GAA/Creatine pathway reprogramming in high-grade glioma (Abdullah et al., 2025)

Paper

• Abdullah KG, Miki K, Edgar CK, et al. “Gliomas phenocopy an inborn error of metabolism to drive neuronal activity and tumor growth.” bioRxiv (2025). DOI: https://doi.org/10.1101/2025.09.15.676412

Background: key genes and metabolites

Creatine synthesis pathway

Note on nomenclature: GATM and GAMT are different genes with similar names.

• GATM encodes the enzyme AGAT (step 1: makes GAA)
• GAMT encodes the enzyme GAMT (step 2: converts GAA to creatine)

Neuronal signaling components

In glioma context

In high-grade glioma (HGG), the creatine pathway becomes dysregulated:

• Tumor cells upregulate AGAT but not GAMT, creating a bottleneck
• GAA accumulates to ~100x normal levels and is secreted
• Tumor-infiltrated neurons downregulate KCC2, raising intracellular chloride
• When secreted GAA activates GABAA receptors, the response becomes excitatory (depolarizing) rather than inhibitory
• This drives neuronal hyperactivity that promotes tumor growth

Literature review

For high grade glioma (HGG), while prior work established specific oncometabolite mechanisms in IDH-mutant gliomas (e.g., (R)-2HG), metabolite alterations that generalize across aggressive glioma subtypes have been less clear. Abdullah et al. (2025) address this gap using deep metabolomics across human surgical specimens, connecting a recurrent metabolic phenotype to tumor-neuron interactions that accelerate tumor growth.

The study’s central claim is that HGGs phenocopy guanidinoacetate methyltransferase (GAMT) deficiency, an inborn error of creatine metabolism, by accumulating the creatine-pathway intermediate guanidinoacetate (GAA) to unusually high levels (~100-fold vs non-malignant brain).

The metabolic bottleneck

This accumulation is mechanistically linked to an imbalance in creatine synthesis pathway enzyme activity:

1. AGAT is upregulated in HGG cells - rapidly converts arginine + glycine into GAA
2. GAMT remains low/unchanged - cannot keep pace with GAA production

This enzyme imbalance creates a metabolic bottleneck with two parallel consequences:

• GAA accumulates because GAMT is saturated and cannot convert it to creatine; the excess GAA is secreted into the tumor microenvironment
• Creatine is depleted because de novo synthesis is blocked at the GAMT step; instead, tumor cells import extracellular creatine (supplied by tumor-associated myeloid cells)

Importantly, GAA accumulation does not cause creatine depletion - both are independent downstream effects of the same AGAT » GAMT imbalance.

Neuroactive signaling

Rather than being a passive metabolic byproduct, secreted GAA functions as a neuroactive signaling molecule, acting as a GABAA receptor agonist. In glioma-infiltrated brain - where neuronal chloride homeostasis is altered (KCC2 downregulation) - GABAA activation becomes depolarizing rather than hyperpolarizing, elevating neuronal firing and strengthening neuron-glioma interactions that promote tumor aggressiveness.

Methodology

The work integrates

1. Multi-omic profiling of primary human tissues (polar metabolites/lipids with matched transcriptomics),
2. Stable-isotope tracing to resolve pathway flux (highlighting strong arginine/glycine-to-GAA labeling uncoupled from creatine synthesis in glioma stem-like cells),
3. Ex vivo explant assays to demonstrate tumor-associated GAA secretion
4. Electrophysiology and multielectrode array recordings to test how GAA modulates neuronal activity in tumor-bearing brain slices. Causality is supported by genetic perturbation (AGAT knockout reduces GAA, microenvironmental neuronal activity markers, and in vivo tumor aggressiveness) and a translationally motivated dietary intervention (arginine restriction + ornithine supplementation, modeled after GAMT deficiency management, extends survival in a xenograft model).

Limitations

I personally like this paper, the results are plausible and coherent with our in-house findings as well, but some caveats to note:

1. Many results are preclinical and rely on xenografts/explants (of course, not everyone can catch brain cancer)
2. “Bottleneck” behavior is inferred from flux/labeling and relative enzyme expression rather than direct enzyme kinetics in human tumors
3. The translational diet signal, while compelling, will require careful evaluation of feasibility and safety in patients with brain tumors (e.g., nutritional status, steroid use, treatment interactions)

Nonetheless, the study provides a strong example of how metabolite signaling (not just bioenergetics) can couple cancer metabolism to circuit-level neurobiology, expanding the conceptual toolkit for therapeutic target discovery in neuro-oncology.

GAA / creatine synthesis pathway

%%{init: {'theme': 'base', 'themeVariables': { 'primaryColor': '#dbeafe', 'primaryTextColor': '#0f172a', 'primaryBorderColor': '#2563eb', 'lineColor': '#475569', 'secondaryColor': '#f1f5f9', 'tertiaryColor': '#fff' }}}%%
flowchart TB
    subgraph legend[" "]
        direction TB
        L1[Metabolite]
        L2[/"Enzyme"/]
        L3["Tumor change"]
        L4(["Neuronal"])
        L5["Imbalance"]
        L1 ~~~ L2 ~~~ L3 ~~~ L4 ~~~ L5
    end

    subgraph pathway["Creatine synthesis (core biochemistry)"]
        direction LR
        Arg[Arginine] --> AGAT[/"AGAT
(gene: GATM)"/]
        Gly[Glycine] --> AGAT
        AGAT --> GAA[GAA
Guanidinoacetate]
        AGAT --> Orn[Ornithine]
        GAA -->|methylation| GAMT[/"GAMT
(uses SAM)"/]
        GAMT --> Cr[Creatine]
        Cr <-->|CK| PCr[Phosphocreatine]
        Cr -.->|spontaneous| Crn[Creatinine]
    end

    GAA -.->|"in HGG"| gaaUp

    subgraph hgg["HGG: metabolic bottleneck"]
        direction LR
        enzymes["AGAT ↑↑ | GAMT low"]
        
        enzymes -->|"rapid production"| gaaUp["GAA accumulates
(~100x vs brain)"]
        enzymes -->|"blocked conversion"| crDown["Creatine ↓
(no de novo synthesis)"]
        
        gaaUp --> secretion["GAA secreted into
microenvironment"]
        crDown -.-> import["Tumor imports creatine
(from myeloid cells)"]
    end

    secretion --> gabaa

    subgraph neuro["Neuronal signaling axis"]
        direction TB
        gabaa(["Extracellular GAA activates
GABAA receptors"])
        kcc2(["KCC2 ↓ in glioma-infiltrated neurons
→ chloride dysregulation"])
        depol(["GABAA becomes depolarizing
(excitatory, not inhibitory)"])
        outcome(["Neuronal hyperactivity
→ promotes tumor growth"])

        gabaa --> kcc2
        kcc2 --> depol
        depol --> outcome
    end

    classDef metabolite fill:#dbeafe,stroke:#2563eb,stroke-width:2px
    classDef enzyme fill:#ffedd5,stroke:#c2410c,stroke-width:2px
    classDef tumor fill:#fee2e2,stroke:#dc2626,stroke-width:2px
    classDef neuro fill:#ede9fe,stroke:#6d28d9,stroke-width:2px
    classDef imbalance fill:#fef3c7,stroke:#d97706,stroke-width:2px
    classDef invisible fill:none,stroke:none

    class Arg,Gly,GAA,Orn,Cr,PCr,Crn,L1 metabolite
    class AGAT,GAMT,L2 enzyme
    class enzymes,L5 imbalance
    class gaaUp,secretion,L3 tumor
    class crDown,import metabolite
    class gabaa,kcc2,depol,outcome,L4 neuro
    class legend invisible

    style pathway fill:#f0fdfa,stroke:#0f766e,stroke-width:2px,color:#0f766e
    style hgg fill:#fef2f2,stroke:#dc2626,stroke-width:2px,color:#dc2626
    style neuro fill:#f5f3ff,stroke:#6d28d9,stroke-width:2px,color:#6d28d9

Creatine pathway metabolites: HGG vs non-malignant brain

The paper reports a striking imbalance in creatine synthesis pathway metabolites in high-grade glioma (Figures 2C-F):

• GAA is ~100-fold higher in HGG vs non-malignant brain - an effect size comparable to 2-HG enrichment in IDH-mutant gliomas
• Creatine and phosphocreatine are decreased in HGG, consistent with a metabolic bottleneck: robust AGAT activity drives GAA accumulation, but insufficient GAMT activity prevents conversion to creatine

The interactive plot below visualizes these reported changes (non-malignant brain normalized to 1; note the log scale).