DREADD Modulation of Transplanted DA Neurons Reveals a Novel Parkinsonian Dyskinesia Mechanism Mediated by the Serotonin 5-HT6 Receptor

Abstract

Transplantation of DA neurons is actively pursued as a restorative therapy in Parkinson's disease (PD). Pioneering clinical trials using transplants of fetal DA neuroblasts have given promising results, although a number of patients have developed graft-induced dyskinesias (GIDs), and the mechanism underlying this troublesome side effect is still unknown. Here we have used a new model where the activity of the transplanted DA neurons can be selectively modulated using a bimodal chemogenetic (DREADD) approach, allowing either enhancement or reduction of the therapeutic effect. We show that exclusive activation of a cAMP-linked (Gs-coupled) DREADD or serotonin 5-HT6 receptor, located on the grafted DA neurons, is sufficient to induce GIDs. These findings establish a mechanistic link between the 5-HT6 receptor, intracellular cAMP, and GIDs in transplanted PD patients. This effect is thought to be mediated through counteraction of the D2 autoreceptor feedback inhibition, resulting in a dysplastic DA release from the transplant.

Figure 1

Characterization of Dopaminergic Fetal Grafts and AAV-Mediated Transduction Efficacy (A) Experimental timeline illustrating the time points for all major surgical procedures and behavioral tests for animals expressing hM3Dq/rM3Ds and hM3Dq/KORD vectors. (B and C) LSM of midbrain dopaminergic neurons, virally transduced with AAV-8 DIO-mCherry and stained for TH (green) and mCherry (red). VTA (B) and SN (C) dopaminergic neurons displayed high specificity for mCherry-labeled neurons, demonstrating that the TH-Cre knockin rat has highly selective expression of Cre recombinase, localized to dopaminergic neurons. (D and E) Motor recovery following grafting of dopaminergic fetal cells in the stepping (D) and cylinder (E) behavior tests. Dashed line in graphs represents the average performance of an unlesioned rat (see also Figure S1). Lesioned animals receiving the fetal midbrain progenitor transplants (TX) recovered significantly, yet not completely, in these tests of advanced motor function, while there was no spontaneous motor recovery observed in lesioned rats that were not transplanted (Lesion Ctrl). (F and G) Dark field microscopy imaging of three-site dopaminergic grafts in horizontal sections, using DAB-amplified immunohistochemical detection of the DA marker TH (F) and characterization of AAV transduction efficacy using HA-tag antibody (detecting the tagged hM3Dq transgene) (G) (see also Figure S2). (H) Stereological quantification of dopaminergic neurons within the grafts of both hM3Dq + rM3Ds (red) and hM3Dq + KORD (blue) transduced animals in relation to intact midbrain dopaminergic neurons of the SN pars compacta (dashed lines ± 1 SD). (I) Quantification of AAV-transduced grafted neurons for the three vector constructs: hM3Dq-HA-tag (blue), rM3Ds-mCherry (tan), and KORD-GFP (cyan) for both groups of hM3Dq + rM3Ds and hM3Dq + KORD transplants, presented as transduction efficiency related to the number of TH+ transplanted DA neurons. (J) LSM image of grafted neurons stained for rM3Ds-mCherry (red), hM3Dq-HA (green), and TH (blue) (see also Figure S2). Scale bar, 500 μm (G) and 100 μm (J). All values reported as arithmetic mean ± 1 SEM. ^∗p < 0.05 in Bonferroni corrected paired Student’s t test; #significantly different from control (p < 0.05) in repeated-measures ANOVA, followed by Bonferroni corrected one-way ANOVA, followed by Dunnett’s T3 post hoc test.

Figure 2

In Vivo Specificity Evaluation of DIO-AAV Vectors to Transplanted TH-Cre-Positive Neuroblasts and Off-Target In Vivo Activation of the DREADD Ligands (A and B) Expression of DIO-hM3Dq and DIO-rM3Ds DREADDs, following viral infusion into the striatum of WT animals (A) and animals grafted with dopaminergic fetal tissue from TH-Cre transgenic rats (B). ^∗Denotes position of viral infusion (see also Figure S3). (C–E) Assessment of virally transduced, grafted mCherry-positive neurons that appeared TH negative under normal laser power conditions. (C) Overview of dopaminergic fetal graft within the striatum with an mCherry-positive neuron that appeared TH negative (white box). When comparing high-magnification z stacks of these mCherry-positive grafted neurons under normal laser power conditions (D) with high laser power conditions (E), it is clear that these mCherry-positive (red) neurons are indeed TH positive (green), although at low levels. We hypothesize based on cell morphology and heterogeneous TH expression that these are grafted neurons originating from the VTA (see also Figure S1). (F) Control measurements in the intact striatum using electrochemical chronoamperometric recordings of DA release. Local KCl administration (red arrow) evoked strong DA release while neither CNO (blue arrow) nor SalB (black arrow) administration resulted in any measurable increase in extracellular DA. Scale bar, 200 μm (B), 100 μm (C), and 10 μm (E).

Figure 3

Bidirectional Modulation of DA Release from Grafted Neurons Using DREADDs (A) Chronoamperometric in vivo recordings in the striatum (in proximity to the transplantation sites), comparing DA release following local administration of KCl (red) and CNO (blue) from the dopaminergic graft. The average peak DA concentration after CNO (1.08 ± 0.51 μM) was comparable to the average KCl-induced DA release (1.60 ± 0.41 μM). (B) In vivo recordings of bidirectional DA modulation from hM3Dq + KORD-expressing graft where CNO (blue arrow)-induced DA release could be silenced following administration of SalB at assumed peak of release (black arrow). The average peak DA concentration on CNO was 1.49 ± 0.20 μM compared to 0.67 ± 0.11 μM when combined with inhibition of the graft using SalB. (C and D) Systemic, noninvasive modulation of the dopaminergic graft during assessment in the cylinder task comparing unmodulated graft baseline (tan) to lesion baseline (cyan) and 15 min following SalB treatment (blue). Inhibition of function occurred in hM3Dq + KORD animals, but not in animals lacking the KORD receptor (C), while CNO treatment (green) significantly increased motor performance in hM3Dq + rM3Ds animals (D). (E and F) Bidirectional modulation of motor behavior in the stepping task with hM3Dq + KORD animals reducing motor recovery following SalB treatment (E) and increasing motor recovery following CNO treatment (F), while the ameliorating effect of the hM3Dq + rM3Ds transplants could be increased by treatment with CNO (F) and remained unaltered after SalB administration (E). (G and H) Sensorimotor performance in response to treatment with SalB and CNO in the corridor task (G), and therapeutic potentiation after CNO in a complex sensorimotor integrative task, measured as the decrease in response latency in the disengage test (H) (see also Figure S3). All values reported as arithmetic mean ± 1 SEM. ^∗p < 0.05 in Bonferroni corrected paired Student’s t test; §p < 0.05 in two-way mixed-model ANOVA followed by Tukey’s post hoc.

Figure 4

GIDs in Animals with CNO-Activated hM3Dq + rM3Ds Receptors in the DA Transplant (A) Rotational asymmetry was prominent when activating the DA neurons in the transplant of hM3Dq + rM3Ds animals with CNO, but not hM3Dq + KORD or Lesion Ctrls. (B) Rotational behavior induced by apomorphine, however, showed a strong contralateral rotation in Lesion Ctrl animals (red), which was comparatively reduced in grafted hM3Dq + rM3Ds (blue) and hM3Dq + KORD (green) animals, indicating a reduced postsynaptic DA receptor supersensitivity, normalized by the graft. (C) When observed in a home-cage-like environment, treatment with CNO induced AIMs in the hM3Dq + rM3Ds animals only. (D) Apomorphine injections, on the other hand, induced strong AIMs in Lesion Ctrl animals, which was significantly reduced in grafted hM3Dq + rM3Ds and hM3Dq + KORD animals. All values reported as arithmetic mean ± 1 SEM. ^∗AUC (area under the curve) significantly different from the Lesion Ctrl group, #AUC significantly different from both treatment groups, p < 0.05 using one-way ANOVA followed by Dunnett’s T3 post hoc. §AUC significantly different from all other groups, p < 0.05 using one-way ANOVA followed by Tukey’s post hoc.

Figure 5

Treatment with 5-HT6 Agonist ST-1936 Induces DA Release and GIDs through Direct Action on the Grafted DA Neurons (A) Rotational behavior following treatment of 5-HT agonists specifically targeting each of the three metabotropic, Gs-coupled, serotonin (5-HT) receptors, with 5-HT4 agonist BIMU8 (green) and 5-HT7 agonist AS19 (black) not inducing any rotational behavior, while 5-HT6 agonist ST-1936 (blue) induced a strong contralateral rotation in grafted hM3Dq + KORD animals (L-DOPA naive and not previously displaying GIDs on CNO), but not Lesion Ctrls (red). (B) Treatment with 5-HT6 agonist ST-1936 induced substantial AIMs in grafted hM3Dq + KORD animals (green) with only a weak, transient AIM score in Lesion Ctrls (red). (C) Comparison of the different manifestations of AIMs between treatments with CNO, apomorphine, and ST-1936. CNO induced mainly locomotor (blue), axial (green), and orolingual (purple) AIMs in hM3Dq + rM3Ds animals. Grafted animals displayed a marked reduction in AIMs following treatment with apomorphine, with limb AIMs (red) being abolished. AIMs following treatment with ST-1936 induced mainly orolingual and axial AIMs, with a relatively small proportion of limb and locomotor AIMs in hM3Dq + KORD-grafted animals. In contrast, Lesion Ctrls mainly exhibited locomotor AIMs with a very small proportion of AIMs being orolingual. (D–G) Electrochemical recordings of DA release in response to local ligand administration, with CNO (D) and ST-1936 (E) inducing a prolonged DA release with very similar kinetics in hM3Dq + KORD-grafted animals, while both KCl (F) and ST-1936 (G) induced a short spike of DA with very different kinetics when applied in the intact, nongrafted striatum (see also Figure S6). (H) LSM image showing strong expression of 5-HT6 (red) colocalized with virally transduced (green) TH (blue)-positive neurons within the dopaminergic fetal graft. (I–L) Magnified z stack of a dopaminergic neuron (taken from dashed square in H) showing the colocalization between virally transduced KORD expression (I), 5-HT6 (J), and TH (K) together with an overlay composite (L) (see also Figure S5). All values reported as arithmetic mean ± 1 SEM. ^∗AUC significantly different from the Lesion Ctrl group, p < 0.05 using Kruskal-Wallis test followed by Bonferroni corrected all-pair comparison, using the Mann-Whitney U test in (A) and using Student’s t test in (B). Scale bar, 100 μm (H) and 10 μm (L).

Figure 6

Second Experimental Group: Validation of the rM3Ds DREADD and Reversal of the 5-HT6-Induced Involuntary Movements by rM4Di DREADD (A) Experimental timeline illustrating the time points for all major surgical procedures and behavioral tests for animals expressing rM3Ds/KORD and rM4Di/eYFP vectors. (B and C) Noninvasive modulation of the dopaminergic fetal graft during assessment in the stepping and cylinder tasks, comparing unmodulated lesion baseline (cyan) to graft baseline (tan) and 1 hr following CNO treatment (green). (B) Comparison between grafted rM3Ds/KORD- and rM4Di/eYFP-expressing animals, as well as nongrafted hM3Dq/rM3Ds transduced controls and nongrafted, nontransduced controls. rM3Ds receptor alone significantly potentiates motor performance in response to CNO treatment, while inert in hM3Dq/rM3Ds nongrafted animals. Inhibition of motor function occurred in animals expressing hM4Di in response to CNO in the stepping task. (C) Comparison between grafted rM3Ds/KORD and hM3Dq/rM3Ds transduced nongrafted controls in the cylinder task. (D) Rotational behavior in response to CNO (10 mg/kg) in grafted rM3Gs/KORD animals compared to hM3Dq/rM3Ds transduced nongrafted controls. (E) Rotational behavior in response to 5-HT6 agonist ST-1936 (20 mg/kg) in grafted animals and inhibition of this behavior by the CNO-dependent hM4Di DREADD. CNO treatment effectively abolished ST-1936-mediated rotational behavior, suggesting that 5-HT6-mediated rotational behavior in grafted animals is DA dependent. All values reported as arithmetic mean ± 1 SEM. ^∗AUC significantly different from the Lesion Ctrl group, p < 0.05 using Student’s t test, #AUC significantly different from both TH-Cre TX rM4Di+eYFP +ST1936 +CNO and Lesion Ctrl WT hM3Dq+rM3Ds +ST1936, p < 0.05 using one-way ANOVA followed by Dunnett’s T3 post hoc.

Figure 7

Host- and Graft-Derived Serotonin Neurons Densely Innervate the DA Neurons of the Transplant (A–D) Overview immunofluorescence staining of serotonin expression in the striatum of animals that received dopaminergic fetal grafts, using LSM, stained for (A) SERT, (B) TH, and (C) DAPI with an overlay composite (D). Serotonin projections were as expected—abundant in the striatum with SERT-positive projections infiltrating the grafted tissue. (E) High-magnification z stack of grafted dopaminergic neuron (red) in close proximity to infiltrating SERT-positive (green) projections (white arrows). SERT-positive projections were found to be within close range around grafted dopaminergic neurons, suggesting communication between the two. Scale bars, 100 μm (A) and 10 μm (E).

Figure 8

The 5-HT6 Receptor Is Highly Expressed in Human Tissue Grafts Originating from Both Fetal and hESC Sources (A–E) Immunofluorescence staining of 5-HT6 expression in a human dopaminergic fetal graft within the rat striatum (A and C) with the phenotypic staining for TH (A and B), and the identification of human NCAM (A and D), imaged using LSM. The 5-HT6 expression was confirmed to reside in DA neurons originating from the transplanted tissue (E). (F–J) Similarly, immunofluorescence staining of a paraffin-embedded section originating from the putamen of a PD patient that received a dopaminergic fetal graft, imaged using LSM, confirmed that the 5-HT6 receptor (F and H) is highly expressed in the TH-positive (F and G, with overlay in J) DA neurons originating from the fetal graft. Nuclear staining (DAPI) confirmed the 5-HT6 expression to be abundant in the neuronal soma (I). (K–O) To confirm if this receptor is also abundant in DA neurons differentiated from hESCs, a graft derived from the H9 hESC line, differentiated using the novel floor-plate-based protocol and transplanted to the parkinsonian striatum of a nude rat, was evaluated using LSM with immunofluorescence for the same genes: 5-HT6 (K and M), TH (K and L), and the human NCAM (K and N). This confirmed that this receptor is highly expressed in DA neurons also differentiated from this cell source, when transplanted to the striatum. Overlay of all three antigens confirmed the 5-HT6 receptor to reside in hESC-derived DA neurons (O). Scale bar, 100 μm (K) and 50 μm (N).