Effects of the Natural β-Carboline Alkaloid Harmine, a Main Constituent of Ayahuasca, in Memory and in the Hippocampus: A Systematic Literature Review of Preclinical Studies

Dos Santos and colleagues introduce harmine as a naturally occurring β-carboline found in Banisteriopsis caapi (a principal component of ayahuasca) and in Peganum harmala (Syrian rue). Earlier preclinical work and observational studies in long-term ayahuasca users have suggested neuroprotective and cognitive-enhancing properties for harmine-rich preparations, but it is unclear whether these effects are attributable specifically to harmine, to other constituents, or to combinations of compounds found in those botanical preparations. This paper sets out to synthesise the preclinical evidence by conducting a systematic literature review of studies that examined the effects of harmine on hippocampal biology and on memory- and learning-related behavioural tasks. The review aims to identify what has been reported in cell-culture and animal models regarding harmine's neuroprotective and cognitive effects and to summarise proposed mechanisms that might underlie those effects.

The investigators followed PRISMA guidelines to identify preclinical studies available up to 29 July 2016. Electronic searches were carried out in PubMed (1966–29 July 2016), LILACS (1982–29 July 2016) and SciELO (1998–29 July 2016) using the keywords harmine AND memory OR learning OR hippocampus. Handsearching of reference lists supplemented the database searches. Only studies published in English were considered. Pre-specified eligibility criteria included original, peer-reviewed experimental investigations in hippocampal cell cultures or in animal models that assessed harmine's effects on memory/learning and hippocampal biochemical parameters. Excluded materials were reviews, case reports, abstracts, letters, book chapters and other non-original reports. Interventions of interest were harmine exposures; main comparators included saline and several psychoactive drugs reported across studies (for example ethanol, piracetam, imipramine, scopolamine, donepezil). Outcomes encompassed biochemical measures in hippocampal cells and behavioural memory/learning endpoints in animals. Two reviewers independently screened studies and extracted data; discrepancies were resolved by consensus. Extracted items included author names, year, study type (cell culture or animal), acute versus chronic treatment, and biochemical or behavioural outcome measures. The authors report that the search returned 59 references for abstract screening, 11 full texts were examined, and all 11 were included in the final sample, comprising two cell-culture studies and nine animal studies. The extracted text does not clearly report sample sizes for individual studies within this section, though tabular material was referenced.

Eleven studies met inclusion criteria: two used hippocampal cell cultures and nine used rodent models. The cell-culture studies reported that harmine attenuated glutamate-induced toxicity and influenced neurite outgrowth. Specifically, hippocampal cells incubated with harmine (50 µM for 8 hours, with glutamate exposure) showed significantly reduced glutamate-induced cell death 24 hours later; cortical neurons exposed to glutamate plus harmine (100 µM for 24 hours) were similarly protected. A second cell-culture report found that harmine (1.6 µM for three days) decreased neurite number in mouse hippocampal neurons, an effect mediated by inhibition of the kinase DYRK1A, a protein implicated in neuronal development and in disorders associated with cognitive deficits. Acute administration studies in animals produced multiple positive signals for cognition and neuroprotection. In mice, intraperitoneal harmine at 1, 2.5 and 5 mg/kg increased exploration of a novel object in an object-recognition test (short-term memory) with P < 0.001 for all doses; harmaline, by contrast, was reported to improve long-term memory. In forced swim test (FST) paradigms, harmine (10 and 15 mg/kg, i.p.) reduced immobility and increased active behaviours, consistent with antidepressant-like effects; the 15 mg/kg dose also raised hippocampal BDNF levels (P < 0.05). Antioxidant effects were reported across several acute-dose studies: harmine decreased lipid peroxidation/TBARS and protein carbonylation and increased catalase and superoxide dismutase (SOD) activity in prefrontal cortex and hippocampus at various doses (significant at P < 0.05 for reported comparisons). One study found that oral harmine (20 mg/kg) increased hippocampal and cortical mRNA of immediate early genes Egr-1, c-Jun and c-Fos, with a significant rise in Egr-1 protein. In older rats, subcutaneous harmine (1 and 5 mg) reduced errors in a delayed-match-to-sample water-maze task (improved short-term/spatial working memory), although some animals given 5 mg had transient motor impairments and were excluded from analysis. Chronic administration studies reported consistent biochemical and behavioural effects. Fourteen days of harmine (5–15 mg/kg/day, i.p.) decreased immobility in the FST and increased active coping behaviours; higher doses (10–15 mg/kg) increased hippocampal BDNF. Chronic dosing reduced markers of oxidative stress in cortex and hippocampus and produced dose-dependent changes in antioxidant enzyme activities. In a chronic mild stress model, harmine (15 mg/kg/day for 7 days) reduced anhedonia and reversed stress-related increases in adrenal size and ACTH and normalised hippocampal BDNF. In Morris water maze paradigms, oral harmine (20 mg/kg by gavage) administered for 2 and 10 weeks improved escape latency and path length when given acutely after scopolamine challenge; after 10 weeks, acute harmine shortened path length in APP/PS1 transgenic mice (an Alzheimer’s model). Chronic harmine inhibited acetylcholinesterase activity in cortex under certain conditions but did not reduce hippocampal Aβ protein increases. In a traumatic brain injury model, five days of harmine (30 mg/kg/day, i.p.) reduced cerebral oedema, improved spatial learning and motor scores, increased GLT-1 expression, reduced caspase-3 and inflammatory cytokine (IL-1β, TNF-α) expression in hippocampus, and increased neuronal survival at 24 hours (all reported as significant at P < 0.05). Across studies, harmine's comparator conditions included vehicle/saline and active drugs such as imipramine and donepezil. The extracted text provides multiple P-values for reported effects but does not consistently report sample sizes for each experiment within the prose.

The review authors conclude that preclinical evidence supports neuroprotective and cognitive-enhancing effects of harmine in cell cultures and in several rodent paradigms of memory and injury. They highlight convergent biochemical findings in hippocampus and cultured neurons—reduced glutamate toxicity, normalisation or increases in BDNF, decreased lipid peroxidation and protein carbonylation, increased antioxidant enzyme activity, upregulation of immediate early genes (Egr-1, c-Jun, c-Fos), increased GLT-1 expression, higher neuronal survival and reductions in apoptosis and pro-inflammatory cytokines—which together align with improved performance on object recognition, water-maze and other memory tasks. Proposed mechanisms synthesised by the authors include monoamine oxidase (MAO) and acetylcholinesterase (AChE) inhibition, GLT-1 upregulation (supporting glutamate clearance), reduced reactive oxygen species, increased neurotrophic signalling (BDNF), anti-inflammatory effects and inhibition of DYRK1A. The authors note that findings from models of amyotrophic lateral sclerosis and global cerebral ischaemia support a role for GLT-1 upregulation in harmine's neuroprotective effects, and that some antioxidant and cholinesterase-related observations mirror results from P. harmala seed extracts. However, uncertainty remains about translation to humans. The authors emphasise gaps in knowledge about oral pharmacokinetics and brain exposure of harmine in humans despite animal data showing rapid brain penetration and rapid elimination; a human PET study of [11C]-harmine showed fast decline in unmetabolised plasma fraction. Human reports of harmine's psychoactivity are mixed, and studies using harmine-rich botanical preparations are challenging to interpret because they contain multiple active constituents, dose composition is often unknown, and pharmacokinetic or pharmacodynamic interactions may occur. The authors acknowledge limitations of their review, chiefly that it is restricted to preclinical studies and that human observational data are retrospective and susceptible to bias. They recommend further preclinical and human research to clarify harmine’s therapeutic potential for cognitive impairment and neurodegenerative conditions.