Does Acute Methamphetamine Change Mouse Memory and 36,376 Hippocampal Nuclei? What the 2026 Experiment Actually Shows

Answer: In a controlled mouse experiment, four intraperitoneal methamphetamine injections of 10 mg/kg each within one day were followed by worse performance on selected novel-object-recognition and Morris-water-maze outcomes. Investigators also sequenced 36,376 hippocampal nuclei from six mice and identified differentially expressed genes, with especially pronounced changes in excitatory neurons. The study supports an association between an intense experimentally administered methamphetamine regimen, particular mouse behavioral outcomes, and cellular transcriptional responses. It does not measure methamphetamine concentrations in human blood, establish a human impairment threshold, prove that a particular person suffered permanent memory damage, or reconstruct a specific accident or criminal event.

The source is An, Lu, Wang and colleagues, Single-nucleus RNA sequencing and functional studies of acute methamphetamine-induced cognitive impairment, Communications Biology 9, article 440 (2026), primary DOI 10.1038/s42003-026-09728-2. The paper combines behavioral experiments, nucleus-level transcriptomics, laboratory pathway analyses, and separate neuronal culture work. These are related but noninterchangeable evidence streams.

The scientifically defensible conclusion in one paragraph

Acute methamphetamine exposure can produce measurable changes in mouse performance and in hippocampal cellular transcription under the conditions tested. After experimental dosing, the authors observed reduced preference for a novel object, slower acquisition or performance on a spatial water-maze task, and less target-quadrant searching and fewer platform crossings. Their nuclear RNA analysis found 619 differentially expressed genes across the analyzed cells and identified especially prominent effects in excitatory neurons and the dentate gyrus. Orthogonal assays supported changes in certain transcripts and proteins, notably Ddx5 and PEX5. Yet transcriptional enrichment is not identical to enzyme activity, neuron death, actual drug concentration, or durable human cognitive impairment. Each inference requires evidence at its own level.

What was administered—and why the dosing model matters

The animal model used male C57BL/6 mice aged roughly six to eight weeks, weighing about 20–25 g. According to the original Methods, the methamphetamine group received four successive 10 mg/kg intraperitoneal injections at intervals of approximately two to three hours, all within 24 hours. Controls received saline. The nominal sum of the four injections was 40 mg/kg. That is administered mass divided by animal body weight, not a measured circulating concentration, oral human dosage, or dose known to reach a particular hippocampal neuron.

Intraperitoneal injection eliminates features of human inhalation, ingestion, prescription use, and self-administration. A high-intensity repeated-dose exposure is useful for probing mechanisms but limits generalization to low-dose medicinal use, a single recreational episode, chronic human dependence, or a disputed driving event. Species, route, formulation, injection spacing, age, sex, physiologic temperature and subsequent testing intervals all affect interpretation. The investigation did not construct a dose-response curve showing how the same molecular findings behave at everyday human exposures.

How many animals—and why 36,376 nuclei do not mean 36,376 mice

The source reports a hippocampal single-nucleus sequencing dataset prepared from six mice, yielding 36,376 quality-controlled nuclei: 18,171 nuclei assigned to methamphetamine-exposed tissue and 18,205 to control tissue. Those are nuclear observations nested within a small number of biological animals. The biological replicate is the mouse, not an isolated nucleus. Treating all nuclei as statistically independent mice would greatly exaggerate precision and obscure animal-to-animal variability.

For the behavioral displays, the original Figure 1 captions identify n = 6 for novel-object and water-maze measurements. The captions do not justify combining those behavioral counts with the sequencing nuclei or with the animals underlying other assays to manufacture an overall cohort size. The investigators also report separate molecular confirmation panels with n = 3. These denominators belong to their respective experiments, not to a hypothetical pooled human-equivalent study.

Not all cell counts have the same meaning either. The authors identified 26 initial clusters and annotated ten broad cell types. A larger proportion of one annotated cell type among captured nuclei can reflect biology, tissue isolation, survival, capture efficiency, normalization, or analytical decisions; it is not automatically a stereologic count of all cells in the intact hippocampus. A marker-defined proportion is not by itself proof that particular cells died.

What the memory and navigation tests measured

The novel-object-recognition test compared time and frequency of exploration after a mouse encountered a familiar and an unfamiliar object. The reported recognition index used novel-object exploration time as a fraction of total exploration time. Methamphetamine-exposed mice spent less time investigating the novel object and showed a lower recognition index. Figure 1 indicates statistically marked differences, but the publication does not provide a table of exact numerical p-values for every behavioral panel. It would be misleading to assign such values from visual impressions alone.

The Morris water maze addressed spatial learning and navigation: finding a submerged platform over repeated training and searching a previously reinforced quadrant when the platform was removed. The methamphetamine group took longer to find the platform, spent less time in the target quadrant, and crossed the former platform location fewer times in the illustrated comparisons. These are operational behavioral endpoints. They are not scores on a human memory examination, observed real-world driving mistakes, or direct measures of hippocampal cell injury.

Crucially, behavior can depend on locomotion, stress, sensory processing, arousal, motivation, visual discrimination and prior training as well as memory. The paper supports particular task differences under its testing protocol; it does not separately prove that every difference arose exclusively from memory encoding. For a courtroom interpretation, the measured task, procedural controls and alternative explanations deserve more weight than a broad label such as “brain damage.”

Evidence table: dosing, experimental units, and measured outcomes

What the sequencing instrument actually analyzed

The method was single-nucleus RNA sequencing, not a forensic drug assay. Hippocampal tissue was processed into nuclear suspensions, sorted and prepared using a 10x Genomics Chromium single-cell 3′ workflow. The investigators performed quality filtering, cell annotation, differential expression analysis, gene set scoring, regulatory network analysis and computational inference about cell-to-cell communications. Biological material was mouse hippocampal tissue. The primary sequenced analyte was nuclear RNA represented by transcript counts.

No mass-spectrometric measurement of parent methamphetamine or amphetamine metabolite in mouse blood, serum, plasma, urine, hair, oral fluid, vitreous or the sequenced hippocampal tissue was reported as part of the principal experiment. “Sequenced nuclei from methamphetamine-treated mice” does not mean “methamphetamine concentration measured in individual nuclei.” Similarly, a sample of hippocampal tissue cannot be treated as equivalent to a human venous whole-blood toxicology specimen.

The analytic outputs are useful for hypothesis generation because they separate classes of hippocampal cells that would be blurred in whole-tissue RNA averages. But a computational cluster is defined by expression patterns, and its pathway score depends on a selected gene set and statistical method. These are not direct measures of synaptic transmission or human psychomotor performance.

Why the difference between nuclei and independent animals is central

A common source of misleading precision in single-cell or single-nucleus research is pseudoreplication. Counting every nucleus as if it came from a different organism ignores the shared genetics, exposure environment and experimental history of nuclei originating from the same mouse. With six source mice, results should be assessed against animal-level reproducibility and the statistical model used for clustered measurements. Very large nuclear counts can make a transcriptomic atlas richly detailed without removing the limitations of a small animal cohort.

The 18,171-versus-18,205 split is close to balanced in counts of captured nuclei, but balance does not establish that the original groups were perfectly matched in age, stress response, hippocampal region sampled, injection handling or quality-control attrition. The paper’s cell annotations and differential gene screen can direct further experiments, not determine a human causal mechanism with clinical certainty. Formal independent replication across larger numbers of animals, sexes and exposure schedules would be needed before broad generalization.

Where the strongest gene-expression changes appeared

The authors identified 619 differentially expressed genes when comparing treatment and control tissue in the reported analysis: 381 increased and 238 decreased. Among annotated broad cell types, excitatory neurons were associated with 252 differential genes, a comparatively large alteration within their analysis. Further subdivision defined five excitatory-neuron groups related to hippocampal territories: CA1, CA2/3, CA4, dentate gyrus and a retrohippocampal region.

In the dentate gyrus subgroup the investigators reported 100 upregulated and 43 downregulated genes. This localization is interesting because the dentate gyrus participates in hippocampal information processing, but the reported pattern is still RNA-level evidence. It does not identify a specific lesion visible on imaging, prove loss of 143 cells, or independently establish that the gene changes caused the observed navigation deficits.

Other annotated cells included inhibitory neurons, astrocytes, microglia, oligodendrocytes, oligodendrocyte precursors, endothelial cells, ependymocytes, Cajal–Retzius cells and fibroblasts. The source reports a higher proportion of microglia and altered relative excitatory/inhibitory populations. Those changes should be interpreted as captured-cell composition and computational annotation, not a validated clinical neuropathology count.

Oxidative phosphorylation: enriched pathway scores are not proof that mitochondria performed better

The investigators described greater oxidative-phosphorylation gene-set activity in selected neuron classes using gene-set variation analysis. That sounds superficially like improved mitochondrial function. Yet they also found reduced ATP5K and COX6B1 messenger RNA in targeted hippocampal qPCR measurements and illustrated damaged mitochondrial ultrastructure in separate microscopy. Those observations address different analytical constructs. Pathway enrichment concerns relative patterns across collections of genes; targeted mRNA quantifies selected transcripts; electron micrographs depict selected physical organelles.

Therefore, “an OXPHOS pathway was enriched” should not be paraphrased as “methamphetamine increased ATP production” or “mitochondria were functioning normally.” Likewise, targeted transcript decreases do not quantify respiration rate, ATP concentration or mitochondrial membrane potential without the appropriate assay. The paper’s combination of findings raises a mechanistic hypothesis about energy stress and mitochondrial disruption, rather than providing a validated biomarker of individual stimulant intoxication.

Peroxisomes and PEX5: why the direction of the measurements must be preserved

A similar distinction applies to the peroxisome pathway. Computational enrichment called attention to peroxisome-related signatures, while targeted qPCR showed lower expression of selected genes including PEX5, PEX1, PEX6, catalase and Abcd1. The researchers also documented a decrease in PEX5 signal at the protein level. It would be incorrect to say that pathway enrichment demonstrated that every peroxisomal gene or protein increased.

PEX5 participates in importing proteins into peroxisomes. A reduced PEX5 measurement is biologically worth testing further, particularly in studies of reactive oxygen metabolism, but this article does not experimentally restore PEX5 in a randomized animal treatment arm and demonstrate rescue of memory performance. A proposed protective or therapeutic target remains a proposal, not proven efficacy. Neither an RNA value nor a western-blot band allows calculation of a case-specific blood methamphetamine concentration.

Ddx5 increases: replication of a signal is not demonstration of causation

The authors observed increased Ddx5 expression across multiple hippocampal cell categories. They used targeted qPCR, immunoblotting and immunostaining to provide corroborating evidence for increased Ddx5 RNA or protein in selected mouse material and in cultured neurons. The regional observations were especially relevant in the dentate gyrus. This is more persuasive than a single computational prediction, because some direction of change was checked with a different laboratory measurement.

Nevertheless, Ddx5 is an RNA helicase with functions extending beyond methamphetamine responses. A rise in its expression does not necessarily mean it caused neuronal death, prevented injury, or specifically identifies a stimulant exposure. Showing correlation after a drug challenge is not the same as altering Ddx5 alone and demonstrating an outcome change. The study did not establish Ddx5 sensitivity, specificity or a forensic reference interval for diagnosing a particular person’s methamphetamine-related neurotoxicity.

Separate fetal-rat neuronal cultures are not equivalent to dosed mice

An additional experimental component used primary hippocampal neurons obtained from embryonic-day-18 Sprague–Dawley rats, maintained in culture and exposed to methamphetamine at a nominal 900 µM for 24 hours. Those cells were grown outside an intact animal. Culture-medium concentration is not measured mouse brain concentration, let alone measured human blood or serum concentration. The culture system removes circulating hormones, pharmacokinetics, complex behavior and much of the network environment.

Cell-culture confirmation can strengthen a narrower proposition that some molecular signals respond to an experimentally applied compound. It cannot establish how much methamphetamine reached the mouse dentate gyrus after intraperitoneal dosing. It also does not prove a clinically relevant exposure in a living person. Mixing a rat culture concentration with mouse dosing and human impairment testimony would be an invalid cross-matrix and cross-species inference.

Evidence table: numbers that should not be collapsed into one claim

What the study did not test, and what “no result” means

The report emphasizes statistically marked differences in mouse behavior, RNA and certain protein indicators. It does not present a controlled human impairment study, quantitative human methamphetamine concentrations, a direct on-road driving trial, an extended recovery trajectory, or a randomized PEX5/Ddx5 rescue of behavioral outcomes. The absence of those tests is not a negative p-value. “Not evaluated” is different from “tested and found no significant difference,” and both differ from “demonstrated no effect.”

Even within the paper, significance stars do not provide exact confidence intervals or a complete corrected multiplicity analysis for every biologic hypothesis. The sequencing differential-gene screen used a false-discovery-rate threshold, but the entire article also includes pathway scoring, regulatory networks, cell communication predictions, laboratory confirmations and multiple behavioral outcomes. The evidentiary status of each layer should remain visible. A selective positive panel should not be promoted into a clinical diagnostic rule.

How to interpret cell-communication and transcription-factor analyses

CellChat inference suggested greater signaling interactions involving excitatory neurons and some other annotated populations. The researchers also described altered information-flow scores for signaling pathways such as SPP1, NPY and INSULIN. Their SCENIC analysis inferred candidate transcription-factor regulatory activity and identified changes involving Ppara and other factors. These computational tools synthesize expression patterns and curated biological relationships; they do not directly record synaptic impulses or measure physical ligand-receptor binding in the behaving mouse.

The source additionally explored coexpression modules linked statistically with task parameters, including a module involving genes such as Grm5, Ryr2 and Gria1. Correlation between a gene module and spatial task performance is a starting point for targeted experiments, not evidence that a proposed gene is necessary or sufficient for behavioral dysfunction. A rigorous causal test would manipulate the candidate independently, include appropriate controls, examine rescue or reproduction, and separate changes in movement and sensory behavior from memory.

Why the distinction from forensic whole-blood toxicology is essential

A forensic methamphetamine result usually begins with a defined physical specimen, collection time, validated instrument, reportable concentration and quality-assurance record. This animal experiment begins instead with administered intraperitoneal mass and ends with behavioral observations, hippocampal tissue transcriptomics and selected molecular confirmation. It supplies neither an individual’s legal venous whole-blood methamphetamine level nor a whole-blood-to-brain conversion factor.

Whole blood, separated serum, plasma, oral fluid, urine and hair are not interchangeable. Urine may support exposure history under appropriate method validation, while a confirmed blood finding answers a different question about a collected specimen at a particular time. Even a measured human whole-blood concentration by itself cannot settle tolerance, timing, functional performance, or event-specific causation. These issues require independently relevant clinical, toxicological and behavioral evidence. For context, see this site’s published guides to biological matrices, confirmatory drug testing and therapeutic drug monitoring.

Why the experiment cannot establish an individual driver’s impairment

No participant drove a vehicle, completed standardized human psychomotor tasks, or underwent a legally calibrated driving evaluation in this study. The reported mouse novel-object and water-maze outcomes are experimentally meaningful but are not equivalent to lane-keeping, divided attention, emergency braking, workplace performance, witness memory or a particular defendant’s cognition. Nor did investigators reconstruct an actual dose, sample collection, prescribed regimen or competing causes in a real forensic case.

An opinion about an individual’s behavior at an alleged event must address what that person took, specimen identity, analytical method, collection interval, known medication and health context, observable behavior, alternative explanations and the specific functional question. A mouse transcriptional signature cannot bridge all those gaps. The sound expert use of this paper is to explain possible biological mechanisms and why stronger, directly applicable evidence would be needed before making an individual causal conclusion.

Strengths and limitations that should appear together

The study’s strengths include controlled saline comparison, an explicit high-intensity exposure protocol, two different behavioral paradigms, a large number of sequenced nuclei, cell-category resolution and orthogonal confirmation of selected RNA/protein signals. Its limits include a small independent animal base relative to nuclear counts, male-only in-vivo animals, a single intensive experimental dosing paradigm, separate fetal-rat cultures, no direct methamphetamine bioanalytical concentration-time data, uncertainty about the causal direction of selected gene signatures, and the lack of a human validation cohort.

The original authors themselves discuss limited capture of cytoplasmic RNA by single-nucleus methods, the focus on only one brain region, incomplete causal validation of differentially expressed genes and the absence of female mice. These limitations do not erase the positive results; they determine the appropriate scope of interpretation. A hypothesis about oxidative metabolism or inflammation can be scientifically plausible without being a validated clinical test or a case-specific legal opinion.

Seven questions about the original evidence

1. Did researchers measure methamphetamine in the 36,376 nuclei?

No. The 36,376 figure is the count of quality-controlled nuclei used for RNA sequencing. It is not 36,376 methamphetamine chemical measurements, and the paper does not report corresponding forensic blood concentrations for individual animals or humans.

2. Were there 36,376 independently treated animals?

No. The sequencing nuclei were obtained from six source mice, while the behavioral figure reports n = 6 for its illustrated tests. Large nested molecular datasets cannot be treated as tens of thousands of independent organisms. Molecular and behavioral denominators should be reported separately.

3. Was methamphetamine administered in a way that models ordinary human use?

Not directly. The protocol used four 10 mg/kg intraperitoneal injections within 24 hours in male mice. It was an intensive mechanistic challenge, not an equivalence trial comparing that dosing with a specific human oral, inhaled or prescription exposure.

4. Did an enriched oxidative-phosphorylation score prove increased ATP production?

No. A gene-set activity score is a relative transcriptomic computation. The authors also reported decreases in certain mitochondrial-related transcripts and changes in organelle morphology. Actual cellular ATP output and respiratory flux would require their own direct experimental assays.

5. Does a decrease in PEX5 or an increase in Ddx5 diagnose methamphetamine brain injury?

No validated diagnostic sensitivity, specificity, reference interval or individual clinical decision rule was established. These are candidate mechanistic signals in the tested model, not approved or qualified biomarkers for deciding whether a particular person was intoxicated or permanently injured.

6. Did the researchers demonstrate that the observed deficits were permanent?

No. The investigation did not provide a longitudinal human recovery study or establish irreversible deficits in the mice. Differences seen during the experimental assessment cannot, without longitudinal evidence, be described as lifetime cognitive impairment.

7. Can this study establish impairment or causation in a specific DUI, accident or criminal case?

No. It does not contain a legal whole-blood concentration, a matched human functional validation study, observed driving, or the facts of any alleged event. It can inform background neuroscience and help frame questions, but it cannot substitute for validated toxicology, contemporaneous behavior and case-specific causal analysis.

Primary-study identification and bottom line

An D, Lu F, Wang Y, Zhang Z, Meng S, Chen Y, You X, Zhu Y, Yuan W, Zhou Z, Huang P, Wang J and Chen X. Single-nucleus RNA sequencing and functional studies of acute methamphetamine-induced cognitive impairment. Communications Biology. 2026;9:440. Primary source: doi:10.1038/s42003-026-09728-2.

Bottom line: A high-intensity methamphetamine challenge was associated with selected mouse memory-test deficits and widespread hippocampal transcriptional differences, including 619 differential genes and changes in PEX5/Ddx5 indicators. The sequencing unit was an isolated nucleus nested within a small animal study; the material tested was neural RNA and protein, not a human forensic drug specimen. The evidence supports a mechanistic research interpretation, not an individualized toxicology concentration, impairment threshold, lasting diagnosis or event-causation determination.