Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The treatment of central nervous system (CNS) diseases with antisense therapy holds great promise. Delivery of antisense to tissues and especially to the CNS is, however, problematic. Phosphorothiolate oligodeoxynucleotides are more enzymatically stable and can even be absorbed orally (Agrawal et al., 1995). Several studies have shown phosphorothiolate oligonucleotides are taken up by many tissues, but the brain is not usually included in these studies (Zhang et al., 1995; Sands et al., 2000). It has been assumed that the blood-brain barrier (BBB) prevents blood-borne oligonucleotides from leaving the brain's vascular space to enter the CNS (Boado et al., 1998). Two studies have examined the question of whether phosphorothiolate oligonucleotides might be taken up by brain (Agrawal et al., 1991; Cossum et al., 1993). One study found the brain had a level of radioactivity lower than any other tissue and the other found the brain took up less than 1% of the injected dose. Neither study determined whether the radioactivity in brain represented intact oligonucleotide nor whether the radioactivity had crossed the BBB or was merely trapped in the vascular space of the brain.

Alzheimer's disease is an example of a condition that could be treated with antisense therapy. An increase in brain levels of amyloid beta protein (A) may be the cause of Alzheimer's disease (Selkoe, 1990). This increase could be due to overexpression of amyloid precursor protein (APP), altered cleavage of APP to A, or decreased clearance of A from the CNS. Decreasing expression of APP with antisense should reverse the increase in brain levels of A caused by any of these mechanisms.

A phosphorothiolate antisense oligonucleotide directed toward the A midregion of APP (Olg) given by i.c.v. administration can reverse the learning and memory deficits seen in aged SAMP8 mice (Kumar et al., 2000). This strain of mouse has spontaneously mutated to overexpress APP as it ages (Nomura et al., 1996; Kumar et al., 2000; Morley et al., 2000). Levels of A are increased about 2-fold by age 12 months (Kumar et al., 2000; Morley et al., 2000). This is much closer to the 50% increase seen in Alzheimer's disease (Rosenberg, 2000) than the 5- to 14-fold increase seen in transgenic mice (Hsiao et al., 1996). SAMP8 mice are normal at 4 months of age, but at 12 months have developed severe deficits in learning and memory (Flood and Morley, 1993, 1998), which are reversed with antibody directed at A (Morley et al., 2000). Olg, but not a random nor two other A-directed phosphorothiolate antisense oligonucleotides, reversed the deficit in learning and memory and decreased the levels of C-terminal APP in hippocampus, amygdala, and septum (Kumar et al., 2000). Here, we determined whether Olg injected intravenously could enter the CNS and reverse the learning and memory deficits in aged SAMP8 mice.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Synthesis. Phosphorothioate oligodeoxynucleotides were custom synthesized by The Midland Certified Reagent Co. (Midland, TX) by use of cyanoethyl phosphoramidite chemistry. Protecting groups were removed by hydrolysis with concentrated ammonium hydroxide and the oligonucleotides purified by reverse phase high-performance liquid chromatography. The trityl-positive fractions were pooled, dried, detritylated, and the oligonucleotides separated from the trityl group by gel filtration. The oligonucleotides were dried and stored as their ammonium salts.

Labeling. The oligonucleotide representing the antisense to A_17-30 (Olg), 5'-(_P=S)GGCGCCTTTGTTCGAACCCACATCTTCAGCAAAGAACACCAG-3', was end labeled by mixing 1 to 5 µg of Olg in 70 mM Tris-HCl (pH 7.6) with 10 mM MgCl₂, 5 mM dithiothreitol, 100 µCi of [³²P]ATP, and 10 units of T₄ polynucleotide kinase in a total volume of 15 µl. The mixture was incubated for 45 min at 37°C in a water bath. At the end of the reaction, the kinase was inactivated by heating the sample to 65°C for 5 min. Unreacted radioactivity was separated from labeled Olg (P-Olg) with a G-50 Sephadex spin column made in a 3-ml plastic disposable syringe. The purity of the sample was assessed by running about 2000 Cerenkov counts on a 5% polyacrylamide gel. The specific activity of P-Olg was determined by counting a known amount of sample in a Beckman scintillation counter and was calculated to be 550 nmol/Ci. Since the molecular mass of Olg was 13,431 Da, there were about 345 pg/10⁵ cpm.

Tissue Distribution. Male ICR mice (Charles River, Wilmington, MA) weighing 20 to 25 g were anesthetized with an i.p. injection of urethane and the left jugular vein and right carotid artery were exposed. Mice were given an injection into the jugular vein of 0.2 ml of lactated Ringer's solution containing 1% by volume of bovine serum albumin (LR-BSA) and 10⁵ cpm of P-Olg (about 345 pg of Olg). At various times between 2 min and 24 h, the arterial blood was collected by severing the carotid artery and the mouse immediately decapitated. The kidney, spleen, testes, brain (free of pituitary and pineal), and pieces of lung, liver, and abdominal muscle were collected and weighed. The blood was centrifuged at 2500g for 5 min and the serum removed. The level of radioactivity in the serum and tissue samples was determined in a beta counter. The results were expressed as (cpm per gram of tissue)/(cpm per microliter of serum) = microliters per gram and plotted against time.

Characterization of Radioactivity in Tissues. Male ICR mice were given an injection into the jugular vein of 0.2 ml of LR-BSA containing 5 × 10⁶ cpm of P-Olg. Arterial serum and tissues were collected as described above at 60 min, 2.5 h, 9 h, and 16 h after i.v. injection. Tissues were homogenized on ice in 3 volumes of Tris borate buffer and the homogenate was centrifuged for 10 min at 4°C at 14,000g. The supernatant was heated at 95°C to denature proteins, centrifuged again, and the resulting supernatant lyophilized. The lyophilized material was dissolved in 100 µl of Tris borate EDTA buffer and 50 µl of this applied to a polyacrylamide gel and electrophoresed with bromophenol blue 12 cm from the origin. The gel was dried and exposed to X-Omat film (Eastman Kodak, Rochester, NY) overnight.

Octanol/Buffer Partition Coefficient. The lipid solubility of P-Olg was measured by determining its octanol/buffer partition coefficient. P-Olg (1.5 × 10⁵ cpm) was added to 1 ml of 0.25 M chloride-free phosphate buffer solution and 1 ml of octanol. This was vigorously mixed for 1 min and the two phases separated by centrifugation at 4500g for 10 min. Aliquots of 100 µl were taken in triplicate from each phase and counted. The mean partition coefficient was expressed as the log of the ratio of cpm (octanol phase) to cpm (chloride-free phosphate buffer solution phase).

Uptake Rate into Brain. Multiple-time regression analysis (Blasberg et al., 1983; Patlak et al., 1983) was used to calculate the blood-to-brain unidirectional influx rate (K_i). The brain/serum ratios were plotted against their respective exposure times (Expt). Expt was calculated from the following formula:

(1)

where Cp is the level of radioactivity in serum and Cpt is the level of radioactivity in serum at time t. Expt corrects for the clearance of P-Olg from the blood. K_i with its error term is measured as the slope for the linear portion of the relation between the brain/serum ratios and Expt. The y-intercept of the linearity measures Vi, the distribution volume in brain at t = 0. This method was also used to measure the uptake rate of P-Olg in young and aged SAMP8 mice.

(2)

Capillary Depletion. To determine whether P-Olg completely crossed the BBB, we performed capillary depletion with the protocol adapted to mice (Gutierrez et al., 1993) from rats (Triguero et al., 1990). ICR male mice anesthetized with i.p. urethane received an i.v. injection of 0.2 ml of LR-BSA and 10⁵ cpm of P-Olg. At either 10, 30, or 60 min after i.v. injection, blood from the carotid artery was collected and the cerebral cortex removed, weighed, and emulsified with a glass homogenizer (10 strokes) in 0.8 ml of physiological buffer (10 mM HEPES, 141 mM NaCl, 4 mM KCl, 2.8 mM CaCl₂, 1 mM MgSO₄, 1 mM NaH₂PO₄, and 10 mM D-glucose adjusted to pH 7.4). Dextran solution (1.6 ml of a 26% solution) was added to the homogenate, which was thoroughly mixed, and homogenized again (three strokes). Homogenization was performed at 4°C in less than 1 min. An aliquot of the homogenate was centrifuged at 5400g for 15 min at 4°C in a Beckman Allegra 21R centrifuge with a swinging bucket rotor. The pellet containing the brain vasculature and the supernatant containing the brain parenchyma were carefully separated and the level of ³²P determined in a gamma counter. The fractions were expressed as volumes of distribution in microliters per gram and as the percentage of P-Olg in the parenchymal fraction. In addition, young SAMP8 mice from our breeding colony were studied at 30 min only.

Uptake into CSF. Mice anesthetized with urethane received an i.v. injection of 2.5 × 10⁶ cpm of P-Olg as described above. Thirty minutes later, the scalp was removed from the posterior aspect of the head, exposing the muscles overlying the posterior fossa. A 30-gauge needle connected to a length of PE-10 tubing was inserted into the posterior fossa with the head in a dependent position. CSF was collected into the PE tubing by capillary action. After collecting about 10 µl of CSF, the tubing was removed, arterial blood collected from the previously exposed carotid artery, and the mouse decapitated and the whole brain removed. The exact amount in microliters of CSF collected was determined by measuring the length in centimeters of PE tubing filled with CSF and multiplying by 0.668. Only CSF that was absolutely clear was analyzed. The CSF, brain, and serum were counted in a gamma counter. The results were expressed as brain/serum (µl/g), CSF/serum (µl/ml), and brain/CSF (ml/g) ratios.

Percentage of Uptake by Brain Regions: i.v. Injection. Mice anesthetized with i.p. urethane received an i.v. injection of 10⁵ cpm of P-Olg as described above. Blood was collected from the abdominal aorta 10 min later. The brain vascular space was then washed free of blood by exposing the heart, clamping the descending thoracic aorta, severing both jugular veins, and infusing 20 ml of lactated Ringer's solution through the left ventricle of the heart in about 1 min. The brain was removed, the cortex and hippocampus separated from the remainder of the brain, and the three brain regions weighed. The level of radioactivity in these regions and for the serum was determined in a gamma counter. The %Inj/g for each region of brain was determined with eq. 2 except that the term Vi was set to equal zero because of the washout of the vascular space.

Intracerebroventricular Injections. A method previously described that accurately quantifies efflux rates was used (Banks et al., 1986, 1997a; Banks and Kastin, 1989). Mice were anesthetized with i.p. urethane, the scalp removed, and a hole made through the cranium 1.0 mm lateral and 0.5 mm posterior to the bregma with a 26-gauge needle. Tubing covered all but the terminal 2.5 to 3.0 mm of the needle, so that the tip of the needle penetrated the brain tissue forming the roof of the lateral ventricle but did not penetrate its floor. One microliter of LR-BSA containing 10⁵ cpm of P-Olg was injected into the lateral ventricle with a 1.0-µl Hamilton syringe.

Effects of i.v. Olg on Learning and Memory. Experimentally naive 12-month-old SAMP8 mice from our colony received 200 µl of saline with or without 6 µg of Olg by tail vein 4 weeks and again 2 weeks before training. Mice were trained in a T-maze footshock avoidance apparatus as previously described (Flood and Morley, 1993; Farr et al., 1999). The maze consisted of a black plastic alley with a start box at one end and two goal boxes at the other. A stainless steel rod floor ran throughout the maze. The start box was separated from the alley by a plastic guillotine door that prevented the mouse from entering the alley until the training trial began. After placing a mouse in the start box, a training trial was begun by simultaneously raising the guillotine door and sounding a buzzer. A footshock was applied 5 s later. The goal box chosen by the mouse on the first trial was designated as "incorrect". Footshock was continued until the mouse entered the other goal box, which on all subsequent trials was designated as the "correct" choice for that particular mouse. At the end of each trial, the mouse was removed from the goal box and returned to its home cage. A new trial began after placing the mouse back in the start box by sounding the buzzer and raising the guillotine door. Footshock was applied 5 s later unless the mouse had entered its correct goal box. The acquisition training conditions used were an intertrial interval of 45 s, a door-bell type buzzer of 65dB as the conditioned stimulus warning, and a foot shock at 0.35 mA (scrambled grid floor shocker model E13-08; Coulbourn Instruments, Allentown, PA). Retention was tested 1 week later and five avoidances in six consecutive trials used as criterion

Statistics. Means are reported with their standard errors and n. Means were compared by ANOVA and when more than two means were compared this was followed by Newman-Keuls multiple comparison test. Regression lines were calculated by the least-squares method with the Prism program (GraphPad, San Diego, CA). Slopes and intercepts are reported with their error terms and n and were compared statistically with the Prism program.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Tissue Distribution. Figure 1 shows the clearance of P-Olg from serum for the first 240 min after i.v. injection. A rapid decline was followed by a steady state that was unchanged between about 30 min to 24 h (data after 240 min not shown). By 30 min, each milliliter of arterial serum contained about 2.5% of the injected radioactivity. The inset of Fig. 1 shows the early phase of clearance followed first order kinetics with a significant correlation between log(cpm/ml) and time (t): log(cpm/ml) = (0.0848)t + 4.531, r = 0.988, n = 5, p < 0.01. This equation gave a half-time disappearance from arterial serum of 3.55 min and a volume of distribution of 2.94 ml.

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Clearance of P-Olg from serum after intravenous injection. An early rapid, linear distribution phase (see inset) was followed by a prolonged period when blood levels were stable. For the early distribution phase, the volume of distribution was 2.94 ml and the half-time disappearance was 3.55 min. For the prolonged period of stable levels, each milliliter of serum contained about 2.5% of the injected dose.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Distribution of P-Olg among tissues after intravenous injection. When expressed on a per gram of tissue basis, spleen took up the most P-Olg and brain the least. , spleen; , liver; , lung; , kidney; , muscle; , testis; , brain.

Characterization of Radioactivity in Tissues. Figure 3 shows that the radioactivity extracted from brain, serum, and other tissues 16 h after i.v. injection corresponded to P-Olg. In some tissues, including brain, a heavier band, presumably P-Olg bound to protein, was seen. The heavier band was more intense at earlier time points (data not shown), suggesting transfer from a bound to an unbound form with time. No obvious fragments of P-Olg were found, demonstrating that P-Olg was taken up intact by brain and other tissues.

View larger version (43K): [in this window] [in a new window]   Fig. 3.   Gel electrophoresis of radioactivity extracted from tissues 16 h after the intravenous injection of P-Olg. The higher the molecular weight band likely represents a bound form of P-Olg. No degradation products were seen.

Octanol/Buffer Partition Coefficient. The mean octanol/buffer partition coefficient was 0.000303 ± 0.0000393, n = 3. This gave a log value of 3.52, showing that P-Olg is hydrophilic.

Uptake Rate into Brain. Figure 4 shows the relation between brain/serum ratios and Expt. In agreement with Fig. 2, brain/serum ratios continued to increase over time. When the results were fitted to a one-site binding model (hyperbola), a maximal brain/serum ratio of 226 ± 44 µl/g was obtained. The K_i estimated from the linear portion of this curve (inset) was 1.40 ± 0.39 µl/g-min with a Vi of 8.52 ± 4.8 µl/g (r = 0.902, n = 5, p < 0.05).

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Uptake of P-Olg by brain after intravenous injection. The unidirectional influx rate (Ki) calculated by multiple-time regression analysis based on the linear portion of the curve (inset) gave a value of 1.40 ± 0.39 µl/g-min. The Vi of 8.52 ± 4.8 µl/g approximates the vascular space of brain. The full data set was fit to a one site binding model (hyperbola), which estimates a maximal value for the brain/serum ratio of 226 ± 44 µl/g.

(3)

View larger version (12K): [in this window] [in a new window]   Fig. 5.   Percentage of intravenously injected dose taken up per gram of brain over time. The maximal value was estimated by a one-site binding model to approach 0.258%Inj/g.

View larger version (28K): [in this window] [in a new window]   Fig. 6.   Saturation of P-Olg uptake by brain. A, self-inhibition of P-Olg by 10 or 200 µg/mouse unlabeled Olg. *p < 0.001 in comparison to P-Olg; n = 7 to 8/group. B, inhibition of P-Olg by 100 µg/mouse of phosphorothiolate oligonucleotides. *p < 0.001 in comparison to P-Olg; n = 9 to 10/group. In addition, the mr42 mer produced less inhibition that the 40 mer or the 10 mer.

Capillary Depletion. Capillary depletion was performed to determine whether the P-Olg taken up by brain had entered the parenchymal space of the brain or was sequestered by brain capillaries. Figure 7A shows that most of the P-Olg taken up by brain appeared in the parenchymal space, demonstrating that P-Olg completely crossed the BBB. A two-way ANOVA for tissue (capillary/parenchymal space) and time as the independent variables showed an effect for both at the p < 0.001 level. Newman-Keuls multiple comparison test showed that the 10-min parenchymal value was lower than the 30-min (p < 0.001) and 60-min (p < 0.001) parenchymal values consistent with time-dependent entry into the parenchymal space. The 30-min capillary value was higher than the 10-min (p < 0.001) and 60-min capillary values, suggesting that flux across the capillaries peaked at about 30 min after injection. The percentage of P-Olg in the parenchymal space was constant over time.

View larger version (32K): [in this window] [in a new window]   Fig. 7.   Distribution of P-Olg taken up by brain between capillary and parenchymal compartments: capillary depletion. A, majority of P-Olg taken up completely crossed the capillaries to enter the parenchymal space in the brain at all times. B, washout of vascular space did not change the distribution space. This demonstrates that P-Olg was not loosely adhering to the luminal surface of capillaries.

Uptake into CSF. Figure 8 shows the results for four mice in which CSF, brain, and serum samples were taken. A mean of 11 µl of CSF was obtained from the mice. The CSF/serum ratio was 23.0 ± 3.8 µl/ml compared with a brain/serum ratio of 22.6 ± 2.4 µl/g. The CSF/brain ratio was 1.05 ± 0.15 ml/g.

View larger version (40K): [in this window] [in a new window]   Fig. 8.   Entry of P-Olg into CSF. About 11 µl of CSF was obtained per mouse. The CSF/serum and brain/serum ratios (left axis) were nearly equal and the brain/CSF ratio (right axis) was about 1.0. This suggests P-Olg enters the CSF as readily as the brain tissue space. n = 4/group.

Percentage of Uptake by Brain Regions: Intravenous Injection. Figure 9A shows the percentage of the i.v.-injected dose of P-Olg taken up per gram of whole brain, cerebral cortex, and hippocampus in five mice. The hippocampus had the highest uptake of 0.865 ± 0.115%Inj/g.

View larger version (24K): [in this window] [in a new window]   Fig. 9.   Percentage of injected dose taken up by brain regions. A, after intravenous injection, the hippocampus took up P-Olg more readily than whole brain or cortex, with a value of 0.865%Inj/g. B, after intracerebroventricular injection, the hippocampus took up P-Olg more readily than the other regions, with a value at 30 min of 78%Inj/g. n = 5 to 6/group.

Intracerebroventricular Injections. No correlation existed between time and the log residual radioactivity in brain after i.c.v. injection for the first 20 min after i.c.v. injection. This shows that minimal brain-to-blood efflux occurred for P-Olg and suggests sequestering by brain tissue. This sequestration had a saturable component, as the %Inj/g of brain decreased from 25.1 ± 1.6 (n = 6) for mice that received P-Olg only to 16.2 ± 1.9 (n = 7) for mice that received P-Olg with unlabeled oligonucleotide [F(1,12) = 12.2, p < 0.005]. Figure 9B shows the distribution of P-Olg in brain regions for five to six mice. The %Inj/g was highest for hippocampus at 30 min with a value of 78.0 ± 15.5. The i.c.v./i.v. ratio for hippocampus was, therefore, 78.0/0.865 = 90.2.

BBB Permeability in SAMP8 mice. Figure 10A compares the uptake of P-Olg for ICR mice, young SAMP8 mice (4 months old), and aged SAMP8 mice (14 months old). The K_i did not differ among ICR (0.442 ± 0.126 µl/g-min), young SAMP8 (0.497 ± 0.061 µl/g-min), and aged SAMP8 (0.578 ± 0.068 µl/g-min) mice. A statistically significant difference did occur among the values for Vi: F(2,24) = 8.47, p < 0.05). Post-testing with Prism showed that both young and aged SAMP8 mice had a lower Vi in comparison to ICR (p < 0.01) but did not differ from each other.

View larger version (22K): [in this window] [in a new window]   Fig. 10.   Uptake of intravenously administered P-Olg by brain in young and aged SAMP8 mice and in young ICR Mice. A, calculation of unidirectional influx rate (Ki). Influx rate of P-Olg did not differ among the three groups. The intercept values (Vi) were lower in both groups of SAMP8 mice. Entry into parenchyma space did not differ between the two strains, but the portion retained by capillaries was lower in SAMP8 mice. n = 4 to 5/group.

Effects of i.v. Olg on Learning and Memory. As shown in the Fig. 11A, i.v. Olg improved acquisition in 12-month-old SAMP8 mice by reducing the mean trials to first avoidance from 12.7 ± 0.6 (n = 9) to 8.6 ± 0.8 (n = 8): F(1,16) = 17.8, p < 0.001. The i.v. Olg also improved retention as shown by a reduction in the mean trials to criterion from 15.2 ± 0.8 to 8.1 ± 0.7: F(1,16) = 53.3, p < 0.001 (Fig. 11B).

View larger version (36K): [in this window] [in a new window]   Fig. 11.   Effects of i.v. Olg on learning and memory. Intravenous Olg reversed the age-related impairments in acquisition (A) and retention (B). *p < 0.001 for control versus antisense. n = 8/group.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The purposes of these studies were to determine whether Olg could enter the brain after i.v. injection and, if it could, whether a dose could be delivered by i.v. injection sufficient to reverse the learning and memory deficits in aged SAMP8 mice, a strain with an age-related overexpression of APP (Morley et al., 2000). Our previous work showed that Olg given i.c.v. reversed the deficits in learning and memory of aged SAMP8 mice (Kumar et al., 2000). Major findings of this study include the following: 1) Olg radioactively labeled with ³²P (P-Olg) crossed the BBB at a moderate rate; 2) Passage across the BBB was facilitated by a saturable transport system, here termed oligonucleotide transport system-1 (OTS-1); 3) P-Olg entered both the parenchymal space of the brain and the CSF, suggesting both the choroid plexus and brain endothelial cells contain OTS-1; 4) Compared with other brain regions, the hippocampus, a region important in learning and memory, had a high uptake of P-Olg after either i.v. or i.c.v. injection; 5) Based on the pharmacokinetics of hippocampal uptake of P-Olg after i.v. and i.c.v. administration, we calculated that the i.v. dose of Olg would need to be 100 times greater than the i.c.v. dose to reverse learning and memory deficits; and 6) We found this dose of Olg administered by tail vein reversed the learning and memory deficits of aged SAMP8 mice.

We first determined whether i.v. P-Olg could cross the BBB in a series of studies that began with the measurement of the unidirectional influx rate (K_i). P-Olg was taken up by brain in a time-dependent manner. Initially, uptake was linear with time, with a K_i of 1.40 µl/g-min. This is a moderate rate of uptake similar to those measured for peptides of about the same size as P-Olg that are transported across the BBB by saturable systems (Banks et al., 1993; Banks and Kastin, 1998). This rate of uptake has been associated with effects on the CNS (Banks and Kastin, 1994; Uchida et al., 1996). This K_i is about 100 times greater than that expected for the residual leakiness of the BBB as measured by albumin. The two mechanisms that can produce this level of uptake are nonsaturable transmembrane diffusion and saturable transport. Transmembrane diffusion depends on the physicochemical properties of the molecule with faster rates of uptake for small, lipid-soluble molecules (Oldendorf, 1974). The poor lipid solubility of P-Olg (log octanol/buffer partition coefficient of 3.52) and its size would limit transmembrane diffusion. Saturable transport was confirmed by finding uptake was inhibited by unlabeled Olg and by other phosphorothiolate oligonucleotides (Fig. 6, A and B).

The enzymatic stability of P-Olg allowed us to observe uptake by brain over an extended time. P-Olg approached equilibrium at a brain/serum ratio of about 226 µl/g (Fig. 4). Capillary depletion and CSF collection showed that P-Olg had crossed the BBB completely (Fig. 7). Capillary depletion showed that at all times examined the majority of P-Olg taken up by brain was in the parenchymal space. About one-fourth was retained by the capillaries and none was loosely adhering to the luminal surface of the BBB. Only the cerebral cortex, a region of the brain free of circumventricular organs and choroid plexus tissue, was used for capillary depletion, and so these results strongly indicate that OTS-1 is located in brain endothelial cells. The appearance of P-Olg in the CSF suggests that OTS-1 is located at the choroid plexus as well (Fig. 8). The brain/CSF ratio of about 1.0 suggests that permeability at the endothelial and choroid plexus barriers is about equal.

The %Inj/g is an important calculation in determining an effective intravenous dose and is a function of BBB permeability and blood concentration. Therefore, prior to estimating %Inj/g, we examined the peripheral pharmacokinetics of P-Olg after i.v. injection. As shown in Fig. 1, there was an initially rapid clearance of P-Olg from blood with a half-time disappearance rate of about 3.55 min. The spleen was a major site of uptake, with the spleen/serum ratio being 10,000 µl/g (Fig. 2). The liver and lung also had high rates of uptake. The testis, a tissue that has blood barriers in series (Plöen and Setchell, 1992), and the brain had the lowest rates of uptake. All tissues reached steady states after about 30 min and had stable levels for 24 h after i.v. injection. The brain/serum ratios at steady state exceeded the vascular space for each tissue by at least 10-fold (Banks and Kastin, 1992; Kastin et al., 1993). Serum also reached a steady state of about 2.5% of the injected dose present per milliliter. Gel electrophoresis showed that the radioactivity even at these late times represented intact P-Olg (Fig. 3). Degradation products readily were evident in any of the gels for any tissue or serum at any time, although all samples showed a higher molecular weight band in addition to the P-Olg band. This band presumably represents P-Olg bound to proteins. Cossum et al. (1993) have shown that phosphorothiolate oligonucleotides bind to albumin and ₂-macroglobulin. The intensity of this high molecular weight band relative to free P-Olg decreased with time, suggesting conversion from the bound to the free form. A reservoir of bound P-Olg could aid its enzymatic stability and explain the rapid uptake phase followed by a long-term steady-state phase. Based on these results, we calculated that the amount of P-Olg entering the brain would approach 0.258%Inj/g (Fig. 5). This compares favorably to values for other substances known to exert effects on the brain after peripheral administration. For example, morphine has a value of about 0.02%Inj/g, domoic acid of 0.002%Inj/g, a cyclic D-Pen analog of enkephalin of 0.08%Inj/g, and PACAP of about 0.11%Inj/g (Preston and Hynie, 1991; Weber et al., 1992; Banks et al., 1993; Banks and Kastin, 1994).

Some indication of the characteristics of OTS-1 is given by Fig. 6. The finding that all of the phosphorothiolate oligonucleotides tested inhibited P-Olg uptake, even the reverse, mirror 42 mer, shows that this is a general transporter for this class of compounds. However, the ability of the random 40 mer and random 10 mer to inhibit P-Olg uptake better than the reverse, mirror 42 mer suggests that the transporter has preferences related to size or primary sequence. The inability of adenosine or uridine to inhibit P-Olg uptake shows that OTS-1 is not one of the known transporters for nucleosides (Davson and Segal, 1996). Phosphorothiolate oligonucleotides are synthetic compounds and so cannot be the endogenous ligand for OTS-1. It seems unlikely that a transporter would exist for transporting oligonucleotides from the blood into the CNS. Endogenous oligonucleotides would probably only be found in blood under pathological conditions and the introduction of such genetic material into the CNS would likely interfere with normal CNS functioning. It is more likely that Olg happens to bind to the transporter for some other class of ligand. Phosphorothiolate oligodeoxynucleotides can bind to heparin binding proteins such as Mac-1, a receptor for fibrinogen (Stein, 1997). The polycationic nature of phosophorthiolate nucleotides then induces adsorptive endocytosis. A similar mechanism has been proposed in the uptake of gp120 and human immunodeficiency virus-1 across brain endothelial cells (Banks et al., 1997b). In blood, binding to albumin and ₂-macroglobulin occurs (Cossum et al., 1993). Therefore, it is possible that OTS-1 is actually a protein transporter co-opted by Olg.

To determine the theoretical dose of P-Olg needed to reverse learning and memory deficits, we determined the %Inj/g of P-Olg for the hippocampus after i.v. and i.c.v. administration, calculated the i.c.v./i.v. ratio, and multiplied the know effective i.c.v. dose by the i.v./i.c.v. ratio. We first determined the %Inj/g of P-Olg for the hippocampus after i.v. injection (Fig. 9) to be 0.865 ± 0.115%Inj/g, a value about 4 times higher than those for the cerebral cortex or whole brain. Such regional variation is not uncommon for peptide and regulatory protein transport into brain and has been noted for some feeding peptides and cytokines. Such regional variation is thought to reflect physiological demand. For example, leptin and insulin transport is highest into the arcuate nucleus and olfactory lobe (Banks et al., 1996, 1999), respectively, the regions with the highest density of receptors for those particular proteins (Fei et al., 1997; Hill et al., 1998).

We next characterized the fate of i.c.v. administered P-Olg. We found that the concentration of P-Olg remained constant for the first 20 min after i.c.v. injection. Most substances show substantial disappearance from the brain during this time. Any compound remaining in the CSF has a half-time efflux rate of about 25 to 45 min, reflecting the reabsorption of CSF back into the blood by the arachnoid villi. Substances with saturable brain-to-blood transport systems have half-time efflux rates substantially shorter (Banks and Kastin, 1984). The absence of measurable efflux during this time is unusual but has been shown to occur when compounds are sequestered from the CSF by the periventricular tissues (Banks and Broadwell, 1994). For P-Olg, this sequestration depends on a saturable mechanism, as 10 µg of Olg injected with P-Olg was able to decrease the percentage of P-Olg retained by the brain by about one-third. P-Olg was still detectable in the brain 24 h after i.c.v. injection. The hippocampus contained more P-Olg than the whole brain or cortex at all time points, with a peak value of 78.0 ± 15.5%Inj/g. The i.v./i.c.v. ratio was, therefore, calculated to be 90. We, therefore, determined whether an i.v. dose 100 times greater than the i.c.v. dose of 60 ng/mouse would be effective in reversing the learning and memory deficits in aged SAMP8 mice.

Before testing this dose, we determined whether SAMP8 mice could also transport P-Olg across the BBB. Some saturable transporters for peptides and regulatory proteins have a variable activity among murine strains. For example, PTS-1 activity is present in some, but not other murine strains (Banks and Kastin, 1997) and leptin transport is attenuated in fat Koletsky and Zucker rats (Kastin et al., 1999; Burguera et al., 2000). The SAMP8 mouse is a natural mutation that increasingly overexpresses APP as it ages (Kumar et al., 2000). By 12 months of age, the mouse has severe deficits in learning and memory and levels of A are increased about 2-fold. This is much closer to the 50% increase seen in Alzheimer's disease than the 5- to 14-fold increase seen in transgenic mice (Hsiao et al., 1996; Rosenberg, 2000). This suggests that the pathophysiology of A toxicity in the SAMP8 mouse may resemble more closely that of Alzheimer's disease seen in humans and may be more easily reversible. In addition, the SAMP8 mouse offers a model for exploring how increases in brain levels of A can occur spontaneously and why increases occur with aging.

We determined whether P-Olg was transported across the BBB in aged and young SAMP8 mice and compared these transport rates to ICR mice (Fig. 10). The K_i was not different among these three groups. Capillary depletion showed that the amount of material entering the parenchymal space was not different between SAMP8 and ICR mice. Both young and aged SAMP8 mice had a lower Vi than the ICR mice. The Vi value reflects material that is in rapid, reversible equilibrium with the vascular space and probably includes reversible association with the luminal surface of the brain endothelial cell. Consistent with this interpretation are the results from the capillary depletion experiment, which showed less association of P-Olg with the capillary fraction from SAMP8 mice. This suggests two classes of binding sites for P-Olg on brain endothelial cells, a reversible one that is decreased in SAMP8 mice and another representing the transporter that is not different between SAMP8 and ICR mice. For the current studies, the most relevant finding is that no difference in transport rates of P-Olg occurred among aged and young SAMP8 and ICR mice.

Based on the above-mentioned findings, 6 µg of Olg given i.v. should be as effective as 60 ng given i.c.v. in reversing the learning and memory deficits in aged SAMP8 mice. Figure 11 shows that i.v. Olg at this dose was indeed effective in dramatically reversing both learning (acquisition) and memory (retention) deficits.

In conclusion, we showed that P-Olg is transported across the BBB by the previously unknown saturable transporter OTS-1 at a moderate rate to enter both the parenchymal space of the brain and the CSF. Uptake of P-Olg by the hippocampus after either i.c.v. or i.v. injection was particularly high in comparison to the cerebral cortex or whole brain. The i.v./i.c.v. ratio indicated that an i.v. dose of Olg about 100 times larger than the effective i.c.v. dose would be needed to reverse the learning and memory deficits of aged SAMP8 mice. We found that this dose, 6 µg/mouse, did indeed dramatically reverse learning and memory deficits. These studies show for the first time that phosphorothiolate oligonucleotides can be delivered to the brain in effective doses by intravenous administration.