See Q&A below with David Holtzman

In many neurodegenerative diseases, pathologic proteins present tough therapeutic targets because they are intracellular—or at least scientists have assumed they were. New evidence from in-vivo microdialysis studies now suggests that at least some of these cytosolic proteins routinely get out of cells in healthy brains. Researchers led by David Holtzman at Washington University in St. Louis, Missouri, developed a microdialysis technique for measuring soluble tau levels in the interstitial fluid (ISF) of awake, active mice. In the September 14 Journal of Neuroscience, the scientists report that the ISF of both wild-type and tau transgenic mice contains soluble tau. Tau levels were much higher in ISF than in cerebrospinal fluid (CSF); surprisingly, the two were uncoupled. High CSF tau levels are widely used as a biomarker of Alzheimer’s disease (see, e.g., ARF related news story). Although it is not yet clear how tau gets out of cells, or what effect extracellular tau has on the brain, one encouraging implication of the finding is that tau might be more accessible to therapies than scientists have thought.

Tau is not the only cytosolic protein to escape from cells into the ISF. Researchers led by Kostas Vekrellis at the Academy of Athens, Greece, used a similar in-vivo microdialysis technique to measure extracellular concentrations of α-synuclein, the major component of the Lewy bodies found in Parkinson’s disease and some dementias. The results, published online July 14 in PLoS One, showed α-synuclein in the ISF of wild-type and transgenic mice. Importantly, Vekrellis and colleagues also found the protein in ISF collected from people with brain injuries. Although such experiments are in early days yet, the tau and synuclein results hint that cytosolic proteins leaking into the ISF could be a common phenomenon.

Scientists have just begun to tap the potential of in-vivo microdialysis, with but a handful of such studies to date. Holtzman’s group used the technique to examine the dynamics of soluble and insoluble Aβ in AD mice (see ARF related news story). With colleagues elsewhere, they have also measured ISF Aβ in injured human brains, finding that Aβ levels rise as brains recover, and suggesting its secretion is part of normal physiology (see ARF related news story on Brody et al., 2008). A prior study has looked at tau: Researchers at Uppsala University Hospital, Gothenburg, Sweden, detected high levels of both interstitial tau and Aβ in people receiving microdialysis after brain injury (see Marklund et al., 2009). Because this was done in injured brains, as was the α-synuclein study, it could not address the question of whether extracellular tau is present in healthy brains.

To get at this question, Holtzman and colleagues used a mouse model. First author Kaoru Yamada adapted the Aβ microdialysis technique for use with tau. As monomeric tau is larger than Aβ, about 50 kDa instead of four kDa, Yamada used a membrane with a pore size of 100 kDa. With such large pores, fluid diffuses from the probe and builds up in the animal’s brain, a hurdle the scientists overcame by using a push-pull pump and a high concentration of albumin protein in the probe to increase the osmotic pressure. After perfecting the method in vitro, the authors slid the tiny probe, 0.5 to 1 millimeter long, through a guide cannula into the mouse hippocampus. Animals remained awake and freely moving around the cage while fluid flowed through the tubing and sampled ISF proteins over the course of two days.

Yamada and colleagues measured tau levels in the collected fluid by enzyme-linked immunosorbent assay (ELISA). The authors found 45 ng/ml soluble tau in the ISF of three-month-old wild-type mice. They found about five times that in young transgenic tau P301S mice, consistent with the five times higher expression level of the protein in these mice. This showed that extracellular tau is present in healthy mouse brain and implied that the assay accurately reflects total soluble tau levels. One limitation of the microdialysis technique is that it captures only monomeric tau, not larger aggregates. In wild-type animals, ISF tau levels were around 10-fold higher than CSF levels.

Next, the authors checked what happens with age. In wild-type mice, tau ISF levels stayed the same, but in the P301S animals, levels began to drop around six months of age, when tau starts aggregating. Conversely, CSF tau levels increased with age in the transgenics. Surprisingly, the ISF and CSF changes did not correlate. The reason is not clear, but it suggests ISF and CSF pools are independently regulated, the authors note. This parallels findings for Aβ, where mouse ISF and CSF levels have also been shown to be unlinked (see Cirrito et al., 2003). The finding that monomeric ISF tau drops when aggregates emerge implies that the two forms might be in equilibrium. Indeed, when the authors injected tau aggregates into the brains of young transgenic mice, soluble tau dropped by 30 percent, suggesting it was getting sequestered onto aggregates.

Microdialysis has several potential applications, Holtzman told ARF. The technique could help to screen the efficacy of tau-lowering treatments. Because researchers could follow change over time, “The pharmacodynamic assessment is potentially much richer” than current methods, Holtzman told ARF. Microdialysis could also provide insight into the normal behavior of tau in the brain, Holtzman suggested. For example, researchers could study how tau gets out of cells by using reverse microdialysis to infuse a drug that inhibits exocytosis and see if that changes tau levels.

Henrik Zetterberg at Sahlgrenska University Hospital in Molndal, near Gothenburg, Sweden, a coauthor on the earlier human tau microdialysis study, suggested that physiological release of tau might be a consequence of synaptic remodeling, noting that tau levels are very high in newborn CSF. “Personally, I believe that extracellular tau in the absence of pathological processes may reflect neuroaxonal plasticity,” Zetterberg wrote to ARF (see full comment below).

One of the crucial questions is what extracellular tau does in the brain. Work from several groups, including Marc Diamond at Washington University and researchers led by Michel Goedert at MRC Laboratory of Molecular Biology, Cambridge, U.K., has shown that misfolded tau can propagate from cell to cell, helping disease spread across the brain (see, e.g., ARF related conference story; ARF related news story; and ARF news story). Extracellular tau could be a culprit in this process. To investigate this, Holtzman’s group is working on adapting microdialysis to detect molecules bigger than 100 kDa and collaborating with Diamond to look for tau aggregates in ISF. If extracellular tau spreads pathology, the good news is that it may also provide an accessible target for tau-lowering therapies. This might help explain preliminary reports that passive and active immunization strategies lower tau pathology in mice (see ARF related conference story and ARF conference story).

The tau findings jibe with those of other proteins implicated in neurodegenerative disease that also seem to spread between cells. For example, deposits of α-synuclein can migrate from diseased tissue into healthy grafts (see ARF related news story; ARF related conference story; and ARF news story). In addition, Aβ has been shown to spread through the brain (see ARF related news story and ARF news story), as does Huntingtin protein (see ARF news story).

In their paper, Vekrellis and colleagues report detecting α-synuclein in the ISF of healthy mouse brains. Vekrellis initially presented this work at the AD/PD 2011 conference in Barcelona, Spain, this past March. First author Evangelia Emmanouilidou used a 100 kDa microdialysis pore size to capture the protein. Although the synuclein monomer is 14 kDa, recent studies suggest the protein may have a native tetrameric form (see ARF related news story), which would be around 56 kDa. After developing an ELISA sensitive between 0.01 ng/ml to 25 ng/ml of α-synuclein, the Greek scientists found about 0.15 ng/ml of ISF α-synuclein in wild-type mice, and about three times that in A53T mutant α-synuclein transgenic mice, matching their threefold overexpression. Extending the study to people, the scientists examined ISF samples from eight patients with severe brain injuries who received microdialysis as part of standard medical monitoring. They found α-synuclein concentrations varying from 0.5 to 8 ng/ml, showing the protein also gets out in human brain. None of the patients had a diagnosis of Parkinson’s or dementia, and the probe sampled healthy brain tissue, not sites of injury. Even so, synuclein levels could have been affected by overall poor brain health in these patients. Although the effect of extracellular α-synuclein in vivo is unknown, in previous cell culture experiments, the authors showed that secreted α-synuclein lowers neuronal survival (see Emmanouilidou et al., 2010).—Madolyn Bowman Rogers

Q&A With David Holtzman

ARF: Why measure interstitial fluid (ISF) tau in mice?

Holtzman: To understand the normal or disease function of a protein, I think it is really important to understand its metabolism. What controls its level? This technique should allow us to make measurements we were never able to make before in the brain of a living animal. The field has not previously been able to make in-vivo brain tau measurements dynamically. Tau is critically important in several diseases, and previous work on Aβ microdialysis led us to believe we can do this for tau, too.

ARF: How did you make the technique work for tau?

Holtzman: We knew the Aβ method would not work for tau. Because tau is a much bigger protein than the microdialysis probe cutoff size, there was no way it would go through the membrane. We obtained probes that have larger cutoff sizes to see if it was feasible. It took a while to get it to work even just in a test tube. We (Kaoru Yamada, post-doctoral fellow) also had to modify the constituents of what was flowing through the microdialysis probe itself. We used albumin to make sure tau did not stick to the plastic. We realized with tau, if you increase the concentration of albumin, you get much better recovery of the protein. That was the biggest trick other than the size. When we went in vivo, the other big adaptation was that we realized we had to use a different pump. With a normal pump you just push fluid through the probe and collect what drips out at whatever rate you set the machine at. If the molecular cutoff is too big, you get a positive flow through the brain, but if you use a push-pull pump, you do not get pressure building up in the brain. The brain, because of the skull, cannot tolerate any increase in pressure.

ARF: What did you find?

Holtzman: The concentration of tau in the extracellular space was a lot higher than in human cerebrospinal fluid (CSF). That was the first big surprise. In human CSF, tau is around 250 pg/ml in a normal person. In wild-type mice, ISF tau is 50 ng/ml, 200 times higher. We thought, “This cannot be right.” But it is true; we have now done many measurements in many mice.

I think what is likely happening is that when proteins get out of the cell, they are getting retained in the brain. There are lots of sticky things in the extracellular space of the brain. Proteins are sticking to membranes, to extracellular matrix, and few reach the CSF. That is a guess.

In tau transgenic mice, tau is higher. Yoshiyama et al. had shown in their paper that tau is overexpressed fivefold in the P301S mice (line PS19), and that is exactly what we found in ISF. That told us the method was likely working reasonably well.

ARF: Does tau change with age?

Holtzman: One of the main points is that when we looked at wild-type up to a year of age, there was no change in tau levels, but in transgenic mice, there was a big drop from 250 ng/ml down to 80 ng/ml. No change in tau expression. We think as tau is aggregating, there is a sink of aggregated material, so a lot of the tau does not float freely around because it is in equilibrium with aggregates.

We do not know if tau is in equilibrium with intracellular aggregates or if there are extracellular aggregates. If aggregates spread from cell to cell, there is likely a phase where they are in the extracellular space before they get into the next cell. We would like to assess if there are extracellular aggregates. If you inject aggregates into the brain, it causes an immediate drop in ISF tau.

ARF: What forms does microdialysis measure?

Holtzman: Not aggregates, probably. I do not know how it could measure even a dimer. In Western blots, it looks like we measure a monomer of full-length tau. The fact that tau level goes down with age in transgenic mice implies there is a change in the aggregation state, and we keep detecting the monomer. I was very surprised to see that tau went down with age. I thought that, as the mice develop neurodegeneration as they get older, tau would go up as the cells were being damaged. But remember, the transgenic mice are a model for frontotemporal dementia. In human frontotemporal dementia, tau does not go up, at least in CSF. In retrospect, maybe it was not surprising that we saw nothing going on.

ARF: Why does CSF tau go up in AD but not in frontotemporal dementia?

Holtzman: I have no answer yet. I have thought about that a lot. One would think if the elevation of CSF tau in AD was due to the presence of neurofibrillary tangles or tau aggregates, then tau should be elevated also in frontotemporal dementias. But that is not the case. While the amount of CSF tau in AD correlates with the amount of neurofibrillary pathology, it also correlates with amyloid deposition. I wonder if the reason tau is elevated higher in AD than in frontotemporal dementia has to do with the amount of neuritic degeneration. I wonder whether it is due to smoldering injury, that tau release from synaptic regions is higher. There is a fair amount of that kind of injury that is occurring in an ongoing fashion in AD. But that is just speculation.

ARF: Tau, unlike Aβ, is a cytoplasmic protein. It is not released.

Holtzman: Yes, but with cytoplasmic proteins, if you look carefully in CSF you can find them. Most of them have been described to be in CSF. These are abundant cytoplasmic proteins in normal people like GFAP, SOD. To me it is unlikely it has to do with a normal function of the proteins. But they do get out. Maybe our cells are normally leakier than we thought.

The concentrations of these proteins inside cells are much higher. It is probably orders of magnitude higher inside, both for tau and α-synuclein.

ARF: Are you doing ISF microdialysis in humans?

Holtzman: Not right now. We are just interested in further understanding the biology of proteins like tau. What regulates it normally?

ARF: What are the therapeutic implications?

Holtzman: It is exciting: Some people want to lower tau as a potential treatment. This would be a nice method to screen treatments, just like we have done for Aβ. For example, if you knock down tau expression with short interfering RNA, do you see ISF tau levels go down, and how fast does that happen? Days? Hours?


  1. This study is a very nice extension of earlier work on brain interstitial tau concentrations performed in humans following traumatic brain injury (Marklund et al., 2009). It is now important to examine tau homeostasis in normal conditions. Personally, I believe that extracellular tau in the absence of pathological processes may reflect neuroaxonal plasticity. This hypothesis is to some degree backed by studies on normal newborns who have very high levels of total and phosphorylated tau in their CSF (Mattsson et al., 2010). The first months after birth are characterized by extensive synaptic and neuronal remodeling, which may involve physiological tau phosphorylation and release of tau proteins from retracting and regrowing axons.


    . Monitoring of brain interstitial total tau and beta amyloid proteins by microdialysis in patients with traumatic brain injury. J Neurosurg. 2009 Jun;110(6):1227-37. PubMed.

    . Converging molecular pathways in human neural development and degeneration. Neurosci Res. 2010 Mar;66(3):330-2. PubMed.

  2. This is an elegant study by the Holtzman group which clearly shows that yet another neurodegenerative disease-associated protein, thought to be cytosolic, is found extracellularly in the brain parenchyma. Using an optimized in-vivo microdialysis approach that causes minimal brain injury, in wild-type and P301S transgenic mice, Yamada and colleagues demonstrated that ISF tau concentration in interstitial fluid (ISF) is a lot higher than that in the cerebrospinal fluid (CSF). Interestingly, the levels between the two pools of tau did not appear to have a positive correlation. This finding is of great importance, especially when considering CSF tau as a biomarker for AD. Nonetheless, the presence of tau extracellularly is exciting since it suggests that a mechanism of propagation of tau aggregates could exist in tauopathies and related disorders. As such, it opens up new avenues for immunotherapy targeting extracellular tau in disease.

    Similar to our studies with α-synuclein, the tau release was found to be physiological and in the absence of neurodegeneration. Still, one important question that remains to be answered is the physiological relevance of extracellular tau. The study proposes the existence of a dynamic equilibrium between different species of intracellular and extracellular tau that is dependent on aggregate formation and accumulation. It is possible that a such a dynamic equilibrium exists to ensure neuronal homeostasis. In this respect, as also implied by our findings with α-synuclein, dysfunctions in the mechanism(s) regulating extracellular levels, such as secretion, re-uptake, or extracellular clearance, may affect neuronal survival. Optimizing in-vivo microdialysis probes to accurately detect and quantify oligomeric forms of such proteins in the ISF will shed more light onto the pathogenesis of Parkinson’s and Alzheimer’s diseases.

    View all comments by Kostas Vekrellis
  3. Dave Holtzman and colleagues have been using the very elegant method
    of microdialysis, which allowed them to collect ISF (interstitial
    fluid) from both wild-type and P301L tau transgenic mice, and
    to infuse tau aggregates. The paper by Yamada et al. further shows
    nicely that full-length tau is released into the ISF already under
    physiological conditions, and that in P301S tau transgenic mice,
    soluble tau levels in the ISF decrease as the mice become older. This
    is likely because soluble tau is trapped by the tau aggregates that
    form as the disease progresses. In CSF, in contrast, the tau
    concentration increases as the mice become older (as there is no
    peripheral aggregated tau that would trap soluble tau). The model that is put forward in Figure 7 explains very well the dynamics of tau ISF.

    A whole series of follow-up experiments come immediately to mind.
    Starting with wild-type mice, one could expose these to all kinds of
    stress (anesthesia, food deprivation, or cold temperature) to
    determine whether these conditions, which cause an altered
    phosphorylation of tau, would
    (permanently) alter extracellular tau deposition. Phosphorylation of
    tau has not been addressed yet, nor whether distinct phospho-tau
    species would reveal a correlation between ISF and CSF tau (a
    relationship not found for total tau). One will be able to determine
    which epitopes of tau are phosphorylated before and which ones only
    after tau aggregation. And obviously, as discussed in the paper, the
    question can be answered whether a tau-directed vaccination works at least in part by depleting extracellular pools of tau.

    View all comments by Jürgen Götz
  4. This is a very ingenious demonstration that significant extracellular human tau levels exist and can be monitored in transgenic mice. It is particularly important because it opens the door to a detailed biochemical characterization of the secretion process in a widely used model system. The finding that CSF tau levels show a different dynamic than does ISF tau is also interesting, since it suggests the possible involvement of ependymal cell transcytosis in the genesis of CSF tau.

    My lab has studied the mechanism of tau secretion for a number of years now. We are using cell-autonomous human tau expression in a non-transgenic lower vertebrate model. We have shown that secretion is modulated by aggregation inhibitors (Hall et al., 2002) and by presence of the P301L tauopathy mutation (Kim et al., 2010a); requires the amino terminus (Kim et al., 2010a); and is associated with the mislocation of membrane-associated tau to the dendrites (Lee et al., 2011). I am particularly interested in seeing how tau secretion mechanisms characterized using microdialysis in mouse models compare with the Lamprey model data, given the complementary advantages of these in-situ systems.

    The differences between extracellular fluid and CSF tau dynamics are also interesting in the context of our recent study (Saman et al., 2011) showing that the typical elevation of CSF Pthr181 tau levels in early AD (Braak Stage 3) is associated with increased ptau in CSF exosome fractions. That study also showed that the exosome pathway was involved in at least some of the tau secretion detected in cultured neuroblastoma cells. Most importantly, our results strongly suggest that passive tau release due to neuron death plays only a minor role in the genesis of CSF tau, especially early in the course of the disease.

    The findings of Yamada et al. raise the possibility that it may be possible to use microdialysis to resolve the roles played by 1) tau secretion patterns, 2) changes to ependymal cell function in the genesis of early AD CSF tau increase, and 3) passive postmortem release from dead neurons or glia in experimentally manipulable systems.


    . Neurofibrillary degeneration can be arrested in an in vivo cellular model of human tauopathy by application of a compound which inhibits tau filament formation in vitro. J Mol Neurosci. 2002 Dec;19(3):253-60. PubMed.

    . Interneuronal transfer of human tau between Lamprey central neurons in situ. J Alzheimers Dis. 2010;19(2):647-64. PubMed.

    . Secretion of human tau fragments resembling CSF-tau in Alzheimer's disease is modulated by the presence of the exon 2 insert. FEBS Lett. 2010 Jul 16;584(14):3085-8. PubMed.

    . Accumulation of vesicle-associated human tau in distal dendrites drives degeneration and tau secretion in an in situ cellular tauopathy model. Int J Alzheimers Dis. 2012;2012:172837. PubMed.

    . Exosome-associated tau is secreted in tauopathy models and is selectively phosphorylated in cerebrospinal fluid (CSF) in early Alzheimer's Disease. J Biol Chem. 2011 Nov 4; PubMed.

    View all comments by Garth Hall

Make a Comment

To make a comment you must login or register.


News Citations

  1. Triple Confirmation: AD Footprint in CSF of Cognitively Normal People
  2. Soluble Aβ: Getting a Grip on Its Fate
  3. Soluble Aβ—Bane or Boon? Real-time Data in Humans Yield New Insight
  4. Keystone: Tau, Huntingtin—Do Prion-like Properties Play a Role in Disease?
  5. Double Paper Alert—Keystone Presentations Now in Press
  6. Traveling Tau—A New Paradigm for Tau- and Other Proteinopathies?
  7. San Francisco: Tau—Time to Shine as Therapeutic Target?
  8. Uppsala: Is Tau Immunotherapy Taking Off?
  9. Dopaminergic Transplants—Stable But Prone to Parkinson’s?
  10. Barcelona: Parkinson’s Treatments on the Horizon
  11. Parkinson's: An Unlikely Proposal Gains Momentum
  12. Aβ the Bad Apple? Seeding and Propagating Amyloidosis
  13. Insidious Spread of Aβ: More Support for Synaptic Transmission
  14. An α-Synuclein Twist—Native Protein a Helical Tetramer

Paper Citations

  1. . Amyloid-beta dynamics correlate with neurological status in the injured human brain. Science. 2008 Aug 29;321(5893):1221-4. PubMed.
  2. . Monitoring of brain interstitial total tau and beta amyloid proteins by microdialysis in patients with traumatic brain injury. J Neurosurg. 2009 Jun;110(6):1227-37. PubMed.
  3. . In vivo assessment of brain interstitial fluid with microdialysis reveals plaque-associated changes in amyloid-beta metabolism and half-life. J Neurosci. 2003 Oct 1;23(26):8844-53. PubMed.
  4. . Cell-produced alpha-synuclein is secreted in a calcium-dependent manner by exosomes and impacts neuronal survival. J Neurosci. 2010 May 19;30(20):6838-51. PubMed.

Other Citations

  1. See Q&A below with David Holtzman

External Citations

  1. P301S mice
  2. A53T mutant α-synuclein transgenic mice

Further Reading

Primary Papers

  1. . In vivo microdialysis reveals age-dependent decrease of brain interstitial fluid tau levels in P301S human tau transgenic mice. J Neurosci. 2011 Sep 14;31(37):13110-7. PubMed.
  2. . Assessment of α-synuclein secretion in mouse and human brain parenchyma. PLoS One. 2011;6(7):e22225. PubMed.