Dietary DHA during development affects depression-like behaviors and biomarkers that emerge after puberty in adolescent rats.

DHA is an important omega-3 PUFA that confers neurodevelopmental benefits. Sufficient omega-3 PUFA intake has been associated with improved mood-associated measures in adult humans and rodents, but it is unknown whether DHA specifically influences these benefits. Furthermore, the extent to which development and puberty interact with the maternal diet and the offspring diet to affect mood-related behaviors in adolescence is poorly understood. We sought to address these questions by 1) feeding pregnant rats with diets sufficient or deficient in DHA during gestation and lactation; 2) weaning their male offspring to diets that were sufficient or deficient in DHA; and 3) assessing depression-related behaviors (forced swim test), plasma biomarkers [brain-derived neurotrophic factor (BDNF), serotonin, and melatonin], and brain biomarkers (BDNF) in the offspring before and after puberty. No dietary effects were detected when the offspring were evaluated before puberty. In contrast, after puberty depressive-like behavior and its associated biomarkers were worse in DHA-deficient offspring compared with animals with sufficient levels of DHA. The findings reported here suggest that maintaining sufficient DHA levels throughout development (both pre- and postweaning) may increase resiliency to emotional stressors and decrease susceptibility to mood disorders that commonly arise during adolescence.

according to the Guide for the Care and Use of Laboratory Animals (8th edition, National Research Council).

Experimental design
The experimental design of the study is illustrated in Fig. 1 . Timed-pregnant Sprague-Dawley rats were obtained from Harlan Laboratories (Indianapolis, IN) at embryonic day 4 and fed either a DHA-defi cient or a DHA-suffi cient diet. Shortly after parturition, DHA is acquired during gestation via maternal stores by way of placental transfer, and it is acquired after birth from mother's milk or formula. As DHA is acquired primarily through the diet after birth, a decline in preformed DHA intake typically occurs immediately following weaning. This circumstance creates some developmental risk because weaning is a critical period for synaptogenesis and myelination. Thus, suboptimal DHA intake may infl uence synapse pruning, myelination, inhibitory synaptogenesis, and so forth, during this time. Indeed, DHA defi ciency throughout development leads to decreased DHA levels in neural tissue and is associated with defi cits in psychomotor development ( 21 ), problem solving ( 22 ), reading skills ( 23 ), visual acuity ( 24 ), and attention ( 25 ). Furthermore, emerging clinical data support a link between suffi cient DHA intake with a reduction in the incidence and/or symptomatic relief of adolescent depression ( 5,(26)(27)(28)(29), and there exists an abundance of clinical evidence supporting a positive role for the omega-3 PUFA EPA in the treatment of adult depression ( 30 ). However, the studies examining the importance of DHA, in particular, are limited in scope and interpretation, and no longitudinal studies have been reported to date. It is therefore unknown whether DHA suffi ciency during development can help to establish optimal emotional resiliency to depressive mood states.
The work reported here sought to examine whether DHA suffi ciency throughout development (gestation, preweaning, prepubescence, and adolescence) in rats could positively affect behavioral measures (forced swim test [FST]) and biomarkers (serotonin, melatonin, and brain-derived neurotrophic factor [BDNF]) associated with mood before or after puberty. The periods of dietary DHA supplementation were designed to include the majority of neural development as well as the transition from adolescence to adulthood in the rat (8-9 weeks). The study design also included the investigation of potential effects of DHA removal at weaning to model the postweaning drop in DHA intake that occurs in some infants. Furthermore, we wanted to determine whether feeding a postweaning diet rich in DHA to offspring weaned from DHA-defi cient dams could affect these measures. Overall, the results described here suggest that DHA supplementation is required throughout development (pre-and postweaning) to positively affect mood-related behavioral measures and biomarkers after puberty in adolescent rats.

Animals
Sprague-Dawley rats were housed individually (dams during gestation and during lactation with offspring) or in pairs (males after weaning) in polycarbonate cages in a temperature and humidity controlled environment, on a 12 h:12 h light:dark cycle, with chow and water ad libitum. All animal protocols were approved by the Institutional Animal Care and Use Committee at the University of Colorado (Boulder, CO) and were performed

FST
Animals were acclimated to the test room overnight prior to testing. The test was performed over two consecutive days. On day 1, the animals were acclimated to the test (0800 h to 1300 h) by placing them in a Plexiglas cylindrical container (45 cm × 20 cm; Stoelting Co., Wood Dale, IL) fi lled with 30 cm of fresh water (25°C) for 15 min, after which they were toweled dried and returned to their home cage. On day 2 (24 h later), the test was performed for a total swim time of 5 min, after which the rats were toweled dried and returned to their home cage. Both trials were recorded by a digital video camera secured to the ceiling above the cylinders. Total time swimming, immobile, and climbing, and number of dives were measured post hoc by an experimenter blind to the group assignments. Total time immobile was measured in real-time by behavioral software (ANY-maze, Stoelting Co.), and confi rmed by the post hoc analysis. Swimming was defi ned as movement of the forelimbs and hind limbs that did not break the surface of the water. Immobility was defi ned as absence of any movement except for slight movements necessary for the animal to keep its head above water. Climbing was defi ned as rapid movement of the forelimbs that did not break the surface of the water. Dives were counted when the animal submerged its head in an effort to fi nd an escape below the surface of the water.

Biomarker analysis
Testosterone. Plasma testosterone concentration was determined via a competitive ELISA (catalog number EIA-1559; DRG International Inc., Mountainside, NJ) according to the manufacturer's specifi cations. Plasma samples were run neat in triplicate. All samples were analyzed in one assay. The intra-assay variance was 7.5%. The limit of detection for this assay is 0.083 ng/ml.
Serotonin. Plasma serotonin concentration was assayed via a competitive ELISA (catalog number RE59121; IBL International Inc., Hamburg, Germany). Plasma samples were centrifuged for 2 min at 10,000 g to ensure a platelet-free sample, and a 100 µl aliquot of the supernatant was taken for the assay. The manufacturer's "Sample B" protocol (for platelet-free plasma) was followed as specifi ed. All samples were analyzed in one assay. The intra-assay variance was 9.6%. The limit of detection for this assay is 0.014 ng/ml.

Melatonin. Plasma melatonin concentration was analyzed via
a competitive ELISA (catalog number RE54021; IBL International on postnatal day (P) 1, pups were sexed, culled to liters of 10, and matched for sex and cross-fostered equally among the dams fed similar diets. At P16, male offspring were weaned from their mothers and fed either a DHA-defi cient or a DHA-suffi cient diet. We chose to examine only the male offspring in this study because including female rats would require more than twice the number of animals placed on study in order to provide suffi cient statistical power given that the day of estrous for each female would be an additional cofactor. This design resulted in four groups that will be referred to as follows: 1 ) defi cient (DHA-defi cient maternal and postweaning diets), 2 ) preweaning suffi cient (DHA-suffi cient maternal diet and DHA-defi cient postweaning diet), 3 ) postweaning suffi cient (DHA-defi cient maternal diet and DHA-suffi cient postweaning diet), and 4 ) suffi cient (DHAsuffi cient maternal and postweaning diets). Offspring were tested in the FST in two separate cohorts at either P39-P40 or P59-P60, but not at both time points. Thirty minutes following the fi nal swim test, each animal was euthanized via decapitation. Whole brain was extracted and frozen in cold isopentane (approximately Ϫ 30°C) and stored at Ϫ 80°C (brains were not perfused with saline or buffer, but rather frozen directly after decapitation and extraction). Trunk blood was collected into K2-EDTA-coated vacutainers (BD Biosciences), inverted to mix, and centrifuged at 1,000 g for 15 min, and the resulting plasma aliquots were stored at Ϫ 80°C until assayed. The cell pellet was then mixed with a 5-fold volume of saline, centrifuged at 1,000 g for 5 min (wash repeated twice), and the resulting pellet containing the red blood cell (RBC) fraction was stored at -80°C until assayed. After necropsy, brains were thawed on ice and bisected sagittally down the longitudinal cerebral fi ssure and cerebellar vermis; one hemisection was used for FA analysis, and one hemisection was regionally dissected and stored at -80°C until further processing.

FA analysis
Tissues (RBCs, plasma, and brain) from the current study were analyzed for FA composition by gas chromatography. Sample preparation was optimized for each tissue matrix. Briefl y, the plasma was aliquoted and dried under evaporative nitrogen; brain tissues were lyophilized, homogenized, and weighed; and Timed pregnant females (embryonic day 4) were provided either a base diet containing no DHA or an equivalent diet containing DHA as ‫ف‬ 1% of the total FAs throughout the gestation and lactation periods. At weaning, male offspring were either provided the base diet or the DHAcontaining diet. Behavior was assessed in the FST before and after puberty (P40 and P60, respectively) in separate cohorts (each animal was only tested once). Tissue was taken at both time points for analysis of biomarkers for mood. t -test with Welch's correction ( t (85) = 11.34, P < 0.0001). There were no signifi cant group differences at either P40 or P60 (data not shown). Therefore, P40 and P60 were likely within the pre-/peripubertal and postpubertal time periods, respectively.

Body and brain weight
There were no differences in body weight between dams on either the DHA-suffi cient or DHA-defi cient diet after parturition or after weaning, and no differences were detected in whole brain weight at the time of euthanization following weaning.

FA analysis
Diet affected the DHA status of peripheral and central tissues in the dam at the time of euthanization (P21; 39 days on diet). Suffi cient dams had higher DHA content in RBCs (4.39 ± 0.09% of total FAs, N = 8) than defi cient dams (1.96 ± 0.11%, N = 9; t (15) = 16.80, P < 0.0001), and higher DHA content in brain (13.50 ± 0.20%, N = 8) than deficient dams (12.59 ± 0.17%, N = 9; t (15) = 3.50, P = 0.0016) as determined by an unpaired one-tailed t -test. In the offspring, supplementation of the maternal or postweaning diets with DHA had profound effects on the FA composition of central and peripheral tissues both before and after puberty. The FA composition of brain, RBCs, plasma, and their statistical analyses are detailed in Tables 2 , 3 , and 4 , respectively. In addition, the data for the primary PUFAs of interest within brain tissue [DHA, 22:6n3; arachidonic acid (ARA, 20:4n6); and docosapentaenoic acid (DPAn6; 22:5n6)] are graphed in Fig. 2 . As found in previous studies, dietary supplementation of DHA during any developmental time period (maternal, postweaning, or throughout) signifi cantly increased brain DHA content both before and after puberty relative to defi cient controls. Furthermore, at P40 suffi cient animals had significantly higher brain DHA content (14.80% of total FAs) than preweaning (12.96%) and postweaning (13.56%) suffi cient cohorts, both of which had higher brain DHA content than the defi cient animals (9.50%). At P60, brain DHA content was similar in the suffi cient and postweaning suffi cient groups (14.01% and 13.51%, respectively), and both groups had signifi cantly greater brain DHA content than both the preweaning suffi cient and defi cient groups (11.89% and 9.71%, respectively).
With the other brain PUFAs of interest, again similar to previous studies, there were differences in ARA and DPAn6 brain content that tracked inversely to the change in brain DHA levels ( Table 2 ). For example, ARA and DPAn6 were Inc.) according to the manufacturer's specifi cations. All samples were analyzed in one assay. The intra-assay variance was 8.2%. The limit of detection for this assay is 1.6 pg/ml. BDNF. Plasma and brain BDNF was determined via a two-site sandwich ELISA (catalog number TB257; Promega Corp., Madison, WI) according to the manufacturer's specifi cations with modifi cations as determined via empirical testing on similar samples. Plasma samples were centrifuged for 2 min at 10,000 g to ensure a platelet-free sample, and a 100 µl aliquot of the supernatant was taken for the assay. Brain tissue samples were homogenized in buffer [100 mM Tris-HCl, 400 mM NaCl, 4 mM EDTA, 0.05% sodium azide, 0.2% Triton-X, 2% BSA (fraction V), protease inhibitor cocktail (Roche Cat# 539137; 1:100 dilution), and 0.1 mM PMSF]. Buffers containing BSA have been shown to improve BDNF recovery from brain tissue samples ( 32 ). Homogenization buffer (10× volume by weight of tissue) was added to each sample, and tissue was homogenized with an ultrasonic tissue disruptor (Misonix XL2000) on setting 4 for 30 s. Homogenates were cleared at 16,000 g for 30 min at 4°C, and 100 µl aliquots of the supernatant were stored at Ϫ 80°C until assayed. Plasma and brain samples were treated with 4 µl of 1.0 M HCl for 15 min, neutralized with 4 µl of 1.0 M NaOH, and diluted with 392 µl sample buffer (1:5) on day 1 of the assay. Changes to the manufacturer's protocol included the following: BDNF standard curve was serially diluted from provided stock standard prior to addition to plate (rather than in plate), initial sample incubation with anti-BDNF coated plate was performed at 4°C for 24 h with no shaking, and incubation with secondary antibody was done at 4°C for 20 h with no shaking. All other portions of the protocol were completed according to the manufacturer's recommendations. Brain BDNF content was normalized to wet weight of each tissue sample because the specifi c homogenization buffer used precluded the ability to measure protein levels in the homogenates due to interference by BSA with standard protein assays. The intra-assay variance was 6.5%, and the interassay variance was 9.7%. The limit of detection for this assay is 15.6 pg/ml.

Statistical analysis
Data were analyzed with SPSS, version 16.0 (IBM, Armonk, NY) and visualized with Prism version 5.4 (GraphPad Inc., La Jolla, CA). Main effects were detected via one-way ANOVA or multivariate ANOVA where ␣ ( P ) levels less than 0.05 were considered statistically different. In the case of a main effect, ANOVA analysis was followed by Tukey's post hoc tests for pairwise comparisons to determine signifi cant differences between groups. Data from the dams (two groups) were analyzed by the Student's t -test with Welch's correction in cases where Levene's test for equality of variances was signifi cant. Correlations were detected via two-tailed Pearson correlation calculations where P levels less than 0.05 were considered statistically signifi cant. Data sets exhibiting skewed distribution frequencies were transformed with log 10 or square-root calculations to improve their frequency distributions prior to analysis. Outlier detection was conducted using Grubb's test prior to any other analyses. All data are expressed as the group mean ± SEM.

Plasma testosterone
Mean plasma testosterone concentration was lower in animals at P40 (0.67 ± 0.06 ng/ml) than those at P60 (2.66 ± 0.16 ng/ml) as determined by an unpaired one-tailed and plasma ( Table 4 ), and this may in part refl ect retroconversion from DHA.

Depression-like behavior
Before puberty (P40), there were no differences among the four groups in measures of passive behaviors (time immobile; Tukey's post hoc pairwise analysis indicated that completely suffi cient animals spent signifi cantly less time immobile ( P = 0.026) and more time climbing ( P < 0.001) and performed more dives ( P = 0.045) than the animals defi cient in dietary DHA throughout development. In addition, preweaning suffi cient animals spent signifi cantly more time climbing ( P = 0.007) than defi cient cohorts, yet were no different from defi cient animals in measures of immobility or dives. Postweaning suffi cient (preweaning defi cient) offspring highest in the brains from animals that were not supplemented with DHA during development, and lowest in those supplemented throughout development. DPAn6 brain content was quicker to change in response to postweaning diet than was DHA, and ARA levels responded at rates somewhat in-between those of DHA and DPAn6. As expected, these changes led to a corresponding decrease in the omega-6 to omega-3 (n6:n3) ratio with improved brain DHA status, while total PUFA content of brain was largely unaffected.
Dietary supplementation with DHA also changed the FA profi les of RBCs ( Table 3 ) and plasma ( Table 4 ) in a fashion similar to changes seen in brain tissue both before and after puberty. However, changes in FA composition following postweaning dietary change (defi cient to suffi cient, or vice versa) were more rapid in both tissues than in brain. For example, the 24 day period from weaning (P16) to prepuberty (P40) was long enough to allow the DHA levels in RBCs from preweaning defi cient (postweaning suffi cient) offspring to reach those of animals maintained on pre-and postweaning DHA-supplemented diets (5.04% vs. 5.38% of total FAs, respectively). EPA was found in very low levels in brain (n.d. to 0.07% total FAs), RBCs (0.06 to 0.21% total FAs), and plasma (0.09 to 0.30% total FAs). However, dietary DHA did cause a small but signifi cant increase in EPA levels in RBCs ( Table 3 )  Data are presented as the mean percent of total FAs ± the SEM. The degrees of freedom and their errors for the analysis of each FA at age P40 were 3 and 68, and at age P60 were 3 and 67, respectively. Signifi cant differences ( P < 0.05) between groups were determined with Tukey's post hoc test and are designated by a lack of common superscripts across each row.
Plasma serotonin was positively correlated to brain DHA levels at P60 (0.346, P = 0.002), but not at P40.
For plasma melatonin, one-way ANOVA identifi ed a main effect of group before puberty at P40 ( Fig. 4E ; F (3,56) = 4.54, P = 0.006) and after puberty at P60 ( Fig. 4F ; F (3,58) = 11.77, P < 0.0001). Before puberty, Tukey's post hoc analy sis showed that only offspring that received DHA throughout development had signifi cantly increased plasma concentrations of melatonin ( P = 0.015). Preweaning suffi cient animals had melatonin concentrations that were approaching those of suffi cient animals, but it appeared that DHA defi ciency during the 24 day window between weaning and P40 was long enough to impair the effect of preweaning dietary DHA in these animals as there was only a trend for signifi cance as compared with defi cient controls ( P = 0.088). After puberty, both groups that received postweaning dietary DHA, regardless of maternal diet, exhibited signifi cantly increased plasma melatonin concentrations as compared with defi cient controls (suffi cient, P < 0.0001; postweaning suffi cient, P = 0.001). Dietary supplementation with DHA after weaning was necessary to enhance plasma melatonin concentrations after puberty, and DHA supplementation during only gestation and lactation was ineffective in modulating plasma melatonin after puberty. Plasma melatonin was positively correlated to brain DHA levels at P40 (0.418, P < 0.001) and at P60 (0.509, P < 0.001).
exhibited behavior similar to the completely defi cient controls. At P60, brain DHA level was negatively correlated to immobility time ( Ϫ 0.293, P = 0.016) and positively correlated to climbing time (0.376, P = 0.002) and number of dives (0.224, P = 0.036). No correlations were seen at P40.

Plasma biomarkers
For plasma BDNF, one-way ANOVA did not identify a main effect of group at P40 ( Fig. 4A ) but did at P60 ( Fig. 4B ; F (3,59) = 2.93, P = 0.041). Tukey's post hoc analysis determined that only animals receiving DHA supplementation throughout development had signifi cantly higher plasma BDNF concentrations after puberty than the defi cient animals ( P = 0.039). Both pre-and postweaning DHA supplementation was required to positively affect plasma BDNF concentration. Additionally, plasma BDNF concentration was positively correlated to plasma, but not brain, DHA levels (0.211, P = 0.049) at P60. No correlation was seen at P40.
With plasma serotonin, one-way ANOVA did not detect a main effect of group at P40 ( Fig. 4C ) but did at P60 ( Fig.  4D ; F (3,64) = 7.55, P < 0.0001). Tukey's post hoc analysis indicated that compared with defi cient control animals, only animals receiving DHA supplementation throughout development had signifi cantly higher plasma serotonin concentrations after puberty ( P = 0.002). DHA supplementation was required during gestation, lactation, and postweaning periods to positively affect plasma serotonin concentration. Data are presented as the mean percent of total FAs ± the SEM. The degrees of freedom and their errors for the analysis of each FA at age P40 were 3 and 68, and at age P60 were 3 and 66, respectively. Signifi cant differences ( P < 0.05) between groups were determined with Tukey's post hoc test and are designated by a lack of common superscripts across each row. that emerged following puberty. Removal of DHA supplementation after weaning (late development) negated nearly all of these behavioral and biochemical effects, whereas starting dietary DHA supplementation after weaning was too late to affect most measures. The circulating biomarkers chosen in this study are clinically reported to be affected by at least one or more successful therapies aimed at treating depression (antidepressant drugs, cognitive behavioral therapy, electroconvulsive treatment, etc.). Overall, we observed an interesting pubertal shift in all of the peripheral biomarkers to lower plasma concentrations postpuberty as compared with prepuberty. This decrease coincided with the overall increase in depressive-like behaviors seen postpuberty in the DHA-defi cient offspring.
Maternal and postweaning dietary supplementation with DHA had an effect on overall FA composition of tissues that depended greatly on the timing and duration of supplementation, the location of the tissue, and the specifi c FA of interest. Addition of DHA to the maternal diet successfully increased brain DHA content in the offspring that was further enhanced or diminished depending on whether DHA was provided in the postweaning diet. Deficient offspring had ‫ف‬ 30% less brain DHA than suffi cient offspring both before and after puberty, or a difference of about 4% to 5% of total FAs. This spread is similar to the difference ( Ϫ 22% on average) that has been noted in postmortem tissue from depressed patients as compared

Brain BDNF
One-way ANOVA identifi ed a signifi cant main effect of group on BDNF content in hippocampus ( Fig. 5A ; F (3,43) = 4.23, P = 0.010) and hypothalamus ( Fig. 5B ; F (3,61) = 3.99, P = 0.012) at P60. Tukey's post hoc analysis indicated that suffi cient animals had signifi cantly higher BDNF content in both the hippocampus ( P = 0.015) and hypothalamus ( P = 0.017) as compared with defi cient controls. These effects were absent in animals from DHA-suffi cient mothers that lacked DHA supplementation after weaning (preweaning suffi cient), although there was a slight trend toward increased hippocampal BDNF in this cohort ( P = 0.194). DHA supplementation after weaning in offspring from DHA-defi cient mothers (postweaning suffi cient) was able to increase hippocampal BDNF levels ( P = 0.030), but not hypothalamic BDNF content. Brain DHA levels were positively correlated to both hippocampal BDNF content (0.454, P < 0.001) and BDNF content in the hypothalamus (0.259, P = 0.019).

DISCUSSION
This study demonstrated that DHA supplementation provided to rats during early and late development increased brain DHA levels considerably and affected depressive-like behaviors and mood-associated biomarkers Data are presented as the mean percent of total FAs ± the SEM. The degrees of freedom and their errors for the analysis of each FA at age P40 were 3 and 68, and at age P60 were 3 and 66, respectively. Signifi cant differences ( P < 0.05) between groups were determined with Tukey's post hoc test and are designated by a lack of common superscripts across each row.

Fig. 2.
Dietary DHA supplementation during development affected the concentration of FAs in brain tissue. A: Before puberty, dietary DHA at any time point resulted in higher brain DHA concentration relative to controls on a DHA-defi cient diet, and DHA throughout development increased brain DHA content to a greater extent than DHA during only the preweaning or postweaning period. B: After puberty, dietary DHA at any time point resulted in higher brain DHA concentration relative to defi cient controls, and DHA throughout development or after weaning increased brain DHA content higher than DHA during only the preweaning period. C: Before puberty, dietary DHA at any time point resulted in lower brain ARA concentration relative to defi cient controls, and DHA throughout development or after weaning decreased brain ARA content to a greater extent than DHA during only the preweaning period. D: After puberty, dietary DHA postweaning, regardless of maternal diet, resulted in lower brain ARA concentration relative to defi cient controls, yet DHA supplementation just during the preweaning period did not affect brain ARA content. E: Before puberty, dietary DHA at any time point resulted in lower brain DPA(n6) concentration relative to defi cient controls, and DHA throughout development decreased brain DPA(n6) content to a greater extent than DHA during only the preweaning or postweaning period. F: After puberty, dietary DHA at any time point resulted in lower brain DPA(n6) content relative to defi cient controls, DHA throughout development decreased brain DPA(n6) content lower than DHA during only the preweaning or postweaning period, and DHA supplementation only after weaning decreased brain DPA(n6) concentration lower than DHA during just the preweaning period. Data are presented as averages ± the SEM. Signifi cant differences ( P < 0.05) between groups were determined by one-way ANOVA followed by Tukey's post hoc tests and are indicated by an absence of shared superscripts. with their healthy counterparts ( 33 ). In fact, DHA was the only FA found to be signifi cantly altered in these postmortem brains. Another interesting phenomenon observed in this study was the effect of diet change at the time of weaning on brain DHA content before and after puberty. Switching preweaning suffi cient animals to a DHA-defi cient diet at weaning caused a slow decline in brain DHA content ( Ϫ 15% relative to suffi cient animals) that did not reach the levels observed in defi cient controls by the end of puberty (44 days). This rate was clearly slower than the rate of increase seen the preweaning defi cient animals (+39% relative to defi cient animals) indicating that the developing brain was able to retain DHA to some extent in the face of dietary defi ciency. Indeed, previous studies show that accretion is relatively faster than depletion of DHA in neural tissue ( 34 ). However, infants are typically weaned to rather DHA-poor foods and children's diets are usually no better than the usual Western adult diet, providing plenty of time throughout childhood and adolescence for depletion of DHA from neural tissue. Unfortunately, longitudinal data on DHA status in children from birth to adolescence are lacking, but in one particular study, healthy term infants who were weaned at 4-6 months of age to formula without DHA had signifi cant losses in RBC DHA content at 1 year of age ( Ϫ 50%) as compared with levels measured at the time of weaning. Conversely, infants fed formula containing DHA not only maintained tissue levels of DHA but had increases in RBC DHA and not tricyclic antidepressants and thereby effectively models the clinical evidence observed in cases of adolescent depression ( 39 ). The data from this study indicated that subsequent to puberty there was a shift in defi cient animals toward a depression-like behavioral phenotype and biomarker profi le. This shift may be akin to the increased incidence of depression following puberty in humans. Because the FST requires two tests separated by 24 h, it is entirely possible that the stress of the fi rst test differentially sensitized the animals to the second test, subsequently infl uencing the behavioral responses seen. Accordingly, prior stressors have been shown to affect behavior in a synergistic fashion with omega-3 defi ciency ( 40,41 ). Animals provided preweaning dietary supplementation with DHA displayed a more active postpubertal coping style than defi cient controls. However, improvements in all active behaviors (climbing and diving) and in passive behavior (immobility) were only observed in suffi cient animals. This indicated that these effects were dependent on early and late organizational (developmental) effects of DHA supplementation, as switching to a DHA-rich postweaning content (24%) relative to baseline levels at weaning and improved measures of visual system development as compared with infants fed DHA-defi cient formula ( 35 ). Thus, the depletion of brain DHA content after weaning in this study may be comparable to what is observed in humans following the postweaning drop in DHA consumption usually seen in infants and continued throughout development into puberty. These data indicate that the time between weaning and puberty was too short to overcome the deficiency induced by the maternal diet, but long enough to cause a substantial loss of brain DHA with a defi cient postweaning diet.
In this study, maternal and postweaning dietary DHA supplementation reduced postpubertal measures of depression-like behaviors that increased in defi cient animals after puberty. We used the modifi ed FST to measure depression-like behaviors before and after puberty ( 36 ). Typical antidepressants limit the amount of passive coping (immobility) and promote the amount of active coping (swimming/climbing) in adult animals ( 37,38 ). In juvenile animals, immobility is particularly sensitive to SSRIs . Similarly, diet did not affect climbing or diving (active behaviors) before puberty (C and E, respectively). After puberty, dietary DHA during the gestation and lactation (preweaning) period, regardless of postweaning diet, increased climbing (D). However, only dietary DHA throughout development increased diving (F). Data are presented as averages ± the SEM. Signifi cant differences ( P < 0.05) between groups were determined by one-way ANOVA followed by Tukey's post hoc tests and are indicated by an absence of shared superscripts.

Fig. 4.
Dietary DHA mitigated postpuberty decreases in plasma biomarkers associated with mood. Diet did not affect plasma levels of BDNF or serotonin before puberty (A and C, respectively), and only DHA supplementation throughout development increased plasma BDNF and serotonin after puberty (B and D, respectively). Before puberty, only DHA supplementation throughout development increased plasma melatonin (E). After puberty, dietary DHA during the postweaning period, regardless of maternal diet, increased plasma melatonin concentration (F). Data are presented as averages ± the SEM. Signifi cant differences ( P < 0.05) between groups were determined by oneway ANOVA followed by Tukey's post hoc tests and are indicated by an absence of shared superscripts. diet in preweaning defi cient offspring was unable to affect these measures. However, climbing behaviors were exclusively dependent on an early organizational effect of DHA, as increased time climbing was seen only in preweaning suffi cient and entirely suffi cient offspring, but not postweaning suffi cient or defi cient offspring. These data show that both maternal and postweaning dietary DHA supplementation were required to effectively buffer the postpubertal rise in passive behaviors and fall in active behaviors seen in the defi cient controls. Conversely, postweaning dietary DHA was too late, and maternal-only DHA supplementation was insuffi cient, to overcome all aspects of this shift in behavior.
In line with the decrease in time immobile, there was a higher plasma concentration of serotonin detected in suffi cient animals from this study relative to their defi cient counterparts. This was only observed in the suffi cient group, and not maintained in preweaning suffi cient or rescued in postweaning suffi cient animals, suggesting that the effect of dietary DHA supplementation on this measure was dependent on its actions in early and late development.
Previous reports suggest that early, but not late, omega-3 PUFA supplementation can restore brain serotonin and dopamine levels in omega-3 PUFA-defi cient rats. Taken with our results, it may be that peripheral serotonin levels are more dependent upon current omega-3 PUFA status, whereas brain serotonin levels are more sensitive to early developmental omega-3 PUFA intake ( 42 ). Interestingly, in our study there was an apparent drop in plasma serotonin levels after puberty in all animals, a phenomenon that has been reported previously in boys ( 43,44 ). Importantly, concentrations of circulating serotonin and its precursor tryptophan have been reported to be lower in depressed adults relative to healthy controls ( 45,46 ). Brain serotonin content is well known to be dependent on plasma tryptophan concentration ( 47 ), and the circulating serotonin level may also be an indicator of brain serotonin concentration ( 48 ). In a recent study, rats supplemented with high-DHA fi sh oil from gestation to adulthood exhibited decreased immobility in the FST and had increased levels of serotonin in the hippocampus and cortex, effects that were inhibited by acute treatment with a 5-HT 1A serotonin receptor antagonist ( 49 ). This suggests that optimal DHA status may enhance serotonergic neurotransmission, thereby improving depressive-like behaviors. Our data extend this previous study to suggest that optimal DHA status during both early and late development is important for measures of juvenile mood and peripheral indicators of serotonin function following puberty. Interestingly, 5-HT 1A receptors are highly expressed at birth in humans but decline rapidly in expression through adolescence ( 50,51 ). Thus, it is intriguing to speculate that this decline is somehow buffered by enhanced levels of DHA in brain, perhaps by facilitating 5-HT 1A receptor activity. Addition of DHA to the postweaning diet, regardless of maternal diet, also increased plasma levels of melatonin relative to defi cient offspring, an effect absent in preweaning suffi cient animals. This indicates that plasma melatonin levels may be more acutely dependent on DHA tissue content than on organizational effects of DHA. Melatonin has an important role in the entrainment of many biological systems (including sleep/wake cycles) to the central circadian clockmaker, the suprachiasmatic nucleus. Mood disorders have been associated with disruptions in circadian rhythms that govern essential function such as sleep, eating, and neuroendocrine function. Depressive symptoms are typically highest in morning, suggesting a phase advance of circadian rhythm, and depressed adolescents have been shown to have altered rest-activity rhythms ( 52 ). Interestingly, we observed a substantial drop in plasma melatonin concentration after puberty, an effect that has been previously reported in humans ( 53,54 ). Circulating melatonin has been shown to be low in depressed patients, and positively affected by SSRI treatment ( 55,56 ). Corroborating our effects reported here, a diet low in omega-3 PUFAs has been shown to decrease plasma melatonin in hamsters ( 57 ), and exogenous melatonin and melatonin receptor agonists administered to rats and mice exert antidepressant-like effects in the FST ( 58,59 ) possibly through a central serotonin-dependent mechanism ( 60 ). Both serotonin and melatonin receptor systems have been linked to the expression of BDNF in the brain ( 61,62 ), an important neural growth factor implicated in the neurotrophin theory of depression ( 63 ).
In accordance with the increases in plasma melatonin and serotonin, there was a higher plasma concentration of BDNF detected in suffi cient animals from this study relative to their defi cient counterparts. As with serotonin, but not melatonin, this increase was only observed in the suffi cient group, and not maintained in preweaning sufficient or rescued in postweaning suffi cient animals. This suggests that the effect of dietary DHA supplementation on plasma BDNF was dependent on its actions in early and late development. Furthermore, in alignment with melatonin and serotonin, plasma BDNF concentrations were lower after puberty, a phenomenon also reported in humans ( 64 ). Circulating levels of BDNF correlate well with the expression of BDNF in brain, at least in rodents ( 65 ), thus providing a convenient indicator of central BDNF status. It also appears that there is evidence, albeit controversial, that BDNF can cross the blood-brain barrier ( 66,67 ). Interestingly, peripheral administration of BDNF in mice reportedly decreased immobility in the FST after 2 weeks of treatment ( 68 ) suggesting that circulating BNDF may have a functional effect on behavioral measures of mood. BDNF has a prominent role in shaping neurotransmission and plasticity in the brain by aiding in the growth, maintenance, and survival of neurons ( 69,70 ), including the normal development and function of serotonin neurons ( 71 ). Defi cits of neural plasticity appear to be associated with depression ( 72 ). Accordingly, peripheral BDNF has been shown to be lower in depressed patients as compared with healthy counterparts, and these levels rise with antidepressant FST immobility compared with unsupplemented controls. On the other hand, McNamara et al. ( 87 ) reported that feeding omega-3 PUFA-suffi cient diets from weaning to adulthood (P21-P90) in female rats had no effect on FST behavior compared with omega-3 PUFA-insuffi cient controls. This suggests that there may be a sex difference in the effect of postweaning omega-3 PUFA supplementation on depressive-like behaviors. However, similar to the females in McMamara et al. ( 87 ), we did not observe a benefi t of postweaning DHA on time spent immobile in male offspring when examined at P60. Taken together with the previous studies, this suggests that perhaps postweaning omega-3 PUFA supplementation is most benefi cial in males when provided past adolescence into adulthood or when behaviors are assessed in adulthood rather than late adolescence. What is unknown is whether pre-and postweaning omega-3 PUFA supplementation in female offspring has similar effects as those seen in our study with the male offspring. Regarding developmental timing of supplementation, Ferraz et al. ( 88 ) fed fi sh oil either during gestation and lactation or only after weaning and found that both groups of male rat offspring had reduced immobility in the FST at 90 days of age as compared with controls on base diet alone. However, the FA composition of the base diet and the tissue levels of omega-3 PUFAs were not reported, making comparisons with our study diffi cult. Furthermore, we examined changes in behavior across puberty, and thus our assessments were performed on adolescents rather than adults, and Ferraz et al. ( 88 ) did not have a group supplemented both before and after weaning. In particular, our study demonstrated that both pre-and postweaning DHA was required to positively affect all measures of depressive-like behavior when measured during postpubertal adolescence. Similar to Ferraz et al. ( 88 ), Chen and Su ( 89 ) provided fi sh oil either only to the dam or to the dam and male offspring after weaning. Unfortunately, they did not have a completely omega-3 PUFA-defi cient group because offspring from dams fed defi cient diets were placed on postweaning suffi cient diets containing ␣ -linolenic acid and DHA. They examined FST behaviors at P70, a time point closer to adolescence, and concluded that only preweaning fi sh oil was effective in reducing immobility in the FST at that age. However, offspring in the postweaning supplemented group were obtained from dams provided fi sh oil; thus, any separate effects of postweaning supplementation beyond those attributable to supplementation during the preweaning period are truly not inferable from data reported in Chen and Su ( 89 ).
Adolescent depression typically follows a recurrent episodic course ( 90 ). In this regard, perhaps DHA defi ciency may be seen as a diathesis, or preexisting vulnerability (akin to a genetic predisposition), that is activated by developmental events such as puberty-dependent changes in brain connectivity and hormone secretion, or by intense psychological stressors (death of a loved one, emotional abuse, parental divorce, etc.). However, there are limited clinical studies that have examined the link between omega-3 PUFAs and juvenile depression. In a case-control study comparing RBC FA profi les in 150 depressed and treatment ( 73,74 ). Furthermore, increased hippocampal BDNF immunoreactivity was detected in postmortem samples from patients who were treated with antidepressants ( 75 ). BDNF protein levels are highest in the hippocampus and hypothalamus, two brain regions that exhibit a high degree of plasticity and are essential elements of the neuroendocrine response to stress ( 76,77 ). The hypothalamus in particular also plays a critical role in the onset, duration, and completion of puberty ( 78,79 ). Thus, increases in the levels of hypothalamic and hippocampal BDNF content may provide an adolescent enhanced resiliency to stressors and negative mood states during a time of great physical, mental, and hormonal change.
In this study, dietary DHA supplementation throughout early and late development increased postpubertal hippocampal and hypothalamic BDNF protein levels relative to defi cient offspring. Loss of dietary DHA after weaning resulted in BDNF protein levels no different than defi cient animals; however, preweaning defi cient animals placed on a postweaning DHA-supplemented diet had increased BDNF protein levels in the hippocampus but not the hypothalamus. The effects seen in hypothalamus are similar to the observations of plasma BDNF concentrations in this study. This suggests that BDNF protein expression in the hypothalamus is dependent upon effects of DHA during both early and late development, but that hippocampal BDNF is sensitive to current DHA tissue status and/or late developmental effects of DHA. Other studies have also reported an effect of an omega-3 suffi cient diet during development on brain levels of BDNF protein. For example, rats fed DHA-adequate diets from gestation to 18 weeks old had elevated BDNF protein in the hippocampus and hypothalamus ( 80 ), and those fed an ␣ -linolenic acid suffi cient diet from weaning to 18 weeks old had elevated BDNF protein expression in the frontal cortex ( 81 ) compared with animals on defi cient diets. In addition, fi sh oil supplementation during gestation and lactation increased hippocampal and cortical BDNF protein levels in 21-dayold and 90-day-old rats ( 49 ). Our data are in line with these reports but suggest that the effects of dietary DHA during development on brain BDNF protein levels are likely dependent on the timing of DHA supplementation in a brain region-specifi c manner. These changes in BDNF protein expression may have contributed to the behavioral phenotype observed in the FST because acute infusion of BDNF into the midbrain or hippocampus has been shown to decrease depression-like behavior in the FST ( 82,83 ).
Our study expands upon those described previously that have examined the effect of dietary omega-3 PUFA consumption, or lack thereof, during development on depression-like behaviors in rats. Both Naliwaiko et al. ( 84 ) and Vines et al. ( 49 ) reported that fi sh oil supplementation throughout gestation, lactation, and after weaning resulted in male offspring that exhibited reduced immobility in the FST compared with controls when tested as adults. In addition, studies by Ferraz et al. ( 85 ) and DeMar et al. ( 86 ) indicated that feeding male rats diets supplemented with fi sh oil or ␣ -linolenic acid, respectively, after weaning (P21) into adulthood for 6 months decreased data, only trace levels of EPA are found in the brain where upon entry it is rapidly ␤ -oxidized and metabolized ( 97,98 ). Furthermore, erythrocyte DHA, but not EPA, content was shown to be lower in SSRI-resistant depressed adolescents compared with healthy counterparts ( 29 ). Given that DHA, but not EPA, is highly enriched in neural tissue and is essential for brain development, it may play a critical role in the developmental factors that shape the etiology of adolescent depression, a supposition in alignment with the results of our study.
The data reported here present a totality of pubertal changes in depression-like behaviors and central and peripheral biomarkers of depression in male rats and strongly suggest that these changes can be positively modulated by DHA supplementation throughout development. These data highlight the particular importance of dietary supplementation with preformed DHA after weaning, a time when most infants are transitioned to foods containing low levels of omega-3 PUFAs, a trend that continues onward into adulthood. The clinical evidence regarding the role of DHA and omega-3 PUFAs on adolescent depression is emerging yet limited, but given the intervention and epidemiological data currently available and potential societal impact, further studies are certainly warranted. 161 healthy adolescents, there was a higher percentage of DHA, but not EPA (20:5n3), present in the RBCs from healthy controls as compared with their depressed counterparts ( 28 ). Interestingly, there was an increase in RBC ␣ -linolenic acid content in depressed adolescents suggesting that the conversion of ␣ -linolenic acid to EPA, which is already ineffi cient in humans ( ‫ف‬ 0.1% conversion), may be impaired in these individuals. In a cross-sectional study, DHA content was lower ( Ϫ 16%) in the RBCs of depressed teenagers with eating disorders than those with eating disorders and no depressive symptoms ( 91 ). In a crosssectional study of Japanese teenagers ages 12 to 15, there was an inverse correlation of EPA plus DHA intake with symptoms of depression in boys, but not girls ( 5 ). The lifetime major depressive disorder prevalence rates in Japan are some of the lowest in the world ( 92 ). However, to achieve these rates it may be necessary to attain the DHA concentrations found in the RBCs of Japanese adults. Such a feat likely requires a daily DHA intake of 400-700 mg by children and 700-1,000 mg by adults in the United States ( 93 ). Along these lines, a handful of intervention studies have been reported to date. Nemets et al. ( 27 ) observed improved scores in tests for symptoms of depression in depressed children ages 6 to 12 that were given 400 mg of EPA plus 200 mg of DHA daily for 1 month as compared with children who received placebo. Clayton et al. ( 26 ) provided 360 mg of EPA plus 1,560 mg of DHA as an adjunct to pharmacotherapy in 18 female bipolar depressed teens (average 16 years old) that had been stabilized for 6 weeks on lithium and valproate. Compared with within-individual baseline data, the supplementation caused a substantial elevation of RBC DHA and EPA content, and a signifi cant decrease in clinical scores of depression. McNamara et al. ( 29 ) provided fi sh oil at doses of 2.4 g/day or 16.2 g/day in a small open-label trial for 10 weeks to adolescents with SSRI-resistant depression and observed symptom remission in 40% and 100% of cases, respectively. These studies provide an initial indication of the effect that optimal dietary DHA consumption might have on adolescent mood. It remains to be determined how maintenance of the optimal tissue levels of DHA throughout pregnancy, nursing, childhood, and adolescence affects the prevalence, duration, or severity of juvenile depression.
It should also be noted that several randomized, clinical trials have indicated that oral supplementation with fi sh oil, which contains substantial (but variable) amounts of the omega-3 PUFAs DHA and EPA, can improve clinical measures of depression in adults [for meta-analysis see ( 30 )]. These effects have largely been attributed to EPA ( 94 ); however, only one direct side-by-side comparison between pure EPA and pure DHA has been reported. Mozaffari-Khosravi et al. ( 95 ) reported that EPA but not DHA was effective as an adjunct therapy to antidepressants in young adults with mild-to-moderate depression. Yet, in this study the effect of antidepressant cotherapy on the effi cacy of either omega-3 PUFA is unknown, and the plasma FA profi les of the participants was not reported. The effect of EPA may be attributable to its anti-infl ammatory actions in the periphery ( 96 ). However, in agreement with our FA