Toward mechanism-based diagnostics and disease interventions

Declan Williams, Mohadeseh Mehrabian, Xinzhu Wang, Gerold Schmitt-Ulms; University of Toronto

(See a list of the products featured in this article.)

The development of models and methods for studying proteins that cause neurodegenerative diseases is the focus of our research at the University of Toronto (Figure 1), with the goal of generating insights that will lead to novel angles for diagnosis or intervention. Within this general theme, we specialize in the study of tauopathies [1], which include Alzheimer’s disease and a subset of frontotemporal dementias (FTDs). We are primarily interested in finding the missing links in aberrant signaling pathways triggered by the formation of oligomeric amyloid beta peptide (oAβ). Binding of oAβ to the cellular prion protein (PrPC) contributes to the detachment of the Tau protein from microtubules and causes proteotoxic stress through a poorly defined chain of events (Figure 2A).

Figure 1. Localization of wild-type and mutant Tau fusion proteins. Co-expression of wild-type and P301L mutant Tau fused to EGFP (green) and ECFP (pseudocolored red), respectively, in African green monkey CV-1 kidney cells. In addition to the profound overlap of both fusion proteins in their localization to the microtubule network (observed in yellow merged color), note the presence of punctate signals only in the red channel, depicting the cellular distribution of P301L mutant Tau. See Figure 4 for more information.

Combining CRISPR-Cas9 model building with mass spectrometry

The use of the CRISPR-Cas9 system [2] has been nothing short of transformative for our work because it allows us to generate relevant models with reasonable effort. There are two CRISPR-Cas9 applications we find particularly useful, namely the generation of knockout models of specific genes of interest, and the introduction of mutations known to underlie human diseases. Once a suitable model has been generated, we interrogate the consequences of these genetic changes on cellular biology through side-by-side comparative analyses with wild-type control models, using quantitative mass spectrometry (Figure 2B).

For such investigations to be meaningful, sample handling and analysis should not introduce inadvertent heterogeneity. One approach we have found useful for minimizing run-to-run variances in protein-directed research projects involving mass spectrometry is to label peptides with isobaric tags in order to facilitate sample multiplexing [3,4]. Following their reversed-phase separation, the mixtures of these isobarically tagged peptides are directed to the orifice of the Thermo Scientific™ Orbitrap Fusion™ Tribrid™ mass spectrometer by electrospray ionization. Next, the mass-to-charge ratios of incoming ions are recorded by an Orbitrap analyzer–based parent scan of exquisite mass resolution and accuracy. The machine then selects, in an automated manner, the most intense ions for mild collisions with an inert gas in order to obtain fragment ions, which later serve as a fingerprint for protein identification. Finally, the 10 most intense of these fragment ions are concomitantly smashed into even smaller pieces to release their isobaric labels. The relative signal intensities of these mass tags, which are specific for each sample, allow us to deduce the relative abundance of a given peptide in each of the multiplexed samples. Here we describe two projects that illustrate the usefulness of combining CRISPR-Cas9 model building with downstream mass spectrometry to address fundamental biomedical research questions.

Figure 2. Schematic of central research theme and workflow for combining CRISPR-Cas9 genome-edited models with mass spectrometry. (A) The Schmitt-Ulms laboratory studies the molecular etiology of tauopathies and prion diseases. Research in the laboratory focuses on signaling downstream of the amyloid beta peptide (Aβ) and the role of the prion protein (PrPC) in these signals, as well as events that lead to cellular toxicity. (B) More recently, the generation of cell models using CRISPR-Cas9 technology has played a major role in our discovery pipeline. A typical analysis compares the effect of a given genome manipulation on the global proteome, as well as on the molecular interactions or posttranslational modifications of a protein of interest.

CRISPR-Cas9–generated knockouts of the gene encoding the cellular prion protein

The first project combined specific gene knockouts with global proteome analyses (Figure 3) and was pursued as part of a broader program aimed at devising mechanism-based strategies to overcome prion diseases, including Creutzfeldt-Jakob disease (CJD) and bovine spongiform encephalopathy (BSE). Despite its discovery more than 30 years ago, the normal function of the cellular prion protein (PrPC), which is known to cause these diseases when it acquires a different shape, remains unknown, leaving uncertain to what extent a perturbation of its normal function contributes to cellular death in the disease.

Using CRISPR-Cas9 technology, we generated knockout cells that no longer can express PrPC [5] (Figure 3A) and compared them with wild-type parental cells using global proteome analyses (Figure 3B). These analyses revealed that the absence of PrPC strongly decreased cellular levels of the neural cell adhesion molecule 1 (NCAM1) [6] (Figure 3C). Western blot–based analyses then led to the surprising discovery that, in addition to profoundly affecting NCAM1 protein levels, the lack of PrPC had abrogated NCAM1 polysialylation (Figure 3D). The polysialylation of NCAM1 is a critical posttranslational modification in the brain that controls specific protein interactions, influences chemotactic guidance, and modulates ion channels, and NCAM1 is the predominant acceptor of this modification in vertebrates. We then became aware of a body of literature documenting that impaired polysialylation of NCAM1 perturbs (i) sleep-wake cycles, (ii) neurogenesis, (iii) neurite outgrowth of specific mossy fiber axon bundles in the hippocampus, and (iv) myelination [7]. These phenotypes are highly reminiscent of independently reported phenotypes observed in mice deficient for the prion protein [8,9], consistent with the interpretation that the contribution of PrPC to NCAM1 polysialylation might be its predominant role [10].

Figure 3. Comparative global proteome analyses of wild-type and PrPC knockout cells identify the role of PrPC in the polysialylation of NCAM1. (A) Generation of PrPC knockout cells by CRISPR-Cas9 technology. Reprinted with permission from Mehrabian M, Brethour D, MacIsaac S et al. (2014) CRISPR-Cas9–based knockout of the prion protein and its effect on the proteome. PLoS One 9:e114594. (B) Design of global proteome analysis with the aim to identify proteins whose levels change upon addition of TGFB1 (a method for inducing epithelial-to-mesenchymal transition) to NMuMG cells (Dataset I) and filter from this list the subset of proteins whose levels are impacted by the presence or absence of PrPC (Dataset II). (C) Box plot depicting relative levels of NCAM1-derived peptides in wild-type and PrPC knockout cells (extracted from Dataset II, see (B)). Note the reduction in mean NCAM1 peptide levels in PrPC knockout cells relative to wild-type levels observed in three biological replicates. (D) PrPC deficiency abrogates NCAM1 polysialylation, identifiable in western blot analyses by the pronounced streaking. In the absence of this posttranslational modification, NCAM1 signals correspond to the relative levels of the three major isoforms of this protein. Panels B–D were reprinted with permission from Mehrabian M, Brethour D, Wang H et al. (2015) The prion protein controls polysialylation of neural cell adhesion molecule 1 during cellular morphogenesis. PLoS One 10:e0133741.

CRISPR-Cas9–generated neuroblastoma cells with inducible Tau expression

The second project highlights a useful application of CRISPR-Cas9 technology for the generation of human cell models that inducibly express a protein of interest fused to a fluorescent affinity-capture tag (Figure 4), allowing the production of in-depth interactome datasets in less than a month. In this project, we were interested in dissecting molecular events that may cause cellular death in a small subset of FTDs caused by specific inherited mutations in the gene encoding the microtubule-associated protein Tau (MAPT). Neural Tau transcripts are subject to alternative splicing events that generate up to six prominent Tau isoforms, which can be further classified as having either three or four repeats (3R or 4R) in the microtubule-binding domains. In the brain, a balanced amount of 3R and 4R Tau is critical for cellular health. Available human cell models exhibit unbalanced isoform ratios, and it has repeatedly been shown that the plasmid-encoded expression of 4R Tau can cause cellular toxicity by itself [11].

To overcome this confounder to cell-based Tau studies, we employed a two-step genome engineering approach to generate human neuroblastoma cell models that express equal levels of 3R and 4R wild-type or mutant Tau [12]. In the first step, we used the double CRISPR-Cas9 nickase technology to introduce a G418 resistance marker flanked by lox sites into the AAVS1 genomic safe harbor, a site known to tolerate insertions without adverse effects on the cell (Figure 4A).The coding sequences for 3R and 4R Tau, packaged in an expression cassette flanked by compatible lox sites, were then switched into the genome via cotransfection of Cre recombinase. To allow the inducible expression of Tau and facilitate its cellular tracking and capture, the Tau coding sequence was placed between a tetracycline response element (TRE) promoter and a C-terminal Enhanced Green Fluorescent Protein (EGFP). Finally, we included in the plasmid expression cassettes a reverse tetracycline transactivator (rtTa) and a puromycin selection marker.

As intended, the cells expressed equal levels of 3R and 4R wild-type or mutant Tau upon induction with doxycycline (Figure 4B). Consistent with expectations, the presence of the mutation caused Tau not only to bind microtubules but also to appear in punctate aggresome-like structures (Figure 4C). The presence of the C-terminal EGFP tag has been shown to have no adverse effect on Tau biology and facilitated the capture of Tau-EGFP fusion proteins on GFP-binding protein (GBP) matrices. Our analyses of these cells are ongoing but have already revealed several interesting insights, including differential binding of wild-type and mutant Tau to certain chaperones and the proteasome [13], consistent with the notion that FTD may be caused by impaired recycling of the Tau protein. The ability to induce Tau and follow it over time will allow us to dissect the chronology of events underlying Tau’s impaired recycling, identify the cellular pathways that are poisoned in cells when Tau aggregates are forming, and uncover abnormal changes to molecular interactions and posttranslational modifications of Tau that facilitate the formation of aggregates.

Figure 4. Comparative interactome analyses reveal compromised binding of mutant Tau to the proteasome and a subset of chaperones. (A) CRISPR-Cas9– based genome engineering approach for the generation of human cell models that can be rapidly manipulated to promote the inducible expression of proteins of interest. (B) Validation of positive clone coding for the inducible expression of Tau-EGFP. (C) Co-expression of wild-type and P301L mutant Tau fused to EGFP (green) and ECFP (pseudocolored red), respectively, in African green monkey CV-1 kidney cells (a model with favorable characteristics for visualizing the microbutule network). In addition to the profound overlap of both fusion proteins in their localization to the microtubule network (observed in yellow merged color), note the presence of punctate signals only in the red channel, depicting the cellular distribution of P301L mutant Tau. Scale bar = 1 μm.

Future directions

Although genome editing is not new, the relative ease with which animal and cell models can be generated with the CRISPR-Cas9 technology has already had a profound impact on biomedical research. The system still suffers from efficacy limitations in regards to achieving a specific genome edit and in its delivery to cells in complex tissues. On the proteome analysis side, we are still limited in the proteome coverage that can be achieved even with the most high-end equipment. The rapid pace of innovation we have witnessed in CRISPR-Cas9 technology and mass spectrometry is not expected to abate any time soon. The rewards that the combination of both approaches promises are only beginning to be realized. Good times for biomedical research are waiting ahead.

Acknowledgments: This article was contributed by Declan Williams, Mohadeseh Mehrabian, Xinzhu Wang, and Gerold Schmitt-Ulms; Tanz Centre for Research in Neurodegenerative Diseases, University of Toronto. The latter three authors are also associated with the Department of Laboratory Medicine & Pathobiology, University of Toronto. Gerold Schmitt- Ulms is the corresponding author; please address correspondence to:


  1. Spillantini MG, Goedert M (2013) Tau pathology and neurodegeneration. Lancet Neurol 12:609–622.
  2. Mali P, Esvelt KM, Church GM (2013) Cas9 as a versatile tool for engineering biology. Nat Methods 10:957–963.
  3. Dayon L, Sanchez JC (2012) Relative protein quantification by MS/MS using the tandem mass tag technology. Methods Mol Biol 893:115–127.
  4. Zieske LR (2006) A perspective on the use of iTRAQ reagent technology for protein complex and profiling studies. J Exp Bot 57:1501–1508.
  5. Mehrabian M, Brethour D, MacIsaac S et al. (2014) CRISPR-Cas9-based knockout of the prion protein and its effect on the proteome. PLoS One 9:e114594.
  6. Mehrabian M, Brethour D, Wang H et al. (2015) The prion protein controls polysialylation of neural cell adhesion molecule 1 during cellular morphogenesis. PLoS One 10:e0133741.
  7. Rutishauser U (2008) Polysialic acid in the plasticity of the developing and adult vertebrate nervous system. Nat Rev Neurosci 9:26–35.
  8. Aguzzi A, Baumann F, Bremer J (2008) The prion's elusive reason for being. Annu Rev Neurosci 31:439–477.
  9. Steele AD, Lindquist S, Aguzzi A (2007) The prion protein knockout mouse: a phenotype under challenge. Prion 1:83–93.
  10. Mehrabian M, Hildebrandt H, Schmitt-Ulms G (2016) ASN Neuro. In press.
  11. Myeku N, Clelland CL, Emrani S et al. (2016) Tau-driven 26S proteasome impairment and cognitive dysfunction can be prevented early in disease by activating cAMP-PKA signaling. Nat Med 22:46–53.
  12. Wang X, Williams D, Wang H et al. (2016) Manuscript in preparation.
  13. Gunawardana CG, Mehrabian M, Wang X et al. (2015) The human tau interactome: Binding to the ribonucleoproteome, and impaired binding of the proline-to-leucine mutant at position 301 (P301L) to chaperones and the proteasome. Mol Cell Proteomics 14:3000–3014.

Learn more about:

Article download

Download a hyperlink-enabled, printer-friendly version of this article.
Download now