The paradigm shift: membraneless organelles are formed by phase separation
Until roughly 2009, most biologists described nuclear and cytoplasmic bodies — stress granules, P-bodies, Cajal bodies, nucleoli, paraspeckles — as ill-defined aggregates or loosely organised protein scaffolds. The reconceptualisation began with studies on intrinsically disordered proteins in Caenorhabditis elegans germ granules (Brangwynne et al., 2009, Science) and reached its canonical statement in the Banani, Lee, Hyman, and Pappu 2017 Nature Reviews Molecular Cell Biology review, which defined biomolecular condensates as assemblies formed by liquid-liquid phase separation (LLPS). The core insight: many cellular compartments are not enclosed by lipid bilayers but instead form when protein (and RNA) concentrations exceed a saturation threshold, causing the system to de-mix into a dense phase and a dilute phase — analogous to oil droplets forming in water, but governed by weak, multivalent molecular interactions rather than hydrophobicity alone.
The thermodynamic framework is the Flory-Huggins lattice model adapted for polymers, in which the interaction parameter chi determines whether mixing or demixing is favoured at a given temperature and concentration. For biomolecular condensates, the relevant interactions are dominated by intrinsically disordered regions (IDRs) — stretches of protein sequence that lack a stable folded structure under physiological conditions.
Intrinsically disordered regions and low-complexity sequences
IDRs are enriched in a small subset of amino acids: glycine, serine, glutamine, asparagine, phenylalanine, tyrosine, and arginine. The low-complexity sequences (LCS) within IDRs often consist of near-repetitive runs of these residues, exemplified by the prion-like domains of RNA-binding proteins FUS (fused in sarcoma), EWS (EWSR1), and TAF15. FUS contains an N-terminal QGSY-rich prion-like domain followed by an RGG repeat region and an RRM (RNA recognition motif). The prion-like domain alone is sufficient to drive phase separation in vitro when concentration or temperature conditions are appropriate.
The molecular interactions underlying LLPS in these systems have been resolved at atomic resolution by solid-state NMR and cryo-EM. Two categories dominate: (1) pi-pi stacking between aromatic residues — tyrosine-tyrosine, phenylalanine-tyrosine, and tyrosine stacking with nucleobase rings of RNA; and (2) cation-pi interactions between arginine (guanidinium group) and tyrosine or phenylalanine aromatic rings. Vernon et al. (2018, eLife) showed that the sequence determinants of phase separation in human proteins are largely captured by the pairwise content of aromatic and arginine residues in IDRs. This is sometimes called the "pi-pi and cation-pi" grammar of condensate formation.
Critically, LLPS exhibits concentration threshold behaviour: below the saturation concentration (c*), proteins remain dispersed in the dilute phase; above c*, condensates nucleate and the dilute-phase concentration remains approximately constant at c* while all additional protein partitions into the dense phase. This threshold behaviour is why small changes in protein synthesis, import, or modification can switch condensate formation on or off — and why cellular metabolic state, which affects protein expression and post-translational modification globally, can exert outsized effects on condensate biology.
Specific condensates and their molecular machinery
Transcriptional condensates
In 2018, two back-to-back papers in Cell (Boija et al.; Cho et al.) described transcriptional condensates at super-enhancers. The Mediator complex co-activator subunit MED1 contains an IDR that phase-separates and recruits RNA Polymerase II via its C-terminal domain (CTD). The CTD consists of 52 heptapeptide repeats (YSPTSPS) in humans; the tyrosine and serine residues participate in the same pi-pi and cation-pi interactions that drive IDR condensation. Phosphorylation of CTD Ser5 by CDK7 (the TFIIH kinase, during initiation) promotes condensate entry; phosphorylation of Ser2 by CDK9/P-TEFb (during elongation) promotes condensate exit, providing a phospho-switch that couples the transcription cycle to phase separation dynamics.
Heterochromatin protein 1 alpha
HP1 alpha (CBX5) phase-separates in vitro and in vivo at pericentromeric heterochromatin. Its hinge region (an IDR between the chromodomain and chromoshadow domain) is required for phase separation. Larson et al. (2017, Science) and Strom et al. (2017, Science) showed simultaneously that HP1 alpha condensates are dynamic, exchange material with the surrounding nucleoplasm, and are disrupted by phosphorylation of the hinge region. The condensate model of heterochromatin provides a mechanism for the remarkable stability and heritability of silenced chromatin domains: once nucleated, the dense phase concentrates HP1 alpha far above c*, maintaining silencing even against dilution during S-phase.
Stress granules
Stress granules (SGs) are cytoplasmic condensates that assemble when translation is inhibited — typically during heat shock, oxidative stress, hypoxia, or viral infection. SG assembly requires two parallel nucleation pathways: (1) the eIF2-alpha kinase pathway, in which phosphorylation of eIF2-alpha (by HRI, PKR, PERK, or GCN2) stalls translation initiation, freeing mRNAs and associated factors; and (2) the G3BP1/G3BP2 scaffold pathway, in which these Ras-GAP SH3 domain-binding proteins act as phase-separation seeds when they are no longer sequestered by USP10 and Caprin-1. The RNA-binding proteins TIA-1 and TIAR (which contain Q-rich prion-like domains), as well as eIF4E (the 5-prime cap-binding protein), HuR, and PABP1 partition into SGs. SG formation is reversible — upon stress removal, kinase inactivation, and G3BP1 dephosphorylation, the condensates dissolve.
Importantly for what follows, SG assembly is redox-sensitive. GAPDH (glyceraldehyde-3-phosphate dehydrogenase) forms oligomers under oxidative stress that co-assemble with TIA-1 in SGs (Arimoto et al., 2008, Nature Cell Biology). Oxidised GAPDH acts as a scaffold that promotes SG nucleation independently of eIF2-alpha phosphorylation. This establishes a direct link between cytoplasmic redox state and SG condensate assembly.
P-bodies
Processing bodies (P-bodies) are distinct from SGs but dynamically exchange components with them. P-bodies concentrate the mRNA decapping machinery (DCP1a/DCP2), the Lsm1-7 complex (which binds deadenylated 3-prime ends), the decapping activators EDC3, EDC4 (also called Ge-1 or Hedls), and the 5-prime-to-3-prime exonuclease XRN1. They also contain the deadenylase complexes CCR4-NOT and PABP-interacting proteins. The IDR in DCP1a and the FDF motif-containing IDRs of EDC3 drive P-body condensation. P-bodies are constitutively present at low abundance but expand dramatically under stress and nutrient deprivation, suggesting that metabolic state regulates their dynamics.
Metabolic regulation of condensate properties
A growing literature documents that condensate formation is sensitive to metabolite concentrations, post-translational modifications driven by metabolic enzymes, and cellular energy charge. Several mechanisms are relevant to the spirulina question:
First, adenylate energy charge (AEC = [ATP + 0.5 ADP] / [ATP + ADP + AMP]) directly affects condensate formation. RNA helicases that dissolve aberrant condensates — notably DDX3X and DHX9 — require ATP hydrolysis. When AEC falls, helicase activity decreases and condensates persist longer. AMPK activation (which occurs as AEC falls) has been shown to phosphorylate several IDR-containing proteins, altering their charge and affecting phase behaviour. Ohn et al. (2008) and subsequent work showed that the AMP-to-ATP ratio modulates SG assembly dynamics independently of the canonical eIF2-alpha kinase pathway.
Second, redox state alters IDR behaviour through oxidation of cysteine and methionine residues. Cysteines within IDRs can form intra- or inter-molecular disulfide bonds under oxidative conditions, cross-linking the condensate scaffold and converting it from a dynamic liquid to a more solid-like gel or aggregate. Methionine oxidation (to methionine sulfoxide) alters the local charge and hydrophilicity of IDRs, shifting their phase-separation propensity. Mateju et al. (2017, EMBO Journal) showed that aberrant condensate solidification underlies amyloid-like SG inclusions in ALS-associated FUS mutants.
Third, iron concentration affects specific condensate-associated proteins. Hsp27 (HSPB1) is a small heat-shock protein that partitions into SGs and modulates their dynamics; its phospho- state (controlled by p38-MK2 signalling) determines whether it promotes or inhibits SG formation. Iron-responsive element binding protein 1 (IRP1/ACO1) is a bifunctional protein: in iron-replete cells it acts as cytoplasmic aconitase; in iron-depleted cells it binds IRE stem-loops in mRNA and affects translation of iron-responsive transcripts. IRP1 has been found in SG fractions, and its iron-sensing conformational switch could directly couple iron availability to SG composition.
Where spirulina connects — and where it does not
Direct studies of spirulina and LLPS do not exist in the published literature as of mid-2026. What follows is mechanistically informed speculation grounded in established spirulina biology and condensate physics.
Phycocyanin (specifically phycocyanobilin, the tetrapyrrole chromophore) is the primary antioxidant effector of spirulina. Phycocyanobilin quenches singlet oxygen and hydroxyl radicals and activates Nrf2/KEAP1 signalling to upregulate catalase, SOD, HO-1, and glutathione peroxidase. If the redox-sensing mechanism of GAPDH-driven SG assembly is correct, then a reduction in cellular oxidative stress by phycocyanin would be expected to reduce GAPDH oligomerisation and attenuate stress-driven SG nucleation. This would predict more efficient mRNA translation (less sequestration in SGs) and faster return to homeostasis after transient oxidative challenge. This is plausible and mechanistically coherent, but it is speculative in the absence of direct LLPS assays with phycocyanin.
Spirulina's well-documented activation of AMPK (demonstrated in multiple in vitro and rodent models; the mechanism may involve the GW-acid component reducing mitochondrial electron transport efficiency, or phycocyanin affecting Complex I, raising AMP:ATP ratio) has direct consequences for condensate biology through two paths. First, AMPK phosphorylates eEF2K (eukaryotic elongation factor 2 kinase), reducing elongation rate and promoting ribosome stalling, which can feed into SG formation under conditions where elongation and initiation are mismatched. Second, AMPK activation of SIRT1 (via NAMPT-driven NAD+ production) leads to deacetylation of numerous IDR-containing proteins, altering their charge and phase behaviour. These are second- and third-order effects, not primary condensate regulators.
The phycocyanobilin tetrapyrrole has a pi-conjugated aromatic system that, in principle, can engage in pi-stacking with aromatic residues in IDRs or with RNA nucleobases. However, the intracellular bioavailability of phycocyanobilin (as a free chromophore rather than covalently attached to phycocyanin protein) is uncertain — phycocyanin is digested in the gut and phycocyanobilin is partially absorbed as a bile pigment analogue, but nuclear concentrations are unknown. Any claim that phycocyanobilin directly modulates condensate formation via pi-stacking would require intracellular co-localisation data that does not yet exist.
Pathological condensates and disease relevance
LLPS biology has clarified the molecular basis of several neurodegenerative diseases. ALS- and FTD-associated mutations in FUS, TDP-43 (TARDBP), and hnRNPA1/A2 cluster in their prion-like IDRs and promote conversion of normal liquid condensates to pathological solid aggregates. The same pi-pi stacking interactions that drive normal LLPS are hijacked by these mutations to create irreversible cross-beta fibrils. Huntington's disease involves polyglutamine (polyQ) expansions that drive pathological condensation. Understanding normal condensate regulation may therefore inform therapeutic strategies for these conditions — though this connection to spirulina remains entirely speculative at present.
Technical note on condensate assays
Researchers studying condensates rely on a battery of techniques: fluorescence recovery after photobleaching (FRAP) to assess internal dynamics (liquids recover quickly; gels do not), droplet fusion assays, optogenetic condensate induction (CRY2-based systems), and in vitro reconstitution with purified IDR proteins. A compelling future experiment would be to measure SG assembly kinetics and dissolution rates in cells pre-treated with phycocyanobilin or spirulina extract under defined oxidative stress conditions, using FRAP to confirm liquid-like behaviour is maintained. No such study has been published.
Summary
Liquid-liquid phase separation provides a unifying physical framework for understanding membraneless organelles — from nucleoli to stress granules to transcriptional condensates. The IDRs of proteins like FUS, G3BP1, and TIA-1 drive condensate formation through pi-pi and cation-pi interactions, with concentration thresholds that make these assemblies acutely sensitive to changes in protein levels, post-translational modifications, and metabolite concentrations. Cellular redox state and AMPK activity — both targets of spirulina's principal bioactives — are established modulators of condensate behaviour. The mechanistic connections are real and worth investigating experimentally. The direct evidence, however, does not yet exist, and readers should treat any specific claims about spirulina modulating condensates as hypothesis-generation rather than established fact.