Mind the GAP—Synaptic Protein in CSF Signals Tau Spread

Some scientists believe that toxic tau fibrils hop between synapses to spread throughout the brain, prompted by Aβ aggregates and overactive neurons. New imaging and biomarker data add weight to this theory. In the January 3 Nature Communications, scientists led by Michael Schöll at the University of Gothenburg in Sweden and Nicolai Franzmeier of Ludwig Maximilian University in Munich reported that synaptic activity, measured by cerebrospinal fluid levels of presynaptic growth-associated protein 43 (GAP-43), seems to drive Aβ-dependent tau spreading. PET amyloid-positive people with high levels of CSF GAP-43 accumulated neurofibrillary tangles faster than their counterparts with low GAP-43. The tau spread mostly through connected regions of the brain.

  • CSF GAP-43 signals synaptic activity.
  • People with high CSF GAP-43 and amyloid load quickly accumulated tangles.
  • Tau seems to travel through neuronal pathways.

“I’m happy to see that tau spreading through neuronal circuits in an activity-dependent manner translates from animal models into humans,” said Marc Suárez-Calvet of the Barcelonaβeta Brain Research Center in Spain. Andréa Benedet of U Gothenburg, who was not involved in this research, noted that this study somewhat challenges the concept that synaptic changes occur after amyloid and tau pathologies in AD. “While not refuting this idea, this study adds to the body of evidence supporting that synaptic alterations are also upstream events, promoting/accelerating the propagation of tau pathology throughout the brain in the presence of amyloid,” she wrote (comment below).

Among the handful of available synaptic markers, GAP-43 caught Franzmeier’s attention because it is upregulated in the CSF of people with AD but not other tauopathies, suggesting the marker’s uptick might be a response to amyloidosis (Qiang et al., 2022; Sandelius et al., 2019). Aβ is a well-known instigator of neuronal hyperexcitability, and Aβ-driven overactivity encourages tau aggregation and propagation in mouse models and in people (Sep 2007 news; Dec 2023 news). The faster neurons fire, the more tau they spew, speeding its spread in mice, the theory goes (Feb 2014 news). Could CSF GAP-43 signal amyloid-dependent spreading of tangles in the early stages of AD?

To find out, first author Franzmeier asked how baseline CSF GAP-43 and amyloid affects tangle accumulation over time. From the Alzheimer’s Disease Neuroimaging Initiative (ADNI), he correlated baseline amyloid PET with longitudinal tau PET for 72 cognitively normal adults, 18 people with mild cognitive impairment, and three with AD for whom CSF GAP-43 data was available. Most participants were in their 70s, and about half were women.

Over an average of three years of follow-up, participants with high amyloid loads and high CSF GAP-43, i.e., above the median of 4.3 ng/mL, accumulated more tau aggregates than those with GAP-43 below the median. This was true for the brain globally and in regions known to accumulate tangles early in the disease, such as the medial temporal lobe (image below). The data hint that GAP-43 pegs people with overactive synapses and the most active amyloid-dependent tangle formation.

More GAP, Faster Tangles. For a given amyloid load (x axis) the rate of tangle accumulation (y axis) throughout the brain (left), and in the medial temporal lobe (right), was faster in people with higher CSF GAP-43 (orange). [Courtesy of Franzmeier et al., Nature Communications, 2024.]

Did tangles propagate through connected regions of the brain? The scientists pinpointed the tangle epicenter in each participant, defined as the area with the strongest 5 percent of the tau PET signal. For most people, this was in the inferior temporal lobe. Then they divvied the rest of the brain into four quartiles based on strength of connectivity to the epicenter: Q1 comprised the temporoparietal and frontal lobes; Q2 the lateral temporal and frontal lobes; Q3 included the insular and somatosensory cortices; and, finally, Q4 was the primary visual and sensorimotor cortices (image below). Lo and behold, people with high CSF GAP-43 accumulated tangles more quickly in regions connected to the tau epicenter (image below). This hints that when neurons were hyperactive, tau had traveled through neuronal pathways rather than by simple diffusion.

“This underscores the role of synaptic integrity and activity in the progression of AD, moving beyond the traditional focus on Aβ and tau accumulation alone,” wrote Qiang Qiang of Fudan University in Shanghai, China (comment below).

Mapping Tau Spread. For most participants, the inferior temporal lobe marked the epicenter of tangle accumulation (top). This makes strongest connections with the temporoparietal and frontal lobes (Quartile 1), then the lateral temporal and frontal lobes (Q2), less so to the insular and somatosensory cortices (Q3), and finally, the primary visual and sensorimotor cortices (Q4). In amyloid-positive people with high CSF GAP-43, tau accumulates fastest in Q1 (bottom left) and slowest in Q4 (bottom right). Rates in Q2 and Q3 fell in between (not shown). [Courtesy of Franzmeier et al., Nature Communications, 2024.]

Tau propagation was also driven by amyloid. In participants with high CSF GAP-43, a high plaque load corresponded with spread through connected brain regions. “Our data support the hypothesis that synaptic changes contribute to tau progression in AD, and GAP-43 is a biomarker that can help us better understand synaptic changes,” Franzmeier said. The authors suggested that CSF GAP-43 increases may mirror Aβ-induced hyperexcitatory synaptic changes, though they acknowledged that this needs to be tested empirically.

Indeed, Douglas Galasko, University of California, San Diego, was skeptical of calling GAP-43 a marker of neuronal hyperexcitability. “To address the question of whether GAP-43 is playing an active role in aberrant excitation or is passively released as a result of synaptic damage (with or without excitation being an important mechanism), model systems will need to be interrogated,” he wrote (comment below). Suárez-Calvet wonders if the association extends beyond GAP-43. “I was surprised that a single such biomarker had a strong association with amyloid load and tau spreading, and I would like to see whether that happens with other synaptic biomarkers,” he said.—Chelsea Weidman Burke

Comments

  1. The recent article by Franzmeier et al. demonstrated, with a series of very elegant analyses, that the synaptic biomarker GAP-43 associates with the spreading and accumulation of Aβ-related tau pathology. The study somewhat challenges the perhaps more “traditional” framework that synaptic changes are downstream of amyloid and tau pathologies in the AD pathophysiological process. While not refuting this idea, this study adds to the body of evidence supporting that synaptic alterations are—also—upstream events, promoting/accelerating the propagation of tau pathology throughout the brain in the presence of amyloid. It could be insightful to explore the Aβ GAP-43 interaction beyond amyloid plaques, proxied by PET, such as with CSF Aβ42/40. Regardless, this work nicely adds pieces to the complex AD puzzle by proposing/supporting information about disease mechanisms.

    Adding to this topic, we, and others, have shown that different synaptic biomarkers may be similarly associated with other AD-related biomarkers, cross-sectionally (Wang et al., 2024). However, to my knowledge, it is not known how comparable these biomarkers are when more mechanistic approaches are considered or how similarly they affect/predict/reflect longitudinal changes in AD pathology. Although I personally do not believe that the synaptic biomarkers alone will, at the current research stage, be able to predict changes in AD pathology, they can be valuable if evaluated in more complex contexts as done here by Franzmeier and colleagues.

    Surprisingly, the results lack validation, despite the availability of similar data in other cohorts. But I am certain this article will spark interest in future studies exploring the role of synaptic biomarkers in AD.

    References:

    Wang YT, Ashton NJ, Servaes S, Nilsson J, Woo MS, Pascoal TA, Tissot C, Rahmouni N, Therriault J, Lussier F, Chamoun M, Gauthier S, Brinkmalm A, Zetterberg H, Blennow K, Rosa-Neto P, Benedet AL. The relation of synaptic biomarkers with Aβ, tau, glial activation, and neurodegeneration in Alzheimer’s disease. https://doi.org/10.21203/rs.3.rs-3797679/v1 (version 1) Research Square

    View all comments by Andrea L. Benedet

  2. The take-home message from this research article is that in AD, synaptic changes, manifested by increased levels of the presynaptic protein GAP-43, are closely associated with the accelerated spread of tau pathology. The study found that higher levels of GAP-43 in cerebrospinal fluid (CSF) are linked to faster Aβ-related tau accumulation and spreading across brain regions closely connected to tau epicenters. This suggests that GAP-43-related synaptic changes play an important role in the progression of AD, and targeting these synaptic changes could be a potential strategy for preventing the spread of tau pathology and, consequently, progression of the disease.

    Previous studies have associated GAP-43 with tau pathology in AD. A surprising result of this study is the specific association of GAP-43 with the spread of tau pathology across connected brain regions, rather than a general, diffuse accumulation of tau. This indicates that GAP-43-related synaptic changes do not lead to a widespread tau accumulation but specifically facilitate its spread along neural pathways. This finding is particularly significant because it underscores the role of synaptic integrity and activity in the progression of AD, moving beyond the traditional focus on Aβ and tau accumulation alone.

    These results emphasize the importance of considering synaptic health and integrity in AD pathophysiology. Furthermore, they suggest that CSF GAP-43 could serve as a potential biomarker for predicting the spread of tau pathology. This could be valuable for the development of therapeutic strategies and for monitoring progression.

    View all comments by Qiang Qiang

  3. In this paper, the authors conducted an interesting secondary analysis of CSF and longitudinal tau PET data from ADNI. CSF levels of GAP-43, a presynaptic protein important in neurodevelopment and in axonal regeneration, are increased in MCI and AD, and higher baseline levels have been shown to predict faster cognitive decline and neurodegeneration e.g., seen as MRI atrophy and FDG PET hypometabolism (Öhrfelt et al., 2023). Although GAP-43 plays roles during brain development, expression remains high in hippocampus and neocortex in adulthood, and increases with aging. The new analysis mapped longitudinal changes in tau PET in an ADNI cohort of amyloid-positive and -negative controls and MCI/mild AD patients, many of whom had more than two serial tau PET images. Higher baseline CSF GAP-43 was a predictor of faster accumulation of tau, after controlling for several additional variables, including CSF P-tau181. Another recent secondary analysis of ADNI data found that APOE e4 carriers had higher levels of baseline GAP-43 and worse trajectories of cognitive decline (Lan et al., 2023).

    The extent to which baseline CSF GAP43 predicted tau spread was statistically significant, although there was a fair amount of variability. Longitudinal studies of CSF GAP-43 have suggested that levels in CSF may plateau, therefore it appears to have potential as an early prognostic marker in AD. The new findings examining tau PET extend the prior work, and align with findings that support the spread of tau through connected tracts and pathways. Since levels of GAP-43 are increased in some models of hyperexcitability (e.g., rodents with seizures), the authors hypothesize that CSF GAP-43 is a marker related to hyperexcitable presynaptic neurons that contribute to spread of pathological tau. To address the question of whether GAP-43 is playing an active role in aberrant excitation or is passively released as a result of synaptic damage (with or without excitation being an important mechanism), model systems will need to be interrogated. 

    References:

    Öhrfelt A, Benedet AL, Ashton NJ, Kvartsberg H, Vandijck M, Weiner MW, Trojanowski JQ, Shaw LM, Zetterberg H, Blennow K, Alzheimer's Disease Neuroimaging Initiative. Association of CSF GAP-43 With the Rate of Cognitive Decline and Progression to Dementia in Amyloid-Positive Individuals. Neurology. 2023 Jan 17;100(3):e275-e285. Epub 2022 Oct 3 PubMed.

    Lan G, Du J, Chen X, Wang Q, Alzheimer’s Disease Neuroimaging Initiative, Han Y, Guo T. Association of APOE-ε4 and GAP-43-related presynaptic loss with β-amyloid, tau, neurodegeneration, and cognitive decline. Neurobiol Aging. 2023 Dec;132:209-219. Epub 2023 Sep 23 PubMed.

    View all comments by Douglas Galasko

  4. This new report by Franzmeier et al. is a beautiful continuation of investigations into the important interplay between Aβ, tau, and neuronal hyperexcitability in the context of Alzheimer's disease. There have long been reports of how the presence of Aβ can impact the types of tau aggregates and tau seeding potential in the brain, though this work has been largely tested in preclinical models. Expanding this concept to humans using the power of in-life longitudinal PET imaging of both AD pathologies helps set the stage for what is arguably one of the most exciting pillars of the manuscript—the use of GAP-43 as a fluid biomarker to understand more fully the disease mechanism(s), instead of solely capturing disease pathologies.

    Coming from a more tau-centric background, I have long been very interested in how Aβ, tau, and sub-epileptiform activity may all be related, not only from a therapeutics perspective but also in the hunt for better pathway engagement and disease modification biomarkers for AD. It is very exciting to now see human data supporting what some preclinical models have previously suggested: that there is indeed an underlying mechanism of Aβ-induced synaptic hyperexcitability that contributes to the disease pathogenesis that is mediated, at least in part, through tau.

    Of course, these analyses will need to be validated with larger datasets, but the data presented herein are a crucial step in the right direction. The authors do a great job of outlining some key next steps, including better understanding the timing of CSF GAP-43 increases as it relates to Aβ and tau pathological loads across different brain regions. I am also very curious and interested in how CSF GAP-43 levels may change in response to certain AD therapeutics, including anti-Aβ monoclonal antibodies and/or MAPT anti-sense oligonucleotides. Looking forward to seeing more GAP-43 biomarker data in the coming years.

    View all comments by Sarah DeVos
    • User Profile Image Jung-Min Pyun
      Seoul National University Bundang Hospital
    • Posted:

    This paper reported very interesting findings that higher CSF GAP43 levels were associated with faster Aβ-related tau accumulation and spreading to connectivity-associated regions. This suggests a clear relation of the presynaptic protein to amyloid (A) and tau (T) pathology in AD. To comprehend the underlying mechanisms of the contribution, examining the expression pattern of GAP43 gene, encoding the protein, according to A/T stages can provide additional clues.

    We recently demonstrated that GAP43 expression levels were significantly lower in A+/T+ and marginally lower in A+/T- compared to A-/T- in brain bulk tissue (Pyun et al., 2023). Moreover, at the cell-type level, GAP43 expression levels were lower in astrocytes, microglia, and oligodendrocytes in A+/T+ compared to A-/T-, but not in excitatory and inhibitory neurons.

    These results highlight the need to elucidate the discrepancy between decreased cortical GAP43 gene expression and increased CSF GAP43 protein expression in AD to discern the specific mechanistic role of GAP43 in AD, which might be situated along complex pathways. Additionally, lower GAP43 expression levels in glial cells of A+/T+, but not in neurons, may indicate that the GAP43-related synaptic pathway in AD could involve glial cells rather than neurons. Further studies on these points would broaden our understanding of the role of GAP43 in AD.

    References:

    Pyun JM, Park YH, Wang J, Bice PJ, Bennett DA, Kim S, Saykin AJ, Nho K. Aberrant GAP43 Gene Expression Is Alzheimer Disease Pathology-Specific. Ann Neurol. 2023 May;93(5):1047-1048. Epub 2023 Mar 21 PubMed.

    View all comments by Jung-Min Pyun

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Do "Silent" Seizures Cause Network Dysfunction in AD?
  2. Plaques Kick Neocortical Neurons into Overdrive, Entangling Tau
  3. Neurons Release Tau in Response to Excitation

Paper Citations

  1. Qiang Q, Skudder-Hill L, Toyota T, Wei W, Adachi H. CSF GAP-43 as a biomarker of synaptic dysfunction is associated with tau pathology in Alzheimer's disease. Sci Rep. 2022 Oct 17;12(1):17392. PubMed.
  2. Sandelius Å, Portelius E, Källén Å, Zetterberg H, Rot U, Olsson B, Toledo JB, Shaw LM, Lee VM, Irwin DJ, Grossman M, Weintraub D, Chen-Plotkin A, Wolk DA, McCluskey L, Elman L, Kostanjevecki V, Vandijck M, McBride J, Trojanowski JQ, Blennow K. Elevated CSF GAP-43 is Alzheimer's disease specific and associated with tau and amyloid pathology. Alzheimers Dement. 2019 Jan;15(1):55-64. Epub 2018 Oct 12 PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. Franzmeier N, Dehsarvi A, Steward A, Biel D, Dewenter A, Roemer SN, Wagner F, Groß M, Brendel M, Moscoso A, Arunachalam P, Blennow K, Zetterberg H, Ewers M, Schöll M. Elevated CSF GAP-43 is associated with accelerated tau accumulation and spread in Alzheimer's disease. Nat Commun. 2024 Jan 3;15(1):202. PubMed.

Annotate

To make an annotation you must Login or Register.