To further establish the translational potential of targeting ΔFosB in LID, we first evaluated whether the correlation previously established in rodents between ΔFosB levels and severity of LID extend to the best non-human primate model of PD, the MPTP-lesioned macaque (18
). We first confirmed by cloning and sequencing that identical splicing of the FosB gene occurs in the macaque brain as reported previously from human and rodent tissues. We found that sequences of FosB and ΔFosB mRNA are 98% conserved with the human variants, and we established a quantitative PCR method to selectively quantify FosB and ΔFosB mRNA from macaque tissues (supplemental figure S1
). We next examined the relationship between the intensity of LID, on one hand, and ΔFosB mRNA and protein levels in striatum, on the other. In clinical settings, only a subset of patients receiving optimized doses of L-DOPA develop dyskinesia within the first months of treatment and similar heterogeneity is also observed among non-human primates (19
). In a cohort of 8 macaques with stable parkinsonian symptoms, and receiving optimal doses of L-DOPA for 6 months, we observed a significant increase in striatal levels of ΔFosB protein only in the subset of animals showing overt LID (). A linear regression analysis conducted in this cohort indicates that, about 70% of the interindividual variability in dyskinesia severity can be predicted from ΔFosB protein levels (). Although an overall increase in ΔFosB mRNA expression with L-DOPA treatment was observed, there was no significant difference in ΔFosB mRNA levels between dyskinetic and non-dyskinetic animals at the same time point (). On the other hand, sensitized induction of pDYN mRNA was observed in dyskinetic compared to non-dyskinetic animals (). As predicted from previous rodent (3
) and primate (26
) studies, mRNA levels of pDYN, a well-validated target gene of ΔFosB, correlated highly with both ΔFosB protein levels and dyskinesia scores (), but not with ΔFosB mRNA levels. Together, these results suggest that enhanced ΔFosB levels in dyskinetic monkeys likely reflect pulses of protein synthesis and accumulation, rather than a sustained increase in ΔFosB mRNA stability, gene transcription or splicing. Having demonstrated that the link between ΔFosB protein levels and dyskinesia severity extends to our primate experimental model, we next tested the causal role of ΔFosB signaling in the long-term maintenance of LID.
Recent clinical studies have indicated that gene therapy using AAV vectors is a well-tolerated approach which holds great promise in PD patients (37
). We adopted a similar approach to manipulate ΔFosB function locally within the macaque motor striatum. To test whether increased ΔFosB signaling in this region augments the severity of LID, we overexpressed ΔFosB using an AAV2-ΔFosB-IRES-hrGFP viral vector. To test the complementary hypothesis, that reduced activity of endogenous ΔFosB may disrupt long-term maintenance of LID and thereby ameliorate them, we opted for a dominant-negative strategy, by overexpressing a truncated form of the JunD protein (20
). The choice of this strategy was based on previous data indicating that JunD, a major in vivo partner for ΔFosB (11
), is the most prevalent Jun family protein found, under dyskinesiogenic conditions, in ΔFosB-containing, AP-1 dimers (5
). To create an AAV-ΔJunD-hrGFP vector, a cDNA encoding an N terminal truncated JunD, devoid of transactivational activity (32
), was inserted into pAAV2-IRES-hrGFP. The effects of ΔFosB and ΔJunD overexpressing vectors were compared to a control vector expressing hrGFP only. In all viral vectors, hrGFP was expressed as a second open reading frame translated from an internal ribosome entry site (IRES) and was used as an internal standard to ascertain the infection efficiency in each animal at the end of the experiment. Based on preliminary infusions and immunohistological evaluations of hrGFP expression, we determined that complete antero-posterior coverage of the macaque motor striatum could be achieved with an overall viral volume of 20μl per side (5μl at 4 different anteroposterior locations in the putamen). Our post mortem estimations, conducted on the brains collected after the completion of the behavioral study, indicated that hrGFP expression was detected across similar volumes of approximately 60 mm3 with each of the 3 vectors.
In previous rodent studies, manipulations of ΔFosB levels have been conducted solely in drug naïve animals, prior to any L-DOPA treatment (15
). This experimental design is not the most clinically relevant as patients do not usually seek dyskinesia relief before starting L-DOPA treatment. In addition, in our preliminary studies, both AAV-ΔFosB-hrGFP or AAV-ΔJunD-hrGFP viruses have proved inefficient when administered prior to the beginning of L-DOPA treatment (data not shown).
Therefore, to directly address the influence of ΔFosB signaling on long-term maintenance of LID, AAV-ΔFosB-hrGFP or AAV-ΔJunD-hrGFP viruses were stereotaxically infused into the motor striatum of parkinsonian macaques, previously rendered dyskinetic by daily administration of L-DOPA for 3 months. In assigning animals to ΔFosB or ΔJunD conditions, particular care was taken to establish balanced groups for pre-infusion severity of both parkinsonian () and LID () symptoms. The control group was composed of parkinsonian macaque infused with AAV-hrGFP, and was previously naïve to L-DOPA. Four weeks after stereotactic delivery of the viral vectors, L-DOPA treatment was resumed in all groups.
In AAV-ΔFosB transduced animals, the therapeutic (i.e., antiparkinsonian) activity of L-DOPA (), as well as dyskinesia severity () were maximal from day 1, and unaltered in comparison to pre-infusion scores (, left side). In contrast, the overexpression of ΔJunD produced a dramatic reduction of dyskinesia severity compared to pre-infusion scores. This effect coincided with a lack of pDYN upregulation by L-DOPA in transduced tissues (), a result we interpret as functional readout of efficient ΔFosB antagonism by ΔJunD (3
). Despite continuous transgene expression (Figure, AAV-ΔJunD-hrGFP treatment did not permanently prevent the reappearance of dyskinesia with repeated L-DOPA administration. Interestingly, the kinetics of the reappearance of LID in AAV-ΔJunD-hrGFP injected animals overlapped with the one of naïve AAV-hrGFP animals experiencing their first L-DOPA course. In both of these conditions, dyskinesia developed from day 8 onwards with a gradual worsening until stabilization on day 18 ().
Our interpretation of these results is that overexpression of ΔFosB in the motor striatum of dyskinetic animals treated with L-DOPA is devoid of behavioral effect, whereas overexpression of ΔJunD resets dyskinesia-related striatal networks to a drug-naïve state, putting the animals back into the so-called “honeymoon” period of L-DOPA treatment. Importantly, the anti-dyskinetic effect of ΔJunD occurs without any alteration in the efficacy of L-DOPA on parkinsonian symptoms (). Dominant-negative inhibition of endogenous ΔFosB activity in the motor striatum, through expression of ΔJunD, thus appears to provide a novel and valuable avenue to interfere with the long-term maintenance of dyskinesia, potentially offering patients a return to the honeymoon period. Our interpretation is that this activity is mediated primarily through an antagonism of endogenous ΔFosB activity. The possibility that ΔJunD may alter additional protein-protein interactions by competing for other partners of endogenous JunD protein cannot be excluded. It will be important in further studies to determine why the antidyskinetic activity of ΔJunD is not sustained. A possibility is that accumulating levels of ΔFosB ultimately supersede ΔJunD antagonism. On the other hand, since levels of viral transgenes were evaluated at a single time point, after completion of the behavioral experiment, we cannot exclude the possibility that these levels may correspond to a fraction of the peak intensity occurring earlier in the study. A full time-course analysis of transgene expression during the length of L-DOPA treatment will be necessary to address this question. The fact that expression of viral transgenes may have dropped during the experiment, could also explain the absence of effect of ΔFosB overexpression in dyskinetic monkeys. This lack of effect, however, was not totally unexpected, as previous studies suggest that full occupancy of AP-1 sites, by endogenous ΔFosB containing complexes, may occur under dyskinesiogenic conditions (5
). Further characterization of the mechanisms underlying ΔJunD de-priming activity are required, as well as ways to extend the period of ΔJunD’s efficacy in reducing dyskinetic responses to L-DOPA.