The folic acid endophenotype and depression in an elderly population

Abstract

Objectives

Folate status and/or genes have been linked to depression in a number of studies. This may be via a direct action (or actions) on neuronal membranes or indirect effects through the metabolism of methyl groups involved in neurotransmitter synthesis. This study examines folate and related thiol metabolism that might underpin either phenomenon.

Design

Cohort study describing the relationship between several genetic and nutritional aspects of folic acid homeostasis and depression assessed by the HADS psychometric index in an elderly cohort.

Setting

New South Wales (Australia) retirement village.

Participants

118 elderly participants (age 65–90 years).

Results

Stepwise multiple regression was used to determine the best statistical model to predict depression; C677T-MTHFR (p=0.0103) was found to be positively associated with depression, while the thiol dipeptide Cys-Gly was negatively associated (p=0.0403). The statistical models used accounted for the major folate related indices (genetic and biochemical) that are most often evaluated in the context of health and disease. When only genetic data were examined for interactions, C677T-MTHFR was found to be negatively associated with the HADS Depression Index Score (p=0.0191).

Conclusion

The potential influence of Cys-Gly on this phenotype is novel, and of considerable interest given that it has been linked to altered spontaneous activity and sedation in an animal model. Cys-Gly is a recognised ligand at the N-methyl-D-aspartatic acid (NMDA) subclass of glutamate receptor, a system associated with depression. In addition, the C677T-MTHFR association adds further support to existing findings underscoring the potential role of folate in depression.

Introduction

Folate status and/or genes have long been associated with mood disorders, with parenteral 5-methyltetrahydrofolate (methylfolate) being used successfully as therapy (1). Although the mode of its pharmacologic action is unclear, potential mechanisms include a direct agonistic action upon neuronal membrane receptors, or an indirect effect on neurotransmitter methyl group metabolism (2, 3). The systemic level of methylfolate is clearly dependent upon the enzyme methylenetetrahydrofolate reductase (MTHFR), which is encoded for by a polymorphic gene (C677T-MTHFR), as well as being dependent on dietary folic acid intake (4). Homocysteine (Hcy), as a methyl group acceptor, is critical for methylfolate dependent remethylation to form methionine and, hence, production of S-adenosylmethionine (AdoMet): Methyltransferases such as catechol-O-methyltransferase, hydroxyindole-O-methyltransferase and phenylethanolamine-N-methyltransferase utilise AdoMet in the biosynthesis of biogenic amine transmitters (3). It is, therefore, interesting that elevated plasma Hcy is associated with depression and the C677T-MTHFR polymorphism is over represented in patients with clinical depression (5). The view that C677T-MTHFR is implicated in depression is supported by a large meta-analysis (10 studies with 11,709 participants involving 1,280 cases of depression) (6). Hcy levels per se also influence neuronal tissue, and this thiol is considered neurotoxic via several divergent mechanisms including binding at ligand gated ion channels (7).

The present study explores the relationship between seven common folate polymorphisms, important indices of folate status, Hcy related thiol metabolism and depression in an elderly population sample (n=118). The single nucleotide polymorphisms (SNPs) include those which are associated with both absorption and uptake of dietary folate [reduced folate carrier (G80A-RFC) and ?-glutamylcarboxy peptidase (C1561T-GCPII)], and methyl group and DNA metabolism [C677T-MTHFR; A1298C-MTHFR; methionine synthase (A2756G-MS); methionine synthase reductase (A66G-MSR); dihydrofolate reductase (19bp deletion DHFR)]. The key thiols related to Hcy metabolism that were examined include Hcy, cysteine (Cys), cysteinyl-glycine (Cys-Gly) and glutathione (GSH). GSH is a transulphuration product of Hcy, and the dipeptide Cys-Gly (a component in GSH metabolism) has been associated with decreased spontaneous activity and increased sedation in an animal model (8).

Materials and Methods

Study Design

Participants: 118 subjects (65-90 years) from a retirement village located on the Central Coast in New South Wales (51 ? and 67 ?) were examined for a range of genetic, metabolic and dietary factors related to folate status.

Ethics Approval: University of Newcastle Human Research Ethics Committee approval – H-782-0304 and Northern Sydney Central Coast Health Human Research Ethics Committee approval – 04/19 apply.

Methods

Thiol Analysis: Following derivatisation with the fluorogenic reagent SBDF, step-gradient HPLC with fluorescence detection (ex 385nm, emm 515nm) was used to measure plasma total Hcy, Cys, Cys-Gly and GSH according to Dufficy et al (9).

Folate/vitamin B12 analysis: Serum and red blood cell folate along with serum vitamin B12 were measured using chemiluminescent immunoassay (Access Immunoassay System, Beckman Instruments, Inc). The laboratory normal reference ranges were 370-1050 nmol/L for red cell folate, 5-21 nmol/L for serum folate and 125-780 pmol/L for serum vitamin B12.

The Hospital Anxiety and Depression Scale: The Hospital Anxiety and Depression Scale (HADS) test is a 14-item questionnaire designed to gauge emotions during the previous week, and consists of two scales. The first is an anxiety subscale (HADS A) and the second a depression subscale (HADS D), each with seven questions. The scores on each question are graded from 0 to 3 and the aggregate scores for the seven questions for each subscale are used to assess the subject's level of depression and anxiety. The aggregate scores for each subscale are also divided into categories defining the level of the participant's depression and anxiety as “Normal range” (0 – 7) “Mild” (8 – 10), “Moderate” (11, 12, 13, 14) and “Severe” (15 – 21) (10). To reduce the interference from possible physical aspects of somatic disorders such as dizziness, headache and insomnia, these symptoms are excluded from the questionnaire (11,12). The HADS screening instrument is commonly used in various settings for patients in non-psychiatric institutions and is considered to provide meaningful results as a psychological screening tool.

PCR and SNP scoring for genotype: All subjects were scored for the common C677T-MTHFR, A1298C-MTHFR, G80A-RFC, A2756G-MS, A66G- MSR, 19bpDHFR and C1561T-GCP II SNPs using the polymerase chain reaction (PCR) followed by restriction enzyme digestion, and gel electrophoresis (excluding 19bpDHFR which uses a nested primer allele-specific strategy for genotype scoring).

C677T-MTHFR: Primers (Sense - 5′-TGAAGGAGAAGGTGTCTGCGGGA-3’ and antisense – 5′-AGGACGGTGCGGTGAGAGTG-3’) were used to amplify a 198-bp MTHFR fragment. The C>T substitution at bp 677 creates a HinfI recognition sequence. If the mutation is present, HinfI digests the 198-bp amplicon into a 175- and 23-bp fragment (13).

A2756G-MS: Primers (Sense - 5′-GGTGTGTTCC CAGCTGTTAGATG-3’ and antisense – 5′-GACACTG AAGACCTCTGATTTGAAC-3’) generate a 265-bp amplicon which can be digested with the restriction enzyme HaeIII to yield a 265-bp fragment for the wild type; 265-, 180-, and 85-bp fragments for the heterozygote; and 180- and 85-bp fragments for the homozygous recessive genotype (14).

A1298C-MTHFR: Primers (Sense - 5'CTTTGGGGAGC TGAAGGACTACTAC -3’ and antisense – 5′- CACTTTGTGACCATTCCGGTTTG-3′) were used to amplify a 163-bp MTHFR fragment. The amplicon was digested with the restriction enzyme MboII to yield 56-, 31-, 30-, 28-, and 18-bp fragments for the wild type and 84-, 31-, 30-, and 18-bp fragments for the homozygous recessive genotype (15).

19bpDHFR: Allele specific primers were used for amplification. Briefly, primer A: 5′-CCACGGTCGG GGTACCTGGG-3′, primer B: 5′-ACGGTC GGGGTGGCCGACTC-3’ and primer C: 5′-AAAAGGGGAAT CCAGTCGG-3’ were used in a nested primer allele-specific scheme to yield either a 113bp amplicon (primers A and C) or 92bp amplicon (primers B and C), and thus permit detection of the 19bp deletion (16).

G80A-RFC: Primers (Sense - 5′-AGTGTCAC CTTCGTCCC-3’ and antisense – 5′-TCCCGCGTGAA GTTCTTG-3′) were used for amplification. If present in the amplicon, the G allele was digested with the restriction enzyme CfoI to yield three fragments; 125 bp, 68 bp and 37 bp. The variant allele abolishes a restriction site leading to two fragments; 162 bp and 68 bp (17).

A66G-MSR: Primers (Sense - 5′-GCAAAGGCCATCGCAGAAGACAT-3’ and antisense – 5′- GTGAAGATCTGCAGAAAATCCATGTA-3′) generate an amplicon with an artificial NdeI restriction site. Following NdeI digestion, a DNA fragment of 66-bp remains uncut in the G allele but is cut into 44- and 22-bp fragments in the presence of the A allele (18).

C1561T-GCP II: Primers (Sense - 5′-CATTCTGGTAGGAATTTAGCA-3′, and antisense 5′-AAACACCACCTATGTTTAACA-3′) generate a 244 base pair amplicon which can be digested with the restriction enzyme AccI. The 1561TT genotype renders two fragments of 141 and 103 base pairs, respectively, while the 1561CC genotype yields only a single fragment of 244 base pairs (19).

Food frequency questionnaire: An interviewer administered food frequency questionnaire (FFQ) was used to estimate daily nutrient intake. The FFQ covered 225 food items and all food groups. Additionally, participants provided a list of all supplements they were taking, and were asked about them specifically during the FFQ interview.

Questionnaires were analysed using Foodworks 2.10.146 (Xyris Software, Brisbane, QLD, Australia). The package utilised a number of food databases to cover the majority of foods consumed by Australians. These included AusFoods (brands), Aus Nut (base foods) and the New Zealand – Vitamin and Mineral Supplements 1999 databases.

The average daily intake of folate included the average daily intake from food and supplements and collectively is referred to as dietary folate. Foodworks™ does not contain vitamin B12 values, therefore, data were evaluated by matching food items from the United States Department of Agriculture (USDA) National Nutrient Database for Standard References which does include vitamin B12 values.

Statistics: All statistical analysis was performed using the JMP program for Windows (version 8.0; SAS Institute Inc., Cary, NC, U.S.A). Relationships between depression index score and other parameters were analysed by multiple and stepwise linear regression. Stepwise regression was performed in a mixed direction with significant probability [0.250] for a parameter to be considered as a forward step and entered into the statistical model or considered as a backward step and removed from the model. Mallow's Cp criterion was used for selecting the statistical model where Cp first approaches p variables.

All data including folate and thiol indices, psychological screening indices, genotype prevalence, allele frequency and carriage of mutant allele have been tabulated and presented as appropriate.

Results

Table 1 provides descriptive statistics for all subjects. It gives dietary and blood folate levels, serum vitamin B12 values and plasma thiol measurements, as well as the HADs depression screening index score. Mean and SD values are provided. Table 2a and 2b give genotype prevalence, allele frequency and carriage of mutant allele for each SNP. Table 3a provides the outcome from stepwise multiple regression analysis (C677T-MTHFR, Cys-Gly, Hcy, red cell folate, dietary folate, Cys, age, gender) to determine the best statistical model to predict depression index score. It clearly shows that C677T-MTHFR, Cys-Gly, Hcy, dietary folate and age were correlated with depression (R² = 0.1507). Table 3b examines only genetic data (C677T-MTHFR; A1298C-MTHFR; A2756G-MS; A66G-MSR; 19bp deletion DHFR; G80A-RFC and C1561T-GCPII). Stepwise multiple regression analysis showed that C677T-MTHFR, A1298C-MTHFR, A66G-MSR, G80A-RFC and C1561T-GCPII provide the best statistical model to predict depression (R2 =0.0969). In this context, R² is derived from a multiple correlation with many effects contributing to each of the linear models described above. It represents the proportion of the variation in the response that can be attributed to terms in the statistical model rather than to random error.

Table 1.

Descriptive statistics (mean, SD and range) for all subjects (n=118)

	Red cell folate (μmol/l)	Serum folate (μmol/l)	Serum B12 (pmol/l)	Hcy (μmol/l)	Cys (μmol/l)	Cys-Gly (μmol/l)	GSH (μmol/l)	Dietary folate (mg/d)	Depression index score	Age (Years)
Mean	774.5	23.3	267.7	9.13	240.7	27.4	10.24	463.8	3.23	74.7
SD	311.4	11.5	119	2.48	26.6	4.9	4.14	195.3	2.812	5.6
Range	254 –1996	4 –46	64 –702	4.57 –17.02	180.9 –333.8	17.1 –46.8	3.17 –35.20	150 –1030	0 –14	65 –90

Table 2.

a) Genotype prevalence (number and %) for all polymorphisms investigated

	Wildtype	Heterozygous	Homozygous recessive
677C>T-MTHFR	54 (45.8)	46 (39.0)	18 (15.3)
1298A>C-MTHFR	59 (50.0)	49 (41.5)	10 (8.5)
80G>A-RFC	33 (28.0)	55 (46.6)	30 (25.4)
2756A>G-MS	59 (50.0)	50 (42.4)	9 (7.6)
66A>G- MSR	18 (15.3)	61 (51.7)	39 (33.1)
19bp deletion DHFR	36 (30.5)	58 (49.2)	24 (20.3)
1561C>T-GCP II	110 (93.2)	8 (6.8)	0 (0)
b) Allele number, frequency and carriage of mutant allele for all polymorphisms examined. Carriage implies possession of at least one mutant allele expressed as % of individuals.

	Wildtype allele Allele number	Mutant allele and (frequency)	Carriage of mutant allele (%)
677C>T-MTHFR	154 (0.65)	82 (0.35)	54.2
1298A>C-MTHFR	167 (0.71)	69 (0.29)	50.0
80G>A-RFC	121 (0.51)	115 (0.49)	72.0
2756A>G-MS	168 (0.71)	68 (0.29)	50.0
66A>G- MSR	97 (0.41)	139 (0.59)	84.7
19bp deletion DHFR	130 (0.55)	106 (0.45)	69.5
1561C>T-GCP II	228 (0.66)	8 (0.34)	6.8

Table 3.

a) Stepwise multiple regression analysis revealed that of the common folate related indices most often associated with health, C677T-MTHFR, Cys-Gly, Hcy, dietary folate and Age were correlated with the HADS depression index score [R² = 0.1507]. The regression coefficient gives the direction of the association.

Variable	Regression coefficient	T value	P value
C677T-MTHFR[CT-CC]	-0.671	-1.25	0.2150
C677T-MTHFR[TT-CT]	1.96	2.61	0.0103
Cys-Gly	-0.107	-2.07	0.0403
Hcy	0.1865	1.69	0.0935
Dietary Folate	-0.00155	-1.19	0.2363
Age	0.0719	1.51	0.1331
b) Stepwise multiple regression analysis revealed that among the polymorphisms examined, C677T-MTHFR; A1298C-MTHFR; A66G-MSR; G80A-RFC and C1561T-GCPII provided the best statistical model correlated with depression index score [R2 = 0.0969].

Variable	Regression coefficient	T value	P value
C677T-MTHFR[CC&CT-TT]	-0.928	-2.38	0.0191
A1298C-MTHFR [AA&AC-CC]	-0.447	-0.97	0.3347
A66G-MSR [AA-AG&GG]	-0.413	-1.17	0.2447
C1561T-GCPII	-1.254	-1.24	0.2189
G80A-RFC [AA&GA-GG]	0.463	1.64	0.1039

Clearly C677T-MTHFR interacts in a fashion that is predictive of depression index score in either of the above linear models. However, in one the relationship of this SNP as an interactive factor with diet and blood metabolic indices gives a positive association with depression (CC vs CT not significant; CT vs TT p = 0.0103), while in the other, the TT genotype interacts with the other folate related SNPs to confer a negative association with depression index score (p = 0.0191). Cys-Gly is also a significant component in depression index score when part of a statistical model that accounts for the major folate related indices that are most often evaluated in the context of health and disease; the resulting association being a negative one. If GSH is added into the statistical model shown in 3a, the outcome from stepwise regression remains essentially the same (C677T-MTHFR p=0.0207; Cys-Gly p=0.0403). If age and gender are added into the complete set of SNP data shown in 3b, again the outcome remains essentially similar (C677T-MTHFR p=0.0325; age p=0.0063).

The data presented therefore shows that both the thiol dipeptide Cys-Gly and the common C677T-MTHFR are major predictors of depression index score in an elderly cohort.

Discussion

The number of clinical conditions that the C677T-MTHFR variant is associated with is quite remarkable, and underscores the role of folate coenzymes in key biosynthetic pathways. These diverse effects are believed to stem from the reduced catalytic activity of the expressed enzyme conferred by carriage of the 677T-MTHFR allele. The resulting reduction of folate based one-carbon units for synthesis of DNA thymidine and/or methionine is sensitive to folate status, and modulates disease risk via molecular mechanisms (altered Hcy, genomic and non-genomic methylation and uracil misincorporation) that are still being elucidated, but ultimately, all likely originate from altered flavin cofactor binding of the polymorphic MTHFR protein (20). In the context of depression, this polymorphism may therefore alter de novo methylation, a particularly important pathway for biogenic amine neurotransmitter synthesis. However, this is not the first report of this variant gene being associated with mood disorders (6). What makes the current study of interest is that Cys-Gly levels seem to be significantly related to the HADS depression score. Cys-Gly is a dipeptide that has recently been recognised as decreasing spontaneous activity and increasing sedation in an animal model (8). It is generated from GSH by ?-glutamyl transpeptidase (21), and is used as a precursor for neuronal GSH (22), and as a supply of both cysteine and glycine for neurons (23). MTHFR is crucial in delivering a one carbon unit for conversion of Hcy into methionine. While methionine may be utilised for neurotransmitter synthesis, the presence of the variant MTHFR can raise Hcy levels. This may be important since much (approximately 50%) Hcy is converted to GSH (24), and both GSH and Hcy along with Cys-Gly are known ligands for the NMDA subclass of glutamate receptor (8). This is interesting since several reports indicate that excitatory glutamatergic neurotransmission may play a role in the pathophysiology of depression (25): Disturbances in glutamate metabolism and NMDA receptors have been associated with depression and suicide tendency, and it has been shown that a single intravenous dose of an NMDA receptor antagonist is sufficient to produce sustained relief from depressive symptoms (26). Although Cys-Gly is a significant component in the linear model predicting depression index score, why it exhibits a negative association and what this might mean in terms of the overall neurobiology of depression is unclear. Equally, it is unclear why the MTHFR-TT genotype interacts with the other folate related SNPs to confer a negative association with depression index score when it has the opposite effect as a component within a linear model that takes account of folate biochemical indices. Certainly the literature is rich in data showing a dichotomy of functional and clinical effects arising from the MTHFR-TT genotype, and this data may simply reflect the complexity of interactions involving this key mutation.

The recognised metabolic link between folate, methionine and GSH metabolism is interesting, may be relevant in depression, and warrants larger studies to specifically examine the disposition of Cys-Gly in patients with affective disorders. This seems particularly relevant given the well established role of C677T-MTHFR genotype in modulating Hcy levels and the critical role of Hcy in the balance between remethylation and transulphuration pathways. The balance between pathways regulates production of methyl groups for neurotransmitters and GSH as an antioxidant and source of neuroactive ligands that have recognition sites on the NMDA receptor. Figure 1 illustrates the various mechanisms by which the folic acid endophenotype, a descriptor that includes thiol metabolism, may influence mood. It shows how the folate/vitamin B12 dependent remethylation of Hcy is dependent upon the activity of MTHFR, itself a vitamin B2 dependent protein; the product, methionine (as AdoMet) being available for methylation of neurotransmitters. It also shows how Hcy itself sits at the nexus of folate dependent Hcy remethylation and vitamin B6 dependent Hcy transulphuration. Although the figure demonstrates how a number of transulphuration products might act as ligands for the NMDA subtype of glutamate receptor, it also draws attention to the fact that early work places methylfolate as a ligand at the kainic acid subtype of glutamate receptor.

Figure 1.

Aspects of folate and thiol metabolism related to neurochemistry and possibly depression

This is clearly a complex area of biochemistry. Within it several common polymorphisms and associated micronutrients such as vitamins B2, B6 and B12 conspire to regulate the folate/Hcy related pathway shown in Figure 1. Furthermore, elevated Hcy is also a product of lifestyle, drug therapy, disease and aging. Ultimately it lowers AdoMet and raises S-adenosylhomocysteine impairing methylation of DNA, myelin, neuronal receptors, membrane phospholipids and neurotransmitters. It is excitotoxic via NMDA receptor activation, and enhances oxidative stress as well as leading to endothelial damage and hence cerebral ischemia (3). An elevated Hcy is usually associated with a low folate status, in itself a corollary of 677T-MTHFR. Any or all of these phenomena may be linked to observations that biogenic amine related compounds in the cerebrospinal fluid, notably 5-hydroxyindoleacetic acid (a metabolite of serotonin), homovanillic acid (an indicator of dopamine turnover) and 3-methoxy-4-hydroxyphenylglycol (an indicator of noradrenaline turnover) were significantly lower in depressed patients with an adverse folate-Hcy ratio (3, 27).

There is much more to discover in regard to the relationship between the folic acid endophenotype and depression. Given the nutritional genetics that are involved, there may even be the potential to utilise novel forms of therapeutic intervention to ameliorate perturbed metabolism and improve clinical symptoms in patients with depression.

Financial disclosure: None of the authors had any financial interest or support for this paper.