Francis Crick's central dogma describes information flowing from DNA to RNA to protein, and for decades the story seemed to end at the ribosome. The polypeptide emerges, the sequence dictates the fold, and the instruction is executed. Anfinsen's experiments with ribonuclease reinforced this picture: denature a protein, remove the denaturant, and it refolds spontaneously. The information appeared to be entirely contained in the sequence.

The cellular interior tells a more precarious story. Macromolecules reach concentrations of 300 to 400 grams per litre of cytoplasm. Nascent chains expose hydrophobic segments for seconds to minutes before their binding partners exist. Many proteins are only marginally stable, with folding free energies of just 5 to 15 kcal/mol. Thermal fluctuation, oxidative damage, and translational errors continually push proteins toward misfolded states. Aggregation is not an exotic failure. It is a thermodynamically accessible fate for much of the proteome.

Cells respond with an integrated system called the proteostasis network. In human cells it comprises roughly 2,000 components, including molecular chaperones, the ubiquitin-proteasome system, autophagy, and stress-responsive transcriptional programs that tune their capacity. This network functions as the error-correction layer of gene expression. It ensures that encoded information becomes functional structure rather than toxic debris. Understanding how it operates, and how it fails with age, reframes neurodegeneration and much of aging itself as problems of information fidelity downstream of the genome.

Chaperones do not encode folds. They reshape the kinetic landscape so that the native state, already specified by sequence, can be reached without detours into aggregation. The first line of defence acts at the ribosome exit tunnel. In eukaryotes, the nascent polypeptide-associated complex and the ribosome-associated complex shield emerging chains, as trigger factor does in bacteria. They delay premature interactions until enough sequence has been synthesized to form a complete domain.

Downstream, the Hsp70 system serves as the network's central hub. Hsp70 binds short hydrophobic stretches flanked by basic residues, motifs that occur roughly every 40 residues in most proteins. An ATP cycle governs its affinity. J-domain proteins deliver clients and stimulate ATP hydrolysis, which closes the substrate-binding domain around the chain. Nucleotide exchange factors such as Hsp110 and BAG proteins then reset the cycle. Repeated binding and release can unfold kinetically trapped intermediates and give them another attempt at the native state.

Co-chaperones provide much of the system's specialization. Humans express around 50 J-domain proteins but only about a dozen Hsp70 paralogs, so client selection is distributed across the J-protein family. Hsp90 acts later, on near-native clients such as steroid hormone receptors and protein kinases. Co-chaperones like Hop, Cdc37, and p23 coordinate a conformational cycle that stabilizes signalling-competent states. Hsp90 therefore buffers phenotypic variation, allowing destabilized mutant kinases and oncoproteins to remain functional.

For proteins with complex topologies, chaperonins provide physical isolation. Bacterial GroEL-GroES and eukaryotic TRiC/CCT form double-ring cages in which a single polypeptide can fold away from the crowded cytosol, an arrangement often called an Anfinsen cage. Actin and tubulin depend obligately on TRiC. Small heat shock proteins work differently. They are ATP-independent holdases that co-assemble with unfolding proteins during stress, keeping them in a reactivation-competent state instead of letting them form irreversible aggregates.

Even aggregation is not necessarily final. Yeast Hsp104 and bacterial ClpB extract polypeptides from aggregates by threading them through a central pore. Metazoans lack Hsp104 homologs. Instead they combine Hsp70, class B J-domain proteins such as DNAJB1, and Hsp110 to disaggregate amyloid fibrils, including those formed by α-synuclein. The overall architecture is layered and redundant, with holding, folding, and unfolding activities feeding into one another. When demand exceeds capacity, the HSF1-driven heat shock response increases chaperone abundance to restore the balance.

Takeaway

Chaperones add no new structural information. They protect the information already written in the sequence from being corrupted by the kinetics of a crowded cell, which makes them noise suppression rather than instruction.

When folding cannot be rescued, the cell must decide to destroy the protein. This triage happens where chaperones meet the degradation machinery. The E3 ligase CHIP binds the C-terminal EEVD motifs of Hsp70 and Hsp90 and ubiquitinates clients that linger too long in the chaperone cycle. Persistence becomes a signal. A protein that repeatedly fails to reach its native state is tagged for removal, so a kinetic property is converted into a molecular mark.

The ubiquitin-proteasome system handles most of this traffic. A cascade of E1 activating enzymes, E2 conjugating enzymes, and roughly 600 E3 ligases in humans attaches ubiquitin to substrates. K48-linked chains of four or more units serve as the canonical degradation signal. The 26S proteasome recognizes these chains through its 19S regulatory particle and removes the ubiquitin. A hexameric AAA+ ATPase ring then unfolds the substrate and threads it into the 20S core, where three classes of protease activity cleave it into short peptides.

This design imposes a constraint. The proteasome's entry channel is roughly 13 Å wide, so substrates must be unfolded and threaded one chain at a time. The system is therefore precise and fast for soluble, short-lived proteins, including regulatory factors whose timed destruction drives the cell cycle. It handles large aggregates poorly. Compartment-specific branches extend its reach. In ER-associated degradation, misfolded secretory proteins are retrotranslocated through Hrd1 complexes and extracted by the p97/VCP segregase before proteasomal digestion.

Autophagy covers what the proteasome cannot. In macroautophagy, a double-membrane phagophore engulfs cytoplasmic cargo and then fuses with lysosomes. During aggrephagy, selective receptors such as p62/SQSTM1, NBR1, and optineurin bind ubiquitinated aggregates. They dock onto LC3 on the growing membrane through LC3-interacting regions. Chaperone-mediated autophagy is more targeted. Hsc70 recognizes KFERQ-like motifs, present in an estimated 30 to 40 percent of the proteome, and delivers substrates to the lysosomal receptor LAMP2A for direct translocation.

The two pathways differ in mechanism but are coupled in function. Proteasome inhibition induces autophagy, while impaired autophagy causes p62 to accumulate, which can slow proteasomal flux. Ubiquitin chain topology helps route cargo between them. This plasticity provides robustness, but it also means that failure in one arm shifts the load onto the other. Drug developers now exploit this machinery directly. PROTACs and molecular glues recruit E3 ligases to disease proteins, turning the cell's quality-control logic into a therapeutic modality.

Takeaway

Degradation is a decision, not a default. Cells convert the time a protein spends failing to fold into a ubiquitin mark, which makes the capacity to destroy as integral to gene expression as the capacity to build.

Proteostasis capacity is not constant across a lifetime. Work in Caenorhabditis elegans by Richard Morimoto's group showed that the heat shock response declines sharply within hours of reproductive maturity, not gradually over the lifespan. The collapse is regulated. Signals from the germline drive chromatin changes at stress-gene loci, including increased repressive histone marks, which reduce HSF1 binding. Once reproduction begins, the organism effectively deprioritizes somatic maintenance, consistent with the disposable soma theory of aging.

As capacity falls, the proteome's latent instability becomes visible. Proteomic surveys in aging worms identified hundreds of proteins that become detergent-insoluble with age, many with no known disease association. In mammals, proteasome activity declines across tissues. LAMP2A levels at the lysosomal membrane also drop, which impairs chaperone-mediated autophagy. Meanwhile, mitochondrial dysfunction and oxidative damage increase the burden of damaged proteins. The network faces rising demand at the same time that its supply contracts.

Aggregation then becomes self-amplifying. Amyloid formation follows nucleation-dependent kinetics: a slow, rate-limiting nucleation step precedes rapid fibril growth, and fibril surfaces catalyse further nucleation. Aggregates sequester chaperones and ubiquitin, and soluble oligomers can impair proteasome function, diverting resources from other clients. Expressing polyglutamine proteins in C. elegans destabilized unrelated metastable proteins carrying temperature-sensitive mutations. The aggregation of one protein can spread dysfunction across the entire folding environment.

Neurons are especially exposed. They are post-mitotic, so they cannot dilute damaged proteins through cell division. Their long axons and distant synapses also depend on trafficking degradative machinery over great distances. This helps explain why aggregation diseases concentrate in the nervous system: Aβ and tau in Alzheimer's disease, α-synuclein in Parkinson's disease, mutant huntingtin in Huntington's disease, and TDP-43 in most ALS cases. Many of these species also spread between cells in a prion-like manner, templating misfolding in recipient neurons.

Genetic evidence points to causation rather than mere correlation. Mutations in quality-control genes produce neurodegeneration. VCP mutations cause multisystem proteinopathy. Mutations in p62, optineurin, and ubiquilin-2 cause ALS, and defects in Parkin and PINK1 impair mitophagy in familial Parkinson's disease. Conversely, enhancing proteostasis extends lifespan in model organisms. Reduced insulin/IGF-1 signalling activates HSF1 and DAF-16 in worms, and rapamycin promotes autophagy across species. Aging appears, in part, to be the slow erosion of buffering capacity.

Takeaway

Protein aggregation diseases may be less about a single bad protein than about a network that has lost its margin of safety. Aging lowers the threshold, and the aggregating protein is simply the first to cross it.

Proteostasis adds a layer to the central dogma that Crick's original scheme left implicit. Sequence specifies structure, but carrying out that specification depends on a dedicated network of folding, triage, and disposal machinery. Gene expression is not complete when translation ends. It is complete when the protein becomes functional, and ultimately when it is cleared.

This framing is already reshaping medicine. Tafamidis stabilizes transthyretin tetramers to prevent amyloid cardiomyopathy. CFTR correctors such as elexacaftor rescue folding-defective channels in cystic fibrosis. Targeted protein degraders redirect E3 ligases toward previously undruggable targets. Each approach treats folding and degradation as programmable variables rather than fixed outcomes.

The harder challenge is systemic. Boosting one arm of the network carries costs, because HSF1 and Hsp90 also sustain tumour growth. Restoring proteostasis in aging tissues will require understanding the network as an integrated control system. Its overall capacity, not just its individual components, determines how long the information encoded in our genomes is faithfully expressed.