LC Morphology What is Established

An epistemic guardrail for the corpus. The locus coeruleus is the anatomical anchor of Phase I, and the corpus explains its vulnerability partly by the size of its axonal arbor — "vast," "prodigiously arborising," "among the most extensive of any neuron in the brain." The vulnerability conclusion is sound. The arbor-size derivation of it is not. This note fixes what the primary literature establishes about this cell's morphology, so the chapters can rest the argument on the parts that hold.

The corpus's Phase I argument has three morphological load-bearing claims: the cell is long and unmyelinated, it is enormously arborised, and that arbor imposes a quantified metabolic burden. These sit at three very different levels of evidence.

The claims, graded

1. Long, thin, unmyelinated or sparsely myelinated axons — Established, and this is the criterion to use.

This is Braak's own formulation, and it is the one the corpus should cite. Braak, Rüb, Schultz & Del Tredici (2006): "select types of projection cells that generate long, unmyelinated or sparsely myelinated axons are particularly susceptible." It is a steady-state architectural property, not a developmental timetable, and it applies to the coeruleus directly. Independently supported for tau by Rubinski et al. (2022), who found higher myelin content associated with lower susceptibility of connected regions to accumulate fibrillar tau.

Direct ultrastructural support exists for the terminal arbor: cortical noradrenergic axons are unmyelinated, fine-calibre (~0.35 µm) and varicose (Descarries et al. 1977; Séguéla et al. 1990). Conduction velocities in rat are consistent with thin unmyelinated ascending fibres (Aston-Jones et al. 1980).

One awkward datum to carry honestly: roughly a third of monkey LC axons conduct faster than 1 m/s (Aston-Jones et al. 1985), which is hard to reconcile with a uniformly unmyelinated 0.35 µm fibre population, and was never followed up with electron microscopy. No EM count of myelinated versus unmyelinated profiles in the human dorsal noradrenergic bundle appears to exist.

2. Exceptionally extensive and branched arbor — Not established. The best current data point the other way.

The claim's usual citation chain does not terminate in a measurement. Theofilas et al. (2015) — "their poor or incomplete myelination" — cites a textbook and Braak & Braak (1996), which is a paper about neocortex, not the coeruleus. Later reviews cite Theofilas. Mather & Harley (2016) is frequently cited for the unmyelinated claim, but that line appears in an uncited highlights box; their substantive argument is about pacemaking, capillary exposure, and fourth-ventricle proximity, not arbor size.

What direct morphometry exists runs against the superlative. Su et al. (2026), reconstructing 130 complete LC-NE morphologies, report axons averaging 35.14 ± 13.60 cm (longest 72.57 cm) but branching less than the average CNS neuron — 17.8 branches/cm against 24.9 across 1,227 MouseLight neurons — and find individual neurons target particular regions rather than projecting pan-cortically. This corroborates older work: Nagai et al. (1981) described "a predominant type [that] possesses few divergent axons innervating a restricted region."

⚠ Su et al. 2026 is a bioRxiv preprint, and mouse. It is sufficient to retire an overclaim. It is not sufficient to assert the opposite: nothing here licenses calling the LC sparsely branched.

The reconciliation with Schwarz & Luo (2015, Nature) — often cited for broad collateralization — is that their finding is at population level, "with several levels of specificity in certain LC-NE sub-circuits." The nucleus reaches the whole forebrain. The individual neuron, on current evidence, does not.

3. A quantified arbor-driven metabolic burden — Established for a different cell.

The calculation the corpus invokes belongs to substantia nigra dopamine neurons. Pissadaki & Bolam (2013) estimated human SNc axons exceeding 4 m in total length with more than a million synapses, and showed energy cost rising with arbor size by a power law. Pacelli et al. (2015) supplied the causal counterpart: reducing arborization reduced both basal oxidative phosphorylation and toxin vulnerability.

There is no locus coeruleus equivalent of this calculation, and no published estimate of total axonal length per human LC neuron. The figure is not merely unverified in the corpus — it does not exist in the literature.

What the LC bioenergetic evidence actually shows is a different mechanism. Sanchez-Padilla, Guzman, … Surmeier (2014) demonstrated that autonomous pacemaking, driven by L-type calcium channels, produces mitochondrial oxidant stress in LC neurons. This is somatodendritic, not arbor-size-driven. Merging the two is an error the corpus has repeatedly made.

Neuron count — a range, and it is per side

The corpus has stated "approximately 50,000 neurons in the human brain" and, elsewhere, "30,000 to 50,000 per hemisphere." These disagree by a factor of two, and the second is closer to right.

The house formulation already exists, in the Cellular Architecture census, and should be propagated verbatim rather than reinvented:

Counted as tyrosine-hydroxylase-positive, neuromelanin-bearing cells, each locus coeruleus holds on the order of twenty thousand neurons in young adulthood (Manaye et al., 1995); counted inclusively — every medium-to-large neuron, pigmented and unpigmented — nearer fifty thousand per side (Theofilas et al., 2017).

Neither number is wrong; they count different things, and the inclusion criterion must be named. German et al. (1988) give a total across both sides falling from ~45,500 to ~19,000 between ages 60 and 104, on n = 5.

Physical dimensions

The nucleus is a thread: ~14.5–16 mm long and 2–2.5 mm across (Fernandes et al. 2012, n = 40 nuclei from 20 brainstems; German et al. 1988 give a 16 mm rostrocaudal extent). Some imaging papers state 3–4 mm wide; the histological figure is the more defensible one. This matters far beyond anatomy — it is why the structure sits below the resolution of the instruments used to build disease-staging models (see Phase Measurability).

The defensible statement

The locus coeruleus neuron carries long, thin, unmyelinated or sparsely myelinated axons reaching broadly across the forebrain, and it is bioenergetically exposed — but the exposure is demonstrated through autonomous pacemaking and catecholamine oxidative chemistry, not through a measured arbor size. No estimate of total axon length per human LC neuron exists, the quantified arbor-burden calculation in the literature is for substantia nigra dopamine neurons, and the best direct morphometry finds LC axons long but less branched than the average CNS neuron. The vulnerability claim survives; the arbor-size derivation of it does not.

Where the arbor argument is sound

Retiring it for the coeruleus does not retire it everywhere, and the corpus should not over-correct:

Substantia nigra dopamine neurons — the argument originates here and is quantified (Pissadaki & Bolam 2013; Pacelli et al. 2015).
The nucleus basalis of Meynert cholinergic neuron — a large projection neuron with a genuinely extensive arbor, and the corpus's parallel between it and the coeruleus is drawn on other grounds (tonic firing, oxidative exposure, early severe loss) that stand independently.
The parvalbumin basket cell — "thousands of targets from a single cell" is a connectivity claim about perisomatic innervation and is not in dispute.

The four-feature vulnerability phenotype in the Cellular Architecture census — extreme metabolic demand, extensive arbor or outsized connectivity load, high oxidative exposure, dependence on a protective envelope — survives intact under the disjunctive reading. It is the LC-specific superlative that fails, not the pattern.

How the corpus should use this

Any chapter asserting that the coeruleus is undone by the sheer size of its arbor should be read as carrying this hedge. The preferred formulations are Braak's density criterion for the anatomy and Surmeier's pacemaking evidence for the energetics. The phrases "vast arbor," "prodigiously arborising," "among the most extensive of any neuron in the brain," "4-metre cumulative axonal arbor," and "cumulative axonal arbors of several meters" are retired.

The audit that raised this is The Three Rivals §IV.4; the correction record is research/CORRECTIONS.md, entries C-002 and P-001.

Primary sources

Braak H, Rüb U, Schultz C, Del Tredici K (2006). Vulnerability of cortical neurons to Alzheimer's and Parkinson's diseases. J Alzheimers Dis 9(3 Suppl):35–44. PMID 16914843
Descarries L, Watkins KC, Lapierre Y (1977). Noradrenergic axon terminals in the cerebral cortex of rat. Brain Res 133(2):197–222. PMID 902092
Séguéla P, Watkins KC, Geffard M, Descarries L (1990). Noradrenaline axon terminals in adult rat neocortex. Neuroscience 35(2):249–264. PMID 2116602
Aston-Jones G, Segal M, Bloom FE (1980). Brain aminergic axons exhibit marked variability in conduction velocity. Brain Res 195(1):215–222. PMID 7397496
Aston-Jones G, Foote SL, Segal M (1985). Impulse conduction properties of noradrenergic locus coeruleus axons projecting to monkey cerebrocortex. Neuroscience 15(3):765–777. PMID 4069354
Nagai T, Satoh K, Imamoto K, Maeda T (1981). Divergent projections of catecholamine neurons of the locus coeruleus. Neurosci Lett 23(2):117–123. PMID 7254696
Schwarz LA, Miyamichi K, … Luo L (2015). Viral-genetic tracing of the input–output organization of a central noradrenaline circuit. Nature 524(7563):88–92. PMID 26131933
Pissadaki EK, Bolam JP (2013). The energy cost of action potential propagation in dopamine neurons. Front Comput Neurosci 7:13. PMID 23515615
Pacelli C, Giguère N, … Trudeau L-É (2015). Elevated mitochondrial bioenergetics and axonal arborization size are key contributors to the vulnerability of dopamine neurons. Curr Biol 25(18):2349–2360. PMID 26320949
Sanchez-Padilla J, Guzman JN, … Surmeier DJ (2014). Mitochondrial oxidant stress in locus coeruleus is regulated by activity and nitric oxide synthase. Nat Neurosci 17(6):832–840. PMID 24816140
Manaye KF, McIntire DD, Mann DMA, German DC (1995). Locus coeruleus cell loss in the aging human brain: a non-random process. J Comp Neurol 358(1):79–87.
Theofilas P, Ehrenberg AJ, Dunlop S, et al. (2017). Locus coeruleus volume and cell population changes during Alzheimer's disease progression. Alzheimers Dement 13(3):236–246.
German DC, Walker BS, Manaye K, et al. (1988). The human locus coeruleus: computer reconstruction of cellular distribution. J Neurosci 8(5):1776–1788. PMID 3367220
Fernandes P, Regala J, Correia F, Gonçalves-Ferreira AJ (2012). The human locus coeruleus 3-D stereotactic anatomy. Surg Radiol Anat 34(10):879–885. PMID 22638719
Rubinski A, et al. (2022). Higher levels of myelin are associated with lower susceptibility to accumulate amyloid-β and tau. Alzheimers Res Ther. PMID 36153607
Su Z, Kosillo P, Jung K, et al. (2026). Complete morphologies of locus coeruleus noradrenergic neurons. bioRxiv — preprint, not peer-reviewed

Converges on

Locus coeruleus as ground zero · Region-selective neuronal vulnerability · Bioenergetics · Mitochondrial Dysfunction · Phase Measurability · Perineuronal Nets