FKBP12 contributes to α-synuclein toxicity by regulating the calcineurin-dependent phosphoproteome

Inhibition of FKBP12 Protects Against α-Syn Toxicity.

Inducing the expression of human α-syn in yeast cells leads to cellular responses relevant to the α-syn–mediated pathologies observed in neurons. These include nitrosative stress (14, 15), defects in vesicle trafficking (16, 17), abnormal mitochondrial function (18), and Ca²⁺ dysregulation (2). In accord with our previous report (2), while in control experiments, concentrations of Tacrolimus did not affect the growth of non–α-syn–expressing yeast cells (Fig. S1A), intermediate concentrations of Tacrolimus rescue α-syn toxicity in yeast (Fig. 1A), in part by reducing calcineurin activation and converting an otherwise toxic response into a protective one. Cyclosporine A (CsA) is a structurally and functionally distinct calcineurin inhibitor that forms a repressive ternary complex between calcineurin and cyclophilin A (CyA), a cis-trans prolyl isomerase different from FKBP12, which is targeted by Tacrolimus. Unlike the yeast experiments with Tacrolimus, CsA did not reduce α-syn toxicity at any of the concentrations tested (Fig. 1B and Fig. S1B). One possibility for the lack of rescue might be the inability of CsA to accumulate in yeast cells. To test this, we challenged the survival of yeast cells in the presence of MnCl₂-induced stress, a calcineurin-dependent process (19). Survival in the presence of MnCl₂-induced stress was severely compromised in our control strain over a wide range of CsA concentrations (Fig. S1C). These experiments reveal that CsA penetrates yeast cells and inhibits calcineurin but does not rescue α-syn toxicity.

Next, we asked if the differential contribution of these calcineurin inhibitors to α-syn toxicity was conserved in mammalian neurons. For these experiments, we used embryonic rat cortical neurons infected with a lentivirus carrying the disease-associated mutation of α-syn, A53T (20). These neurons are more abundant than highly vulnerable dopaminergic neurons and are also affected in patients with PD (21). In agreement with our yeast data, intermediate, subsaturating concentrations of Tacrolimus rescued neurons from the toxic effect of α-syn expression assessed by an ATP assay as well as by counting the remaining MAP2-positive neurons in the well (Fig. 1 C and E). Higher concentrations of Tacrolimus, which completely inhibit calcineurin, were not protective and in fact, increased toxicity. In control experiments, these concentrations did not compromise the viability of neurons not expressing α-syn (Fig. S1D). In addition, we found that the bell-shaped curve of Tacrolimus rescue was observed even at higher levels of α-syn toxicity (Fig. S1E). Importantly, the differences in calcineurin activation between control and α-syn–expressing neurons were not caused by differences in calcineurin expression but rather, calcineurin activity (Fig. S1F).

Consistent with results in yeast, at concentrations where neuronal viability was not compromised (Fig. S1G) and irrespective of the extent of α-syn toxicity, CsA did not rescue. In fact, CsA enhanced α-syn toxicity at concentrations of 1 μM assessed by an ATP assay as well as by counting the remaining MAP2-positive neurons in the well (Fig. 1 D and E). Since CsA and Tacrolimus directly interact with distinct prolyl isomerases to inhibit calcineurin, these observations suggest that FKBP12 plays an important and specific role in modulating α-syn toxicity.

To further test this hypothesis, we individually deleted each member of the FKBP and cyclophilin gene families in yeast (this experiment is difficult to interpret in mammalian cells, since deletion of some of the FKBP and cyclophilin members is embryonic lethal). These deletions had no effect on viability of yeast cells in the absence of α-syn (Fig. 1F) in accord with previous studies (22). Remarkably, deletion of the yeast FKBP12 gene (FPR1) and to a lesser extent, the closely related yeast FKBP13 gene (FPR2) conferred protection against α-syn toxicity (Fig. 1F). In contrast, deletion of the yeast CyA gene (CPR1) actually enhanced α-syn toxicity (Fig. 1F). Deletion of other members of the FKBP and cyclophilin families either had no effect (cpr 4, 5, 7, and 8) or increased toxicity (cpr 2, 3, and 6) (Fig. 1F). These genetic experiments mirror the pharmacological data and indicate that CyA plays a protective role, whereas FKBP12 contributes to α-syn toxicity.

Inhibition of FKBP12 Confers Protection Against α-Syn Toxicity Through Calcineurin-Dependent and -Independent Pathways.

We previously showed that calcineurin activity is a critical determinant of α-syn toxicity (2). In this report, our data suggest that FKBP12 is also a critical mediator of α-syn toxicity based on the protection afforded by genetic deletion of FKBP12 as well as the rescue of toxicity by Tacrolimus but not CsA. To investigate whether the neuroprotective properties of low doses of Tacrolimus against α-syn toxicity are caused by inhibition of FKBP12 and/or calcineurin, we took a genetic, pharmacological, and biochemical approach.

Deletion of calcineurin’s regulatory subunit, Cnb1p, eliminates calcineurin function in yeast without affecting the growth in the absence of α-syn (Fig. 2A and Fig. S2). The deletion also has little net effect on α-syn toxicity. Conversely, the deletion of the yeast FKBP12 conferred strong protection against α-syn toxicity. Importantly, the double deletion of calcineurin and FKBP12 markedly reduced, although did not completely eliminate, the protective effects of the FKBP12 deletion (Fig. 2A and Fig. S2). This result suggests that FKBP12 contributes to α-syn toxicity in a calcineurin-dependent as well as a calcineurin-independent manner.

To further dissect the relationship between these two proteins, we took advantage of the binding mode of the pharmacological agent Tacrolimus. One side of Tacrolimus binds to the catalytic pocket of FKBP12 and inhibits its ability to catalyze cis-trans isomerization of proline amino acids. The other side of Tacrolimus forms a composite surface with FKBP12 and calcineurin, thus inhibiting the phosphatase in a noncompetitive manner (23). Therefore, while Tacrolimus can inhibit FKBP12 in the absence of calcineurin, calcineurin can only be inhibited by Tacrolimus in the presence of FKBP12. We reasoned that we could determine the contribution of each protein to this inhibitory activity if we deleted calcineurin and FKBP12, individually and in combination, in the α-syn–expressing yeast strain and assess growth over a broad range of Tacrolimus doses.

In cells carrying the calcineurin deletion (calcineurin KO), Tacrolimus provided substantial but not full rescue at all concentrations tested (Fig. 2 B and C). Importantly, Tacrolimus can still inhibit FKBP12 in these calcineurin-deleted cells, thereby conferring some protective effect. This fact suggests that FKBP12 can exert some of its toxic effects independent of calcineurin. Conversely, KO of FKBP12 (FKBP12 KO) produced substantial rescue in the presence of calcineurin (Fig. 2D). This was not affected by the addition of Tacrolimus at any concentration, showing that the protective effects of Tacrolimus require the presence of FKBP12 and are not caused by off-target effects. Notably, the deletion of FKBP12 did not produce the optimal rescue effect of Tacrolimus observed in WT α-syn–expressing cells (Fig. 2 B and D). These data suggest that the maximal protective effects of Tacrolimus against α-syn toxicity are achieved by partial inhibition of both calcineurin and FKBP12. To confirm this possibility, we tested the effect of the calcineurin and FKBP12 double deletion in α-syn−expressing yeast cells. The double deletion conferred some but not full protection against α-syn toxicity, and this rescue was not affected by Tacrolimus (Fig. 2 B and E). These pharmacological and genetic findings reveal a physiological interaction between calcineurin and FKBP12 largely responsible for the toxic effects of α-syn. To determine whether calcineurin and FKBP12 can physically interact in the absence of Tacrolimus, we turned to a mammalian system. Indeed, in HEK 293 cells, calcineurin can interact with FKBP12 in the absence of Tacrolimus (Fig. 2F). As previously reported, addition of Tacrolimus enhances this interaction by stabilizing while inhibiting the enzymatic properties of both enzymes (Fig. 2F). However, addition of the Ca²⁺ chelator EGTA disrupts the endogenous interaction between calcineurin and FKBP12, suggesting that this interaction occurs under high Ca²⁺, such as the conditions triggered by α-syn.

Together, these genetic, pharmacological, and biochemical findings establish that FKBP12 can interact with calcineurin in the absence of Tacrolimus and contribute to α-syn toxicity in a calcineurin-dependent pathway.

FKBP12 Modulates the Calcineurin-Dependent Phosphoproteome.

FKBP12 has been shown to play a pathological role in α-syn toxicity, which is independent of calcineurin (24). This is in agreement with our pharmacological and genetic data; however, our findings also support the notion that FKBP12 contributes to α-syn toxicity through a calcineurin-dependent mechanism. This suggests that FKBP12 could exert part of its toxic effects by modulating calcineurin function. Phosphatases exert their effects by dephosphorylating substrates, which ultimately results in the activation and/or repression of downstream pathways. If FKBP12 contributes to α-syn toxicity by enabling calcineurin to dephosphorylate substrates, with dephosphorylation that contributes to cell death, we reasoned that inhibition of the functional interaction between FKBP12 and calcineurin should restore the phosphorylation state of those substrates. In doing so, this restoration may confer protection against α-syn toxicity. To test this hypothesis, we performed a phosphoproteomic analysis of yeast cells expressing empty vector (control), α-syn alone, and α-syn in cells deleted for (i) calcineurin, (ii) FKBP12, or (iii) both calcineurin and FKBP12 (Fig. 3A). Using a label-free approach, we quantified 13,641 phosphopeptides across all five conditions after normalizing for protein abundance (Materials and Methods has complete details). To detect differentially phosphorylated peptides, we compared control and α-syn–expressing yeast cells. We focused on those peptides showing at least a twofold (log2) change in abundance with a false discovery rate-adjusted P value (q value) < 0.05 (Fig. 3A and Fig. S3). These phosphopeptides were further annotated into unique phosphosites. For example, a given phosphopeptide could have multiple specific phosphorylation sites; hence, the data are reported as unique phosphosites within a given protein. Phosphorylation sites passing these criteria numbered 5,250.

Since calcineurin is a phosphatase highly activated in α-syn–expressing yeast cells, we concentrated our analysis on the twofold hypophosphorylated subset (527 of 5,250) to define phosphosites that may be dephosphorylated by calcineurin. We reasoned that any hypophosphosites that gained phosphorylation in the α-syn yeast-expressing cells with the calcineurin deletion are potentially calcineurin-dependent phosphosites. Likewise, if FKBP12 contributes to α-syn toxicity by enabling calcineurin to dephosphorylate specific substrates, then hypophosphosites that gained phosphorylation in the absence of FKBP12 should be similar and/or sensitive to the absence of calcineurin. The criteria for gain of phosphorylation (i.e., reversion) were established as those hypophosphosites with phosphorylation status that did not significantly change compared with the empty vector negative control cells (i.e., non–α-syn–expressing) but with status that was significantly more phosphorylated in the positive control α-syn–expressing cells. Based on this stringent test, the phosphosites that reverted were scored as positive for two thresholds and annotated as putative calcineurin-dependent phosphosites (Fig. 3B and Fig. S3).

In the calcineurin KO, ∼4% of all of the hypophosphorylated sites satisfied the reversion criteria (Fig. 3A and Dataset S1). Strikingly, deletion of the yeast FKBP12 had a more profound effect on phosphorylation than deletion of calcineurin alone. When FKBP12 was deleted, ∼34% of all of the hypophosphorylated sites fulfilled the reversion criteria. A minimal subset of these phosphosites was not affected by the absence of calcineurin and hence, is unique to the deletion of FKBP12 (Fig. 3A, Dataset S1, and Fig. S3). However, most of the phosphosites that reverted to control in the FKBP12 KO strain were dependent on calcineurin, given that the reversion was lost in the double-KO strain of calcineurin and FKBP12 (Fig. 3A and Fig. S3).

To establish which of these phosphosites are dependent on the functional interaction of calcineurin and FKBP12, we focused on those with (i) reversion pattern that was shared between the calcineurin KO, FKBP12 KO, and the double-KO strains and (ii) reversion pattern in the single-KO strain that was either lost or gained in the double-KO strain (Fig. 3B). With these criteria, we retrieved 175 of 527 unique calcineurin- and FKBP12-dependent phosphosites that were subsequently mapped to their corresponding proteins, representing a list of 66 putative calcineurin- and FKBP12-dependent substrates (Fig. 3C). The canonical calcineurin substrate docking site LxVP can determine calcineurin substrate recognition (2). Within the 66 putative calcineurin and FKBP12-dependent substrates, we found a 2.3-fold enrichment for LxVP-containing motif proteins, further supporting their reliance on calcineurin (P = 0.036) (Dataset S2). Notably, these 66 substrates, one-half of which have human homologs, are within biological processes, which have been shown to be critical in the pathobiology of α-syn (25). Examples include YKT6 and YCK3 (16) in vesicle trafficking, SLM1 (2) in actin cytoskeleton organization, PBS2 (26) in mitochondrial function, and TC089 (27) and NMA2 (28) in aging and energy metabolism.

It is known that toxic levels of α-syn impair vesicle trafficking, resulting in the aggregation of α-syn into functionally diverse large vesicles readily visualized as foci (16, 17). If it is true that FKBP12 enables calcineurin-dependent dephosphorylation of substrates, which impairs vesicle trafficking, we reasoned that low doses of Tacrolimus should restore the block in vesicle trafficking caused by α-syn. Indeed, low doses of Tacrolimus reduced these α-syn–induced vesicle foci (Fig. 3D). An alternative way to assess vesicle trafficking defects is by following the mobility on a gel of the carboxypeptidase Y (CPY), the best studied soluble protein targeted to the vacuole (29). The block from endoplasmic reticulum (ER)–Golgi caused by α-syn causes an accumulation of the glycosylated form of CPY in the ER as manifested by its higher mobility in an acrylamide gel (Fig. 3E). Importantly, low protective doses of Tacrolimus restored the trafficking of CPY from the ER to the secretory pathway.

Taken together, these data coupled to the MS phosphoproteomic studies strongly support an endogenous functional interaction between FKBP12 and calcineurin, whereby FKBP12 has the ability to alter the calcineurin-dependent phosphoproteome toward proteins involved in vesicle trafficking, endocytosis, and actin cytoskeletal organization among other functional ontologies.

Subsaturating Doses of Tacrolimus Protect Against α-Syn Toxicity in Vivo.

Our findings thus far establish a central role of FKBP12 in contributing to α-syn toxicity through at least two mechanisms: (i) by shifting the calcineurin-dependent phosphoproteome toward proteins involved in vesicle trafficking, endocytosis, actin cytoskeleton organization, etc. and (ii) by affecting other pathways independent of calcineurin, which have been previously implicated by other groups (24). Importantly, both of these toxic properties of FKBP12 can be pharmacologically rescued with low subsaturating doses of Tacrolimus. Our data suggest that subsaturating doses of Tacrolimus are key to maintain a minimal but essential level of calcineurin activity, which is critical to engage the potential protective substrates described above, while at the same time, inhibiting FKBP12 through a calcineurin-independent manner.

To investigate whether the protective properties of low doses of Tacrolimus are relevant in vivo, we turned to a rat model of PD. This model is based on unilateral stereotactic injection of an adeno-associated virus encoding α-syn directly into the substantia nigra pars compacta (SNc), a brain region highly affected in PD. Importantly, α-syn–induced damage in the SNc is reflected in the striatum, the region of the brain that receives all of the projections from the dopaminergic neurons in the SNc. These defects include loss of DAT, loss of DA, and increases in DA’s breakdown products, such as homovanillic acid (HVA) and 3,4-Dihydroxyphenylacertic acid (DOPAC) (13). The dysregulation of trafficking of the DATs as well as DA secretion at presynaptic terminals are responsible for the behavioral deficits in these animals (13).

Using this rat in vivo model of PD, we tested whether inhibiting calcineurin and FKBP12 over a range of Tacrolimus doses could ameliorate any of the neurological defects caused by α-syn (Fig. 4A). Indeed, the lower doses but not the higher doses of Tacrolimus rescued the behavioral deficits caused by α-syn as assessed by a paw asymmetry assay (Fig. 4B). The restoration in behavioral deficits caused by α-syn overexpression was accompanied by a rescue in the deficits of DATs (Fig. 4C), DA (Fig. 4D), and DA breakdown products DOPAC and HVA at the striatal terminals (Fig. S4A). Importantly, like in cell culture experiments, the differences in calcineurin activation between control and α-syn–expressing neurons were not caused by differences in calcineurin expression but rather, calcineurin activity (Fig. S4C).

Normal functioning at these terminals is highly dependent on biological processes, such as vesicle trafficking, endocytosis, and actin reorganization, and largely regulated by phosphorylation (30). Therefore, it is reasonable to consider that low doses of Tacrolimus protect against α-syn toxicity by regulating the phosphorylation of proteins implicated in boosting vesicle trafficking at the terminals, thereby facilitating DA transport and secretion.

To explore whether low doses of Tacrolimus can restore the phosphorylation of calcineurin- and FKBP12-dependent substrates important for the regulation of vesicular traffic at SNc presynaptic striatal terminals, we utilized Tandem Mass Tags (TMT) MS (31). This unbiased isobaric mass method allows one to quantitate phosphopeptides across animals and conditions postlysis. We reasoned that this is the best available approach, despite the fact that significant changes in the phosphorylation state of substrates at the SNc terminals would be masked by the rest of the other neuronal types present in the striatum, which are unaffected by α-syn. Despite these limitations, we detected 51 of 3,526 phosphopeptides that significantly changed comparing control and α-syn animals. Of these, one-half were significantly hypophosphorylated in the α-syn animals. We retrieved four phosphosites, two from each of two presynaptic proteins, with phosphorylation that increased in the animals treated with low doses of Tacrolimus (Fig. 4E and Dataset S3). Specifically, these proteins are GAP-43 and BASP1. Both proteins have critical roles in assisting neuronal path finding and branching during regeneration and contribute to presynaptic changes leading to neurotransmitter release, endocytosis, and synaptic vesicle recycling (32, 33). Importantly, their activity has been shown to be affected by calcineurin, although the exact mechanism of regulation is not known (34, 35). While these proteins do not have homologs in yeast, they share common features with the putative calcineurin- and FKBP12-dependent proteins retrieved from the yeast model. Specifically, these substrates regulate processes linked to endocytic, vesicle trafficking, and actin cytoskeleton dynamics, and they also contain an unusually high number of prolines, making them an ideal substrate for cys-trans prolyl isomerases, such as FKBP12.

Together, these data support the findings in yeast that targeting the endogenous functional interaction between calcineurin and FKBP12 with subsaturating doses of Tacrolimus is neuroprotective. Specifically, low doses of Tacrolimus can inhibit a fraction of active calcineurin/FKBP12 complexes, thereby modulating the phosphorylation of proteins critically involved in the restoration of DAT trafficking and DA secretion at presynaptic terminals to ultimately ameliorate the behavioral deficits caused by α-syn in a rat model of PD.

Mammalian Constructs.

All neuronal constructs were cloned into the pLENTI6/V5 DEST (Invitrogen) lentivirus expression vector using the Gateway system. These included the genetic-encoded β-gal (LacZ) and α-syn A543T.

Rat Primary Cortical Cultures.

Cultures were prepared based on Lesuisse and Martin (44). Embryos were harvested by Cesarean section from anesthetized pregnant Sprague–Dawley rats at embryonic day 18. Cerebral cortices were isolated and dissociated with ACCUMAX digestion for 20 min at 37 °C and trituration with Pasteur pipettes. Polyornithine- and laminin-coated 96-well plates were seeded with 4 × 10⁴ cells in neurobasal medium (Life Technologies) supplemented with B27 (Life Technologies), 0.5 mM glutamine, 25 μM β-mercaptoethanol, penicillin (100 IU/mL), and streptomycin (100 μg/mL). One-third of the medium was changed every 3–4 d. Tacrolimus (Ontario Chemicals) and CsA (Sigma) at the indicated concentrations were added to the lentivirus-transduced cultures in 96-well plates at day in vitro 18 (DIV18), keeping the amount of DMSO constant (vehicle). As a surrogate marker of cell viability, cellular ATP content was measured using the ViaLight Plus kit (Lonza). Lentiviral constructs were packaged via lipid-mediated transient transfection of the expression constructs and packaging plasmids (pMD2.G and psPAX2) into 293 cells. Lentivirus were purified and concentrated using the Lenti-X Maxi Purification kit and LentiX Concentrator (Clontech) according to the manufacturer’s protocol. Lentivirus titer was determined using the QuickTiter Lentivirus titer kit (lentivirus-associated HIV p24; Cell Biolabs) according to the manufacturer’s protocol. Rat cortical cultures were transduced with high multiplicity of infection (25) or low (15) of lentivirus at DIV5.

Yeast Deletions.

All deletions were performed by homologous recombination according to the Saccharomyces Genome Deletion Project using a one-step gene disruption. Briefly, a PCR-based strategy was performed to amplify a product that contains an antibiotic resistance cassette (either kanamycin or hygromycin) flanked by 45 bp of sequence corresponding to the 5′ and 3′ UTRs of the target gene (sequence available at the Saccaromyces Genome Deletion Project). Yeast strains were transformed with the PCR product using a standard lithium acetate protocol, allowed to recover overnight in yeast extract-peptone-dextrose media at 30 °C, and then, selected on plates with the corresponding antibiotic-resistant marker. Colonies were screened for the deletion using PCR confirmation primers. Results are representative of at least three independent colonies. All transformations in yeast for deletion purposes were performed at least three independent times and analyzed with at least three independent transformants each time.

Spotting Assays.

Cells were grown overnight at 30 °C in 3 mL synthetic dropout (SD) medium lacking the relevant amino acids and containing glucose. Cell concentrations (OD₆₀₀) were adjusted to the lowest concentration, and then, they were fivefold serially diluted and spotted onto SD medium plates containing glucose (uninduced) or galactose (induced). Plates were incubated at 30 °C for 2 d (glucose) or 3 d (galactose).

MS Shotgun Approach.

TMT MS.

Rat in Vivo Studies.

Motif Enrichment Analysis.

ORFs from the genome of the ScS288c strain containing LxVP, PxIxIT, or both patterns were identified using the Yeast Genome Pattern Matching tool provided by the SGD; for each list of phosphosites identified in a specific experimental condition, the corresponding list of ORFs was analyzed to obtain counts for ORFs containing LxVP, ORFs containing PxIxIT, and ORFs containing both patterns, respectively. Enrichment was calculated as the percentage of ORFs containing each specific pattern (or both patterns) in the experimental list divided by the percentage of ORFs containing the same pattern(s) in all ORFs of the genome. P value for the null hypothesis (i.e., there is no enrichment) was calculated using Pearson’s χ² test.