Safety and pharmacological characterization of the molecular tweezer CLR01 – a broad-spectrum inhibitor of amyloid proteins’ toxicity

Mice

All procedures were compliant with the National Research Council Guide for the Care and Use of Laboratory Animals, and approved by the University of California at Los Angeles (UCLA) Institutional Animal Care Use Committee. Two-month old WT C57BL/6J mice for toxicity studies were purchased from Jackson Laboratory (Bar Harbor, Maine, Stock 000664). 3×Tg and WT mice with the same genetic background [14] for BBB studies were bred at UCLA. Mice were housed 2–4 per cage under standard conditions and maintained on a 12-h dark and 12-h light cycle with ad libitum access to rodent chow and water.

CLR01

CLR01 was produced and purified as described previously [7]. ³H-CLR01 was prepared by Moravek Biochemicals (Brea, CA) using a method that provides ³H incorporation into the hydrocarbon skeleton (i.e., non-labile protons) [20] yielding pure ³H-CLR01 with specific activity 1.3 Ci/mmol.

Inhibition of tubulin polymerization

The effect of CLR01 on tubulin polymerization [21, 22] was analyzed using a commercial kit (Cytoskeleton, Inc., Denver, Colorado). Three mg/ml porcine brain tubulin (~18 μM) were allowed to polymerize at 37 ºC in the absence or presence of CLR01 concentrations ranging from 10–1,000 μM. The turbidity of the solution was measured as absorbance at λ = 340 nm using a Synergy HT microplate reader (BioTek, Winooski, VT). The data are an average of three independent experiments with two wells per condition.

Toxicity evaluation

For acute-toxicity studies, 2-m old C57BL/6J mice were administered saline-vehicle, 10 mg/kg, or 100 mg/kg CLR01 by a single intraperitoneal (IP) injection. The mice were sacrificed 24-h after the injection. For chronic toxicity studies, 2-m old C57BL/6J mice were administered saline-vehicle, 3 mg/kg, or 10 mg/kg CLR01 by daily IP injection for 30 days. Acute-study mice were visually monitored for 1 h after injection and then every 50 min for 10 min over the first 6 h of the experiment for changes in activity and behavior. The mice also were monitored every 110 min for 10 min during the last 6 h of the experiment until they were sacrificed. Chronic-study mice were monitored for 1 h after injection and then 3 times a day for 10 min each, every day of the first week. During that week there were no appreciable changes in the behavior, appearance, or weight of the mice. Therefore, monitoring was reduced to twice a day during the remainder of the experiment. On all occasions, the mice were monitored for any signs of severe toxicity, including bruising or bleeding, pale mucous membranes or extremities, diarrhea, dehydration, neurological signs, such as difficulty ambulating or paralysis, tachypnea or dyspnea, or abdominal distension.

Following the treatment, mice were anesthetized with pentobarbital and blood was collected by cardiac puncture and placed in tubes containing a clot activator for serum separation (Capiject T-MG tubes, Terumo Medical Products, Somerset, NJ). Then, the lungs were filled through the trachea with 4% paraformaldehyde to prevent collapse, and tissues (brain, heart-lung, liver, kidney, and spleen) were collected and fixed for 72 h in 4% paraformaldehyde at a ratio of ~1:10 tissue:fixative (v/v). Tissues then were transferred into a 70%-ethanol solution and transferred to the UCLA Mouse Pathology Core for paraffin embedding, sectioning, and tissue histopathology analysis. Serum was analyzed by the UCLA Division of Laboratory Animal Medicine (DLAM) Animal Serology & Molecular Diagnostic Laboratory for an 11-panel serum chemical analysis using the ACE Alera Clinical Chemistry system (Alfa Wassermann Diagnostic Technologies, West Caldwell, NJ). The panel included: alanine aminotransferase, aspartate aminotransferase, albumin, alkaline phosphatase, creatinine, total bilirubin, lactate dehydrogenase, blood urea nitrogen, cholesterol, total protein, and glucose.

Plasma concentration and blood–brain barrier permeability

For studies of plasma concentration, CLR01 was administered by either SC or intravenous (IV) injection at 1 mg/kg, or by oral gavage at 10 mg/kg, and plasma was collected at time points between 0.33 - 24 h. Three mice were used per time point. The concentration of CLR01 in plasma was determined by Wolfe Laboratories Inc. (Watertown, MA) using liquid chromatography-mass spectrometry (LC-MS) by interpolation of sample peak area data into the calibration curve.

The following groups of mice were used for CLR01 BBB penetration studies: 3×Tg and the corresponding WT mice at 2-m, 12-m, and 22 - 24-m (hereafter referred to as 22-m) of age. The groups were: 2-m WT, 2-m Tg, 12-m WT, 12-m Tg, 22-m WT, 22-m Tg. Mice were administered ³H-CLR01 intravenously. Two μCi per gram of mouse body weight, which are equal to 11.86 μg/g of CLR01 in which ³H-CLR01 made up 10% of the total CLR01, were injected into the jugular vein. Blood and brain were collected at 0.5, 1, 3, 8, 24, or 72 h (not all time points were collected for all groups, see the Results section). For times ≤ 3 h, mice were anesthetized by IP injection of ketamine and xylazine. The mice remained anesthetized following the injection until the specified time point, at which point they were given a lethal dose of pentobarbital. Then, blood was collected via a cardiac puncture, and the brain harvested with or without a perfusion step Pharmacokinetics of CLR01 in vivo. For time points 8–72 h, mice were not anesthetized and ³H-CLR01 was injected into the tail vein. This change was due to the difficulty of keeping mice anesthetized for longer than 3 h. No differences were observed between mice given anesthesia and jugular vein injections and mice receiving tail vein injections.

Euthanasia procedures were the same as described above. For all mice, one hemisphere of the brain and 100–350 μl of blood were separately digested following instructions from Perkin-Elmer (document: Scintillation Cocktails and Consumables) with 1 ml Solvable (Perkin-Elmer, Waltham, MA), added to Ultima Gold Liquid Scintillation Cocktail (Perkin-Elmer) and read in a Triathler Liquid Scintillation Counter model 425–034, (Hidex, Turku, Finland). Brain permeability percentage was calculated as counts per minute (CPM) per g of brain relative to CPM per ml of blood. The data are an average of values from three mice per genotype/age/time combination.

For CLR01 transport-saturation studies using 5× the CLR01 dose, ³H-CLR01 was kept at 10% of the total CLR01 mixture and a total of 59.3 μg of CLR01 (10 μCi) per g of mouse body weight was injected IV (22-m WT 5× dose). For CLR01 brain-accumulation studies, two 11.86-μg/g injections were administered at equal time intervals (22-m WT 2× inj). In these experiments, mice were injected at time = 0 and at t = ½ of euthanasia time. For example, in the original, single-injection experiments, a mouse would receive an injection at t = 0 and then be euthanized at t = 1 h. In this experiment, a mouse received one injection at t = 0, a second injection at t = 0.5 h, and then was euthanized at t = 1 h.

In experiments using ³H-CLR01, urine was collected when possible over the period between injection and euthanasia. In all cases, we found that the urine was radioactive. However, comparison among mice proved to be difficult. We could not normalize the radioactivity because the amount of urine in the bladder prior to injection and the volume produced during the experiment could not be calculated. Thus, we can simply conclude qualitatively that CLR01 is excreted through the urine, but cannot provide quantitative measures of what percentage of the compound is excreted this way.

In vitro metabolism

Potential dephosphorylation of CLR01 was analyzed by incubating 100 nmol CLR01 with 0.08 units of alkaline phosphatase (ALP; calf intestinal alkaline phosphatase, Promega, Madison, Wisconsin) for 60 min at 60°C. One enzymatic unit is defined as the amount of enzyme required for catalyzing the hydrolysis of 1 μmol of p-nitrophenylphosphate per minute. p-Nitrophenylphosphate disodium salt at 5–50 nmol (Fisher, Waltham, Massachusetts) was used for generation of a standard curve. The amount of inorganic phosphate generated was measured spectrophotometrically using an EnzChek Phosphate Assay kit (Life Technologies, Carlsbad, California) according to the manufacturer’s instructions on a DU-640 spectrophotometer (Beckman Coulter, Brea, California) at λ = 360 nm. Baseline values were subtracted from readings and compared to the standard curve resulting from serial ALP reactions to calculate the amount of inorganic phosphate. Similarly, potential dephosphorylation of CLR01 by brain homogenates was measured. For these experiments, one brain hemisphere was homogenized by sonication in the presence of cOmplete protease-inhibitor cocktail (Roche, Penzberg, Germany). Protein concentration was measured using a BCA Protein Assay Kit (Pierce, Rockford, Illinois). A “phosphate-mop” system was used according to the EnzChek Phosphate Assay kit instructions to sequester inorganic phosphates naturally present in 1.5 mg of brain, and then 50-nmol CLR01 or different concentrations of p-nitrophenylphosphate disodium salt were added and incubated for 60 min at 60°C.

Statistics

Data are shown as mean ± SD or mean ± SEM as appropriate. Statistical analysis was performed using Prism 6.0c (GraphPad, La Jolla, CA). For all experiments, 2-way analysis of variance followed by Sidak’s multiple comparisons test post-hoc analysis were used. The level of significance was set at p < 0.05.

In vitro examination of the process-specific mechanism of CLR01

As stated above, the mechanism by which CLR01 remodels the assembly of amyloidogenic proteins into non-toxic assemblies that can be degraded by normal clearance mechanisms is by its specific binding to Lys residues. The mechanism is “process-specific” because it is postulated to affect only the aberrant assembly of proteins that leads to toxic oligomers and aggregates, but not normal protein structure, function, or assembly as happens, e.g., in tubulin polymerization. To test whether this indeed is the case, we examined the effect of CLR01 on tubulin polymerization [21, 22]. Three mg/ml (~18 μM) porcine brain tubulin, which contains 3.8% Lys and 4.8% Arg, was allowed to polymerize in the absence or presence of CLR01 concentrations ranging from 10–1,000 μM.

In the absence of CLR01 or in the presence of up to 300 μM of the compound, the change in turbidity followed a typical sigmoidal curve, starting at 0.05-0.09 absorbance units (Figure 2). The absorbance remained unchanged for the first 10–15 minutes, which is a typical lag phase in this reaction, and then increased gradually up to ~60 min, at which point the rate of increase began to decline, and the reaction was followed for another 10 min. The only concentration at which significant modulation of the polymerization was observed was 1,000 μM (Figure 2, blue curve), i.e., at a tubulin:CLR01 concentration ratio ~1:55. At this high ratio, a high absorbance, 0.15, was observed immediately, followed by a slight gradual decline during the lag phase. Then, the absorbance began to increase for 30 min, followed by a slow decline for the rest of the experiment. One interpretation of these data is that at the high concentration used, 1,000 μM, binding of CLR01 to tubulin induced immediate self-assembly into irregular aggregates. Similar immediate induction of self-assembly was observed with 4 of the 9 amyloidogenic proteins tested by Sinha et al. [3], suggesting that this reaction occurs with some, but not all proteins. In all the cases studied by Sinha et al., these aggregates were non-amyloidogenic and non-toxic.

Presumably, following the immediate aggregation in the presence of 1,000 μM CLR01, the irregular tubulin aggregates observed at t = 0 partially disassembled as the polymerization reaction progressed, between 10–30 min. At that point, the high CLR01 concentration appeared to interfere with the polymerization reaction and the tubulin polymers gradually disassembled. Validation of this interpretation will require further investigation, yet it was not the focus of the current study.

The motivation for this experiment was to test whether the concentration of CLR01 needed to interfere with a controlled self-assembly process was substantially different from that required for modulation of aberrant self-assembly, which was found indeed to be the case. Most of the protein:CLR01 concentration ratios needed for inhibition of amyloidogenic protein aggregation were in the range 1:1–1:3 [3], compared to the 1:55 tubulin:CLR01 concentration ratio at which disruption of tubulin polymerization was observed. These results support the specificity of CLR01 for inhibition of aberrant aggregation as opposed to controlled polymerization.

CLR01 safety

If CLR01 indeed operates by a process-specific mechanism and remodels the abnormal aggregation of amyloidogenic proteins at substantially lower concentrations than concentrations that would perturb normal physiological processes, one would expect the compound to have a high therapeutic index. To calculate the therapeutic index, a lethal dose must be reached. The in vitro data described above suggested that disruption of tubulin polymerization occurs at concentration ratios 20–50 times higher than those needed for inhibition of aggregation of amyloidogenic proteins. In addition, cell culture experiments indicated that CLR01 began to show toxicity at concentrations 1–3 orders of magnitude higher than those required for inhibition of toxicity by different amyloidogenic proteins [3, 4]. The next rational step was to test the safety margin of CLR01 in vivo. Based on the in vitro and cell culture data, we expected that 100 mg/kg would be lethal to mice and therefore used it as the highest dose in our safety-evaluation experiments. We assessed the safety of CLR01 in 2-m old, male, WT mice either 24 h following a single IP injection of 10 or 100 mg/kg (acute administration) or after daily IP injection of 3 or 10 mg/kg for 30 days (chronic administration). Following euthanasia, serum was collected for chemical analysis and tissues were harvested for histopathology evaluation.

All CLR01-treated groups, except for the 100-mg/kg acute-administration group, behaved indistinguishably from control mice in terms of levels and type of activity and grooming. The administration of 100-mg/kg CLR01 caused obvious signs of distress immediately, which lasted for ~30 min following the injection. For most mice, activity level decreased and eyelids became droopy. Some of the mice exhibited arching of the back, sporadic gasping, lying down, dragging one leg, and twitching. These signs of distress diminished after the first 30 min, at which point the mice resumed grooming and sitting on hind legs. Some mice showed decreased activity and droopy eyelids for up to 2 h following the injection. No symptoms of severe toxicity, as defined by the UCLA DLAM veterinarians, were observed for any mice, including bruising, bleeding, pale mucous membranes or extremities, diarrhea, paralysis, tachypnea or dyspnea, or abdominal distension.

Liver, kidney, spleen, heart, lung, and brain were collected for histopathology analysis. Tissue samples from heart, lung, spleen, and brain of all acutely CLR01-administered mice were indistinguishable from those of control mice. In all 100-mg/kg-dosed mice and one of eight 10-mg/kg-dosed mice of the acute-administration groups, liver degeneration and necrosis was detected in centrilobular and midlobular regions (Figure 3). Zonal nature of liver toxicity is common in drug-toxicity studies and was expected in the high-dose group.

The fact that all the mice in the high-dose group survived meant that the actual therapeutic index could not be calculated because contrary to our expectation, 100 mg/kg was under the lethal dose. However, we considered the observation of obvious liver toxicity at this high dose as sufficient for determining the maximal dose in future efficacy experiments and therefore did not treat mice with higher doses. Rather, we conducted next a 30-day, chronic-toxicity experiment in which mice were administered IP either 3 or 10 mg/kg/day of CLR01. Because one mouse of the eight used in the 10-mg/kg acute-administration group showed signs of liver toxicity, 10 mg/kg/day was chosen to be the highest dose in this experiment.

Heart, lung, spleen, and brain from both chronically CLR01-treated groups of mice were indistinguishable from vehicle-treated mice and were free of signs of malformation, degeneration, necrosis, or inflammation within normal variability among mice. A few mice in the 3-mg/kg group showed signs of mild-to-moderate multifocal extramedullary hematopoiesis in the liver. The consulting veterinary pathologist concluded that this was possibly immune-stimulated but not pathogenic. Mild pancreatitis also was observed in one of the mice showing liver hematopoiesis and one additional mouse in the 3-mg/kg group. In contrast, no signs of tissue pathology or liver necrosis were detected in any of the mice in the 10-mg/kg dosed group. Thus, it is unlikely that the hematopoiesis or inflammation found in the low-dose group were related to the CLR01 treatment.

Pharmacokinetics of CLR01 in vivo

The plasma concentration of CLR01 was measured by LC-MS in 2-m old WT mice following administration by a SC or IV injection or by oral gavage. The SC bioavailability was found to be identical, within experimental error, to the IV administration, which was considered as 100% bioavailable (Figure 4). In both routes, ~30% of the administered dose was detected in the blood at the earliest time point measured – 20 min, and the plasma half-life was found to be ~2.5 h. Approximately 5% of the initial CLR01 levels were found in the plasma 8 h following either SC or IV administration. Oral bioavailability was negligible, suggesting that CLR01 gets metabolized in the gastrointestinal tract and/or does not pass from the gut to the blood.

Next, we asked what percentage of the administered CLR01 penetrates through the BBB and gets into the CNS. Our first attempt was to measure CLR01 in brain extracts using LC-MS. However, this proved to be difficult. Due to the multiple negative charges of CLR01, its partial protonation at physiologic pH, and the presence of various counter-ions in biological fluids, the MS signal splits into multiple peaks resulting in low signal-to-noise ratio. The difficulty to observe the CLR01 signal in brain extracts using LC-MS suggested that the concentration was low and detection would necessitate considerable optimization of the extraction and LC-MS methods, which would require substantial effort and high costs. Therefore, we decided to test first whether CLR01 could be found in the CNS by using a radiolabeled derivative of the compound.

As the permeability of the BBB has been shown to be dependent on age and morbidity, and in particular to be increased in AD [17] and in mouse models of AD [24, 25], we assessed how age and disease progression affected the brain penetration of CLR01 by using WT and 3×Tg mice at three different ages. The 3×Tg model was chosen because it was used in a previous study, in which CLR01 was found to reduce AD-like pathology in the brain [5]. Mouse ages were chosen to correspond with: 1) a stage before Aβ burden and cognitive deficits are found at 2-m of age [14, 26]; 2) a stage with mild-to-moderate plaque and tangle pathology but with observable memory deficits at 12-m of age [14, 27]; and 3) a stage of abundant plaque and tangle pathology with consistent behavioral deficits at 22-m of age [28]. Mice were administered ³H-CLR01 IV, blood and brain were collected at time points between 0.5 - 72 h following CLR01 administration, and radioactivity levels were measured by liquid scintillation counting. Radioactivity is presented as CPM/g of brain or CPM/ml of blood.

To correct for the radioactivity associated with blood-borne ³H-CLR01 in the brain vasculature, we performed both perfusion and subtraction analyses. In perfusion experiments, WT and 3×Tg mice at each of the three ages analyzed (n = 3 per group) were perfused with phosphate buffered saline following euthanasia. Perfusion lasted for either 5 min or until the liver changed color from a red to yellow, whichever was longer. In other experiments, mice were not perfused, but radioactivity associated with 10 μl of blood per g of brain [29, 30] was calculated based on brain weight and blood radioactivity levels and subtracted from brain radioactivity levels. At 1 h post injection, perfusion-corrected brain values were statistically similar to subtraction-corrected brain values (Figure 5). Due to difficulties associated with the perfusion analysis, specifically liver color being used as an indirect readout of brain perfusion level, and because including a perfusion step could increase variability among experiments, the rest of the experiments utilized the subtraction method, which is a common practice in BBB-permeability studies [29, 30].

At 0.5 h following injection, blood radioactivity levels in 12-m old mice were 39 ± 13% and 40 ± 6% of the injected levels, for WT and 3×Tg mice, respectively. These values were in agreement with the CLR01 concentration levels detected in plasma by LC-MS. About 5 - 10% of the radioactivity observed at time 0.5 h remained in the blood after 8 h (Figure 6).

Brain-radioactivity levels, calculated as a percentage of blood-radioactivity levels (CPM/g)/(CPM/ml) at 1 h following the injection ranged from 0.86–3.09% depending on age and genotype (WT versus 3×Tg, Figure 7). Analysis of brain penetration levels at 1 h by absence or presence of AD transgenes and by age showed a statistically significant effect of age but not of genotype. Interestingly however, 2-m old 3×Tg mice significantly differed from 12-m and 24-m old 3×Tg mice (2-m: 3.09 ± 0.55%; 12-m: 1.43 ± 0.17%; 24-m: 1.45 ± 0.28%; p < 0.05), whereas in the WT group, the only significant difference was between the 2-m and 24-m old mice (2-m: 2.68 ± 0.31%; 12-m: 2.11 ± 0.69%; 24-m: 0.86 ± 0.17%; p < 0.05). This suggests that changes in BBB permeability occur earlier and more sharply in 3×Tg mice compared to WT mice.

Surprisingly, although blood radioactivity levels declined rapidly (Figure 6), the radioactivity levels measured in the brain did not change significantly up to 72 h post-injection (Figure 8). Brain radioactivity levels were insensitive to genotype or time after injection and thus the 24-h time point was assessed only in the 22-m old mice (both 3×Tg and WT) and the 72-h time point was assessed only in the 22-m old WT mice. Differences were statistically insignificant and within experimental error.

To explore further the mechanics of CLR01 transport across the BBB, we asked whether the transport system was saturated. To answer this question, we injected 5-times the amount of total CLR01, keeping the ratio of ³H-CLR01:CLR01 at 1:9, into 22-m old WT mice. This experiment resulted on average, in 5-times the absolute amount of radioactivity detected in the brain. The percentage of brain penetration at 1 h following the injection did not change significantly (1× CLR01 brain penetration: 0.86 ± 0.30% of blood; 5× CLR01 brain penetration: 0.97 ± 0.28% of blood; Figure 8). This result suggests that the transport mechanism, whether active or passive, is concentration-dependent because there was an increase in the absolute value but not the relative value of CLR01 entering the brain.

To begin to explore whether additional dosing would increase the effective CLR01 concentration in the brain, we injected 22-m old WT mice twice over two equal time intervals and compared brain levels to mice that received one injection. On average, over the 1-, 3-, and 8-h time points measured, the amount of radioactivity found in the brain following the double-injection was twice the amount measured following the single-injection protocol (1 h: 3.3× compared to one injection, 3 h: 1.6×, 8 h: 1.9×; Figure 8). These data suggest that upon continuous dosing, as with the SC osmotic mini-pumps used in our previous efficacy study [5], CLR01 could reach sufficiently high brain concentration levels to inhibit Aβ aggregation even though the dose was relatively low — 40 μg/kg/day – when brain penetration levels are taken into account (see Discussion).

In vitro catabolism of CLR01

The BBB permeability experiments described above used radioactivity as an indirect readout of CLR01 concentration levels, which could have reflected the parent compound, CLR01 itself, or its metabolites. The question of the source of radioactivity seemed particularly important in view of the surprising persistence of radioactivity attributed to CLR01 in the brain. The most likely metabolism of CLR01 is cleavage of one or both phosphate groups resulting in monophosphate and hydroquinone derivatives, respectively (Figure 9). Each such dephosphorylation would decrease the polarity of the compound and increase its potential partition into the lipophilic brain parenchyma environment relative to the blood. In particular, the hydroquinone product is insoluble in aqueous solutions, in contrast to CLR01 and its monophosphate metabolite, which are soluble at millimolar concentrations. Thus, double dephosphorylation could result in precipitation and accumulation of the hydroquinone in the brain, potentially leading to misinterpretation of the BBB permeability data. Complete analysis of CLR01 metabolism in the brain was beyond the scope of the study described here. However, to evaluate the potential for dephosphorylation, we incubated CLR01 in vitro with ALP or brain extracts and measured the release of inorganic phosphate.

ALP is a widely distributed plasma enzyme found in many tissues which can be released into body fluids [31]. The enzyme received its name because it shows optimal activity at pH ~9. There are four isoforms of ALP: intestinal, placental, germ cell, and tissue non-specific. All four isoforms are non-specific enzymes that catalyze the hydrolysis of a wide range of phosphate esters [32]. Tissue non-specific ALP concentration levels increase in both brain and plasma of patients with familial or sporadic AD relative to age-matched healthy individuals [33], possibly as a compensatory mechanism because the enzyme catalyzes tau dephosphorylation [34].

Because of its promiscuous hydrolysis activity, we tested whether calf intestinal ALP catalyzed CLR01 dephosphorylation by incubating the molecular tweezer with ALP and comparing the amount of inorganic phosphate released to a standard curve obtained by incubating ALP with increasing concentrations of a common substrate, p-nitrophenylphosphate. This standard curve had a detection sensitivity limit of 5 nmol. Incubation of up to 100 nmol CLR01 with ALP resulted in undetectable levels of inorganic phosphate, suggesting that despite its promiscuity, ALP did not catalyze dephosphorylation of CLR01.

To test whether CLR01 dephosphorylation might be catalyzed by brain phosphatases other than ALP, we incubated 50 nmol CLR01 with 1.5 mg of mouse-brain homogenate. The brain homogenate dephosphorylated the positive control substrate, p-nitrophenylphosphate, at 99 - 130% of the activity of 0.8 enzymatic units of ALP. In contrast, similarly to the reaction with ALP, no release of inorganic phosphate was detected when the brain homogenates were incubated with CLR01 under the same conditions. Based on these results, dephosphorylation of CLR01 likely did not happen in our BBB permeability experiments and the radioactivity measured in mouse brains plausibly reflected CLR01 itself.