Selective expression of PINK1 kinase in the human and monkey brains
Our recent study showed a remarkable reduction of PINK1 protein in CRISPR/Cas9-targeted monkeys (Yang et al., 2019b), which is apparently different from the previous findings in PINK1 knockout rodent models, as none of these rodent models were able to demonstrate the difference in PINK1 protein expression between wild type and Pink1 knockout animals (Kitada et al., 2007; Gispert et al., 2009; Xiong et al., 2009; Akundi et al., 2011). Whole genome sequencing verified the specific targeting of the PINK1 gene in mutant monkey brains (Yang et al., 2019b). In order to compare the expression levels of endogenous PINK1 in mouse and monkey brains, we also generated a new Pink1 KO mouse model by disrupting exon 2 and exon 4 in mouse Pink1 via CRISPR/Cas9 (Fig. S1A and S1B) with the same strategy for deleting the monkey PINK1 (Yang et al., 2019b). We obtained multiple mouse founders with different types of PINK1 mutations (Fig. S1C). Of these mutant mice, mice carrying the targeted exon 2 (Δ38/Δ38) and exon 4 (Δ7/Δ7) are homozygous Pink1 KO mice (Δ38/Δ38; Δ7/Δ7), which were found to have disrupted two alleles of the Pink1 gene (Fig. S1C) and used for further characterization. Although the new Pink1 KO mice lived normally without any obvious phenotypes or neurodegeneration, similar to other previously reported Pink1 KO mice, they provided a rigorous control to validate the expression of Pink1 in mice.
Using RT-PCR with primers specific for exons 1–2 and 6–7 (Fig. S1D), we found that PINK1 transcripts were undetectable in the brain cortical tissues of Pink1 KO mice (Fig. S1E and S1F), suggesting a complete knock-out of the mouse Pink1. In a PINK1 targeting monkey (M1), which was generated in our recent study (Yang et al., 2019a, b), different degrees of reduction of transcripts of exon 1–2 and exon 6–7 were seen (Fig. S1E and S1F), consistent with the mosaic PINK1 targeting by CRISPR/Cas9. We then used six commercially available antibodies (rabbit BC100-494, sheep S086D, sheep S085D, rabbit Ab23707, sheep S460C, and sheep S774C) that recognize different regions in PINK1 (Figs. 1A and S2A) to examine PINK1 expression in the mouse, monkey, and human tissues. Because all PINK1 antibodies label some non-specific bands, it is necessary to use the mutant monkey brains (M1 and M6) that were targeted on the PINK1 gene (Yang et al., 2019b) to verify the specificity of the PINK1 antibodies. CRISPR/Cas9 targeting created mosaic mutations that do not completely delete the PINK1 gene but could reduce its expression. The rabbit antibody BC100-494, which is against human PINK1 (175–250 aa), appeared to react strongly with PINK1 at ~55 kDa (PINK1-55) that was markedly reduced in the M1 monkey brain, indicating that PINK1-55 is a protein product of the targeted PINK1 gene (Fig. 1B). However, BC100-494 did not detect any specific band that was only present in the wild type mouse brain and absent in Pink1 KO mouse brain (Fig. 1B), supporting the idea that endogenous Pink1 in the mouse brain is expressed at an undetectable or very low level. Comparison of the monkey brain and peripheral tissues showed that PINK1-55 was selectively expressed in the brain tissues (Fig. 1C). This brain selective expression was validated by comparing wild type and mutant monkey (M6) tissues, showing that PINK1-55 was specifically expressed in the monkey brain (Fig. S2B). The selective expression of PINK1 in the primate brain was further verified by Western blot of human peripheral and brain tissues and additional PINK1 mutant monkey (M6) that showed a reduction in PINK1-55 expression because of PINK1 targeting (Fig. 1D). We also used a sheep antibody (S086D) against mouse Pink1 (175–250 aa) and obtained similar results that additionally revealed the selective brain expression of PINK1-55 in human tissues (Fig. 1E). To verify that PINK1-55 is only present in the primate brain, we also used three other antibodies (S774C, Ab23707 and S460C) for Western blot. The results demonstrated that all these antibodies could detect PINK1-55 in the human and monkey cortex and that PINK1-55 was reduced in the M6 monkey brain (Fig. S2C). Again, none of these antibodies could detect a specific band that was only present in wild type mice (Fig. S2C). We further used five anti-PINK1 antibodies to compare PINK1-55 expression in different monkey brain regions. The results consistently demonstrated that PINK1-55 was expressed at a higher level in the substantia nigra (SN) (Fig. S2D). Given that the SN is the most vulnerable brain region in PD, the unique expression pattern of PINK1 in the primate brain regions is consistent with its involvement in the selective neurodegeneration in PD.
Comparing different antibodies revealed that BC100-494 reacted strongly with PINK1-55 in the primate brain so that we used BC100-494 for further characterization of PINK1 expression. We found that PINK1-55 is undetectable or at the very low level in several cell lines (Fig. S2E). BC100-494 has been used to detect full-length PINK1 at ~70 kDa, which is localized to damaged mitochondria in human and mouse cell lines (Okatsu et al., 2012). Indeed, BC100-494 was able to recognize full-length PINK1 at ~70 kDa, which was induced by a mitochondrial membrane potential disruptor, CCCP (carbonyl cyanide 3-chlorophenylhydrazone), in human (HEK293) and mouse (N2A) cell lines and is larger than PINK1-55 (Figs. 1F and S2F). These results indicate that PINK1-55 lacks the N-terminal PINK1 region that can associate with the mitochondria. Consistently, fractionation of wild-type monkey brain tissues revealed that PINK1-55, which was detected by both BC100-494 and S086D, is mainly present in the cytoplasm (Fig. 1G).
Using isolated white matter that is enriched in glial cells and gray matter that is enriched in neurons for Western blot, we found that PINK1-55 is expressed in both neuronal and glial cells (Fig. S3A). Immunocytochemical staining revealed abundant distribution of PINK1-55 in the cytoplasm of neuronal and glial cells (Fig. S3B and S3C). Using the postmortem brain tissues from humans, we also found that PINK1 was detected in neurons and astrocytes (Fig. 1H). Taken together, PINK1-55 is a cytoplasmic form of PINK1 that is selectively expressed in the human and monkey brains.
Acute deletion of PINK1 in the adult monkey brains causes neurodegeneration
Our recent studies have shown that targeting the PINK1 gene in embryos leads to neuronal loss in monkey brains (Yang et al., 2019a, b), therefore, next we wanted to know whether directly removing PINK1 in the brains of wild-type adult monkeys could cause neurodegeneration. We previously used stereotaxic injection of AAV9 vector expressing CRISPR/Cas9 and gRNA (AAV-gRNA) to target the mutant huntingtin (HTT) gene in adult mouse brain, which efficiently and permanently eliminated mutant HTT-mediated neuronal toxicity without affecting neuronal viability (Yang et al., 2017). We used the same AAV-gRNA vector to target PINK1 exon 2 and exon 4 in the monkey brain (Figs. 2A, S4A and S4B) and performed stereotaxic injection of this viral vector into the brains of rhesus monkeys using the same method in our previous studies (Yang et al., 2015). Because AAV-gRNA also expressed RFP, immunohistochemistry with anti-RFP was able to detect RFP signal in the injected area one month after injection (Fig. S4C). Using Western blot with BC100-494, we confirmed the reduction of PINK1-55 in the AAV-PINK1 gRNA/Cas9-injected monkey cortex, and this reduction was also revealed by a different sheep anti-PINK1 antibody (S085D) (Fig. S4D).
To examine whether PINK1 deletion in adult monkey brains also causes neurodegeneration in different brain regions, we injected AAV-PINK1 gRNA/Cas9 into the prefrontal cortex and substantia nigra of adult monkeys (one 3-year-old, two 12-year-olds, and one 10-year-old) and then isolated their brain tissues 3 months later for histological examination. We first verified that targeting PINK1 could reduce PINK1 expression by performing double immunofluorescent staining of the injected monkey brain regions with antibody to RFP to detect PINK1 gRNA expression and BC100-494 to detect PINK1. Targeting PINK1 by AAV PINK1-gRNA/Cas9 apparently significantly reduced PINK1 expression as compared with the AAV control-gRNA/Cas9 injection (Fig. 2B). High magnification micrographs showed that AAV-infected neuronal cells, which showed both RFP and NeuN expression, were present in the AAV control-gRNA/Cas9-injected brain cortex but were rarely seen in the AAV-PINK1-gRNA/Cas9-injected cortex (Fig. S5A and S5B). We then used antibodies to NeuN, PINK1 and RFP to further assess neuronal loss in the injected monkey brains (Fig. 2C–F). We found that PINK1 depletion in the cortex and substantia nigra in adult monkeys resulted in a remarkable loss of neuronal cells, evidenced by decreases in both NeuN staining (Fig. 2C and 2D) and RFP labeling (Fig. 2E and 2F) in the AAV-PINK1-gRNA/Cas9-injected area. However, no evident degeneration was found in AAV-GFP or AAV-control-gRNA/Cas9-injected brain regions (Fig. 2C and 2E). The selective neuronal loss in the AAV-PINK1-gRNA/Cas9-injected area was also supported by the abnormal and difficult movement of the left limbs of the injected monkey (Video S1), which is controlled by the contralateral right side of the substantia nigra that was injected with the AAV-PINK1-gRNA/Cas9, when compared with the right limbs that are controlled by the left substantia nigra injected with AAV-control gRNA/Cas9.
Loss of PINK1 did not alter mitochondrial proteins and morphology in the monkey brain cells
In the substantia nigra, we also found a remarkable reduction of large-sized and NeuN-positive neurons when PINK1 was depleted (Fig. 3A). Interestingly, in the survived neurons, immunostaining of the mitochondrial protein (TOM20) did not reveal any obvious alteration in the mitochondria density when comparing with control neurons infected by AAV-control gRNA/Cas9 (Fig. 3B). These results led us to use electron microscopy to explore whether loss of PINK1 can affect mitochondria. Degenerated neurons were seen in the substantia nigra that had been injected with AAV PINK1-gRNA/Cas9 when comparing with the AAV-GFP control (upper panel in Fig. 3C). The degenerated neurons displayed increased cytoplasmic density to dark profiling without clear nuclear and organellar structures and often contained increased lysosomal and phagocytic vacuole-like structures. Interestingly, in the degenerated neurons, the morphology and number of mitochondria appeared normal compared with the control gRNA-injected area (low panel in Fig. 3C). Quantification of ultrastructurally identifiable mitochondria also showed that the percentage (346 out of 387 or 89.4%) of mitochondria with normal morphology in AAV PINK1-gRNA/Cas9-injected substantia nigra was similar to that (279 out of 307 or 91.2%) in AAV-control gRNA/Cas9-injected substantia nigra.
Cultured monkey brain cells would allow us to monitor mitochondria dynamics. Thus, primary cultures were obtained from a fetal money brain cortex at embryonic day 90. Double immunostaining of the cultured monkey neuronal and glial cells at DIV14 did not show that PINK1 and the mitochondrial protein TOM20 were closely colocalized (Fig. 4A). The results are consistent with fractionation data (Fig. 1G) and suggest that the majority of endogenous PINK1 is at least not localized in mitochondria under physiological conditions. We then used electroporation to transfect primary cultures with PINK1-gRNA-RFP/Cas9 plasmids. After transfection for 14 days, we observed that many PINK1-gRNA-RFP/Cas9 transfected neurons (59.91% ± 12.11%, n = 145 cells) showed fragmented and short neurites, a degeneration phenomenon. In contrast, a small fraction of control transfected cells expressing control-gRNA-RFP/Cas9 (14.75 % ± 4.09%, n = 137 cells, P < 0.001 vs. PINK1 gRNA) were degenerated, perhaps because of electroporation, while the majority of control neurons developed long and intact neurites (Fig. 4B). Importantly, Western blot confirmed that deletion of the PINK1 gene by its gRNA and Cas9 reduced the expression of PINK1 and neuronal proteins NeuN and PSD95. However, no obvious alterations in mitochondrial proteins (VDAC1, TOM20, Complex-II, -III, -V) were seen when compared with the control transfected neurons (Fig. 4C).
Because neuronal degeneration and cell death can affect mitochondrial dynamics and because PINK1 deficiency did not affect the survival of astrocytes, we used astrocytes to examine the impact of PINK1 deficiency on mitochondria dynamics. Consistent with the in vivo finding that PINK1 targeting did not obviously affect glial cells in the monkey brain (Fig. 2C–F), GFAP staining showed the normal morphology and numbers of astrocytes after PINK1 targeting as compared with control astrocytes (Fig. 4D). We then used PINK1-targeted live astrocytes to measure mitochondrial dynamic changes in their size, lengths, and motility (Videos S2 and S3). This assay showed that mitochondrial dynamics in control and PINK1-targed astrocytes are similar (Fig. 4E–G).
Our continuous study in generation of PINK1 mutant monkeys yielded a prenatal PINK1 targeted monkey (M7) that was aborted at the gestation days 135. Western blot analysis of the M7 brain tissues confirmed the reduction of PINK1 and neuronal proteins (Fig. 5A). However, no alterations in mitochondrial proteins were seen in the prenatal M7 monkey brain and newborn PINK1 mutant monkey (M1–M4) brain tissues (Fig. 5B). Our early study also obtained a 3-year-old PINK1 mutant monkey (M6) (Yang et al., 2019b), which provided us with enough fresh brain tissues for isolation of mitochondria fraction. Western blot analysis of the mitochondrial fraction isolated from the M6 monkey brain cortex showed that mitochondrial proteins were expressed at the similar levels as an age-matched WT monkey (WT6) (Fig. 5C). Quantitation of the ratios of mitochondria related proteins (TOM20, VDAC1, CV, CI, CII, CIII) to the loading control verified no difference between PINK1 mutant and the age-matched control monkeys (Fig. 5D). We also performed Western blot to analyze the levels of proteins that participate in mitochondria fission/fusion (OPA1, NdufA10, and Mfn1) and found their levels were similar in wild type and PINK1 mutant monkey cortex (Fig. 5E). The primary cultures from the fetal monkey brain also showed no obvious changes in DJ-1, OPA1, and Mfn1 after PINK1 was knocked down by PINK1 gRNA/Cas9 (Fig. 5F). Electron microscopy was then used to compare the morphology of mitochondria in M6 and the age-matched monkey brain (WT6) (Fig. 5G). Counting the percentage of different types of mitochondria (Type I: normal appearing cristae; Type II: swollen, irregular or whirling cristae; Type III: discontinuous outer membrane or deficient cristae; Type IV: both discontinuous outer membrane and swollen cristae) showed no difference in the striatum and substantia nigra between M6 and WT6 monkeys (Fig. 5H).
If PINK1 mutations severely affect mitochondrial function, brain metabolomics may show alterations in the metabolites that reflect mitochondrial function. Since the 3-year-old monkey provided us with enough fresh brain cortical tissues, we used its tissues for metabolomics analysis but did not observe significant differences in most metabolites between M6 monkey and the age-matched wild type monkeys (Fig. S6A). This result was also supported by comparing the relative levels of metabolites such as creatine, phosphocreatine, ATP, high-energy phosphates (AMP, NADPH) and TCA cycle (citric acid), which are closely related to mitochondrial function, in the brain cortical tissues of M6 and WT monkeys (Fig. S6B).
PINK1 deficiency reduces protein phosphorylation in the monkey brain
Since PINK1 is a serine/threonine kinase, we next examined protein phosphorylation in PINK1-deficient monkey brain cortex using the same approach for cell cultures in our previous study (Wan et al., 2018). Our earlier study revealed that CRISPR/Cas9 mediated differing extents of PINK1 deletion and neuronal loss in the brain cortical tissues of newborn monkeys (M1–M4) and adult monkeys (M5 and M6), with the greatest reduction of PINK1 in M1 and M2, a modest reduction in M3 and M4, and the least reduction in M5 and M6 (Yang et al., 2019b). These brain tissues had more homogenous PINK1 reduction and less variability in neuronal loss than viral-injected brain tissues and therefore were used for protein phosphorylation assays. Western blot analysis also revealed that the reduction in protein phosphorylation was most dramatic in M1 and M2, less remarkable in M3 and M4, and less noticeable in M5 and M6 (Fig. 6A). Consistent with the reduced protein phosphorylation, there was also a decrease in the neuronal protein NeuN. In PINK1 KO mouse brains, however, there was no evident reduction of protein phosphorylation or neuronal proteins (synapsin-1, NeuN) (Fig. 6B).
We then performed mass spectrometry analysis of the proteomics and phosphoproteomics in cortical tissues available from PINK1 mutant (M1, M2, and M4) and three WT age-matched postnatal monkeys (WT1-3). While the amounts of total proteins between WT and PINK1 mutant cortical tissues do not appear to be different, there was a significant reduction in the amount of phosphorylated proteins (Fig. 6C and 6D). Importantly, there was decrease in the phosphorylation of many proteins that are important for neurogenesis and neuronal function in PINK1 mutant monkey brains (Fig. 6E–G). We analyzed a total of 46,685 peptides and 11,286 phosphorylation sites (Fig. 6H). The results demonstrated far more changes in phosphorylation reduction (91%) than up-regulation (9%) (Fig. 6I). Loss of PINK1 caused significant downregulation of phosphorylation of a number of proteins (83.15%) important for neurogenesis (Figs. 6J and S7). In contrast, fewer proteins (4.7%) related to gliogenesis showed a decrease in phosphorylation (Fig. 6J). Analysis of genes for immune response and endolysosomal sorting and trafficking only revealed that the phosphorylation of some endosomal function related proteins, which are also important for neuronal function, was reduced significantly in PINK1 mutant monkey brain (Fig. S8A–C). We also compared phosphorylation profiling of mitochondria function related genes as well as early PD associated genes but found that their differences between WT and PINK1 mutant monkeys are limited and not obvious when compared with those for neuronal function (Fig. S8D and S8E). All these data collectively show that loss of PINK1 reduces the phosphorylation of various proteins that are important for neurogenesis and neuronal function.
PINK1 mutations affect protein phosphorylation in the monkey brain and human cells
CRISPR/Cas9 targeting can cause various mutations, leading to different extents to which the targeted gene expression is reduced. We have shown that PINK1 is dramatically reduced in the brains of newborn M1 and M2 monkeys but is not significantly decreased in the newborn M3, M4, 1.5-year-old (M5), and 3-year-old (M6) monkey brains, perhaps because the in-frame or mosaic mutations rather than a large deletion did not significantly affect the level of PINK1 in M3–M6 monkeys (Yang et al., 2019b). Previous in vitro studies have shown that PINK1 phosphorylates Parkin at Ser65 and ubiquitin and mutations in PINK1 can affect its kinase activity (Lazarou et al., 2015; Ordureau et al., 2015; Gladkova et al., 2018; Wan et al., 2018; Han et al., 2020). Western blot revealed that loss of PINK1 in M1 and M2 could reduce the phosphorylation of Parkin-Ser 65 but PINK1 mutations in M3 and M4 did not (Fig. S9A). Similar to Parkin phosphorylation, ubiquitin phosphorylation was reduced in M1 and M2 monkey brains but did not show obvious reduction in M3 and M4 monkey brains (Fig. S9B), which was confirmed by the phosphate-affinity (Phos-tag) PAGE that can retard the migration of phosphorylated proteins to distinguish them from unphosphorylated proteins (Koyano et al., 2014) (Fig. S9C). It is clear that loss of PINK1 can induce a more dramatic and broad reduction of protein phosphorylation and affects cellular function to a greater extent than PINK1 mutations that do not decrease the level of PINK1. In support of this idea, Western blot analysis showed that caspase-3 cleavage, a process for apoptosis, was more pronounced in M1 and M2 than in other PINK1 mutant monkey brains (Fig. 7A).
Since the majority of PINK1 mutations in PD patients are point mutations that may not affect the level of PINK1 (Pickrell and Youle, 2015), it would be important to investigate the effect of PINK1 mutations in those monkey brains that do not show significantly reduced expression of PINK1. By looking at phosphoproteomics data, we found that the phosphorylation of BAD, a proapoptotic member of the Bcl-2 family whose phosphorylation is important for neuronal survival (Datta et al., 2002), was also reduced in the M4 mutant monkey brain (Fig. 7B). We used a specific antibody to the phosphorylated BAD for Western blot and found that BAD phosphorylation at S112 and S136 was decreased in the cortical tissues of M3 and M4 monkey brains in which the PINK1 level was not significantly reduced (Fig. 7C). However, in the mouse brain, no obvious alteration of BAD phosphorylation was observed in the Pink1 KO brain (Fig. S9D), again suggesting that endogenous Pink1 in the mouse brain is expressed at a very low level. In PINK1 mutant monkey brains (M5 and M6), Western blot with available antibodies revealed selective phosphorylation reduction of CRMP2 and STXBP1, but not AKT (Fig. 7D), which confirmed mass spectrometry results and supported the idea that loss or mutations of PINK1 by CRISPR/Cas9 targeting can impair its kinase activity to reduce phosphorylation of some neuronal proteins.
If PINK1 mutations affect protein phosphorylation, we may observe this defect in human cells expressing mutant PINK1. To test this idea, we examined fibroblast cells from patients with PINK1 mutations (R492X and T313M) (Han et al., 2020). Because fibroblast cells express endogenous PINK1 at a very low level, we induced PINK1 expression by treating them with 10 µmol/L CCCP for 12 h and then analyzed the phosphorylation of BAD and Drp1, two substrates for PINK1 phosphorylation (Wan et al., 2018; Han et al., 2020). As R492X mutation leads to a truncated PINK1, R492X cells did not produce full-length PINK1 and showed a reduction in BAD and Drp1 phosphorylation without affecting the expression of mitochondrial proteins (CV, CIII, and TOM20) when compared with wild-type (WT) cells (Fig. 7E, left panel). Also, PINK1 mutation (T313M) reduced the phosphorylation of BAD and Drp1, though the reduction of Drp1 phosphorylation was not great as that in R492X cells lacking the expression of full-length PINK1 (Fig. 7E and 7F). Based on these results, we propose that PINK1 (PINK1-55) is selectively expressed in the human and monkey brains to phosphorylate a large number of proteins. Loss of PINK1 can broadly affect phosphorylation of various proteins while mutations may more selectively affect phosphorylation of some proteins. Because of this unique and critical function of PINK1 in the primate brains, depletion of PINK1 in the monkey brains can lead to neuronal loss at different ages. Mutations in PINK1 found in PD patients may also partially affect its kinase activity, and such kinase dysfunction can contribute to the late-onset neurodegeneration in Parkinson’s disease (Fig. 7G).