Characterization of α-syn phosphorylation in TG models of synucleinopathies and human brain from AD, LBD, and MSA patients
In order to determine the physiological and pathological relevance of S87 phosphorylation, we generated a rabbit polyclonal antibody, anti-S87-P, against a synthetic peptide that corresponds to amino acid residues 81–93 of human α-syn, with S87 being phosphorylated. demonstrate that our anti-S87-P antibody specifically detects serine 87 phosphorylated forms (S87-P) of α-syn, and reveals the absence of immunoreactivity against either unphosphorylated, S129A or S129-P α-syn. To rule out the possibility that the lack of immunoreactivity of S87A is due to a loss of the epitope rather than lack of phospohorylation, the specificity of the antibody was verified by probing its immunoreactivity toward diphosphorylated α-syn (S87-P/S129-P) after phosphatase (CIAP) treatment. Phosphatase treatment resulted in the loss of immunoreactivity toward both anti S87-P and anti-S129-P antibodies (), consistent with the loss of the phosphate group on both residues. Thus, providing further verification of the specificity of both antibodies.
Specificity of anti-S87-P antibody
To probe the specificity of our S87-P antibody and determine if phosphorylation at this residue occurs in vivo
, we characterized the level of S87-P in slices from rat brains injected on one side with an AAV2/6 viral vector over-expressing the S129A variant of α-syn A30P (, n=3). As a control, we also examined tissues from rat brains injected a mutant form of the same protein, in which phosphorylation at S87 was blocked by a serine to aspartate substitution at position 87 (S87D/S129A–A30P) (, n=3). In both cases, the uninjected side served as an additional control to assess the level of background signal due to non-specific interactions and antibody cross-reactivity with other proteins. We detected a strong staining against human A30P α-syn in the injected substantia nigra () compared to the non-injected side (). However, anti-S87-P staining was exclusively detected in the substantia nigra over-expressing human S129A α-syn () and no labelling was detectable, neither in the non-injected side () nor in the substantia nigra over-expressing the S87D/S129A mutant (). These findings demonstrate conclusively that phosphorylation at serine 87 does occur in vivo
, at least under conditions of increased α-syn expression (corresponding to an approximate doubling of the total α-syn level in nigral tissue extracts (Azeredo da Silveira et al. 2009
)) and validate the specificity of our antibody on brain tissue. However, it is not possible to determine the overexpression levels in individual neurons.
α-Syn in LBs is phosphorylated at S87-P and S87-P levels are increased in human brains from AD, LBD, MSA patients
Having established the specificity of the antibody, anti-S87-P, together with anti-α-syn and anti-S129-P antibodies, was tested in human brain (frontal cortex) homogenates from control, AD, LBD and MSA cases. Immunoblot analysis of cortical samples, from control or diseased human brains, showed a native α-syn band at 14 kDa in the cytosolic fractions (), which was significantly increased in the AD, LBD, and MSA cases relative to the control (). In the membrane fractions increased α-syn levels is detected in the LBD and MSA cases, with higher molecular weigh species being present in the LBD cases only (). Anti-S87-P antibody detected a weak band at 14 kDa in the control cases. This band became more pronounced in the diseased brains notably in the LBD cases (), where it was accompanied by a weaker band at around 17 kDa. The antibody against S129-P detected a band at 14 kDa as well as higher MW bands (). With both antibodies against phosphorylated α-syn most of the immunoreactivity was detected in the membrane fractions in the LBD and MSA cases but not in the controls and AD cases (). Several α-syn immureactive bands at molecular weights of 38 kDa and higher were detected in the control samples from LBD cases. These bands were more abundant in the membrane fractions and showed clear immureactivity towards the anti-S129-P, but at best weak reactivity to the anti-S87-P antibody.
Comparison of the levels of phosphorylated α-syn immunoreactivity by immunoblot in human brains from controls, AD, LBD and MSA cases
To further investigate the pathological relevance of S87 phosphorylation, we assessed S87-P levels in Lewy body samples from three diffuse Lewy body (DLB) cases. Lewy bodies were intensely labelled by the sheep anti-α-syn antibody. Therefore, the signal gain was set low (50% of S87-P α-syn signal) to show characteristic homogenous and concentric patterns, with little background or neuroile labelling ( left column). The S87-P α-syn immunoreactivity was less intense but readily detected in all Lewy bodies (more than 500 Lewy bodies were analyzed by confocal microscopy). The S87-P α-syn labelling appeared smooth within Lewy bodies (). In addition, granular S87-P α-syn labellings of 0.2–0.5µm size were also detected, more obvious in Lewy body periphery, neural processes, and neuropil (). No such labelling was seen when anti- S87-P α-syn antibody was omitted. These granular labellings may be complexes or cellular particles containing more concentrated S87-P α-syn. illustrate the relative intensities for α-syn (red line), S87-P α-syn (green) and DNA (blue) labelings across a Lewy body. However, it must be noted that fluorescence intensity depends on many factors, including the dye used, the affinity of each antibody for its epitope, the accessibility of the epitope, etc.
Analysis of the colocalization of S87-P and α-syn in Lewy bodies isolated from fresh brain of DLB cases
The localization of S87-P α-syn in Lewy bodies suggests it is an integrated part of the inclusions. We further examined whether S87-P α-syn is enriched in buffer insoluble fractions in brains of synucleinopathies and Alzheimer’s disease. The buffer insoluble fractions contained numerous Lewy bodies and glial inclusions and were further solublized in 5%SDS/Urea. As shown in , S87-P α-syn reactivity was more prominent in 5%SDS/Urea fraction, compared to total α-syn reactivity. demonstrates that the level of S87-P in these fractions is higher in diseased brains relative to control cases.
The levels of S87-P are increased in brains of transgenic (TG) models of synucleinopathies
To further assess if S87 phosphorylation is a pathological event, we assessed its levels in brain homogenates from TG mouse experimental models of PD/LBD and MSA (). As expected the highest levels of α-syn expression were observed in the mThy1 α-syn TG mice followed by the PDGF α-syn and the MBP-α-syn TG both in the cytosolic () and membrane fractions (). Bands reflecting monomers (at 14 kDa) and the oligomers (at 38–62 kDa) were detected. Consistent with the finding in the human brain homogenates, antibodies against S87-P detected a band at 14 kDa, which were significantly more abundant in the membrane fraction of the mThy1 α-syn TG mice (). With antibody against S129-P, a strong 14 kDa band was detected in the cytosolic fraction of both the mThy1 α-syn and MBP α-syn TG mice (). S129-P α-syn was detected at 14 and 42 kDa in the membrane fractions of the α-syn TG mice (). Consistent with the western blots, immunocytochemical analysis with the antibody against total α-syn showed immunoreactivity in the neuropil corresponding to nerve terminals in the non TG mice (). In the PDGF and mThy1-α-syn TG mice abundant accumulation of α-syn was detected in the neuronal cell bodies while in MBP α-syn TG mice immunoreactivity was associated with oligodendrocytes (). With S129-P antibody abundant α-syn immunoreactivity was detected in the three lines of α-syn TG mice (). With the S87-P antibody abundant immunoreactivity was observed in the neuronal cell bodies of the mThy1-α-syn TG (but not in the PDGF-α-syn) mice and in the inclusions in the oligodendrocytes in the MBP-α-syn TG mice ().
Comparison of the levels of phosphorylated α-syn immunoreactivity by immunoblot in brains of α-syn models of LBD
Patterns of phosphorylated α-syn immunoreactivity in brains of α-syn models of LBD
S87-P colocalization with CK1
Among the various kinases reported to phosphorylate α-syn in vitro
, CK1 and the dual specificity tyrosine regulated kinase 1A (Dyrk1A) are the only two that phosphorylate α-syn at S87. To determine the co-localization between S87-P immunolabeled-neurons and CK1, 40 µm-thick vibratome sections from the temporal cortex of α-syn TG mice and LBD/PD patients were immunolabeled with the antibodies against S87-P and CK1. demonstrates a great degree of colocalization of S87-P and CK1 in neuronal inclusions in Tg mice () and in Lewy body like structures in LBD/PD diseased brains (), consistent with in vitro
observations and suggesting that CK1 may by directly involved in modulating α-syn phosphorylation in vivo
. These findings are consistent with previous cell culture studies implicating CK1 in the phosphorylation of α-syn at S129 and S87 (Okochi et al., 2000
Colocalization of S87-P and CK1 within inclusions in the thy1-α-syn TG mice and LBD/PD diseased brains
Serine→Glutamate substitution or phosphorylation at S87 inhibits the fibrillization of WT and mutant (S129A and S129E) α-syn
Having verified the pathological relevance of S87 phosphorylation, we sought to understand the role of S87 phosphorylation in modulating the structure and aggregation properties of α-syn. Towards this goal, we compared the structural, oligomerization, fibrilization and membrane binding properties of monomeric WT α-syn to those of the phosphorylation mimics (S87E) as well as the purified in vitro S87-phosphorylated form of α-syn using NMR, CD, SEC, SDS-PAGE, ThT, and TEM.
Phosphorylation at S87 is sufficient to block α-syn fibrillization
Recent studies from our laboratory demonstrated that CK1 mediated phosphorylation blocks α-syn fibrillization (Paleologou et al., 2008
). In vitro
CK1 phosphorylation of WT α-syn followed by tryptic digestion and mapping of phosphorylation sites revealed that CK1 phosphorylates α-syn at multiple sites (S87, T92, S129), with S87 and S129 being the major phosphorylation sites (supplemental material and figures 2 and 3
). To determine the relative contributions of phosphorylation at S87 to the CK1-induced inhibition of α-syn fibril formation, we examined the effect of CK1-mediated phosphorylation on the fibrillization of the S129A and S129E, both of which cannot be phosphorylated at S129. Prephosphorylation of both variants with CK1 results in significant retardation of α-syn fibrillization and a reduction in amyloid fibril formation relative to the unphosphorylated forms of both proteins (supplemental Fig. 3
). S129A, S129E and the WT α-syn formed significant amounts of amyloid fibrils, whereas prephosphorylated forms of S129E (S129E/S87P) and WT (S129P/S87P) showed predominantly soluble oligomeric species. S129A rapidly forms short protofibrillar/fibrillar structures, consistent with its marked increased propensity to fibrillize (Paleologou et al., 2008
). However, the amount of fibrils formed by prephosphorylated S129A is significantly less than that observed for unphosphorylated S129A.
To further prove that inhibition of α-syn fibrillogenesis is due to phosphorylation at S87, we prepared and purified the S87-P phosphorylated forms of S129A and S129E α-syn and compared their aggregation properties to those of the corresponding unphosphorylated forms of α-syn. The phosphorylated (S87-P) and unphosphorylated α-syn forms of each protein were separated by RP-HPLC and their purities were verified by MALDI-TOF mass spectrometry. We consistently observed that the mono (S87-P) and diphosphorylated (S87-P/T92-P) forms of S129A and S129E species did not form fibrils even after 48 h, whereas the unphosphorylated forms of these proteins exhibited extensive fibril formation after 24 h of incubation (). These observations were confirmed by SEC and TEM studies, which revealed the absence of any fibrillar aggregates and presence of predominantly monomers in the S129A/S87-P and S129E/S87-P samples (). Interestingly, monophosphorylated S129E/S87-P appears to form significantly less oligomeric/protofibrillar-like aggregates than S129A/S87-P, suggesting that the presence of both phosphate and Glu substitutions at S87 and S129, respectively results in a greater inhibition of α-syn fibrillization.
Phosphorylation at S87 inhibits S129A α-syn aggregation propensity
In the fibrillar state, only S129 undergoes phosphorylation by CK1 in vitro
According to the solid state NMR data α-syn residue 87 is in a region that is part of the rigid structure of the fibrils and within one of the beta-strands participating in the formation of the fibril structure (Kloepper et al., 2007
; Heise et al., 2008
; Vilar et al., 2008
), suggesting that it might not be subject to phosphorylation after fibril formation has occurred. To determine if S87 is accessible and can undergo phosphorylation in the fibrillar state, we generated monomer-free fibrillar samples of WT α-syn and subjected them to in vitro
phosphorylation with CK1. Phosphorylation was assessed by western blotting using antibodies against α-syn, S87-P and S129-P. We observed that only S129 undergoes phosphorylation by CK1, suggesting that S87, within α-syn fibrils, is not accessible or exists in a conformation that is not recognized by CK1 (supplemental Fig. 5
The phosphomimic S87E aggregates slower than WT and S87A α-Syn
Selective phosphorylation and/or overexpression of S87-P in vivo
is currently not possible, as CKI phosphorylates at both S87 and S129 and DYRK1A mediated phosphorylation at S87 is less efficient. Therefore, future efforts to assess the role of S87 phosphorylation in modulating the normal biology and aggregation of α-syn in vivo
are likely to rely on the use of the phosphomimics S87E and S87A. Recent studies from our group (Paleologou et al., 2008
) and others (Waxman and Giasson, 2008
; McFarland et al., 2009
) demonstrate that substitution of S129 by glutamate or aspartate does not reproduce the effect of phosphorylation at this site on α-syn structure and aggregation properties in vitro
. Having established that S87-P is sufficient to inhibit α-syn oligomerization and fibrillogenesis, the 87 mutants S87A and S87E were generated to assess whether the phosphorylation mimic S87E is likely to mimic S87 phosphorylation in vivo
. Comparison of the fibrillization of WT, S87A and S87E α-syn showed that S87A aggregated at a similar rate and to similar levels as WT α-syn, whereas for S87E the observed aggregation rate was slower than those of the WT and S87A proteins (). TEM images further confirmed these findings, as both WT and S87A formed dense networks of fibrils as opposed to S87E, which formed a few short fibrils after 72 h of aggregation (). In accordance with these findings, the consumption of monomeric protein over time monitored by SEC and SDS-PAGE analysis also revealed a significant decrease in monomeric WT and S87A (), whereas the levels of the monomeric S87E α-syn remained virtually unchanged over the 72 h of aggregation, providing further evidence that S87E does not aggregate.
α-Syn is disordered independent of phosphorylation at S87
Serine 87 lies in the hydrophobic NAC region of α-syn, suggesting that the introduction of a negative charge through the substitution of serine by glutamate at this position is unfavourable and may alter the secondary structure and hydrodynamic properties of monomeric α-syn. demonstrates that WT, S87A and S87E α-syn exhibit virtually identical random coil and α-helical CD spectra in solution and upon binding to synthetic membranes, respectively. Given that α-syn exists predominantly in a random coil conformation, subtle changes in the secondary structure of the protein are unlikely to be discernable by CD studies. Therefore, to further elucidate the consequences of phosphorylation on the structure and dynamics of monomeric α-syn, high-resolution NMR studies were also conducted. For all proteins, the resonances in 1
N HSQC spectra were sharp and showed only a limited dispersion of chemical shifts, reflecting a high degree of backbone mobility (). Upon phosphorylation of WT α-syn by CK1, the resonances of S87 and S129 were strongly attenuated at the position seen in the proton-nitrogen correlation (HSQC) of the unphosphorylated protein, but new signals appeared in the region, in which resonances of phosphorylated amino acids are usually found (supplemental Fig. 4A
). For S129D α-syn, phosphorylation at S129 was blocked so that only the resonance of S87 was attenuated at its original position and appeared at its phosphorylated position (). Other chemical shift changes induced by phosphorylation were generally small. In addition, steady-state heteronuclear 15
H]-NOEs (), 15
relaxation rates and 3
) couplings (supplemental Fig. 4
) were very similar for unphosphorylated and phosphorylated α-syn, indicating that phosphorylation at S87 has no apparent effect on the secondary structure of α-syn.
α-syn is disordered independent of phosphorylation at S87
Pulse field gradient NMR experiments allow accurate determination of the diffusion coefficient of a molecule. From the diffusion coefficient, a hydrodynamic radius Rh
can be calculated that provides an estimation of the overall dimensions of a biomolecule (Farrow et al., 1994
). For WT and S129D α-syn, we determined hydrodynamic radii of 28.2 Å and 28.1 Å, respectively (). The phosphorylation mimic S87E did not change the hydrodynamic radius of α-syn. On the other hand, phosphorylation of S129A and S129D α-syn at S87 increased Rh
by 2.6 and 1.9 Å (). Addition of 8 M urea to phosphorylated S87A α-syn further increased the Rh
value to 36.2 Å. If α-syn were a true random coil, a hydrodynamic radius of 36.9 Å would be expected (Kohn et al., 2004
). Interestingly, phosphorylation of S129 appears to have a more dramatic effect on the hydrodynamic radius of monomeric α-syn as it results in an increase of Rh
by ~ 5.0 Å. This result is consistent with our previous studies demonstrating disruption of the intramolecular interactions between the C and N-terminal regions of α-syn upon phosphorylation at S129 and suggest that phosphorylation at S87 may only partially disrupt this interaction (Paleologou et al., 2008
Phosphorylation at S87 results in changes in protein conformation upon membrane binding
In the presence of SDS micelles, the lipid-binding domain of α-syn adopts a conformation consisting of two helical segments, which have been previously characterized using NMR (Eliezer et al., 2001
; Bussell and Eliezer, 2003
; Chandra et al., 2003
; Ulmer and Bax, 2005
). A comparison of proton-nitrogen correlation spectra of micelle-bound phosphorylated WT (S87-P and S129-P) α-syn and its unphosphorylated counterpart shows that the resonances of S87 and S129 of the phosphorylated protein are dramatically shifted. In addition to local changes, many resonances of amino acids in the vicinity of S87-P were also shifted (79, 81, 83, 84, 85, 86, 88, 89, 91, 92, 93, 94, 95, 96) (). Most of these amino acids are part of the second helix that α-syn adopts upon binding to micelles. Relatively less extensive shifts were observed in the resonances of amino acids close to S129, consistent with our previous studies of the effects of phosphorylation at this site (Paleologou et al., 2008
). Similar changes in the vicinity of S87-P were made when comparing the HSQC spectra of S129A/S87-P () and S129E/S87-P () and their unphosphorylated counterparts, strongly suggesting that changes occur in the environment of the second helix are a result of S87 phosphorylation.
Phosphorylation at S87 results in changes in protein conformation upon membrane binding
In order to quantify the chemical shift changes between the phosphorylated and unphosphorylated micelle-bound proteins, the resonances of each form of S129A α-syn were directly assigned using triple resonance methods. Graphs of the resulting weighted average of the amide proton and nitrogen chemical shift changes between the phosphorylated and unphosphorylated proteins are shown () for S129A /S129A–S87-P. A similar graph was generated for S129E/S129E–S87-P () by transferring the S129A assignments. The data demonstrate clearly that relatively large chemical shift changes take place at sites close to S87 in both phosphorylated proteins, with more minor changes spanning the region between residues 60 and 80. Chemical shift changes observed upon comparing micelle-bound unphosphorylated S129A or S129E with WT α-syn were restricted to the immediate sites of the mutation and were essentially identical to those previously observed for the free protein (Paleologou et al., 2008
), indicating that the mutations themselves do not cause long-range effects.
To better understand the effects of S87 phosphorylation on the micelle-bound structure of α-syn, we analyzed both sequential HN-HN NOE and Cα chemical shift data. The NOE data () reveal a decrease in signal intensity, when compared with those previously obtained for the WT protein, in the second half of the second micelle-bound helix, coinciding with the location of the phosphorylated S87 residue and suggesting a possible decrease in the stability of the helical structure in this region. Accordingly, secondary Cα chemical shift data () also show smaller positive amplitudes in this region when compared with previous observations for the unphosphorylated WT protein. Interestingly, these structural effects are similar to those observed for the protein β-syn, a close homologue of α-syn, which is however missing 11 residues from the NAC region (Sung and Eliezer, 2006
The effects of S87 phosphorylation on the micelle-bound structure of α-syn
To further probe the effect of phosphorylation on α-syn-membrane interactions, we compared the CD spectra of mono- (S129A/S87-P) and diphosphorylated S129A (S129A/S87-P,T92-P) to that of S129A and WT α-syn in solution and upon binding to POPG vesicles. Although the effect of phosphorylation at S87 and T92 at the monomer is not discernable by Far-UV/CD (), both mono- (S129A/S87-P) and diphosphorylated (S129A/S87-P,T92-P) exhibit reduced α-helical propensity in the presence of synthetic vesicles with diphosphorylation having the most dramatic affect ().