Technical Comments

Response to Comments on “ApoE-Directed Therapeutics Rapidly Clear β-Amyloid and Reverse Deficits in AD Mouse Models”

Science  24 May 2013:
Vol. 340, Issue 6135, pp. 924
DOI: 10.1126/science.1234114


The data reported in the Technical Comments by Fitz et al., Price et al., Tesseur et al., and Veeraraghavalu et al. replicate and validate our central conclusion that bexarotene stimulates the clearance of soluble β-amyloid peptides and results in the reversal of behavioral deficits in mouse models of Alzheimer’s disease (AD). The basis of the inability to reproduce the drug-stimulated microglial-mediated reduction in plaque burden is unexplained. However, we concluded that plaque burden is functionally unrelated to improved cognition and memory elicited by bexarotene.

The data provided in the Technical Comments (14) replicate and validate the principal conclusions of our work published in Cramer et al. (5). Specifically, we concluded that the retinoid X receptor (RXR) agonist bexarotene promoted the apolipoprotein E (ApoE)–dependent clearance of soluble β-amyloid (Aβ) peptides from the brain, and the reduction in Aβ levels was associated with cognitive improvement in murine models of Alzheimer’s disease (AD). In each of the Comments, the investigators report that bexarotene treatment acted on astrocytes to elicit the expression of ApoE and ABCA1. The induction of these genes leads to the production of ApoE-containing high-density lipoprotein (HDL) particles, which, in turn, promotes the proteolytic degradation of soluble Aβ peptides (6).

Importantly, Fitz and colleagues report the effect of oral bexarotene treatment of two independent lines of APP/PS1ΔE9 mice in which the human APOE3 or APOE4 gene was knocked into the murine locus. They found that drug treatment resulted in the reduction in soluble brain Aβ levels in interstitial fluid and improved behavior in two different behavioral tasks. A new finding in this study was that bexarotene treatment also reduced the levels of Aβ oligomers by ~50%, suggesting the preferential clearance of these species. Notably, these studies provide direct evidence that bexarotene-mediated induction of the human APOE isoforms is associated with enhanced clearance of soluble forms of Aβ in interstitial fluid, consistent with the findings reported in Cramer et al. (5). Strikingly, Veeraraghavalu et al. demonstrate that bexarotene treatment resulted in a reduction of soluble Aβ species, and the reported effect sizes were greater than those reported by us or by Fitz et al. Lastly, Tesseur et al. found that bexarotene treatment resulted in behavioral improvement, even in compromised mice.

The principal issue raised in the Comments is the inability to replicate the bexarotene-stimulated, microglial-mediated reduction in plaque burden reported in Cramer et al. We found that bexarotene treatment of two different mouse models of AD resulted in reduction in plaque burden, reflective of the stimulation of microglia phagocytosis of deposited forms of amyloid. We reported the efficient clearance of amyloid deposits in 6-month-old APP/PS1ΔE9 mice [figure 2 in (5)]. However, reduction in plaque levels was less efficient in older APP/PS1ΔE9 mice [figure S4 in (5)] and in a more aggressive amyloidogenic model [figure S6 in (5)]. We reported that upon chronic treatment of APP/PS1ΔE9 mice with bexarotene, there was no plaque loss [figure S5 in (5)]. Importantly, improved cognition and memory was consistently observed in all the murine models tested. We concluded, “the behavioral improvements were poorly correlated with the microglial-mediated removal of insoluble, deposited forms of Aβ.”

There is a clear discrepancy between the outcomes reported by us and those contained within the Comments on the ability of bexarotene to reduce plaque burden, and it is important to understand their basis. The ability of nuclear receptor agonists to induce changes in microglial phenotype and to stimulate phagocytosis is well documented, but mechanistically unclear (7). The conversion of macrophages/microglia into “alternative activation” states is the subject of intense investigative interest, and exactly how nuclear receptors act to govern this phenotypic conversion is poorly understood (8). The phenotypic changes in microglia and macrophages in response to nuclear receptor agonists are reliant upon a coordinated transcriptional response involving both transrepression of inflammatory genes and transactivation of genes associated with phagocytosis and tissue repair. These changes in microglial gene expression exhibit a much different drug dose and time dependency than those required for the induction of the astrocytic reverse cholesterol transport genes (9). Evaluation of the microglial response to nuclear receptor agonists in vivo is hampered by the lack of robust and reproducible measures of drug action in this cell type. The variable effect of the RXR agonist bexarotene on microglial phenotype and induction of phagocytosis is directly analogous to that observed with agonists of the liver X receptor (LXR). There have been six published papers investigating the actions of LXR agonists in mouse models of AD (6, 1014). These studies report behavioral improvement after drug treatment. However, plaque reduction varied from 0 to 65%. This led to the conclusion that plaque burden was not correlated with cognition and memory. The basis for the inconsistent induction of microglial phagocytosis by nuclear receptor agonists is unclear but appears to vary as a function of age, genotype, diet, drug administration schedules, and other unknown variables. For example, the absorption efficiency of bexarotene varies dramatically (by up to 48%) depending on dietary fat content (15, 16). Thus, one or more of these variables may explain the differences in plaque clearance observed in Cramer et al. and in the studies within the Comments. However, it is important to note that Price et al. mischaracterize the principal conclusions from our studies, stating that clearing Aβ deposits was a “key endpoint used to justify human studies.” Rather, we explicitly stated that plaque burden was poorly correlated with improved cognition and memory. Indeed, these conclusions are similar to those recently published by these same authors (17). It is also notable that Heneka and colleagues have recently reported that bexarotene induces microglial phagocytosis of Aβ and that this effect is reliant upon RXRα (18), directly supporting the findings reported in Cramer et al.

A critical difference in the studies reported by Price et al., Tesseur et al. and Veeraraghavalu et al. is that these investigators used unconventional formulations of bexarotene, which alters its pharmacokinetics and bioavailability, and this may underlie the reported differences in experimental outcomes. Targretin is a micronized form of bexarotene delivered orally in an aqueous vehicle to preserve its microcrystalline structure. The pharmacokinetics of orally administered micronized bexarotene are well described in rodents, dogs, and humans (15, 19). The utility of this formulation is documented in the U.S. Food and Drug Administration (FDA) filing (15) in ~80 studies in rodents and more than 200 clinical studies. To verify appropriate administration of the drug, we performed a pharmacokinetic study after a single oral dose of either 25 or 100 mg per kg of weight (mg/kg) of bexarotene (Fig. 1) and obtained results that were similar to those reported in the literature (19) and in the FDA filing (15). We observed time to maximum drug concentration (Tmax) of 1 to 2 hours with dose-proportionate levels of bexarotene in plasma and brain (Fig. 1). Thus, bexarotene is fully blood-brain-barrier permeant. We have never observed weight loss or other negative effects of drug treatment, consistent with the extensive literature on this formulation of bexarotene.

Fig. 1

Bexarotene pharmacokinetics. Mice (C57BL/6) were orally gavaged with bexarotene (Targretin) suspended in water and delivered at a final dose of 100 mg/kg. The brains were collected at the indicated intervals between 0 and 8 hours and were then homogenized in phosphate-buffered saline (PBS) (2:1, PBS:brain by weight). Bexarotene and an internal standard were extracted from plasma (open square) or brain (open circle) homogenate by protein precipitation and processed for liquid chromatography–tandem mass spectrometry analysis. The kinetic analysis of bexarotene levels in plasma and brain is shown in (A) and tabulated in (B). Elimination half-life values for plasma and brain were similar and were also similar between dose groups. Within each dose level, exposure (Cmax and AUC0-∞) was similar between plasma and brain. Plasma and brain exposure (Cmax and AUC0-∞) increased in a linear manner with increasing dose. The 4-fold increase in dose from 25 to 100 mg/kg, Cmax increased 4.2-fold for plasma and 4.8-fold for brain, and AUC0-∞ increased 3.7-fold for plasma and 3.4-fold for brain.

It is unclear to us why these investigators chose to first solubilize the drug, which differs from the formulation employed by Cramer et al. and alters its known pharmacokinetic properties. This is clearly demonstrated by Tesseur et al., who report that formulation of the drug in a cyclodextrin vehicle resulted in frank toxicity. They report plasma and brain drug levels in mice 20-fold greater than those we observed (at 7 to 8 hours) (Fig. 1). Similarly, administration of bexarotene to dogs in a different cyclodextrin vehicle resulted in high and non–dose proportionate levels of bexarotene in plasma. This was associated with a 2 to 5-fold increase in drug half-life (15), reflecting the effect of high drug levels on saturation of drug elimination mechanisms. The latter is a known mechanism of toxicity for many drugs (20, 21); thus, it is not surprising that they observed overt toxicity and weight loss. Moreover, the high and sustained levels of bexarotene in the mouse and dog studies may have led to compensatory changes in RXR-modulated pathways that muted the effects of the drug. The extremely high drug levels and overt toxicity preclude any interpretation of the presence or absence of biological effects. The pharmacokinetics and bioavailability of bexarotene in the studies reported in Price et al. and Veeraraghulau et al. are unknown. The statement by Tesseur et al. that the toxicity of bexarotene cautions against testing in AD patients is not supported by the clinical experience with this drug. The FDA-approved formulation of bexarotene, Targretin, is safe, even in the elderly, as evidenced by its chronic use by thousands of individuals since 1999. It is this formulation of the drug that is employed in phase I and II clinical trials that are currently under way.

References and Notes

  1. Acknowledgments: This work was supported by the Blanchette Hooker Rockefeller Foundation, Thome Foundation, Roby and Taft Funds for Alzheimer’s Research, Painstone Foundation, American Health Assistance Foundation, Coins for Alzheimer’s Research Trust, and the National Institute on Aging (NIA) (grant AG030482-03S1 to G.E.L.); National Institute on Deafness and Other Communication Disorders (grant DC003906, RO1-AG037693 to D.A.W); NIA (grants K01 AG029524 and P50-AG005681), Shmerler family, and the Charles F. and Joanne Knight Alzheimers Disease Research Center at Washington University (to J.R.C.); and Marian S. Ware Alzheimer Program (to K.R.B.). G.E.L. is an officer of ReXceptor, Inc., a company designed to further RXR agonists as treatments for AD. P.E.C. was on the board of ReXceptor, Inc. until October 2012. G.E.L. and P.E.C. are listed as inventors on a patent application by Case Western Reserve University involving the use of bexarotene in Alzheimer’s disease.
View Abstract


Navigate This Article