Alongside the higher neuron density, treated mouse brains showed less neuroinflammation and less amyloid-beta [15]. That complicates the reading of the behaviour. Nesting and maze performance in an amyloid model can improve when plaque load and inflammation come down, without a single support cell becoming anything. The behavioural readout measures function; it does not report where cells came from. Xu is candid on the point, telling Live Science that the dish result has not been confirmed in living mice and that other neural stem cells might also have converted [17].
Delivery gives a second reason for care. New Scientist reports the drug enters brain cells generally, including astrocytes, depletes PTBP1 inside them, and has its antibodies recycled by the cell after about a week [21]. PTBP1 is described as the master switch holding astrocytes back from a neuronal fate [6], so depleting it broadly leaves more than one plausible origin for the new neurons the team found sprouting in the hippocampus [16].
Then the denominator. Twelve engineered mice, half dosed and half given saline [11], which works out to six animals per arm [12]. Within that, the reported effect is not subtle: treated animals nested and ran the maze at the level of mice without the condition, while the saline group did not change [13]. Andras Lakatos at the University of Cambridge told New Scientist there is "clearly an improvement, which is very thought-provoking" [22]. Six per arm can carry a yes-or-no signal. It cannot carry durability or dose response, and it says nothing about how new neurons behave in a live circuit over months, which is what Benedikt Berninger at King's College London asks about [23].
The two write-ups also disagree on schedule. Live Science describes two intravenous doses eight days apart followed by four weeks of behavioural testing [14], while New Scientist describes two injections across two weeks with testing two weeks after that [11][13][31]. Anyone lining this up against the earlier literature needs the Aug. 26 Cell Biomaterials paper itself [1], not the coverage.
What is genuinely new here is human and in vitro: cultured human astrocytes and stem-cell organoids growing axons and firing synchronously [9], with no edit to the genome [3]. The 2020 work that opened this line used CRISPR to deplete PTBP1 in a Parkinson's mouse model and reported converted astrocytes [5]; tracing experiments afterwards could not connect the new neurons back to astrocytes [7]. In mice, this paper offers density plus behaviour, which is the same class of evidence that tracing overturned, so the in vivo burden has not moved. Christiane Wrann of Harvard Medical School, who was not involved, makes the same point about needing to follow an individual astrocyte [18]. In dishes of human cells, the burden has moved, and that is evidence the original dispute never had.
