Tau therapeutics for alzheimer’s disease

Tau Therapeutics for Alzheimer’s DiseaseThe Promise and the Challenges

Michael Gold*

Global Clinical Research and Development, CNS/Analgesia, Johnson & Johnson Pharmaceutical Research and Development, Titusville, NJ 08560

Received June 7, 2002; Accepted October 28, 2002

Abstract

The pathological diagnosis of Alzheimer’s disease (AD) depends on the presence of plaques consisting ofthe β-amyloid peptide as well as neurofibrillary tangles consisting of paired helical filaments (PHFs) of the tau(τ) protein. The role of each type of pathology in the pathogenesis and progression of AD remains unclear. Pre-vious hypotheses suggested that these two processes were independent, whereas more recent data suggest thatthere may be a bidirectional interaction between these two pathological processes.

The identification of the neurotoxic effects of β-amyloid and the discovery of mutations responsible for early-onset Alzheimer’s disease (EOAD) and their linkage to β-amyloid overproduction, has made the amyloid hypoth-esis of AD the predominant influence for therapeutic targets. Several approaches have emerged from preclinicaltesting and have entered early phases of clinical developments.

The recent identification of τ mutations and their linkage to progressive neurodegenerative disorders pro-vides a counterbalancing influence on the search for therapeutic targets for AD. Therapeutic approaches thatare targeted to either β-amyloid or τ share certain features at the level of pharmacology and will face many ofthe same challenges as they progress through drug development paradigms. The aim of this article is to pro-vide a brief overview of some of the commonalities and the challenges faced by τ-related therapeutic strategies.The issues discussed in this article are not exhaustively dealt with in either scope or detail.

Index Entries: Tau; Alzheimer’s disease; drug development; clinical research; pharmacology.

Pathological Model

Current models of Alzheimer ’s disease (AD)pathogenesis related to τ propose that some alter-ation in τ (i.e., phosphorylation) causes it to disso-ciate from microtubules, to adopt some pathologicalconformation that predisposes it to fibril formationand to initiate the process of paired helical filament(PHF) formation (Hagestedt et al., 1989; Brion et al.,2001). Downstream effects of PHF formation includevarious processes culminating in cell death. Thesemodels have significant mechanistic parallel to theamyloid hypothesis for AD and lead to testablehypotheses for targeting τ (see Table 1).

Each one of these approaches faces unique chal-lenges that will have to be overcome before anypotential therapy enters into clinical trials. In the fol-lowing sections, I will delineate some of these chal-lenges and provide some examples of thesechallenges.

Pharmacological IssuesSpecificity/Selectivity

Whether the approach is to inhibit phosphoryla-tion or accelerate dephosphorylation, the fact thatmultiple kinases (Steiner et al., 1990; Flaherty et al.,2000) and phosphatases (Billingsley and Kincaid,

Journal of Molecular NeuroscienceCopyright © 2002 Humana Press Inc.All rights of any nature whatsoever reserved.ISSN0895-8696/02/19:331–334/$11.00

Journal of Molecular Neuroscience 331 Volume 19, 2002

*Author to whom all correspondence and reprint requests should be addressed. E-mail: [email protected]

JMN19-3,CH12pgs329-334,6pgs 11/20/02 11:03 AM Page 331

332 Gold

Journal of Molecular Neuroscience Volume 19, 2002

1997) are involved in modulating τ suggests that targeting a specific enzyme may be insufficient to“normalize” τ. One is reminded of the multiple mechanisms activated during cellular ischemia andof the many drugs targeting one or more of thesemechanisms that have been tested against ischemicstroke. The fact that every neuroprotectant tested inpatients with ischemic stroke has failed bespeaksour inability to effectively interfere with multiplepathways. The challenge here will be to decidewhether to target the kinase or phosphatase that hasthe largest effects on τ or to target a large subset ofthese enzymes but inhibit them to a lesser degree.

The rationale for attempting to achieve some sortof specificity in target selection is anchored in thebelief that nonselective activity would probably leadto unacceptable safety problems. The rationale for anonselective approach is to prevent compensatorychanges in kinases/phosphatase activation thatwould nullify the drugs initial effect.

The same general concerns apply to approachesthat are designed to interfere with τ polymerization(Guttmann et al., 1995). There are multitudes of phys-iological polymers whose functions need to be pre-served (i.e., myosin, microtubules, neurofilaments).Therefore, any approach aimed at preventing ordestabilizing τ filaments will also have to resolve theselectivity dilemma.

Degree of InhibitionA complementary concern for any of these

approaches relates to the degree to which the patho-logical processes need to be altered. For example,will it require 10% or 90% inhibition of some kinase(s)to alter τ phosphorylation and prevent PHF forma-tion? The concern here relates to the therapeuticindex of a drug because clinical experience gener-ally dictates that higher doses of drugs are often asso-ciated with increased side effects. It will be criticalto establish dose/exposure–response relationshipsin order to provide estimates of the doses requiredto have different degrees of effect. An additional

rationale for these type of data is the notion thatinterfering with AD pathology at earlier stages mayrequire less pharmacological intervention thanattempting to alter the course of AD when the pathol-ogy is rampant.

Tonic/Phasic InhibitionAn implicit assumption for altering AD pathol-

ogy is that whatever approach is taken, sustained(tonic) interference with the process is required. Thealternative hypothesis, that phasic inhibition maywork as well, needs to be tested. The rationale fordetermining whether tonic or phasic inhibition isrequired relates to the issue of preventing unrelatedsafety problems and increasing the therapeuticindex.

A recent example relates to the safety of γ-secretase inhibitors and Notch inhibition. It is notknown whether tonic or phasic inhibition ofγ-secretase activity is required to reduce β-amyloidand it is also not known whether such phasic inhi-bition would provide a wider safety margin withrespect to Notch or other substrates for γ-secretasesthat could be source of dose-limiting toxicity.Another example where tonic versus phasic dosinghas clinical relevance is in the treatment of Parkin-son’s disease (PD), where phasic exposure to highconcentrations of L-dopa may accelerate the loss ofdopaminergic cells.

Another reason for investigating the tonic versusphasic aspect of pharmacology relates to the fact thatmany of the drugs being used for disease modifica-tion effects (i.e., statins or steroids) have long-lasting clinical effects that are unrelated to plasmahalf-life. Whether this is related to second-messenger systems, gene activation or some long-lasting modulation of a receptor, the key message isthat many central nervous system (CNS)-relatedcompounds may have a PK/PD dissociation. Such dissociation implies that traditional plasma-concentration-driven dosing paradigms may not beapplicable.

Table 1Proposed Mechanisms for Altering Tau as a Therapeutic Strategy in AD

1. Inhibition of τ phosphorylation.2. Stimulation of τ dephosphorylation.3. Prevention of τ aggregation.4. Stimulation of τ dis-aggregation.5. Increasing clearance of τ from the central nervous system.6. Inhibiting downstream effects of τ.


Tau Therapeutics for Alzheimer’s Disease 333


Clinical Development IssuesWhat Is Adequate Pharmacological Proof

of Concept Study?One of the critical stage gates for clinical devel-

opment is the achievement of pharmacological proofof concept. Although there is no single standard def-inition, at a minimum, a proof of concept (POC) studyneeds to demonstrate that the lead compoundreaches its designated target, has the predicted effecton target, and has the predicted effect on the rele-vant pharmacodynamic markers. In the case of akinase inhibitor, it would be necessary to establishbinding to a specific kinase, inhibition of the kinase,and reduction in tau phosphorylation.

Difficulties arise in translating data from what areusually acute POC studies into useful informationfor the design of human studies. For example, thereare questions related to the comparable exposure ofa rodent or other animal model versus the expectedexposure in man as a result of metabolic differences.Other questions relate to the extrapolation of plasma,brain, or cerebrospinal fluid (CSF) concentrations inanimals to the target concentrations in man givenspecies-dependent brain metabolic rates, CSF flow,and brain structure.

What Is an Adequate Animal Model?Current pharmaceutical research into disease

modification of AD is heavily reliant on transgenicmodels. There is a clear trade-off between the needfor such models to test the pharmacological effect ofcompounds versus the predictive value of suchmodels for both efficacy and safety purposes. Therecent experience with active immunization (Lam-bert et al., 2001) illustrates the lack of predictive valid-ity with regard to safety and the failure of three recentstroke studies illustrate the lack of predictive valid-ity with regard to efficacy. Some of the controversiessurrounding the β-amyloid transgenic models suchas the pathology of the plaques, the degree of inflam-mation, and behavioral deficits are likely to be

encountered with τ transgenic models or doubletransgenic models (Lewis et al., 2001). These modelsare based on mutations that cause very rare andfamilial forms of AD (amyloid precursor protein[APP] or PS-1 mutations [Raskind et al., 1995]) andfrontotempmoral dementia (FTD) (FTDP-17 muta-tions [Mack et al., 2001]). The degree to which oneshould rely on these models to predict efficacy in thesporadic form of AD remains unknown because thefamilial and sporadic forms of AD differ in naturalhistory and phenotype and by inference, pathology.

Clinical Trial DesignsTherapeutic strategies aimed at reducing path-

ology are inherently disease modifying in nature.Although we have acquired some expertise in thedesign, conduct, analysis, and interpretation of clin-ical trials for symptomatic treatments of AD, we havevirtually no experience with disease modification inCNS disorders. A number of hypotheses (see Table2) will have to be tested in phase II studies beforethere is sufficient data to justify large-scale clinicaltrials. It is not clear what clinical, surrogate, phar-macoeconomic, or regulatory end points wouldallow a claim of disease modification. Methodolog-ical hurdles in these types of clinical trials includethe very large sample sizes required to detect smalleffects on the rate of progression of AD or on the rateof conversion from mild cognitive impairment (MCI)to AD, the potentially confounding effects ofapproved or off-label use of medications that mayaffect AD pathology, the lack of a surrogate marker,and the reluctance of study sponsors to validate sur-rogates without some return on the investmentrequired for such validation.

ConclusionsTherapeutic approaches to AD that are based on

either β-amyloid or τ modulation share certaincommon challenges if they are to successfullymigrate from drug discovery into clinical develop-ment. There are likely to be challenges related to the

Table 2Phase II Studies

1. What AD population should be investigated first (MCI, mild, moderate, or severe)?2. Are the doses dependent on severity, duration of disease, age, gender, and apolipoprotein E (APOE)?3. What is the latency between dosing and effects on PD markers?4. Do we need to demonstrate an effect on more than one PD marker?5. What is a reasonable dosing period to allow PD markers to come to steady state?6. Is there a dose/exposure–response relationship?7. If CNS penetration is necessary, can we reach necessary levels and can we reach the appropriate areas?


334 Gold


safety profile of such compounds, as they involvealterations in highly conserved and multifunctionalsystems. There are challenges in translating datafrom preclinical models into testable hypotheses inman and there are clinical trial methodology chal-lenges. Clearly, we are at the threshold of a new erain CNS therapeutics and such a moment requires usto balance our enthusiasm with a healthy respect forthe vast amounts of knowledge we still lack.

ReferencesBillingsley M. L. and Kincaid R. L. (1997) Regulated phos-

phorylation and dephosphorylation of tau protein:effects on microtubule interaction, intracellular traf-ficking and neurodegeneration. Biochem. J. 323 (Pt. 3),577–591.

Brion J. P., Anderton B. H.. et al. (2001) Neurofibrillarytangles and tau phosphorylation. Biochem. Soc. Symp.67, 81–88.

Delaere P., Duyckaerts C., et al. (1989) Tau, paired helicalfilaments and amyloid in the neocortex: a morphome-tric study of 15 cases with graded intellectual status in aging and senile dementia of Alzheimer type. ActaNeuropathol. (Berl.) 77(6), 645–653.

Flaherty D. B., Soria J. P., et al. (2000) Phosphorylation ofhuman tau protein by microtubule-associated kinases:GSK3beta and cdk5 are key participants. J. Neurosci.Res. 62(3), 463–472.

Guttmann R. P., Erickson A. C., et al. (1995) Tau self-association: stabilization with a chemical cross-linkerand modulation by phosphorylation and oxidationstate. J. Neurochem. 64(3), 1209–1215.

Hagestedt T., Lichtenberg B., et al. (1989) Tau pro-tein becomes long and stiff upon phosphorylation:correlation between paracrystalline structure anddegree of phosphorylation. J. Cell Biol. 109(4 Pt. 1),1643–1651.

Lambert M. P., Viola K. L., et al. (2001) Vaccination with soluble Abeta oligomers generates toxicity-neutralizing antibodies. J. Neurochem. 79(3), 595–605.

Lewis J., Dickson D. W., et al. (2001) Enhanced neuro-fibrillary degeneration in transgenic mice expres-sing mutant tau and APP. Science 293(5534),1487–1491.

Mack T. G., Dayanandan R., et al. (2001) Tau proteins withfrontotemporal dementia—17 mutations have bothaltered expression levels and phosphorylation profilesin differentiated neuroblastoma cells. Neuroscience108(4), 701–712.

Raskind M. A., Carta A., et al. (1995) Is early-onsetAlzheimer disease a distinct subgroup within theAlzheimer disease population? Alzheimer Dis. Assoc.Disord. 9(Suppl. 1), S2–S6.

Steiner B., Mandelkow E. M., et al. (1990) Phosphoryla-tion of microtubule-associated protein tau: identifica-tion of the site for Ca2(+)-calmodulin dependent kinaseand relationship with tau phosphorylation inAlzheimer tangles. EMBO J. 9(11), 3539–3544.


Documents

Tau therapeutics for alzheimer’s disease