Huntington PRD swap reshapes toxicity in cortical organoids

What the work claims

Iennaco, Maffezzini, Maestri, Cattaneo and colleagues use a recombination-mediated cassette exchange (RMCE) platform to replace the endogenous mouse Htt exon 1 locus with defined mouse or human HTT exon 1 sequences.¹ In this isogenic background, human mutant huntingtin exon 1 produces more severe neural cyst and neuronal abnormalities than the equivalent mouse mutant sequence. The difference maps to the proline-rich domain (PRD): inserting the human PRD into mouse mutant exon 1 increases toxicity, while replacing the human PRD with the mouse PRD in human mutant exon 1 rescues dendritic architecture, stimulus-induced cFOS activation, and the proteomic profile. The rescue operates primarily at the protein level and converges on the actin cytoskeleton regulator MKL2/MRTFB.

How it works

The authors started from an RMCE-competent mouse ES cell line derived from E14 cells. The endogenous Htt exon 1 on one allele was replaced with an RMCE cassette; the other allele was either deleted (RMCE/-) or left wild-type (RMCE/+). Into this site they integrated human or mouse exon 1 variants, creating lines that express murine exon 1 with 7Q (mWT/+) or 107Q (mHD/+), human exon 1 with 20Q (hWT/+) or 72Q (hHD/+), and reciprocal chimeras: mHD hP/+ (mouse HD 104Q carrying the human PRD) and hHD mP/+ (human HD 72Q carrying the mouse PRD). Correct targeting, expression, and pluripotency were confirmed by PCR, sequencing, western blot, and OCT3/4/SOX2 staining.

To model early neurogenesis, they generated neural cysts: three-dimensional polarized neuroepithelial organoids grown for six days in single-cell suspension. High-content imaging and principal component analysis of size and shape parameters showed that human HD cysts (hHD/+) were more abnormal than mouse HD cysts (mHD/+). The chimeric lines moved in opposite directions: mHD hP/+ cysts shifted toward the more severe hHD/+ profile, while hHD mP/+ cysts clustered closer to the milder mHD/+ profile. The PRD therefore behaves as a bona fide species-specific modifier of early developmental toxicity.

The same pattern persisted in mature cortical neurons differentiated to DIV28. The neurons were predominantly dorsal cortical (TBR1+, CTIP2+, VGLUT2+). Human HD neurons had a median total dendritic length (TDL) of 93.7 µm, compared with 130.6 µm for mouse HD neurons. Replacing the human PRD with the mouse PRD increased TDL by 51 percent relative to hHD/+. After three-hour glutamate stimulation, cFOS induction was more strongly impaired in hHD/+ neurons than in mHD/+ neurons; the human PRD inserted into mouse exon 1 worsened cFOS activation (55.8 percent MAP2+/cFOS+ cells versus 67.0 percent in mHD/+), while the mouse PRD in human exon 1 improved it by 23.6 percent relative to hHD/+.

Global proteomics of mWT/+, hHD/+, and hHD mP/+ neurons identified 389 differentially expressed proteins. Hierarchical clustering placed hHD mP/+ closer to mWT/+ than to hHD/+, and 311 proteins (80 percent of the differentially expressed set) were restored toward wild-type levels by the mouse PRD. In contrast, bulk RNA-seq identified 2,522 differentially expressed genes but found no significant transcriptomic difference between hHD/+ and hHD mP/+ neurons. The rescue is therefore post-transcriptional, consistent with the PRD's known role in protein-protein interactions rather than direct transcriptional control.

Bioinformatic analysis of the rescued proteins pointed to actin cytoskeleton organization as the top pathway. Among the most strongly normalized proteins were MKL2 (also called MRTFB), SCMH1, and UBE2B. Western blotting across five independent neuronal differentiations confirmed that MKL2 protein was reduced in hHD/+ neurons and restored in hHD mP/+ neurons. MKL2 was also reduced in the cortex of R6/2 Huntington's mice (n = 3 animals per genotype) and in post-mortem cortex from Huntington's patients (n = 4 individuals per group). Overexpressing HA-tagged MKL2 at DIV14 rescued dendritic length in hHD/+ neurons to wild-type levels by DIV28 (n = 4 independent replicates), establishing MKL2 as functionally sufficient to reverse at least part of the structural defect.

Where a skeptic should push

The system is built in mouse embryonic stem cells, not human induced pluripotent stem cells. Human exon 1 is being studied in a murine genomic and cellular context, so some of the observed species-specific effects could reflect differences in mouse versus human flanking regulatory sequences or interacting partners rather than the PRD alone.

The neurons are dorsal cortical, whereas Huntington's disease primarily kills striatal medium spiny neurons. Neural cysts model very early neurogenesis, so the relevance to adult-onset neurodegeneration is indirect. Sample sizes are also modest for some key validations: three animals for the R6/2 cortex, four human post-mortem subjects per group, and five or four replicates for the neuronal western blots and rescue experiments.

The proteomic rescue is broad, but the authors have not shown exactly how the PRD modulates MKL2. Possible mechanisms include altered protein stability, subcellular localization, or ribosome stalling at the polyQ-PRD junction. Until the direct physical link is mapped, MKL2 should be treated as a strong correlate and a promising target, not a proven causal intermediary.

What this means for HD organoid model fidelity

For organoid-based Huntington's research, this study is a warning and a toolkit. The warning is that mouse models carrying chimeric or fully murine huntingtin exon 1 may underestimate the toxicity of the human sequence. A humanized exon 1 knock-in is not the same as the human disease, but it is closer than rodentized constructs. Groups building patient-derived or isogenic human iPSC organoids should pay attention to the PRD length and composition, not just CAG repeat number, when comparing models.

The toolkit is the domain-swap strategy itself. The ability to exchange a single protein module in an otherwise isogenic ES or iPSC background, then read out morphological and functional phenotypes in neural cysts and neurons, gives a practical route to map other cis-modifiers of polyglutamine toxicity. This is exactly the kind of mechanistic dissection that organoid platforms are supposed to enable: controlled genetic variation in a human-relevant three-dimensional tissue context.

The opportunity is a new candidate target. If MKL2 reduction is consistently observed across mouse models, human post-mortem tissue, and humanized organoids, restoring MKL2 activity becomes a plausible therapeutic strategy. The threat is premature translation: MKL2 is a transcriptional coactivator with broad targets, and indiscriminately raising its activity could have off-target effects. Organoid and assembloid screens that measure both structural rescue and circuit-level function will be needed before any MKL2-directed approach moves toward the clinic.

The bottom line

This preprint uses an elegant isogenic ES-cell platform to show that the human huntingtin proline-rich domain is a species-specific modifier of neuronal toxicity, acting largely through post-transcriptional mechanisms that converge on MKL2 and the actin cytoskeleton. The neural cyst and cortical neuron organoid assays provide a scalable readout for domain-level structure-function studies. The next test is whether the same PRD-driven, MKL2-dependent phenotypes appear in human iPSC-derived striatal organoids and patient tissue, where the disease actually occurs.