Progressive supranuclear palsy, a rare tauopathy, slowly destroys a person's cognition, balance, and movement. PSP ranks second behind Parkinson’s disease as a cause of Parkinsonism. Beyond variants in the tau gene, a handful of other risk genes have turned up in about 10 genome-wide association studies. Now, scientists led by Wan-Ping Lee and Gerard Schellenberg of the University of Pennsylvania in Philadelphia; Daniel Geschwind at the University of California, Los Angeles; Günter Höglinger of the Ludwig-Maximilians University Hospital in Munich; and Dennis Dickson at Mayo Clinic, Jacksonville, Florida, report the first whole-genome sequencing of PSP in a preprint uploaded to medRxiv on January 30.

  • Scientists sequenced the genomes of 1,700 people with PSP.
  • They found six known and six new risk loci.
  • APOE2 seemed to increase risk.

They uncovered new structural and rare variants in six genes previously linked to PSP plus 12 novel loci, including APOE. Curiously, APOE2 seemed to up a person’s odds of PSP, in contrast to its protective role in Alzheimer’s disease.

James Rowe of the University of Cambridge, U.K., thinks this study is important. “These discoveries can now be mapped to pathogenic pathways to guide much-needed therapeutic strategies,” he wrote to Alzforum (comment below).

While previous GWAS uncovered common risk variants, GWAS generally fall short of finding rare mutations, including small insertions and deletions (indels), and larger structural variants (Jun 2011 news; Nov 2023 news). To get a sense of how such variants influence PSP, co-first authors Hui Wang of UPenn and Timothy Chang at UCLA analyzed whole-genome sequencing data of 1,718 people with PSP and 2,944 controls from the National Institute on Aging’s Alzheimer’s Disease Sequencing Project, a collection of genetic data from 40 cohorts worldwide. On average, people with PSP were in their mid-60s, while controls were in their 80s. All participants were of European ancestry and half were women. PSP was autopsy-confirmed in 1,441 cases and clinically diagnosed in the rest.

The WGS turned up six loci previously linked to PSP, including the tau gene. MAPT harbored common single-nucleotide variants (SNVs), indels, and a 238-bp deletion within intron 9. MAPT was the strongest risk gene for PSP in this and previous studies (image below). Common SNVs and indels in the myelin-associated protein gene MOBP and in STX6, which encodes a SNARE protein involved in intracellular trafficking, also reached genome-wide significance, i.e., a p-value of less than 5 x 10-8.

Falling short with p-values below 1 x 10-6, which the authors deemed "suggestive of significance,” were the phosphatase gene DUSP10, the transcription factor gene SP1, and the ion transporter gene SLCO1A2.

Genetic Skyline. This Manhattan plot shows common single-nucleotide variants and indels in six known loci (black) and six new ones (red). Only three known genes—MAPT, MOBP, and STX6—and the new loci APOE reached genome-wide significance (red horizontal line). The rest were suggestive (blue horizontal line). [Courtesy of Wang et al., medRxiv, 2024.]

Wang and Chang also found common SNVs and indels in six novel loci: APOE, the lipid elongase gene ELVOL1, the immunity gene FCHO1, the microtubule motor protein gene KIF13A, the extracellular matrix glycoprotein gene TNXB, and the transcription factor gene TRIM24 (image above). Six structural variants not previously linked to PSP occurred in five other loci. The most notable was a 619-bp deletion within PCMT1, which encodes a protein carboxyl methyltransferase highly expressed in the brain. Having the deletion in one copy of PCMT1 upped a person’s odds of having PSP by 4.2-fold; two copies increased the risk by 8.4 times.

The other four loci comprised the sperm protein gene SMCP, the histone acetyltransferase complex gene KANSL1, the immunoglobulin heavy gene, and a 1.5-kb stretch between the cytochrome P450 genes CYP2F1 and CYP2A13. KANSL1 was previously linked to tauopathies (Nov 2018 conference news; Mar 2022 conference news). As for rare variants, missense or splice mutations turned up in the zinc finger protein gene ZNF592. It has not been associated with PSP but controls the transcription of genes within the cerebellum, an area of the brain that regulates movement and balance (NIH gene entry).

ELVOL1 produces neurotoxic lipids in astrocytes, the cells that crank out most of the brain's APOE (Oct 2021 news). This led Lance Johnson and Lesley Golden of the University of Kentucky, Lexington, to wonder if the two PSP risk alleles conspire. “Does this suggest a common astrocyte lipoprotein-based mechanism of neurodegeneration in PSP?” they asked (comment below).

Most striking was the direction of APOE’s relationship to PSP: APOE2 seemed to make it go up. People with PSP were four times as likely to carry two copies of APOE2 than controls. Guojun Bu of Hong Kong University of Science and Technology, China, found the same in a cohort of 2,500 people with PSP. APOE2 carriers with PSP also had more tau pathology postmortem than did PSP cases with other APOE alleles. Bu thinks two cysteine residues within ApoE2 form disulfide bonds with two cysteines in tau (Zhao et al., 2018). This might shift the shape of tau’s fibrils, possibly helping explain the distinct cryo-EM structures of tau seen in PSP versus AD (Oct 2021 news).

While APOE4 seemed protective against PSP—again the opposite of AD—this did not hold when Lee excluded from his analysis cohorts with a higher prevalence of APOE4 carriers than in the general population. Because some of the cohorts' control groups still had unusual APOE allele frequencies after excluding outliers, Yadong Huang, University of California, San Francisco, thinks the direction of the APOE risk data needs to be confirmed in large cohorts.

Indeed, David Holtzman, Washington University, St. Louis, suspects the APOE associations might reflect selection bias. Because APOE4 increases the risk of amyloid deposition starting in mid-life and APOE2 reduces it, many people carrying APOE4 will go on to develop AD pathology, while APOE2 carriers will not. “I wonder then, if the pool of individuals who might get PSP is relatively enriched for APOE2,” he wrote (comment below).—Chelsea Weidman Burke

Comments

  1. This important paper represents a strong international cross-cohort collaboration to identify genetic associations of autopsy-confirmed PSP. There are some old culprits (MAPT, STX6, SLCO1A2, etc.), and some new suspects (TRIM24, TNXB, etc.), reflecting a diversity of types of genetic variation, e.g., single-nucleotide variants, small insertions and deletions, or larger structural variants. The new study complements Farrell et al. Together, these studies’ discoveries can now be mapped to pathogenic pathways, to guide much-needed therapeutic strategies. The novel APOE2 deleterious association (and APOE4 beneficial association) is interesting. It is not clear yet whether this is biologically meaningful, or a residual selection bias. However, the authors took steps to mitigate the risk of bias, and such antagonistic pleiotropy occurs for other diseases. Replication will be key, along with data on co-pathologies in the autopsies, in future collaborative studies.

    The novel APOE2 deleterious association (and APOE4 beneficial association) is interesting. It is not clear yet whether this is biologically meaningful, or a residual selection bias. However, the authors took steps to mitigate the risk of bias, and such antagonistic pleiotropy occurs for other diseases. Replication will be key, along with data on co-pathologies in the autopsies, in future collaborative studies.

  2. For those of us studying APOE in the background of AD, this new preprint from Wang and Chang et al. certainly caught our attention. Here we have a situation where the “good guy” in AD (the APOE2 allele) is suddenly the “bad guy" in PSP. Although few and far between, studies that include E2 carriers or E2 model systems generally suggest that it decreases AD-associated pathologies in a stepwise fashion with E2 > E3 > E4.

    However, as is the case in this new PSP GWAS, E2 does not follow this trend for all disorders or pathologies. In fact, E2’s relationship with tau appears to be one of these scenarios where E2 may not be “better” than E3 or E4. Zhao et al. showed that E2 worsened tau pathologies when they administered AAV-Tau with the P301L mutation into the brains of E2-, E3-, and E4-targeted replacement mice (Zhao et al., 2018). Specifically, TauP301L-E2 mice had increased AT8 and Thioflavin-S tau aggregates and increased insoluble fractions of tau.

    Additionally, Zhao et al. first discovered that APOE2 increases risk for PSP, as they reported that E2 carriers had significantly higher tau pathology scores than E3/E3 individuals or E4 carriers (across multiple tau lesion subtypes). Interestingly, they note that these findings appear to have been driven by the E2/E3 genotype, which was associated with a greater severity of tau pathology than what was noted for the E2/E2 genotype.

    Seemingly at odds with these findings, there have been several papers that suggest that E4 was associated with a higher tau burden than E2 or E3 (Wang et al., 2021; Shi et al., 2017; Young et al., 2023; Koutsodendris et al., 2023; Nelson et al., 2023). The current paper by Wang et al. further fuels the uncertainty about the ApoE-tau association by identifying E2 as a risk-associated gene for PSP, a 4R tauopathy.

    How could this be? The authors mention differences in tau species between PSP and AD, and perhaps the ApoE isoforms variably interact with different tau species in a manner that somehow explains the “reverse risk effects” we see in PSP versus AD. Additionally, brain region-specific mechanisms and/or the presence of additional pathologies could also attribute to E2’s increased risk in PSP, but not AD. Young et al. showed that E2 carriers had less tau in all brain regions compared to E4 carriers and suggests an amyloid-dependent component to tau accumulation in AD—in other words, perhaps amyloid changes the way ApoE interacts with or influences tau.

    Lastly, ApoE is secreted by several different cell types within the CNS (with different post-translational modifications) which could indicate a cell-type specific ApoE-tau interaction. This is particularly interesting when one thinks about the different cell types displaying tau pathology in PSP vs AD.

    For instance, ELOVL1, another newly identified PSP GWAS hit, encodes an elongase that lengthens saturated fatty acids—fatty acids shown to be secreted by reactive astrocytes and toxic to neurons when transported by ApoE-containing particles (Guttenplan et al., 2021). Does this suggest a common astrocyte lipoprotein-based mechanism of neurodegeneration in PSP?

    Several recent publications (Chemparathy et al., 2024; Koutsodendris et al., 2023; Wang et al., 2021), including the current study, provide a strong foundation for depleting ApoE4 from the brains of E4 carriers, rather than an approach in which the brain is supplemented (or ApoE4 is replaced) with ApoE2. Acknowledging the flashing PSP warning light, the latter approach still has merit in our opinion, considering the general need for multiple therapeutic options and that it could also be an approach for the 40 percent or so of individuals with AD who are non-E4 carriers.

    Along these lines, the authors urge caution when considering germ-line gene editing of APOE4 to APOE2 as a therapeutic modality for AD. These concerns are valid for the reasons we note above and the fact that while E2 is protective against AD, it does confer increased risk for other disorders such as age-related macular degeneration, PTSD, and in the case of some E2/E2 individuals, Type 3 hyperlipoproteinemia (although the strength of APOE’s association with these disorders varies and generally does not approach the level of impact as AD risk).

    However, it is this same concern that drives many of the questions our group and others are actively working on. For example, if one were to replace APOE4 with APOE2 (or another protective isoform such as the Christchurch or Jacksonville variants), how can one maximize benefits while minimizing side effects? What cell type would best do this? At what time point in the disease course could/should we intervene?

    These questions, and others, will all be critical in determining which approaches, if any, are the optimal APOE-directed therapeutic strategies.

    References:

    . APOE ε2 is associated with increased tau pathology in primary tauopathy. Nat Commun. 2018 Oct 22;9(1):4388. PubMed.

    . Selective removal of astrocytic APOE4 strongly protects against tau-mediated neurodegeneration and decreases synaptic phagocytosis by microglia. Neuron. 2021 May 19;109(10):1657-1674.e7. Epub 2021 Apr 7 PubMed.

    . ApoE4 markedly exacerbates tau-mediated neurodegeneration in a mouse model of tauopathy. Nature. 2017 Sep 28;549(7673):523-527. Epub 2017 Sep 20 PubMed.

    . APOE effects on regional tau in preclinical Alzheimer's disease. Mol Neurodegener. 2023 Jan 4;18(1):1. PubMed.

    . Neuronal APOE4 removal protects against tau-mediated gliosis, neurodegeneration and myelin deficits. Nat Aging. 2023 Mar;3(3):275-296. Epub 2023 Feb 20 PubMed.

    . The APOE-R136S mutation protects against APOE4-driven Tau pathology, neurodegeneration and neuroinflammation. Nat Neurosci. 2023 Dec;26(12):2104-2121. Epub 2023 Nov 13 PubMed.

    . Neurotoxic reactive astrocytes induce cell death via saturated lipids. Nature. 2021 Nov;599(7883):102-107. Epub 2021 Oct 6 PubMed.

    . APOE loss-of-function variants: Compatible with longevity and associated with resistance to Alzheimer's disease pathology. Neuron. 2024 Jan 29; PubMed.

    . Selective removal of astrocytic APOE4 strongly protects against tau-mediated neurodegeneration and decreases synaptic phagocytosis by microglia. Neuron. 2021 May 19;109(10):1657-1674.e7. Epub 2021 Apr 7 PubMed.

  3. This GWA study shows that APOE2 is associated with higher risk and E4 with lower risk for PSP, a 4R tauopathy. That said, I don’t think we can say from this genetic analysis that it means that with PSP, E2 increases and E4 decreases tauopathy.

    Further biological studies would be needed to understand why this genetic association is seen. The authors don’t discuss this, but I wonder if these genetic findings may be due to how APOE4 influences risk for AD. APOE4 strongly increases risk for amyloid deposition starting in midlife and APOE2 strongly decreases risk for amyloid deposition. Thus, by the ages of 50 to 60, many individuals with E4 will already have amyloid and then, over the following years, will develop AD tauopathy. The opposite is true for APOE2, i.e., many fewer will get amyloid deposition. I wonder then if the pool of individuals who might get PSP is relatively enriched for E2 over E4? I don’t know if this is a possibility, but just a thought.

    In an animal model of 4R tauopathy, i.e., a model of tau-related FTD (PS19 mice), E4 predisposes to worse tauopathy and neurodegeneration, while E2 is linked with less tauopathy and neurodegeneration.

  4. Wang et al. performed whole-genome sequencing in more than 1,700 PSP patients, most of them confirmed by neuropathology. The authors confirm results from previous GWAS analyses and found other new relevant loci related to PSP risk. Among the multiple results presented, a surprising one is that the authors found an increased risk of developing PSP in APOE2 carriers. The APOE2 allele is the least frequent allele in the general population but it is known to be protective for Alzheimer’s disease. In particular, APOE2/3 individuals have a lower risk (0.4-0.6 OR) of developing Alzheimer’s pathology compared to APOE3 homozygotes; and APOE2/2 even less (0.1-0.5 OR, Reiman et al., 2020). 

    On the other hand, the APOE4 allele is the most common genetic risk factor for Alzheimer’s disease, but seems to be protective for PSP, based on this manuscript. These results, if confirmed in another sample, might have important implications for those AD treatments that attempt to change APOE e4 to e2.

    This, together with another recent paper suggesting that the detrimental effect of the APOE-e4 allele is its gain of abnormal function (Chemparathy et al., 2024), may suggest that the optimal measure would be lockdown of the APOE4 allele instead of its conversion to an APOE2 allele. This is also supported by the risk associated with APOE2 for other diseases, such as cerebral amyloid angiopathy and hyperlipoproteinemia (Li et al., 2020).

    Although these findings seem to be consistent with previous results (Zhao et al., 2018), further research is needed to understand the biological underpinnings of these apparent contradictory results of these APOE genotypes and two neurodegenerative diseases related to tau pathology. Given the lower frequency of PSP than AD, worldwide efforts should be put to confirm these results in a bigger sample with comparable controls and other neurodegenerative disease patients.

    References:

    . Exceptionally low likelihood of Alzheimer's dementia in APOE2 homozygotes from a 5,000-person neuropathological study. Nat Commun. 2020 Feb 3;11(1):667. PubMed.

    . APOE loss-of-function variants: Compatible with longevity and associated with resistance to Alzheimer's disease pathology. Neuron. 2024 Jan 29; PubMed.

    . APOE2: protective mechanism and therapeutic implications for Alzheimer's disease. Mol Neurodegener. 2020 Nov 4;15(1):63. PubMed.

    . APOE ε2 is associated with increased tau pathology in primary tauopathy. Nat Commun. 2018 Oct 22;9(1):4388. PubMed.

  5. In this comprehensive and detailed study, Wang et al. completed one of the most thorough analyses to reveal novel genetic underpinnings of PSP. They used whole-genome sequencing to analyze single-nucleotide variants, indels, and structural variants in nearly 1,800 PSP cases and approximately 3,000 controls.

    An important finding was the corroboration of results from by Sawa et al. and Zhao et al. that ApoE2 significantly associated with increased risk for PSP in a Japanese cohort. Given the vast number of cases, these data from Wang et al. highlight a more robust association of ApoE2 with PSP, a higher frequency of homozygosity in this tauopathy, and a much broader population reach expanding from Japanese to a large cohort of cases with European ancestry. Another unexpected finding was that ApoE4 had a protective effect. This intriguing result underscores the importance of the APOE gene in neurodegeneration.

    Newly identified loci with risk for PSP highlight two microtubule-associated proteins: MAP1S and KIF3A. Together with MAPT, these findings further implicate errors in microtubule dynamics as pathogenic events in PSP. However, how these pathological mechanisms contrast with those for Azheimer’s disease, especially in the context of ApoE isoform risk, remains puzzling. Further studies linking these findings will reveal important molecular mechanisms of PSP, distinguishing them from other tauopathies, and focus on new targets.

    References:

    . Apolipoprotein E in progressive supranuclear palsy in Japan. Mol Psychiatry. 1997 Jul;2(4):341-2. PubMed.

    . APOE ε2 is associated with increased tau pathology in primary tauopathy. Nat Commun. 2018 Oct 22;9(1):4388. PubMed.

Make a Comment

To make a comment you must login or register.

References

Mutations Citations

  1. APOE R176C (ApoE2)

News Citations

  1. GWAS Fingers Tau and Other Genes for Parkinsonian Tauopathy
  2. In Progressive Supranuclear Palsy, Risk Loci Converge on Oligodendrocytes
  3. International Symposium Puts PSP/CBD on the Map
  4. Tau Haplotypes Hint at Transcriptional Changes, Ferroptosis
  5. ELOVL Hurts—Enzyme Makes Lipids That Turn Astrocytes Toxic
  6. Flock of New Folds Fills in Tauopathy Family Tree

Paper Citations

  1. . APOE ε2 is associated with increased tau pathology in primary tauopathy. Nat Commun. 2018 Oct 22;9(1):4388. PubMed.

External Citations

  1. Alzheimer’s Disease Sequencing Project
  2. NIH gene entry

Further Reading

Primary Papers

  1. . Whole-Genome Sequencing Analysis Reveals New Susceptibility Loci and Structural Variants Associated with Progressive Supranuclear Palsy. 2024 Jan 30 10.1101/2023.12.28.23300612 (version 2) medRxiv.