Merkel Cells: Touch Receptors That Are Not Neurons

If a cell in the skin is not a neuron, how can it be the receptor for a nerve that keeps firing as long as you press?

The touch and pain lesson names the slowly adapting type I (SA1) afferent, the fibre that keeps firing during a held touch, and notes that Piezo2 is needed in Merkel cells. This lesson starts at the question that sentence leaves open. If a non-neuronal cell transduces force, which part of the SA1 response does it make, and how does it pass the signal to the nerve? The answers come mostly from mice, and the second one is still disputed.

An epithelial cell that feels force

A Merkel cell-neurite complex is a Merkel cell in the basal epidermis paired with the terminal of one Aβ low-threshold afferent. In hairy mouse skin the Merkel cells cluster in touch domes beside guard hairs, and the afferent ends in disc-like contacts on them. The cell looks neuronal. It carries neuronal markers and, as you will see, it fires spikes. But genetic lineage tracing in mouse embryos (Van Keymeulen and colleagues, 2009) showed that Merkel cells arise from epidermal progenitors. For mouse skin, then, a Merkel cell is an epidermal cell. It belongs with the cells described in epithelia, not with the neural crest derivatives of neural crest, a distinction that matters again below, because the lineage of the cell and the lineage of the afferent differ.

The afferent’s output has two phases, defined in mouse touch-dome recordings (Maksimovic and colleagues, 2014). The dynamic phase is the high-frequency firing at the onset of a touch. The static phase is the sustained firing that follows, with variable interspike intervals. Adaptation and rate coding are covered in sensory coding; here the two phases are two measurements, and the lesson asks which cell is responsible for each.

Three papers, three preparations, 2014

Three papers appeared in 2014, and their value lies in how different their preparations were. Read the preparation before the result.

Maksimovic and colleagues recorded single touch-dome afferents in an ex vivo mouse skin-saphenous nerve preparation. Selective optogenetic depolarization of Merkel cells was sufficient to evoke SA1 action potentials. Optogenetic hyperpolarization suppressed sustained touch-evoked firing; in the reported experiment (three units) median firing fell tenfold. In mice with a developmental Atoh1 deletion that leaves no Merkel cells, the afferents still responded to touch, but their dynamic firing was lower. Atoh1 loss is developmental, so it cannot say what the cells do acutely in an adult.

Woo and colleagues, from the Patapoutian laboratory, worked with Merkel cells purified from the dorsal skin of newborn mice (P0 to P5), cultured overnight. They found Piezo2 transcripts in the purified cells. When Piezo2 was deleted from skin epithelium, none of 13 cultured Merkel cells had a detectable mechanically activated current. The afferents that contact Merkel cells still expressed Piezo2, and the complexes appeared anatomically normal. A Cre driver for epithelium is not specific to Merkel cells, so the experiment removes the channel from the whole epidermal lineage, and only the Merkel cells were recorded.

Ikeda and colleagues, from the Gu laboratory, patch-clamped Merkel cells in place inside excised rat whisker follicles (age not stated in the accessible text). Tetrodotoxin did not block the action potentials evoked by depolarizing the cell, and cadmium, which blocks voltage-gated calcium channels, abolished them. Mechanically activated currents were rapidly adapting with a small steady component. Knocking down Piezo2 in the follicle reduced slowly adapting responses recorded from the afferent bundle. A rat whisker follicle is not a mouse touch dome, so a difference between this paper and the other two could be a difference of preparation.

Excitability does not make the Merkel cell a neuron. Many non-neuronal cells fire calcium spikes, and the lineage evidence puts the cell in the epidermis.

Splitting the response: what the cell adds and what the neurite keeps

Put the three preparations together and you get the two-receptor model. The Merkel cell transduces force through Piezo2 and shapes the sustained, static firing; the afferent ending has its own mechanosensitive machinery and can still signal touch without Merkel cells. The model has limits, and the numbers show them.

In the cultured mouse Merkel cells of Maksimovic and colleagues (dorsal skin, P2 to P5), the mechanically activated current averaged 370 pA at its peak (±80 pA) and 20 pA at steady state (±6 pA), and it inactivated with a time constant of 8 ms. Taking the ratio of the two group means, the steady current is 20/370, about 5.4% of the peak. The cell’s own current is mostly transient, yet the afferent it drives fires for as long as the touch lasts, with irregular intervals: under prolonged photostimulation the interspike-interval coefficient of variation averaged 1.17 (12 units). A sustained afferent response is therefore not a copy of a sustained receptor current. How the transient current becomes sustained firing is not explained by these data.

Epithelial Piezo2 deletion in the same laboratory’s touch-dome recordings gave a clean split: dynamic responses were similar to controls, and normal sustained firing was lost. That is the evidence behind the model, and it comes from Maksimovic and colleagues’ skin-nerve recordings. Woo and colleagues’ cultured-cell experiments did not test it, although their own skin-nerve recordings from these knockouts also found lower static firing. The Atoh1 result is messier, as the next problem shows.

Worked problem: how much does losing Merkel cells cost the dynamic phase?

In female mice aged 8 to 15 weeks, in an ex vivo skin-nerve preparation, Maksimovic and colleagues recorded touch-dome afferents in mice with a developmental Atoh1 deletion that removes Merkel cells. Peak SA1 firing was 79 ± 40 Hz in the knockouts and 238 ± 69 Hz in controls (mean ± SD).

1. Remaining fraction: 79 Hz / 238 Hz = 0.332, so the knockout afferents reached 33.2% of the control peak.
2. Reduction: (238 - 79) Hz / 238 Hz = 159/238 = 0.668, a 66.8% fall. As a check, 33.2% + 66.8% = 100%.
3. Scope of the number: this is a ratio of two group means, not the change within any one afferent. The standard deviations (40 and 69 Hz) tell you the groups are spread widely.

reduction = (control - knockout) / control = (238 - 79) / 238 = 66.8%

What follows from it? The popular summary that Merkel cells do only the static phase is refuted: peak firing, which belongs to the dynamic phase, fell by about two thirds. But the number does not say that Merkel-cell Piezo2 accounts for that whole 66.8%. Atoh1 deletion removes the cells during development, with whatever knock-on effects that has on the skin and on the afferent. The Piezo2 deletion left dynamic responses similar to controls, which is a different result. The honest summary is that Merkel cells affect dynamic firing, and a developmental deletion cannot say through which mechanism.

The synapse nobody agrees on: noradrenaline, serotonin or protons

If the Merkel cell drives the afferent, it must send something across the gap. Blocking vesicle release in the epidermis with tetanus toxin reduced both dynamic and static SA1 firing in mouse touch domes, so release matters (Hoffman and colleagues, 2018; ex vivo hindlimb skin-saphenous nerve, 7 to 10 weeks). Three accounts say what is released, and each holds in its own preparation.

Noradrenaline, mouse touch dome. Hoffman and colleagues, from the Lumpkin laboratory, found transcripts for presynaptic machinery and adrenergic biosynthesis in purified adult Merkel cells. The cells loaded a fluorescent vesicle tracer, FFN206, and calcium-channel blockers prevented its release on depolarization. Local noradrenaline evoked spikes in touch-dome afferents without touch (half-maximal concentration 2.5 mM in this bath assay), and deleting Adrb2, the gene for the β2-adrenergic receptor, in neural-crest derivatives with Wnt1-Cre attenuated dynamic and static firing. That Cre driver is broad and the deletion is developmental. The authors left the channel that β2 signalling opens for future work. A β2 receptor is a G protein-coupled receptor (see GPCRs and G proteins and cAMP and PKA), but nothing here shows which conductance follows it.

Serotonin, mouse whisker. Chang and colleagues (2016) reported mechanically triggered serotonin release from mouse whisker Merkel cells.

Protons, rodent whisker. Yamada and colleagues (2024) recorded with a pressure patch at the afferent’s heminode, the terminal region just before its first node, in rat (24 to 60 days) and mouse (6 to 13 weeks) whisker complexes. A β2 blocker did not change the mechanically evoked fast EPSCs, and ASIC3 deletion diminished them. They attributed fast transmission to neuritic ASICs, which are channels gated by protons.

These findings are not contradictory in any simple way. The endpoints differ (firing in one case, fast postsynaptic currents in the other), as do the sites and the manipulations. Calling the adrenergic synapse settled is wrong, and so is calling it refuted. The experiment that would decide it measures both endpoints at both sites in the same hands.

What SA1 input is for, and a cancer that borrowed the name

Knowing that the cell transduces does not tell you what the brain does with SA1 spikes. In anaesthetized rhesus macaques, Weber and colleagues (2013) recorded 15 SA1, 13 RA and 7 PC fingertip afferents while textures moved at 40, 80 and 120 mm/s, and compared them with roughness judgments by human volunteers aged 18 to 31. In this texture set, SA1 firing represented millimetre-scale features spatially, and fine texture was carried by the timing of RA and PC responses. So Merkel-associated signalling should not be taught as the sole code for fine natural texture. In human microneurography at the fingertip, Pruszynski and Johansson (2014) found that first-order neurons signalled edge orientation through both response intensity and temporal structure; the vibration-sensitive afferents are the subject of vibration corpuscles.

A skin cancer, Merkel cell carcinoma, is named after the cell, but the name is not evidence that the cell of origin is a touch-receptor Merkel cell.

Design problem

Hoffman’s mouse touch-dome data support β2 receptors, and Yamada’s rodent whisker data show fast EPSCs that β2 block did not change and ASIC3 deletion reduced. Design experiments that could decide whether these results conflict.