Proteomics Lays Groundwork for FTD Biomarkers

Scientists are urgently seeking fluid biomarkers for frontotemporal dementia, the most common dementing illness among 40- and 50-year-olds. Two large collaborations studying genetic FTD cohorts have now cracked open the door. In one, scientists led by Jonathan Rohrer at University College London and Johan Gobom and Henrik Zetterberg at the University of Gothenburg, Sweden, measured proteins in cerebrospinal fluid samples from 162 people with familial FTD mutations who participated in the Genetic Frontotemporal Initiative (GENFI). As reported in the February 5 Science Translational Medicine, for each of the three most common forms of familial FTD, they identified specific proteins that were more or less abundant than in controls, many of them changing before symptoms began. First author Aitana Sogorb-Esteve at UCL said the distinct proteomic signatures of genetic forms of FTD imply that different biological processes can underlie FTD pathology.

  • GENFI CSF analysis hints at prognostic and diagnostic markers for specific mutations.
  • Other proteins may distinguish FTD from AD. Follistatin-4, anyone?
  • ALLFTD CSF analysis emphasizes change in synaptic and lysosomal proteins.
  • RNA-processing proteins may be promising markers for TDP-43 pathology.

Charlotte Teunissen at Amsterdam University Medical Center called the study impressive. “There are not so many proteomics studies published yet for FTD, and having a benchmark such as this is therefore very valuable,” she wrote to Alzforum. Agustín Ibañez at the Global Brain Health Institute agreed. “This is a very impactful paper… [It] advances the field toward precision biomarkers,” he wrote (comments below).

In the other paper, scientists led by Rowan Saloner, Julio Rojas, Adam Boxer, and Kaitlin Casaletto at the University of California, San Francisco, presented data from 116 mutation carriers in the North American ALLFTD genetic cohort. In a preprint posted to Research Square, they reported that the biggest differences from controls were in CSF proteins related to RNA processing, synapses, and lysosomal function. Synaptic and extracellular matrix proteins stood out for FTD with tau accumulation, while RNA-processing proteins were prominent in FTD with TDP-43 pathology. TDP-43 binds RNA. Rojas noted that both cohort studies had overall similar findings, largely replicating each other’s work.

GENFI Top Hits. CSF proteomics linked abundance of CALB2, GRN/NAGA, and PEA15/SEMA6A to C9ORF72 (gold), GRN (red), and MAPT (blue) forms of FTD, respectively. Solid circles indicate statistical significance. [Courtesy of Sogorb-Esteve et al., Science Translational Medicine/AAAS.]

GENFI had previously analyzed small cohorts, turning up several candidates (Remnestål et al., 2020; Bergström et al., 2021; Sogorb-Esteve et al., 2022). For the new study, they went bigger, not only including more participants, but also switching from an antibody-based method to high-resolution mass spectrometry, allowing for identification of lower-abundance proteins.

Sogorb-Esteve and colleagues measured 1,981 proteins in CSF from 55 people with symptomatic FTD, 107 with presymptomatic, and 76 controls. This is the largest genetic FTD cohort yet analyzed. The carriers had mutations in one of three genes: C9ORF72, progranulin, or MAPT. Altogether, there were 71 people with a C9ORF72 expansion, 55 with a GRN, and 36 with a MAPT mutation. The first two typically result in TDP-43 pathology; MAPT mutations cause tauopathy.

The authors found proteins distinct for each genetic form (image at right). For C9ORF72, the intracellular calcium sensor calbindin 2 correlated most strongly. Weaker hits, such as GPI and HK1, related to glycolysis, implicating metabolic disruption in this form of FTD. Concentrations of all these C9ORF72 hits rose along with worsening disease.

For GRN FTD, the biggest change was a drop in progranulin, as expected from prior work. The lysosomal enzyme NAGA and the ubiquitin ligase RNF13 went up as disease progressed, though RNF13 missed statistical significance.

For MAPT FTD, the biggest changes in CSF were a boost of the astrocytic protein PEA15, which helps regulate apoptosis, and a drop in signaling receptor SEMA6A. Intriguingly, several lysosomal proteins were low in MAPT carriers but not the other groups, even though GRN and C9ORF72 mutations are known to affect lysosomes. This suggests the mutations might affect distinct aspects of lysosomal processing, Sogorb-Esteve noted.

Already at presymptomatic stages of disease, changes were measurable in these mutation-specific proteins. Teunissen noted that such early changes allow scientists to study “cleaner” biology, before mutation-specific alterations are masked by neurodegenerative changes.

To probe that biology, the authors looked for proteins that correlated with each other to form networks regulating specific processes. Here, symptomatic carriers showed the biggest changes. Unsurprisingly, neurodegenerative networks featuring markers such as NfL and YKL40 ticked up in all symptomatic carriers, as did protein networks involved in the cellular stress response, actin binding, and immunity. Synaptic proteins dropped in every symptomatic participant.

Cognition Markers? As CDR score worsens (left), most (red) core neurodegenerative markers (top) and actin-binding proteins (middle) rise, while most synaptic proteins (bottom) fall. Altogether, nine protein modules were linked to cognitive decline (right); solid circles indicate statistical significance. [Courtesy of Sogorb-Esteve et al., Science Translational Medicine/AAAS.]

Both the neurodegenerative and the synaptic protein changes correlated with measures of disease severity, such as plasma NfL, brain atrophy, baseline cognitive scores, and cognitive decline (image above). Daniel Geschwind at the University of California, Los Angeles, called the correlation with cognitive decline a strength of this study.

Many of these proteins, especially in the neurodegeneration and synaptic networks, are not specific for FTD. They also rise or fall in the CSF in other diseases, such as Alzheimer’s. Indeed, when the authors compared their data to two AD proteomic data sets, they found 1,192 proteins in common between all three, and only 221 specific for FTD. Among the latter, only two were different from controls in all three types of mutation carriers, but not in either AD data set: the extracellular matrix (ECM) protein follistatin-like 4, which went down, and the cell-surface glycoprotein CD44, which went up. However, Sogorb-Esteve cautioned that more work is needed to show that these are truly specific for FTD.

In the ALLFTD study, first author Saloner used aptamers to measure 4,138 CSF proteins, the largest number of any FTD study so far. Samples came from 47 people with a C9ORF72 expansion, 37 with a MAPT mutation, 32 with a GRN mutation, and 39 controls. Almost half of mutation carriers were symptomatic.

In people with symptomatic FTD, synaptic and lysosomal proteins were down compared with controls, while RNA-processing spliceosome proteins were up. The degree of change in all these protein modules correlated with symptom severity, and spliceosome proteins also correlated with CSF NfL, a marker of neurodegeneration. In people with presymptomatic FTD, the only consistent change was a drop in ion transport proteins compared with controls.

When the authors stratified CSF findings by mutation, they found that spliceosome and lysosomal proteins were most associated with C9ORF72 and GRN mutations, suggesting they relate to TDP-43 pathology. MAPT mutations were more closely linked to a drop in synaptic proteins and a rise in ECM and innate immune proteins. Spliceosome, synaptic, and ECM proteins were all associated with cognitive decline.

The authors compared their findings with CSF data from 35 people with progressive supranuclear palsy, a primary tauopathy. The findings were quite similar to those in MAPT carriers, hinting that all tauopathies may be alike. Conversely, when the authors compared ALLFTD findings with those from 87 AD patients in the BioFINDER 2 cohort, they found that FTD patients had more pronounced changes in synaptic, lysosomal, and spliceosome proteins, suggesting some of these could help distinguish the diseases.

The authors noted that although these findings suggest fruitful places to look for biomarkers, they will need larger samples from each type of mutation carrier to drill down to the level of individual proteins. Rojas told Alzforum that both the GENFI and ALLFTD studies captured only a fraction of all CSF proteins, and so could have missed promising biomarkers. In addition, these studies quantified only native proteins. Modified proteins, such as phosphorylated tau in AD, often make better disease markers (comment below).

Sogorb-Esteve noted that consistent differences between C9ORF72/GRN and MAPT carriers suggests that candidate biomarkers to identify the two major pathologies of FTD, TDP-43 and tau, might be on the horizon. These are sorely needed, because there is currently no reliable way to distinguish these pathologies in living people (Dec 2023 conference news).—Madolyn Bowman Rogers

Comments

  1. This is impressive work. The relatively large number of individuals per genetic subtype included explains the large number of proteins identified. The paper is a very valuable source for other proteomics studies. There are not so many proteomics studies published yet for FTD, and having a benchmark such as this study is therefore very valuable.

    The within-genetic subtype analyses compared presymptomatic changes with those of symptomatic changes. The proteins observed for the presymptomatic phase are interesting as they reveal “cleaner’’ biology, i.e., before the genetic subtype-specific biology is masked by overt neurodegenerative pathology occurring in the symptomatic stage. The protein changes, and the modules identified in these presymptomatic changes, are to me the most germane findings of the study for understanding the biology of each subtype.

    It is remarkable that the neurodegeneration markers that are increased across all stages highly overlap with those observed in AD, although there are clearly different symptoms between these diseases. This says to me that even with the very good coverage of proteins in the paper, there are likely still a lot of proteins to be identified. Perhaps affinity based -omics arrays such as PEA and Somascan can add to the full biological picture.

    View all comments by Charlotte Teunissen

  2. This is a very impactful paper.

    The authors found distinct and shared proteomic alterations across genetic FTD subtypes. Each genetic FTD mutation (C9ORF72, GRN, MAPT) displayed unique CSF proteomic changes. Some proteins were altered across all three subtypes and in Alzheimer’s Disease, suggesting common neurodegeneration mechanisms. Lysosomal protein levels were markedly reduced in MAPT carriers but not in the other groups. Several biomarkers were common to all FTD subtypes and AD, including Neuronal Pentraxin 2, Fatty Acid Binding Protein 3, UCHL1, and 14-3-3 proteins (YWHAZ, YWHAG, YWHAE).

    Some biomarkers were unique to specific FTD subtypes. Some of these displayed early proteomic changes in presymptomatic FTD mutation carriers. Moreover, they found strong correlations between disease severity and cognitive decline, with NfL showing strong associations with disease severity and brain atrophy. WGCNA revealed protein clusters linked to synaptic dysfunction, glial responses, and lysosomal impairment.

    These clusters were strongly correlated with Mini-Mental State Examination (MMSE) scores, disease progression, and brain volume loss. The identified proteomic signatures in presymptomatic carriers that could be early biomarkers for tracking disease onset, with candidates for monitoring disease progression. The authors also found distinctive protein changes that differentiate FTD subtypes from AD. 

    Proteomic comparisons with AD may refine differential diagnoses in ambiguous cases. Their findings highlight lysosomal dysfunction as a potential therapeutic target. Additionally, integrating proteomic data with imaging and genetic datasets could enhance biomarker discovery.

    One of the most notable and (a bit unexpected) findings was the proteomic similarities between FTD and AD.

    This study provides the largest and most comprehensive proteomic analysis of genetic FTD, identifying shared and unique markers across genetic subtypes, advancing the field toward precision biomarkers.

    View all comments by Agustin Ibanez
    • User Profile Image Julio Rojas
      Memory and Aging Center, Weill Insititute for Neurosciences, University of California, San Francisco
    • Posted:

    This is an important study for fluid biomarker development in FTD. The results could support further diagnostic and therapeutic development in FTD, a condition with no cure and for which no specific diagnostic test exist.

    The study in this European cohort replicates the findings of its North American counterpart ALLFTD, which also identified similar mutation-specific CSF proteomic signatures in familial and pathology-confirmed sporadic cases (e.g., extracellular matrix, synaptic function, and autophagy), as presented in the last International Society for Frontotemporal Dementia scientific meeting in September 2024 (Saloner et al., 2024). The ALLFTD study interrogated the levels of twice as many CSF targets and identified an additional strong signal from RNA splicing proteins. The changes in MAPT mutation carriers involving synapse formation and axonal guidance suggest a major role of these neuronal pathways in tauopathies, which has been previously seen in progressive supranuclear palsy, a sporadic and relatively more common FTD tauopathy (Wise et al., 2024). Similar to the previous CSF proteomics studies in FTD, the study by Sogorb-Esteve and colleagues showcases the power of WGCNA and other novel data analysis techniques, that are becoming the standard for the analysis of proteomics signatures in FTD.

    A limitation of this and the previous proteomics studies in FTD mentioned above are that the quantitative platforms only capture a fraction of the CSF proteome (about 10 percent in Sogorb-Esteve et al.’s study). There may be biases imposed by what analytes are measured. For example, many nuclear or mitochondrial pathways and proteins were not represented in this mass-spectrometry study. Also, these proteomic platforms quantify native proteins, but it is possible that a stronger signal would be obtained when measuring proteins with post-translational modifications, like Aβ and phosphorylated tau, which are useful in Alzheimer’s disease as opposed to the native proteins.

    Future proteomics studies should also tackle products of cryptic splicing, a fundamental pathobiological process, especially in forms of FTD with FTD-43 pathology. Also, although findings in CSF are great progress, the field needs blood-based biomarkers, which are more easily implemented on clinical grounds, especially in clinical trials. There is also a need to run proteomics studies in diverse populations of non-European descent.

    The field of clinical FTD research is living a golden era and more progress on biomarkers and therapeutics will likely be seen in the upcoming years for this devastating group of diseases.

    References:

    Saloner R, Staffaroni A, Dammer E, Johnson EC, Paolillo E, Wise A, Heuer H, Forsberg L, Lago AL, Webb J, Vogel J, Santillo A, Hansson O, Kramer J, Miller B, Li J, Loureiro J, Sivasankaran R, Worringer K, Seyfried N, Yokoyama J, Seeley W, Spina S, Grinberg L, VandeVrede L, Ljubenkov P, Bayram E, Bozoki A, Brushaber D, Considine C, Day G, Dickerson B, Domoto-Reilly K, Faber K, Galasko D, Geschwind D, Ghoshal N, Graff-Radford N, Hales C, Honig L, Hsiung GY, Huey E, Kornak J, Kremers W, Lapid M, Lee S, Litvan I, McMillan C, Mendez M, Miyagawa T, Pantelyat A, Pascual B, Paulson H, Petrucelli L, Pressman P, Ramos E, Rascovsky K, Roberson E, Savica R, Snyder A, Sullivan AC, Tartaglia C, Vandebergh M, Boeve B, Rosen H, Rojas J, Boxer A, Casaletto K. Large-scale network analysis of the cerebrospinal fluid proteome identifies molecular signatures of frontotemporal lobar degeneration. Res Sq. 2024 Mar 28; PubMed.

    Wise A, Li J, Yamakawa M, Loureiro J, Peterson B, Worringer K, Sivasankaran R, Palma JA, Mitic L, Heuer HW, Lario-Lago A, Staffaroni AM, Clark A, Taylor J, Ljubenkov PA, Vandevrede L, Grinberg LT, Spina S, Seeley WW, Miller BL, Boeve BF, Dickerson BC, Grossman M, Litvan I, Pantelyat A, Tartaglia MC, Zhang Z, Wills AA, Rexach J, Rojas JC, Boxer AL, as the 4-Repeat Tauopathy Neuroimaging Initiative. CSF Proteomics in Patients With Progressive Supranuclear Palsy. Neurology. 2024 Aug 13;103(3):e209585. Epub 2024 Jul 3 PubMed.

    View all comments by Julio Rojas

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Second Holloway Summit Showcases Intense Search for FTD Biomarkers

Paper Citations

  1. Remnestål J, Öijerstedt L, Ullgren A, Olofsson J, Bergström S, Kultima K, Ingelsson M, Kilander L, Uhlén M, Månberg A, Graff C, Nilsson P. Altered levels of CSF proteins in patients with FTD, presymptomatic mutation carriers and non-carriers. Transl Neurodegener. 2020 Jun 23;9(1):27. PubMed.
  2. Bergström S, Öijerstedt L, Remnestål J, Olofsson J, Ullgren A, Seelaar H, van Swieten JC, Synofzik M, Sanchez-Valle R, Moreno F, Finger E, Masellis M, Tartaglia C, Vandenberghe R, Laforce R, Galimberti D, Borroni B, Butler CR, Gerhard A, Ducharme S, Rohrer JD, Månberg A, Graff C, Nilsson P, Genetic Frontotemporal Dementia Initiative (GENFI). A panel of CSF proteins separates genetic frontotemporal dementia from presymptomatic mutation carriers: a GENFI study. Mol Neurodegener. 2021 Nov 27;16(1):79. PubMed.
  3. Sogorb-Esteve A, Nilsson J, Swift IJ, Heller C, Bocchetta M, Russell LL, Peakman G, Convery RS, van Swieten JC, Seelaar H, Borroni B, Galimberti D, Sanchez-Valle R, Laforce R Jr, Moreno F, Synofzik M, Graff C, Masellis M, Tartaglia MC, Rowe JB, Vandenberghe R, Finger E, Tagliavini F, Santana I, Butler CR, Ducharme S, Gerhard A, Danek A, Levin J, Otto M, Sorbi S, Le Ber I, Pasquier F, Gobom J, Brinkmalm A, Blennow K, Zetterberg H, Rohrer JD, GENetic FTD Initiative. Differential impairment of cerebrospinal fluid synaptic biomarkers in the genetic forms of frontotemporal dementia. Alzheimers Res Ther. 2022 Aug 31;14(1):118. PubMed.

External Citations

  1. GENFI
  2. ALLFTD

Further Reading

Primary Papers

  1. Sogorb-Esteve A, Weiner S, Simrén J, Swift IJ, Bocchetta M, Todd EG, Cash DM, Bouzigues A, Russell LL, Foster PH, Ferry-Bolder E, van Swieten JC, Jiskoot LC, Seelaar H, Sanchez-Valle R, Laforce R, Graff C, Galimberti D, Vandenberghe R, de Mendonça A, Tiraboschi P, Santana I, Gerhard A, Levin J, Sorbi S, Otto M, Pasquier F, Ducharme S, Butler CR, Le Ber I, Finger E, Tartaglia MC, Masellis M, Rowe JB, Synofzik M, Moreno F, Borroni B, Genfi, Blennow K, Zetterberg H, Rohrer JD, Gobom J, GENFI. Proteomic analysis reveals distinct cerebrospinal fluid signatures across genetic frontotemporal dementia subtypes. Sci Transl Med. 2025 Feb 5;17(784):eadm9654. Epub 2025 Feb 5 PubMed.
  2. Saloner R, Staffaroni A, Dammer E, Johnson EC, Paolillo E, Wise A, Heuer H, Forsberg L, Lago AL, Webb J, Vogel J, Santillo A, Hansson O, Kramer J, Miller B, Li J, Loureiro J, Sivasankaran R, Worringer K, Seyfried N, Yokoyama J, Seeley W, Spina S, Grinberg L, VandeVrede L, Ljubenkov P, Bayram E, Bozoki A, Brushaber D, Considine C, Day G, Dickerson B, Domoto-Reilly K, Faber K, Galasko D, Geschwind D, Ghoshal N, Graff-Radford N, Hales C, Honig L, Hsiung GY, Huey E, Kornak J, Kremers W, Lapid M, Lee S, Litvan I, McMillan C, Mendez M, Miyagawa T, Pantelyat A, Pascual B, Paulson H, Petrucelli L, Pressman P, Ramos E, Rascovsky K, Roberson E, Savica R, Snyder A, Sullivan AC, Tartaglia C, Vandebergh M, Boeve B, Rosen H, Rojas J, Boxer A, Casaletto K. Large-scale network analysis of the cerebrospinal fluid proteome identifies molecular signatures of frontotemporal lobar degeneration. Res Sq. 2024 Mar 28; PubMed.

Annotate

To make an annotation you must Login or Register.