Review Papers Claim by Claim

Fast but frequency-tuned membrane retrieval & clathrin division of labor

Based on the evidence you provided, the paper most directly supporting the “ultrafast endocytosis / synaptic recycling” framing is the Neuron study (10.1016/j.neuron.2014.05.007), which argues for a frequency-dependent split: clathrin-independent endocytosis (CIE) dominated by dynamin-1/3 + endophilin at high-frequency stimulation, while CME via clathrin/AP-2 is important for synaptic vesicle (SV) membrane reformation from endosome-like vacuoles (i.e., a role in the reformation step, not necessarily the initial high-frequency retrieval step).

1) Core claim map (what the model says)

• High-frequency activity (40 Hz) primarily uses clathrin-independent endocytosis (CIE), with recruitment/function tied to dynamin-1/3 and endophilin. Neuron 2014: high-frequency retrieval is CIE-tuned
• Clathrin/AP-2 is not obligatorily required for initial membrane retrieval at high frequency, but it supports SV membrane reformation from endosome-like vacuoles (a “stepwise” model where CME matters downstream for making functional SVs). Neuron 2014: AP-2 in SV reformation
• A parallel kinetic modeling framework supports simultaneous CME and CIE contributions, with CIE rate constants reported to be about ~2× CME. Neuron 2014: parallel kinetics (CIE faster)

Evidence anchor for the above: the provided extract states “5 Hz set to 100%; 40 Hz ~1.08% during-stimulus endocytosis.” Neuron 2014: 5 Hz vs 40 Hz during-stimulus fraction

3) Time-to-endocytosis shifts with clathrin/AP-2 depletion

The extract reports time-to-endocytosis during stimulation (t) under multiple perturbations. Below are two key comparisons: clathrin depletion vs control, and AP-2(m) conditional KO (EtOH vs tamoxifen induction).

The extract provides: for clathrin knockdown, t_ctrl = 24.0 ± 3.0 s vs t_CHC = 49.0 ± 7.0 s at 5 Hz, and t_ctrl = 37.0 ± 3.0 s vs t_CHC = 34.0 ± 2.0 s at 40 Hz (no significant block at high frequency). Neuron 2014: clathrin effects are frequency dependent

Extract summary: AP-2(m) loss slows endocytosis at 5 Hz (EtOH 24.6 ± 2.5 s → Tmx 63.3 ± 12.7 s; p=0.007), but at 40 Hz differences are not significant (EtOH 41.7 ± 2.9 s → Tmx 47.0 ± 3.3 s; p=0.232). Neuron 2014: AP-2 KO is 5 Hz-selective for endocytosis kinetics

4) Ultrastructural “endosome-like vacuoles” and SV density changes in vivo

The paper’s central reconciliation is: even if high-frequency retrieval can proceed via CIE, AP-2 is needed for SV reformation from endosomal-like structures. Below are the extract values for SV density and endosome-like vacuole (ELV) density in vivo.

Extract: AP-2(m) KO reduces SV density (WT 193 ± 7.30 vs KO 113 ± 14.61; p=0.006) and increases ELV density (WT 1.54 ± 0.47 vs KO 11.34 ± 1.23; p=4.2e-6). Neuron 2014: AP-2 KO shifts ELVs up, SV density down

5) Mechanistic interpretation (what is strong vs what remains uncertain)

6) How this fits into the broader synaptic recycling literature (context)

The contextual framing (frequency-tuned ultrafast vs clathrin-dependent steps) is consistent with the broader SV recycling perspective that treats ultrafast endocytosis as an initial clathrin-independent step followed by clathrin-dependent regeneration from endosomes. Annual Review 2025: ultrafast retrieval + endosomal regeneration

A separate review (J Neurosci 2019) emphasizes that multiple SV recycling modes—CME, kiss-and-run, ultrafast endocytosis, ADBE—can be engaged depending on stimulation strength, temperature, and synapse type, which matches the “frequency-dependent division of labor” logic. J Neurosci 2019 review: multiple modes are condition-dependent

7) What experiments would most sharply falsify the model?

8) Publication metrics caveat

You requested an “author review” link at the bottom for all valid full-name authors. The author list is not present in the data you provided, so I cannot reliably generate correct author-specific links without risking false/malformed names.