By transducing extracellular signals carried by neuromodulators, the cyclic adenosine monophosphate (cAMP) signaling plays a crucial role in the regulation of neuronal activities in the brain. Multiple guanine nucleotide-binding protein (G-protein)-coupled receptor (GPCR) cascades regulate the intracellular levels of cAMP, which activates its key effector protein kinase A. Seven-transmembrane domain receptors can transmit extracellular signals to the intracellular signaling cascades through the activation of heterotrimeric G-proteins, which are composed of the guanine nucleotide-binding Gα subunit and the dimeric βγ subunits (Pierce et al., 2002). Gα_s is the predominant stimulatory G-protein subunit in the brain. However, in the striatum, Gα_s is replaced by the olfactory type G protein α subunit (Gα_olf), which is encoded by the GNAL gene (Jones and Reed, 1989). Cellular Gα_olf/cAMP signaling pathway represents a principal regulator for the striatal functions in normal physiological processes and pathological conditions (Hervé, 2011). It is worth noting that mutations in the GNAL gene have been identified as a cause for generating dystonia (Fuchs et al., 2013; Pelosi et al., 2017), suggesting that the Gα_olf function might participate in the brain circuit involving motor control.

The motor symptoms of Parkinson’s disease (PD) are caused by striatal dopamine (DA) deficiency, predominantly in the putamen, resulting from a progressive degeneration of nigrostriatal DA-producing cells (Kish et al., 1988; Goto et al., 1989). Although the DA replacement therapy remains the mainstay to treat PD symptoms, long-term exposure to dopaminergic drugs, particularly to the DA precursor levodopa, eventually causes adverse effects such as motor fluctuations and levodopa-induced dyskinesia (LID; Jenner, 2008; Calabresi et al., 2010; Bastide et al., 2015). LID is a major cause of disability in patients with PD, and occurs in approximately 80% of patients after 5 years of treatment with a daily administration of levodopa (Obeso et al., 1989; Luquin et al., 1992; Rascol et al., 2000). Importantly, once LID has been primed (or established), its severity progressively increases despite even when the used dosage of dopaminergic drugs is not increased (Brotchie, 2005). LID is known to be closely linked to the altered function of the DA signaling pathways in the striatum (Brotchie, 2005; Jenner, 2008; Bastide et al., 2015; Calabresi et al., 2016). It has also been suggested that LID is associated with the hypersensitivity of striatal MSNs to DA receptor stimulation and with ongoing deregulation of corticostriatal inputs, which activate striatal glutamate receptors, such as N-methyl-D-aspartate (NMDA) receptors (Brotchie, 2005; Jenner, 2008; Bastide et al., 2015; Calabresi et al., 2016). In this hypothesis article, we primarily considered the levodopa-induced changes in cellular Gα_olf protein levels in the DA-denervated striatum as the key mechanism to increase the striatal responsiveness to DA receptor stimulation in LID.

As being innervated by massive dopaminergic afferents originating from the midbrain, the striatum is highly enriched in DA receptors, which belong to a superfamily of GPCRs and are classified into two subtypes, D₁- and D₂-type receptors. Through their specific targeting of G proteins, the D₁-type receptors (D₁Rs) elicit the adenylyl cyclase type (AC) to increase the cAMP production, whereas the D₂-type dopamine receptors (D₂Rs) inhibit the cAMP production (Kebabian and Calne, 1979; Missale et al., 1998). Medium spiny neurons (MSNs) constitute more than 90% of the neuronal types in the striatum (Graybiel, 2008; Kreitzer, 2009; Gerfen and Surmeier, 2011). Anatomically, they are roughly halved with the MSN group to form the “direct” striatonigral pathway by the MSN group to from the “indirect” striatopallidal pathway (Crittenden and Graybiel, 2011; Gerfen and Surmeier, 2011; Calabresi et al., 2014). The striatonigral and striatopallidal MSNs express D₁Rs and D₂Rs, respectively. Moreover, the striatopallidal MSNs, but not the striatonigral MSNs, are enriched in adenosine A_2A receptors (A_2ARs), which are prototypical Gs-coupled receptors that elicit AC to increase cAMP production (Svenningsson et al., 1999; Schwarzschild et al., 2006; Fuxe et al., 2007). Figure 1 depicts the cell-type specific localization of Gα_olf, D₁R and A_2AR among the striatal MSNs that constitute the basic circuits of the basal ganglia.

Gα_olf is highly expressed in all striatal MSNs including those expressing the D₁Rs and A_2ARs (Kull et al., 2000; Hervé, 2011; Morigaki et al., 2017; see Figure 2). As Gα_olf positively couples with D₁R and A_2AR to activate the AC type 5 (AC5) and, thereby, increase the intracellular cAMP levels, it serves as the rate-limiting factor for both the D₁R- and A_2AR-dependent cAMP production in striatal MSNs (Kull et al., 2000; Corvol et al., 2001). The Gα_olf protein level plays a key role in regulating the D₁R/cAMP- and A_2AR/cAMP-signal activities of striatonigral and striatopallidal MSNs, respectively. The D₁R/Gα_olf-mediated increases in the cAMP levels cause the activation of the striatonigral MSNs (Hervé, 2011). On one hand, as D₂R activation inhibits AC5 through G_i/o proteins but A_2AR activation elicits AC5 through G_s/olf proteins (Kull et al., 2000), the A_2AR/Gα_olf-signal stimulation functionally opposes the actions of D₂Rs on the striatopallidal MSNs (Schwarzschild et al., 2006; Fuxe et al., 2007).

Quantitative immunohistochemistry (IHC) has shown that the Gα_olf protein is unevenly distributed within the mouse striatum, where it is highly concentrated in the dorsolateral striatum (Morigaki et al., 2017). Since the dorsolateral portion of the mouse striatum corresponds to the motor territory in rodents and is analogous to the putamen in primates (Graybiel, 2008), this strategic expression of Gα_olf protein indicates that Gα_olf may function as the stimulatory G protein that has a tight link to the basal ganglia “motor” circuit (Alexander and Crutcher, 1990) at the striatal level. With respect to the striatal compartments, there was a differential localization of Gα_olf with higher densities of Gα_olf proteins in the striosomes relative to the matrix compartment (Sako et al., 2010; Ruiz-DeDiego et al., 2015; Morigaki et al., 2017). This suggests that Gα_olf may be a key molecule that determines differential responses between the striosome and matrix compartments to the D₁R or A_2AR activation in the striatum at maturity.

Rodent animal models for PD (Iderberg et al., 2012; Francardo and Cenci, 2014) have so far been used to elucidate the regulatory mechanism for the striatal expression of Gα_olf. In line with the evidence that there is a significant increase in Gα_olf protein levels in the putamen of patients with PD (Corvol et al., 2004), a dramatic increase in Gα_olf protein levels has been identified in the DA-depleted striatum of rats (Hervé et al., 1993; Marcotte et al., 1994; Penit-Soria et al., 1997; Corvol et al., 2004; Rangel-Barajas et al., 2011) and mice (Alcacer et al., 2012; Ruiz-DeDiego et al., 2015; Morigaki et al., 2017) with nigrostriatal 6-hydroxydopamine lesions. However, this upregulation of the Gα_olf protein levels is not associated with a parallel increase of the Gα_olf mRNA expression. Accordingly, the homeostatic regulation of Gα_olf protein levels is thought to occur through post-translational mechanisms in the striatum, where the altered expression of the Gα_olf protein depends directly on its usage rate (Hervé, 2011). The persistent lack in the use of D₁R and Gα_olf could lower the Gα_olf degradation rate and thereby result in the accumulation of Gα_olf protein in the DA-denervated striatum of PD models. In agreement with this hypothesis, a total lack of D₁Rs by D₁R gene targeting induces a significant increase of the Gα_olf protein levels without any changed expression of Gα_olf mRNAs in the striatum of mutant mice (Hervé et al., 2001). In contrast, the decreased levels of striatal Gα_olf proteins were found in mutant mice lacking the DA transporter (Hervé et al., 2001), which exhibit a marked increase in the extracellular DA levels leading to persistent activation of D₁Rs in the striatum (Giros et al., 1996). Importantly, the lack of A_2ARs in homozygous A_2AR knock-out mice (Ledent et al., 1997) also results in an upregulation of Gα_olf proteins with no obvious changes in the levels of Gα_olf transcripts (Hervé et al., 2001). Collectively, the agonist-induced activation of D₁Rs (Hervé et al., 2001; Corvol et al., 2004, 2007; Alcacer et al., 2012; Ruiz-DeDiego et al., 2015) or A_2ARs (Hervé et al., 2001) might lead to the degradation of Gα_olf proteins in striatal MSNs through posttranslational usage-dependent mechanism (see Figure 2).

On the hypothesis that the upregulation of the Gα_olf protein levels results from the disuse of the D₁Rs in the DA-depleted striatum in rodent models for PD, several studies with IHC and western blot analyses revealed that the Gα_olf could be returned to normal levels by DA replacement with a daily exposure to levodopa in rodent models for PD with LID (Corvol et al., 2004; Rangel-Barajas et al., 2011; Ruiz-DeDiego et al., 2015; Morigaki et al., 2017). With respect to the striosome-matrix system, IHC studies revealed that the Gα_olf levels were normally found in both the striosome and matrix compartments in PD with LID, although they were markedly increased in the matrix compartment, but not or only mildly increased in the striosome compartment, in PD (Ruiz-DeDiego et al., 2015; Morigaki et al., 2017). This novel finding indicates that there is a difference in the dopaminergic regulation of the Gα_olf expression between the striosome and matrix compartments.

In situ proximity ligation assay (PLA) for dual-antigen recognition disclosed cell-type specific changes in the Gα_olf levels in the DA-depleted striatum of mice with and without LID (Morigaki et al., 2017). The in situ PLA technique can indicate the presence of the Gα_olf protein in close proximity to the D₁R protein (D₁R-Gα_olf) or A_2AR protein (A_2AR-Gα_olf). Quantitative in situ PLA showed that DA depletion caused a marked (~90%) increase in the striatal levels of D₁R-Gα_olf PLA signals, which were downregulated by a daily administration of levodopa. However, there remained a significant (~50%) increase in the striatal D₁R-Gα_olf PLA signals in mice with LID when compared with normal controls. On one hand, quantitative in situ PLA also disclosed that a daily exposure to levodopa, but not DA depletion per se, caused a significant (~40%) decrease in the striatal A_2AR-Gα_olf PLA signals in the DA-depleted striatum of mice with LID. These findings indicate that, in the DA-depleted striatum, DA replacement could induce the downregulation of the Gα_olf protein levels not only in the striatonigral MSNs but also in the striatopallidal MSNs.

An intriguing question is how the Gα_olf protein levels are decreased in the striatopallidal MSNs in LID. In animal models with nigrostriatal 6-OHDA-lesions, persistent (chronic) DA depletion per se has been shown to cause no apparent changes (Ballarin et al., 1987; Herrera-Marschitz et al., 1994; Nomoto et al., 2000) or mild decrease (Pinna et al., 2002) in the extracellular levels of adenosine in the DA-denervated striatum. However, evidence shows that the striatal adenosine levels are elevated by the activation of NMDA receptors (Delaney and Geiger, 1998; Delaney et al., 1998), which can be enhanced by D₁R activation (Cepeda and Levine, 2012; Morigaki and Goto, 2015; see Figure 3). Interestingly, a pulsatile exposure to the D₁R agonist reportedly facilitated the NMDA receptor-evoked increase in the extracellular adenosine release in the rat striatum (Harvey and Lacey, 1997). This evidence suggests that, in the DA-depleted striatum with D₁R hypersensitivity, a repeated administration of levodopa may exert a pulsatile activation of D₁Rs, which subsequently facilitates the NMDA receptor-evoked increase in the extracellular adenosine levels. Moreover, in the DA-depleted striatum, the activation of NMDA receptor could lead to a marked increase in the extracellular adenosine levels and, then, indirectly activate A_2ARs (Nash and Brotchie, 2000). Thus, it is likely that the downregulation of the Gα_olf levels in striatopallidal MSNs in LID might result from an increased usage of Gα_olf proteins through the A_2AR activation subsequent to the daily pulsatile activation of striatal D₁Rs. This notion also suggests that the striatal D₁R signals might play a critical role in the regulation of the Gα_olf protein levels not only in the striatonigral MSNs, but also in the striatopallidal MSNs in the DA-denervated striatum. This consideration may corroborate the general concept that increased activities of striatal D₁Rs are requisite for the genesis of LID (Westin et al., 2007; Darmopil et al., 2009; Alcacer et al., 2012).

Figure 4 shows the hypothetical representation of the Gα_olf protein levels in striatonigral and striatopallidal MSNs in the DA-denervated striatum under the conditions of both PD with and without LID. In PD, there is a dramatic increase in the Gα_olf protein levels in the striatonigral MSNs, but not in the striatopallidal MSNs. Because of no apparent changes in the striatal D₁R levels (Shinotoh et al., 1993; Turjanski et al., 1997; Hurley et al., 2001) and other principal mediators of the D₁R signaling cascades (Girault et al., 1989; Nishino et al., 1993) in patients with PD, the marked increase in the Gα_olf protein levels in the striatonigral MSNs may be a principal cause for generating striatal D₁R hypersensitivity to levodopa exposure in PD. This notion corroborates the evidence that there is a marked increase in the responsiveness of the striatonigral MSNs to D₁R activation in PD, as determined by the fos induction experiments (Engber et al., 1989; Asin et al., 1995; Kashihara et al., 2000; Xu et al., 2003; Morigaki et al., 2017).

In PD with LID, there is an important decrease in the Gα_olf protein levels in the striatopallidal MSNs after a prolonged and pulsatile administration of levodopa. This leads to the facilitation of the effects of DA on striatopallidal MSNs by reducing the A_2AR/Gα_olf signal-mediated cAMP production and subsequently to the increase in the responsiveness of striatopallidal MSNs to D₂R activation. Indeed, it was importantly noted that, during the increasing phase of dyskinesias, an abnormal lowering of intracellular cAMP levels transiently occurred in the DA-denervated striatum in rat model of LID (Sancesario et al., 2014). These novel findings parallel the evidence that a repeated exposure to levodopa results in a significant increase in the responsiveness of striatopallidal MSNs to dopaminergic stimulation, as determined by fos induction experiments (Engber et al., 1989; Asin et al., 1995; Kashihara et al., 2000; Xu et al., 2003; Morigaki et al., 2017). In addition, there is a significant increase in the Gα_olf protein levels in striatonigral MSNs in PD with LID as compared to normal controls. Because Gα_olf is the regulator of cAMP signal-dependent activities in the striatum, an increase in the responsiveness of both striatonigral and striatopallidal MSNs to levodopa exposure, which depends on the Gα_olf protein levels, serves as a principal cause for generating LID.

Since the intracellular cAMP signaling cascades serve as a determinant of striatal cell activities (Girault, 2012), maladaptive change in Gα_olf protein levels is thought to be closely linked to the pathophysiology of PD (Hervé, 2011). Here, we hypothesized that DA depletion might cause a marked upregulation of the Gα_olf protein levels in striatonigral MSNs, which results in a crucial hypersensitivity of the striatum to D₁R stimulation in PD. A prolonged and pulsatile exposure to levodopa might lead to a usage-dependent decrease in the Gα_olf protein levels not only in the nigrostriatal MSNs but also in the striatopallidal MSNs in PD with LID. This levodopa-induced decrease in Gα_olf protein levels, which might be due to a pulsatile activation of postsynaptic D₁Rs and NMDA receptors, could result in reduced A_2AR/Gα_olf/cAMP signal levels in striatopallidal MSNs. This might cause an increase in the responsiveness of striatopallidal MSNs to D₂R activation, and thereby develop LID in PD. Our hypothesis corroborates the long-lasting concept that LIDs are associated with a decreased activity of the “indirect” striatopallidal pathway (Crossman, 1990; DeLong, 1990; Brotchie, 2005).

As an important cellular mechanism to regulate the activities of striatal MSNs, the recurrent collateral connections between the MSNs have also been identified (Bolam et al., 1983; Yung et al., 1996). The activities of striatopallidal MSNs can be inhibited by the GABAergic collateral axon branches from neighboring MSNs (Taverna et al., 2008; Lalchandani et al., 2013; Dobbs et al., 2016; Wei et al., 2017). Thus striatal D₁ hypersensitivity could lead to an increased responsiveness of striatopallidal MSNs to D₂R activation in the conditions of PD with and without LID, although only a small population of the striatonigral MSNs has been found to form collateral axon connections with striatopallidal MSNs in the mouse striatum (Taverna et al., 2008). However, this notion per se could not explain the progressive increase in the severity of LID, which occurs in the PD patients treated with unaltered dosages of given dopaminergic drugs (Brotchie, 2005), because there is an ongoing decline in striatal responsiveness to D₁R activation along a repeated exposure to levodopa under the conditions of PD, as determined by fos induction experiments (Saka et al., 1999; Kashihara et al., 2000; Xu et al., 2003; Morigaki et al., 2017).

Finally, we suggest that the pharmacological concomitant therapy to increase Gα_olf protein levels in the striatum might be useful in the management of LID and motor fluctuations in patients with PD treated with DA replacement therapy. The normalization of the decreased Gα_olf protein levels in the striatopallidal MSNs might suppress LID. On one hand, the elevation of the Gα_olf protein levels in the striatonigral MSNs could increase the striatal responsiveness to D₁R activation and, thereby, facilitate the therapeutic efficacy of dopaminergic drugs. In considering the possible involvement of the activated NMDA receptors in lowering striatal Gα_olf levels in LID, NMDA receptor antagonists (e.g., amantadine or memantine) might attenuate LID, as already shown in clinical practice (Rascol et al., 2015). Because A_2AR activation, which could reduce the Gα_olf protein levels in the striatopallidal MSNs leading to LID, might be required for the “priming” of LID (Brotchie, 2005; Xiao et al., 2006), it is suggested that A_2AR antagonists (e.g., istradefylline) might be effective in dampening the “priming” of LID. However, after the establishment of LID, the adjunct use of A_2AR antagonists might exacerbate the dyskinetic symptoms as shown in clinical practice (Kondo and Mizuno, 2015).

SG wrote the manuscript.

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.