Human glia can both induce and rescue aspects of disease phenotype in Huntington disease

Abstract

The causal contribution of glial pathology to Huntington disease (HD) has not been heavily explored. To define the contribution of glia to HD, we established human HD glial chimeras by neonatally engrafting immunodeficient mice with mutant huntingtin (mHTT)-expressing human glial progenitor cells (hGPCs), derived from either human embryonic stem cells or mHTT-transduced fetal hGPCs. Here we show that mHTT glia can impart disease phenotype to normal mice, since mice engrafted intrastriatally with mHTT hGPCs exhibit worse motor performance than controls, and striatal neurons in mHTT glial chimeras are hyperexcitable. Conversely, normal glia can ameliorate disease phenotype in transgenic HD mice, as striatal transplantation of normal glia rescues aspects of electrophysiological and behavioural phenotype, restores interstitial potassium homeostasis, slows disease progression and extends survival in R6/2 HD mice. These observations suggest a causal role for glia in HD, and further suggest a cell-based strategy for disease amelioration in this disorder.

Figure 1. Mice may be generated with striata chimeric for human HD ESC-derived glia.

The striata of neonatally engrafted rag1 null mice were efficiently colonized with donor hESC hGPCs, a proportion of which differentiated as astroglia in both striatal grey and white matter. (a) Mice engrafted with hESC GPCs expressing either normal Htt (GENEA19; 18Q) or mHtt (GENEA20, 48Q; a sibling to GENEA19) manifested striatal chimerization by 20 weeks of age, which was denser at 40 weeks. Mice engrafted with GENEA20-derived glia (48Q) manifested striatal chimerization analogous to that of GENEA19-derived normal HTT (18Q) glia. (b) GENEA19-derived glia identified by their expression of the human-specific nuclear antigen (hNA; red) interspersed with host cells (DAPI, blue), revealing extent of striatal and cortical human glial chimerization at 40 weeks. (c,d) GENEA19 hGPC-engrafted striatal sections at 20 (c) and 40 (d) weeks post graft, stained for human and mouse NG2, showing the progressive domination of the striata by human NG2-defined GPCs. (e,f) 20 (e) and 40 (f) week post-graft striata, stained for hNA (red) and human PDGFRA (green) similarly showing the progressive domination of the striata by hGPCs. (g,h) GENEA19 hGPC-engrafted striata at 20 (g) or 40 (h) weeks, stained for hNA (red) human GFAP (green), showing the maturation and age-dependent increase in fibre complexity of human astroglia in the host striatum. Arrows indicate graft-derived OPCs (e,f) and astrocytes (g,h). Scale bars, 1 mm (a,b); 50 μm (c–g); 25 μm (h). DAPI, 4,6-diamidino-2-phenylindole.

Figure 2. HD ESC-derived glial chimeras exhibit impaired motor coordination.

Mice engrafted with GENEA20-derived glia expressing mHtt demonstrated significantly impaired motor coordination compared with littermates chimerized with either GENEA19-derived normal HTT GPCs or control mice (both sham-treated and untreated). In particular, the mHTT glial chimeras (GENEA20, with 48Q; n=31) manifested significant age-dependent decrements in motor coordination relative to their normal HTT (GENEA19, 18Q; n=28) hESC GPC-derived chimeric controls, as well as relative to sham-treated (n=11) and untreated controls (n=21). Two-way ANOVA revealed both a significant treatment effect (F(3, 593)=39.6, ***P<0.0001) and time effect (F(7, 593)=5.47, P<0.0001); mean±s.e.m.

Figure 3. Striatal neurons are hyperexcitable in mice chimerized with mHTT-transduced hGPCs.

Human glial chimeric striata were established with fetal human glia transduced to overexpress mHTT. (a) Example of a host striatal neuron, filled with Alexa-594 after recording, surrounded by EGFP-tagged donor-derived glia. (b) Representative action potentials recorded in response to current injection in host neurons in striata chimerized with 73Q mHTT (n=8)- and 23Q HTT (n=12)- expressing hGPCs. Striatal neurons of mice engrafted with 73Q mHTT human glia required significantly fewer current injections to achieve voltage thresholds for firing, than did those engrafted with 23Q HTT-transduced or either EGFP-only (n=8) transduced or untreated (n=8) control glia. (c) Current–voltage curves (I–V curves) of neurons in 23Q HTT- and 73Q mHTT-expressing hGPC-engrafted mice, reflect the typical inwardly rectifying currents of MSNs. (d) Representative traces of injected current (20-pA steps)-induced voltage changes are shown for the four treatment groups. The waveforms of stimulus injection are shown below the tracings. (e) The relative hyperexcitability of striatal neurons in the mHtt glial environment was also reflected by the higher resting membrane potential of those neurons, relative to both Q23 hGPC-engrafted and unengrafted controls. (f) The input resistances at negative current injection (R_Neg, −40 pA hyperpolarization currents) were compared with those with positive current injection (R_Pos, 40 pA depolarization currents), and confirmed the higher input resistance of striatal neurons in 73Q glial chimeras, relative to both 23Q and GFP control glia-engrafted mice. (g,h) Comparison of frequency (g) and amplitude (h) of sEPSCs and miniature EPSCs (mEPSCs). Despite their relative hyperexcitability, striatal neurons within 73Q glial chimeric striata manifested sEPSC frequencies and amplitudes that did not differ significantly from those of either 23Q glial-engrafted or unengrafted striatal neurons. Scale bar, 50 μm (a); (e,f) **P<0.01; **** P<0.0001 by ANOVA with post hoc t tests; means±s.e.m.

Figure 4. CD44-sorted hGPCs colonized and replaced endogenous glia within the R6/2 × rag1^−/− striatum.

Striatal engraftment of the R6/2 mice by CD44-sorted hGPCs was robust and dense. (a,b) Fetal derived cells expanded to colonize the striata and ventral forebrain of engrafted mice by 20 weeks. (c) Donor-derived cells in the striata of transplanted mice increased as a function of time (means±s.e.m.). (d–g) By 20 weeks after neonatal graft, the donor hGPCs (human nuclear antigen, red) integrated as astrocytes (d; GFAP, green) or persisted as GPCs (e,f; PDGFαR and olig2, green), but did not give rise to neurons; no overlap was ever seen of hNA and NeuN expression (g; NeuN, green). (h) Resident human glia did not manifest detectable nuclear Htt aggregates, as assessed by EM48 immunostaining; the staining patterns of host Htt and donor human nuclear antigen were always entirely non-overlapping (h). Scale bars, 1 mm (a,b); 25 μm (d–h).

Figure 5. Chimerization with normal glia slows motor loss and extends survival of R6/2 mice.

(a) Linear regression revealed that the rate of rotarod-assessed motor deterioration of R6/2 mice was significantly slower in mice engrafted with hGPCs (n=15) than in untreated mice (sham-treated, n=11; untreated, n=10) (F(3,608)=41.87; P<0.001). (b) R6/2 (120Q) × rag1^−/− mice whose striata were engrafted with human GPCs survived significantly longer than unengrafted mice (n=29 hGPC-engrafted; n=28 untreated; P<0.01 by Mantel–Cox Log-rank test). (c) Striatal volumes were estimated stereologically (Stereo Investigator, MicroBrightfield). hGPC-engrafted R6/2 mice manifested larger striatal volumes than unengrafted R6/2 mice by 16 weeks of age, which were restored to levels no different than those of WT controls. Means±s.e.m.; **P<0.01 and ****P<0.0001 by one-way ANOVA with post hoc t-tests.

Figure 6. Treatment with hGPCs partially rescues disease signatures in SmartCube and NeuroCube and improved cognitive deficit in T-maze.

To build a two-dimensional representation of the multidimensional space in which the groups are best separated, we first find statistically independent combinations of the original features, pick the two new composite features (axes 1 and 2) that best discriminate between the groups, and used them as x and y axes. Each dot represents a mouse. The centre, small and large ellipses are the mean, s.e. and s.d. of the composite features for each group. The distance and overlap between the groups are used to calculate the discrimination index, which indicates how reliably a classifier can be trained to discriminate between the two groups. (a) SmartCube showed a significant difference between sham-treated WT and R6/2 mice (91%) at 8 weeks of age. There was a significant functional preservation (36%) in hGPC-treated versus untreated R6/2 mice. (b) At 11 weeks of age there was also a significant R6/2-associated deficit (92%), with a marginally significant recovery (19%) in response to hGPC treatment. (c) In NeuroCube, R6/2 mice had a marginally significant deficit (66%) and therefore no significant recovery by hGPC treatment could be measured. (d) However, at 11 weeks of age, a significant deficit (88%) and recovery (46%) by hGPC treatment was noted. The improvement of motor and cognitive behaviour of hGPC-engrafted R6/2 mice was also manifest in the T-maze test. (e) R6/2 mice chose the correct arm fewer times than WT mice over training at 8 weeks of age and again when retrained at 13 weeks of age (as compared with the sham-treated WT mice). GPC-treated R6/2 mice showed better performance than sham-treated R6/2 mice in specific sessions at both ages. (f) Fewer R6/2 than WT mice reached criterion (6 out of 8 correct trials per day for 3 consecutive days) during training and again during retraining at the older age. hGPC treatment improved acquisition in R6/2 mice during the initial training phase. (g) R6/2 mice were slower than WT mice to reach the platform during training and retraining. hGPC treatment improved performance during retraining. Asterisks denote significant main effects or post hocs and (means±s.e.m.; *P<0.05; **P<0.01; ***P<0.001.)

Figure 7. Chimerization with normal glia partially normalizes MSN physiological function.

(a) The current–voltage relationship (I–V) derived from whole-cell V-clamp recordings in WT and R6/2 × rag1^−/− (R6/2) mice. (b) Representative whole-cell I-clamp recordings from rag1^−/− WT, CD44 hGPC-engrafted rag1^−/− WT's, R6/2 × rag1^−/− mice and CD44-engrafted rag1^−/− mice. Lines below each group of traces indicate the current injection steps. (c) The input resistance R_input was significantly higher in R6/2 × rag1^−/− striatal neurons than in WT × rag1^−/− controls, but was partially restored to normal in R6/2 mice chimerized with normal CD44-sorted hGPCs. (d) Representative traces of sEPSCs from striatal neurons recorded in rag1^−/− control (black), CD44-engrafted rag1^−/− (yellow), R6/2 × rag1^−/− (purple) and CD44-engrafted R6/2 × rag1^−/− (green) mice. (e,f) The frequency (e) and amplitude (f) of sEPSCs and miniature EPSCs (mEPSCs). (e) The sEPSC frequency was significantly lower in R6/2 striatal neurons than in WT rag1^−/− controls, but was restored in CD44-engrafted R6/2s to levels not significantly different from control. (f) In contrast, the EPSP amplitude of R/2 striatal neurons was unaffected by chimerization. (g) Cumulative distribution of sEPSCs. The lower frequency of sEPSCs in the R6/2 MSNs, and partial restoration by hGPC engraftment, was consistent across EPSC amplitudes. WT-untreated, n=11; WT-hGPC, n=8; R6/2-untreated, n=11; R6/2-hGPC, n=8. Means±s.e.m.; *, ** and *** indicates P<0.05, 0.01 and 0.001, respectively, by one-way ANOVA with post hoc t-tests.

Figure 8. Normal glial engraftment reduces interstitial K⁺ levels in the R6/2 striatum.

Potassium electrodes were used to measure the interstitial levels of striatal K⁺ in both WT mice and their R6/2 littermates at 16 weeks of age (±4 days), with and without neonatal intrastriatal transplants of CD44-sorted hGPCs. Untreated R6/2 mice manifested significantly higher levels of interstitial K, which were restored to normal in R6/2 mice neonatally engrafted with hGPCs (P<0.01 by one-way ANOVA). In contrast, hGPC engraftment did not influence the interstitial K⁺ levels of WT mice. All values graphed as means±s.e.m.