure began again with a new pair of objects until a total of 4 reversals were accomplished with this new pair. Following completion of both pairs of object reversals, spatial reversal testing was administered.

Spatial Reversal Learning Task:

For spatial reversals, both the left and right wells were simultaneously covered with identical plaques. Either the left or right well was pseudo-randomly selected, for each monkey, to be the rewarded side for the initial acquisition. Once the monkey reached criterion on the initial acquisition, reversal testing began the following day. If reversal criterion was met, a reversal was administered (the previous rewarded well now became the unrewarded well and the previously unrewarded well now became the rewarded well) beginning on the 21^st trial. Testing continued until another reversal criterion was reached and then the 2^nd reversal was administered (the initially rewarded side was once again rewarded). This pattern continued for 2 additional reversals.

Data Analysis:

For both the object and spatial Reversal Learning Tasks, the total number of trials and errors to reach criterion for the initial acquisition and for each reversal were calculated. To further assess patterns of learning on these tasks, a learning stage analysis was performed for each phase of testing, excluding trials in which a reversal criterion was met. A similar analysis was performed previously by Lai et al., 1995, in which they demonstrated that a staging analysis of blocks of ten trials is more sensitive at detecting differences between young and old monkeys. Briefly, based on the number of errors made within each block of ten trials, each block of 10 was then characterized as a Stage I, II or III error. Within a block of ten trials, Stage I was defined as 7–10 errors, Stage II was defined as 4–6 errors, and Sage III was defined as 0–3 errors. Stage I errors are indicative of a failure to reverse established stimulus reward contingencies and are analogous to making preservative responses. Stage II errors are characterized as being near chance levels of selection of either presented stimuli, while Stage III errors demonstrate a shift in selection of toward the new positive stimulus. Separate one-way between subjects’ ANOVAs were conducted to compare the effect of treatment on the initial acquisition for both trials and errors, for both object pairs and spatial reversal testing. Comparison of the two groups in trials, errors and each type of stage error were analyzed for each reversal using separate repeated measures ANOVA grouped by treatment with Greenhouse-Geisser corrections when appropriate. Post hoc testing was done by pairwise comparison with Bonferroni corrections when appropriate. All statistics were done using SPSS software (IBM Corp., Armonk, NY).

Results:

Object Reversal Learning Task

Initial Acquisition – 1^st and 2^nd Object Pairs

The total trials and errors to criterion during the acquisition phase (prior to beginning reversals), for the 1^st and 2^nd object pairs, were separately analyzed using one-way between subjects ANOVA comparing the CUR and VEH groups. As shown in Table 2, no significant group differences were observed during the acquisition phase for either object pair (Object Pair 1 Trials: [F(1,6) = 0.11, p=0.75)], Errors: [F(1,6) = 0.01, p=0.92)], Object Pair 2 Trials: [F(1,6) = 0.36, p=0.57)], Errors: [F(1,6) = 0.49, p=0.51)].

Table 2.

Total Trials and Errors to Criterion for Initial Learning for each Reversal Task

Group	Monkey	Object Pair 1 Trials	Object Pair 1 Errors	Object Pair 2 Trials	Object Pair 2 Errors	Spatial Trials	Spatial Errors
Control (n=4)	AM352	102	18	164	41	124	26
	AM347c	141	69	202	62	30	3
	AM350c	56	4	53	10	97	23
	AM340	76	14	96	24	73	17
Mean		94	27	129	34	81	17
SD		37	29	67	22	40	10
Treated (n=4)	AM344c	150	49	105	29	63	7
	AM349c	75	15	51	9	91	29
	AM312c	144	36	146	34	50	10
	AM310c	48	12	120	30	77	13
Mean		104	28	106	26	70	15
SD		51	18	40	11	18	10

Trials required to learn initial positive stimulus prior to onset of reversals, for both rounds of object reversal and spatial reversal testing.

Reversals – 1^st and 2^nd Object Pairs

For each of the two object pairs, the total trials and errors to criterion during each reversal were separately analyzed using a repeated measures ANOVA, with reversal as the within subjects factor, and treatment set as the between subjects factor. For the 1^st object pair, there was no significant effect of reversal or treatment on either trials or errors (Table 3, Figure 1). (Reversal: Trials: Mauchly’s Test: [X²(5) = 18.40, p = 0.003, Greenhouse-Geisser corrected: [F(1.32,7.90) = 1.14, p = 0.34)], Errors: [F(3,18) = 1.40, p = 0.28); Treatment: Trials: [F(1,6) = 0.20, p = 0.67)], Errors: [F(1,6) = .433, p = 0.54)]). No significant interactions between reversal and treatment were observed for either Trials: [F(1.32,7.90) = 1.18, p = 0.35] or Errors: [F(3,18) = 0.93, p = 0.45].

Table 3.

Performance on Object Pair 1 Reversal Testing

Group	Monkey	Rev 1 Trials	Rev 1 Errors	Rev 2 Trials	Rev 2 Errors	Rev 3 Trials	Rev 3 Errors	Rev 4 Trials	Rev 4 Errors
Control (n=4)	AM352	179	79	140	70	290	160	200	68
	AM347c	350	144	410	192	260	99	320	150
	AM350c	110	59	50	28	110	35	80	42
	AM340	80	55	50	34	80	24	80	37
Mean		180	84	163	81	185	80	170	74
SD		121	41	170	76	105	63	115	52
Treated (n=4)	AM344c	566	259	80	36	50	21	110	38
	AM349c	140	60	140	65	80	44	110	39
	AM312c	140	60	50	32	140	50	140	57
	AM310c	80	40	230	102	50	25	80	22
Mean		232	107	125	59	80	35	110	39
SD		223	102	79	32	42	14	24	14

Total trials and errors taken during each reversal for Object Pair 1 testing.

Figure 1. Object Pair 1 Reversal Errors.

Errors made during 1^st object pair reversal testing. Monkeys treated with vehicle control are in blue and monkeys treated with CUR are in orange. A line is drawn through the 50-error mark for visual clarity. No significant group differences were observed in performance.

For the 2^nd object pair, there was no significant effect of reversal or treatment on trials to criterion (Table 4, Figure 2). (Reversal: Trials: [F(3,18) = 1.5, p = 0.24)]; Treatment: Trials: [F(1,6) = 3.13, p = 0.13)]). However, there was a significant effect of reversal [F(3,18) = 4.74, p = 0.01] but not treatment [F(1,6) = 4.60, p = 0.08)] on the totals errors to criterion. Pairwise comparison with Bonferroni correction showed that fewer errors were made by both groups during the 3^rd reversal compared to the 1^st (p < 0.006). No significant interactions between reversal and treatment were observed for either Trials: [F(3,18) = 0.99, p = 0.42] or Errors: [F(3,18) = 2.85, p = 0.07].

Table 4.

Performance on Object Pair 2 Reversal Testing

Group	Monkey	Rev 1 Trials	Rev 1 Errors	Rev 2 Trials	Rev 2 Errors	Rev 3 Trials	Rev 3* Errors	Rev 4 Trials	Rev 4 Errors
Control (n=4)	AM352	140	68	50	32	80	33	230	102
	AM347c	260	128	230	102	260	82	200	110
	AM350c	140	65	140	65	80	41	110	46
	AM340	170	86	80	102	110	35	140	54
Mean		176	86	123	59	133	48	170	78
SD		57	29	79	32	86	23	55	33
Treated (n=4)	AM344c	80	20	20	11	46	11	20	4
	AM349c	140	44	110	47	170	51	110	51
	AM312c	80	35	80	24	80	28	80	35
	AM310c	110	60	110	47	92	45	110	43
Mean		103	40	80	32	97	34	80	32
SD		29	17	42	19	52	18	42	20

Total trials and errors taken during each reversal for Object Pair 2 testing. Repeated measures ANOVA revealed an overall effect of reversal but not treatment [F(3,18) = 4.74, p=.0.01], pairwise comparisons with Bonferroni corrections show that fewer errors were made during the 3^rd reversal compared to the 1^st (p < 0.006).

Figure 2. Object Pair 2 Reversal Errors.

Errors made during 2^ndt object pair reversal testing. Monkeys treated with vehicle control are in blue and monkeys treated with CUR are in orange. A line is drawn through the 50-error mark for visual clarity. Repeated measures ANOVA found no significant effect of treatment, however an overall significant difference was found due to reversal, where less errors were made during the third reversal in comparison to the first (reversal [F(3,18) = 4.74, p=.0.01)], R1 v R3 p < 0.006).

Staging Analysis – 1^st and 2^nd Object Pairs

For both object pairs, each type of stage error (I, II, III) was analyzed using a repeated measures ANOVA, with reversal as the within subjects factor and treatment as the between subjects factor (Figure 4). No significant effects of either reversal or treatment were observed for the 1^st object pair (Reversal Stage I: Mauchly’s Test: [X²(5) = 15.63, p = 0.009, Greenhouse-Geisser corrected: [F(1.92,11.53) = 0.82 p = 0.46)], Stage II: Mauchly’s Test: [X²(5) = 16.23, p = 0.007, Greenhouse-Geisser corrected: [F(1.24,7.45) = 1.21, p = 0.32)], Stage III: [F(3,18) = 0.04, p = 0.99)]. Treatment on Stage I: [F(1,6) = 1.93, p = 0.21)], Stage II: [F(1,6) = 0.05, p = 0.83)], Stage III: [F(1,6) = 0.28 p = 0.62)]). No significant interactions between reversal and treatment were observed for any of the stage errors (Stage I: [F(1.92,11.53) = 1.52, p = 0.26] , Stage II: [F(1.24,7.45) = 0.31, p = 0.64], Stage III: [F(3,18) = 1.97, p = 0.16]).

Figure 4. Staging Errors by Task and Reversal.

Total blocks of trials spent in each stage of error per reversal per subject. Top Row: Object Pair 1 Reversals. Middle Row: Object Pair 2 Reversals, * denotes that treated monkeys made significantly fewer Stage I errors during the 1^st (mean difference: 3.75, p < 0.02) and on the 4^th reversal (mean difference: 2.75, p< 0.03) reversal in comparison to VEH treated monkeys. ** Denotes within group, VEH monkeys made fewer Stage I errors during the 3^rd reversal compared to their performance on the 1^st (mean difference: 5, p < 0.02) and 2^nd reversals (mean difference: 2.5, p < 0.05). Bottom Row: Spatial Reversal. No significant group differences were observed in the Object Pair 1 or Spatial Reversal testing.

In contrast, for the 2^nd object pair, significant differences were found in the total Stage I errors as the main effects of both reversal [F(3,18) = 6.26, p=.0.004)] and treatment, [F(1,6) = 8.31, p=0.03)], were significant. Further, a significant interaction between reversal and treatment was also observed [F(3,18) = 4.54, p=0.02)]. To explore this interaction, an analysis of simple effects followed by pairwise comparisons with Bonferroni corrections was performed. This revealed that within the VEH group, VEH monkeys made fewer Stage I errors on the 3^rd reversal compared to the 1^st reversal (mean difference: 5, p < 0.02) and 2^nd reversal (mean difference: 2.5, p < 0.05). A similar effect of reversal within group was not observed for the CUR monkeys.

To explore the effect of treatment on Stage I errors, an analysis of simple effects followed by pairwise comparisons with Bonferroni corrections was conducted. This revealed that CUR monkeys made fewer Stage I errors on the 1^st (mean difference: 3.75, p < 0.02) and the 4^th reversal (mean difference: 2.75, p< 0.03) in comparison to VEH monkeys.

No significant effects of either reversal or treatment were found for either Stage II (Reversal [F(3,18) = 1.06, p = 0.39)], Treatment [F(1,6) = 2.01, p = 0.20)] or Stage III errors (Reversal [F(3,18) = 1.17, p = 0.35)], Treatment [F(1,6) = 1.36, p = 0.29)]). No significant interactions between reversal and treatment were observed for either Stage II errors: [F(3,18) = 0.41, p = 0.76] or Stage III errors: [F(3,18) = 0.42, p = 0.74]).

Spatial Reversal Learning Task

Initial Acquisition - Spatial

The total trials and errors to criterion during the acquisition phase (prior to beginning reversals) were separately analyzed using one-way ANOVA comparing the CUR and VEH groups. As shown in Table 2, no significant group differences were observed during the acquisition phase of spatial reversals (Trials: [F(1,6) = 0.243, p = 0.640)], Errors: [F(1,6) = 0.125, p = 0.736)]).

Reversals - Spatial

The total trials and errors to criterion during the spatial reversals were separately analyzed using a repeated measures ANOVA, with reversal as the within subjects factor, and treatment as the between subjects factor. There were no significant effects observed for either reversal or treatment (Figure 3 and Table 5) (Reversal: Trials: [F(3,18) = 0.51, p = 0.68)], Errors: [X²(5) = 18.16, p = 0.003, Greenhouse-Geisser corrected: [F(1.49,8.92) = 1.70, p = 0.23)]; Treatment: Trials: [F(1,6) = 0.49, p = 0.51)], Errors: [F(1,6) = 0.45, p = 0.53)]). No significant interactions between reversal and treatment were observed for either Trials: [F(3,18) = 0.58, p = 0.63] or Errors: [F(1.49,8.92) = 0.89, p = 0.41].

Figure 3. Spatial Reversal Errors.

Errors made during spatial reversal testing. Monkeys treated with vehicle control are in blue and monkeys treated with CUR are in orange. A line is drawn through the 50-error mark for visual clarity. No significant group differences were observed in errors made during other reversals.

Table 5.

Performance on Spatial Reversal Testing

Group	Monkey	Rev 1 Trials	Rev 1 Errors	Rev 2 Trials	Rev 2 Errors	Rev 3 Trials	Rev 3 Errors	Rev 4 Trials	Rev 4 Errors
Control (n=4)	AM352	170	120	80	33	170	77	140	57
	AM347c	170	72	200	47	200	66	130	52
	AM350c	80	37	80	20	80	25	80	17
	AM340	110	36	80	18	20	8	70	17
Mean		133	66	110	30	118	44	105	36
SD		45	40	60	13	83	33	35	22
Treated (n=4)	AM344c	74	20	20	9	260	119	170	71
	AM349c	110	48	110	42	50	20	140	29
	AM312c	80	26	80	20	80	28	80	25
	AM310c	110	41	50	16	50	13	80	15
Mean		94	34	65	22	110	45	118	35
SD		19	13	39	14	101	50	45	25

Total trials and errors taken during each reversal for Spatial Reversal testing.

Staging Analysis - Spatial

Each type of stage error (I, II, III) was analyzed using a repeated measures ANOVA, with both reversal and treatment as between subjects factors (Figure 4). No significant effects were found resulting from either reversal or treatment on any type of Stage Error (Reversal: Stage I: [F(3,18) = 1.68, p = 0.21)], Stage II: [F(3,18) = 1.91, p = 0.34)], Stage III: [F(3,18) = 0.61, p = 0.62)]; Treatment: Stage I: [F(1,6) = 0.71, p = 0.43)], Stage II: [F(1,6) = 1.1, p = 0.34)], Stage III: [F(1,6) = 0.40 p = 0.55. No significant interactions between reversal and treatment were observed for any of the stage errors (Stage I: [F(3,18) = 1.32, p = 0.30] , Stage II: [F(3,18) = 0.37, p = 0.77], Stage III: [F(3,18) = 1.70, p = 0.20]).

Discussion:

Summary of Results:

During the 2^nd object pair, a significant effect of reversal but not treatment, with no interactions, was observed, such that monkeys of both groups made fewer errors during the 3^rd reversal compared to the 1^st. To more closely examine the errors made by monkeys in each group a learning stage analysis was preformed and the results demonstrated that long term, daily oral CUR treatment enhanced the performance of middle age rhesus monkeys on object but not spatial reversal learning. Specifically, the learning stage analysis revealed that during reversal on the 2^nd object pair the CUR treated monkeys made fewer Stage I errors than the VEH treated monkeys (Figures 2, 4, Table 4). Within the 2^nd object pair, VEH monkeys show improvement as a group by the 3^rd reversal, making fewer stage I errors on the 3^rd reversal compared to the 1^st and 2^nd. However, this performance level was not maintained on the subsequent reversals (Figure 4) and did not differ from monkeys in the CUR group. Overall, these findings demonstrate that CUR improved performance on object reversal learning and suggest that CUR facilitates frontal-cortical functioning in the middle-aged rhesus macaque compared to monkeys that received vehicle control.

Patterns of Performance

Though a group difference did not reach statistical significance, an interesting pattern emerged in task performance between the two groups of monkeys. Starting with the third reversal of the first object pair, monkeys treated with CUR on average reached a reversal criterion having made fewer than 50 errors during a reversal. This average error rate of approximately 50 errors/reversal by the treated monkeys was maintained during all subsequent object and spatial reversals. VEH monkeys did not reach this 50 errors/reversal rate until the third reversal of the second object pair, four reversals after the CUR monkeys. Additionally, unlike the CUR treated monkeys this error rate was not maintained and only re-established by VEH monkeys until the 2^nd round of spatial reversals, two reversals later (Tables 3–5). This pattern of performance is interesting and two possible explanations for this pattern include 1) the performance by the CUR group may represent a ceiling level in performance with no room for additional improvement and therefore it appears as though they are not significantly different from the VEH monkeys. 2) the monkeys in this study were for the most part in the early middle-age range, a time in which cognitive decline begins and is only mild in severity. As such, it is likely that the monkeys are not significantly impaired on the task to begin with and therefore it is difficult to determine any level of improvement. Further studies to explore these patterns of performance and the possible benefits of curcumin, should include monkeys in older age ranges when cognitive decline is more prevalent and therefore more likely to demonstrate treatment related improvements in performance.

Cortical Regions of Executive Function and Reversal Learning:

Working memory and executive function are both mediated in large part by the prefrontal cortices, and are among the first cognitive domains to exhibit age related decline that first manifests near midlife (Albert, 1984; Albert and Wolfe 1990 Chodosh et al., 2002; Drag and Bieliauskas; 2010; Fisk and Sharpe 2004; Hara et al., 2012; Kwon et al., 2016; Lai et al., 1995; León et al., 2016; Light 1991; Moore et al., 2003, 2006; Rapp 1990; Simen et al., 2011; Wecker et al., 2000; Voytko 1999; Zeamer et al., 2012; Zeamer et al., 2011, Park and Reuter-Lorenz, 2009). Specifically, in the rhesus monkey, it has been demonstrated that aging results in deficits in abstraction, set shifting, working memory and rule learning. In our laboratory, we have demonstrated that both middle-aged and aged monkeys are impaired on our Category Set Shifting Task (CSST) and the Delayed Recognition Span Task (DRST) both of which assess PFC function (Moore et al., 2003, 2006). Both, middle-age and aged monkeys require more trials, while also making more errors in learning these tasks compared to young controls. Furthermore, middle and aged monkeys have a greater tendency to perseverate in their response patterns than young monkeys. This pattern of impairment has also been observed by other groups. For example, Bartus et al., 1979 and Rapp 1990 have both demonstrated that middle-aged and aged monkeys have impairments in shifting response patterns, establishing new response patterns, demonstrated preservative response patterns and finally were more susceptible to interference on Reversal Learning Tasks. More recently Gray et al., 2017 and Munger et al., 2017 reported that aged monkeys and marmosets were impaired in attentional updating, response shifting and committed more perseverative errors than young monkeys on reversal tasks.

Our previously published data from our CUR supplemented cohort demonstrated that CUR improved performance on the Delayed Recognition Span Task spatial (DRSTsp), but had no significant effect on the DNMS (Moore et al., 2017). These results led us to hypothesize that given the age group of our monkeys (mean age 17 years), which are likely in the earliest stages of age-related cognitive decline, we would best observe the effects of CUR supplementation on tasks necessitating frontal-cortical activation.

Reversal learning tasks have been used extensively across species in a multitude of modalities including object, spatial, auditory and olfactory reversal paradigms to assess cognitive flexibility, perseveration and rule learning. The specific cortical region most involved in mediating performance on reversal learning tasks however appears to depend on the modality of stimuli used in testing. For example, lesions within the orbitofrontal cortex (OFC) have been shown to cause impairments in reversal tasks using visual and auditory stimuli, while medial PFC lesions result in reversal learning impairments when using spatially orientated stimuli (Meunier et al., 1991; Shaw et al., 2013; Young and Shapiro 2009). Further, Iverson and Mishkin 1970 demonstrated that lesions of the inferior frontal convexity can impair auditory frequency differentiation, and increase perseverative interference during object and spatial reversal learning. Jones and Mishkin 1972 show that lesions in the OFC or a lesion encompassing the temporal pole and amygdala (TPA) resulted in impairments in both object and spatial reversals, while hippocampal lesions resulted in only in spatial reversal impairments. Importantly they showed that lesions to the OFC resulted in greater perseverative or stage I errors during object reversals than either TPA or fusiform-hippocampal gyrus and hippocampus (FHH) lesions. Further, Jones and Mishkin showed a double dissociation on performance during object reversals where monkeys with OFC lesions made greater Stage I errors, while TPA lesioned monkeys made greater Stage II errors, the FHH group did not differ from controls at any stage. Mahut 1971 and 1972 which reversed the order of testing between spatial and object reversals in comparison to Jones and Mishkin 1972 yielded similar results despite the difference in testing order. Lai et al., 1995 report that aged non-human primates make more errors on spatial but not object reversals in comparison to young monkeys. Furthermore, Lai et al., 1995 report that the aged monkeys made increased Stage I on both spatial and object reversal tasks in comparison to young monkeys. More recent studies performed by Rudebeck et al., 2013, show that excitotoxic lesioning of OFC neurons alone does not cause object reversal deficits, but further they show that lesions encompassing nearby fiber tracts are needed to induce object reversal deficits. These data from Rudebeck et al., 2013 highlight the difficulty in attributing reversal deficits to a singular region. From these studies it is clear that reversal tasks engage multiple brain regions, however aging in monkeys appears to cause similar deficits on reversal tasks to those observed in monkeys with lesions in the OFC, such that aged monkeys exhibit greater perseverative responding than young monkeys during object reversals. While it is difficult to pinpoint a singular region as definitively benefitting from CUR supplementation, the lessening of perseverative errors in object reversals, our previously published data on improved DRSTsp performance (Moore et al., 2017), and the age of the monkeys who would be in the earliest stages of cognitive decline, combined would suggest that CUR supplementation may be improving frontal cortical function. While the PFC is likely being affected by CUR supplementation in our model, other cortical regions are likely also affected. Future studies should include administration of additional cognitive tasks that assess a wider range of cognitive functions, such as attention, memory and visuospatial skills. In addition, histological and biochemical analysis of brain tissues from various regions from treated monkeys should be conducted to provide additional insight into the mechanism of action of CUR in specific brain regions and white matter tracts.

Aging and Inflammation

The PFC is associated with age-related myelin pathology and decreased white matter volume (Wisco et al., 2008; Grady 1998; Makris et al., 2007; Peters et al., 1994; Raz et al., 1997; West 1996). In addition, there is a loss of myelin integrity as measured by decreased fractional anisotropy in PFC white matter (Makris et al., 2007). This age-related increase in myelin damage can have downstream effects as myelin debris can inhibit oligodendrocyte differentiation and remyelination (Kotter et al., 2006). Finally, there is evidence of a rise in age related inflammation within the frontal white matter demonstrated by increased phagocytic and ameboid microglia, (Shobin et al., 2017). Together, these findings suggest an age-associated vulnerability of the PFC to alterations in the myelin sheath. It is hypothesized that changes in white matter and myelin are associated with age-related immunosenescence that contributes to a chronic neuroinflammation (Di Benedetto et al., 2017; Ownby 2010). In support of this idea, is evidence of an age-related increase in microglial proinflammatory activation (Shobin et al., 2017). Further support for this notion comes from data showing increased circulating proinflammatory cytokines that negatively impact the CNS, specifically the myelin and white matter (Cornejo and von Bernhardi 2016; Robillard et al., 2016; Safaiyan et al., 2016; Xie et al., 2013).

Aging is also associated with a decrease in intrinsic antioxidant and anti-inflammatory capability, with a shift towards proinflammatory immune response (Ye and Johnson 2001, London et al., 2013). Microglia, the macrophages of the brain, have been demonstrated to have a loss in phagocytic capability, a priming toward pro-inflammatory activation, and an increased density within the frontal white matter with age (Plowden et al., 2004; Safaiyan et al., 2016; Shobin et al., 2017). The age-related increases in white matter pathology, inflammation and immune activation suggests that a cycle of inflammation induced myelin damage leads to a persistent and inefficient immune response. This inflammaging effect can then further exacerbate myelin pathology and may drive the cognitive decline observed in aging. Finally, changes in myelin ultrastructure, increased microglial activation, and increased markers of inflammation are all strongly correlated with age-related cognitive decline and therefore are potential targets for anti-inflammatory compounds, such as CUR (Simen et al., 2011). Many studies have demonstrated that CUR is a potent anti-inflammatory agent (Lee et al., 2007; Nahar et al., 2015; Parada et al., 2015; Tegenge et al., 2014; Yang et al., 2014) observed to inhibit of the activity the NF-κB transcription factor, a key regulator of the inflammatory process which can promote the production of many pro-inflammatory cytokines, such as TNF-α, IL-1β, and IL-6 (Jin et al., 2007; Kure et al., 2017; Lee et al., 2007). Together, these actions of CUR may have downstream effects that result in improved performance on our Reversal Learning Task in monkeys receiving long-term daily administration of CUR. Future studies, examining brain tissue and CSF from these monkeys, will help to elucidate the potential mechanisms of CUR in these monkeys.

Study Limitations

Human clinical trials of CUR treatment with aged individuals, show that control subjects exhibit noticeable decline in performance on the Montreal Cognitive Assessment in comparison to those who received CUR (Rainey-Smith et al., 2016). These data suggest that while CUR is not improving performance, it is perhaps delaying or mitigating aging effects on cognition. A key limitation of this study, in addition to the small group size, is that perhaps the cohort of monkeys tested have not developed significant cognitive impairment such that a similar mitigation of decline by CUR treatment can be easily observed. Further, glaring improvements following supplementation may be partially masked by the lack of pronounced decline in cognitive ability in the cohort tested. For this reason, we believe that a thorough understanding of CUR effects would be more readily assessed in longitudinal studies with treatment beginning in middle age and continuing through to advanced old age (15 to 25 years+). A longitudinal study as described, may provide evidence that CUR has a beneficial role in delaying age-related cognitive decline. Further studies in which CUR treatment begins an older cohort of monkeys, which likely will have a poorer baseline cognitive ability, would address if CUR treatment can reverse age related cognitive decline. This current study only tested a subset monkeys from our larger study of CUR, as several animals in the larger study had already been euthanized when the spatial and object reversals were added to the testing paradigm. These more “frontal” based tasks were added for the 2^nd cohort based on the findings that curcumin improved performance on the DRSTsp task, a task of PFC function. While this study did provide valuable insight in the impact of curcumin on reversal learning, a limitation of this study, lies within our inability to definitively state that the positive effects of CUR are on a specific brain region. Though given the ages of the monkeys, our previous reported results of DRSTsp, and our findings here would suggest CUR is affecting, at least in part, the frontal cortex. Future studies to further explore region specific effects of curcumin could include administration of additional frontal cortical dependent task such as the Conceptual Set Shifting Task (Moore et al, 2005). In addition, histological studies quantifying markers of inflammation and oxidative damage in the PFC would provide additional insight into regional effects of curcumin.

A second limitation to this study was that while both groups can be considered middle aged it should be noted that the CUR treatment group was significantly older than our VEH group. This did not impact the initial acquisition of any of the tasks and as age is a known to negatively impact reversal learning, it is plausible that any effect of age on task performance would act as a disadvantage to the CUR treatment group. That the CUR treated group made fewer Stage I then younger VEH controls provides even further evidence of the beneficial effects of curcumin.

Conclusions and Future Directions

The present study combined with previous investigations by our laboratory have demonstrated that long-term daily administration of CUR to middle-aged rhesus monkeys enhances performance on tasks of spatial working memory and motor function (Moore et al., 2017; 2018). Here we show that CUR treated monkeys make fewer perseverative type errors in contrast to monkeys given VEH and it is feasible that this may have contributed to the enhancement in cognitive performance observed earlier on tests such as the DRSTsp. We did find that VEH monkeys only show improvements relative to themselves during the reversal testing of the 2^nd object pair. The CUR group quickly reaches and maintains a steady state of errors made during reversals, in contrast the VEH group does not, and because the CUR group are making fewer errors a significant improvement within group may be difficult to detect. The improvement in performance by middle aged monkeys on object but not spatial reversals tasks and improvement in DRSTsp but not DNMS, taken together support the hypothesis that CUR supplementation is likely affecting the frontal cortex in our model. However, the precise role of CUR in delaying and/or improving age-associated cognitive and motor decline requires further exploration. The demonstration that CUR supplementation has a beneficial role on cognition in humans and monkeys offers a nutritional approach to reducing age-related cognitive decline