Introduction

Exendin-4 is in clinical trials for the treatment of type II diabetes mellitus and associated obesity. Originally isolated from the saliva of the Gila monster, this 39 amino acid peptide has slightly more than 50% sequence homology with glucagon-like peptide (GLP)-1. GLP-1, however, is rapidly metabolized in blood, less than half remaining intact after 2 min.¹

Exendin-4 is much more potent in inducing satiety and weight loss in both lean and obese mice and rats, particularly after peripheral administration.^2,3 The action of exendin-4 in reducing food intake is considered to be mediated centrally in the brain,³ where its binding sites in places like the hypothalamus and thalamus are identical to those of GLP-1.⁴ If the central effects of peripherally administered exendin-4 are direct, then it must cross the blood–brain barrier (BBB). Accordingly, we investigated whether exendin-4 crosses the BBB in intact form and whether this crossing is saturable.

Materials and methods

Animals

Adult male CD1 mice (Charles River, Wilmington, MA, USA), weighing about 22 g each, were anesthetized with urethane (4 g/kg, i.p.) and used as approved by the Institutional Animal Care and Use Committee.

Radiolabeling of the compounds

Exendin-4 (Heloderma Suspectum) was purchased from Phoenix Pharmaceuticals, Belmont, CA, USA. This 39 amino acid peptide has a molecular weight of 4184. Although it does not contain a Tyr, the N-terminal amino acid is His, permitting radiolabeling with ¹²⁵iodine (¹²⁵I) by Iodo-Beads (Pierce, Rockford, IL, USA). The iodination mixture was purified by high-performance liquid chromatography (HPLC). The specific activity of ¹²⁵I-exendin-4 was 95.8 Ci/mmol. Bovine serum albumin (BSA) was labeled with ^99mTc. ¹²⁵I-[Tyr³⁹]exendin-4 retains full binding activity.⁴ In the few situations in which it has been compared with tritium, iodination did not interfere with the BBB transport of a small tripeptide like MIF-1⁵ or a larger polypeptide like leptin.⁶

HPLC

¹²⁵I-exendin-4 was injected onto a reversed-phase Vydac C18 column. The gradient was 0–60% acetonitrile with 0.1% trifluoroacetic acid (B) for 40 min followed by 60–80% for 5 min and then 80% B for an additional 5 min. Peak fractions were pooled and aliquots were dried in a Speed Vac concentrator (Savant, Holbrook, NY, USA) and stored at -80°C. To test whether these peptides were stable in vivo, they were injected into the jugular vein of anesthetized mice. Blood was collected from the carotid artery followed by immediate decapitation. Serum was tested by the same HPLC conditions as for purification.

The brain was homogenized in the presence of a protease inhibitor cocktail (Sigma, St Louis, MO, USA) as well as 5.8 mM EDTA and 1.1 mM 1,10-phenanthroline to retard degradation induced by the homogenization procedure, and the supernatant was eluted with the same HPLC program. To compare the degradation occurring during tissue processing, the iodinated peptide was added to freshly harvested brain and blood of the anesthetized mouse, and brain supernatant and serum were processed and analyzed by HPLC.

Capillary depletion

To study the compartmental distribution of ¹²⁵I-exendin-4 in brain, a mixture of ¹²⁵I-exendin-4 and ^99mTc-albumin in 200 l of lactated Ringer's solution containing 1% albumin was injected into the jugular vein of anesthetized mice (n=4/group). Blood was collected from the descending abdominal aorta and the mouse was decapitated with or without intracardial perfusion to clear the cerebral vascular space. The cerebral cortex was harvested, weighed, and homogenized in capillary buffer with the addition of 26% dextran. The capillaries and brain parenchyma were separated by centrifugation of the mixture in a swinging bucket rotor at 4°C, 5400 g, for 15 min, and radioactivity was measured in a dual-channel -counter for both ¹²⁵I-exendin-4 and ^99mTc-albumin.

Blood-to-brain transfer after i.v. injection

For multiple-time regression analysis, anesthetized mice (n=7/group, repeated four times for a total of 28/group) were injected with iodinated peptide in lactated Ringer's solution with 1% BSA (pH 6.5) through the left jugular vein (i.v.). At various time intervals between 1 and 10 min after i.v. injection, blood was collected from the right carotid artery and mice were decapitated immediately afterwards. The radioactivity of the serum and brain samples was measured in a -counter. Exposure time was calculated for each time point as the theoretical steady-state value, and it represents the integral of serum radioactivity from time 0 to time t divided by the serum radioactivity at time t.⁷ The unidirectional influx constant (K_i, expressed in ml/g min) was determined from the linear relationship between tissue/serum ratios and exposure time using the equation K_i=slope ((cpm/g brain)/(cpm/l serum)) vs exposure time in minutes.

For the self-inhibition studies with ¹²⁵I-exendin-4, 5 g unlabeled exendin-4 was included in the injectate. A separate control group received ¹²⁵I-exendin-4 only. The injection solution also contained ^99mTc-albumin.

For the study of the effect of food deprivation, food, but not water, was removed 24 h before the experiment. The control group had free access to food and water.

For study of the possible role of circumventricular organs (CVOs), cortex was separated from the rest of the brain and counted separately. The counts for cortex and those for the remainder of the brain were added for each mouse at each time to arrive at the cpm for the total brain. The K_i for each of these three regions was determined as described above.

Perfusion with blood-free buffer

In situ brain perfusion was performed in anesthetized mice in which the abdominal aorta was clamped and the bilateral jugular veins disrupted so as to generate an excorporeal system. The blood-free perfusate buffer was oxygenated for 10 min with 95% O₂ and 5% CO₂. To this was added ¹²⁵I-exendin-4 at 0.6 10⁶ cpm/ml and ^99mTc-albumin at 1.6 10⁶ cpm/ml. The perfusion syringe was driven by a Harvard pump (model 975; Millis, MA, USA) at a rate of 2 ml/min. Two groups of mice were studied (n=7/group, repeated three times for a total of 21/group). One group received radiolabeled compounds only; the other group received unlabeled exendin-4 at 2 g/ml in addition to the radioactive compounds. The mice were decapitated at different time intervals between 1 and 5 min, and brain radioactivity was measured in a dual-channel -counter for both ¹²⁵I-exendin-4 and ^99mTc-albumin. The ratios of radioactivity of brain over perfusate for each compound were plotted against perfusion time to derive the influx rate, which is the slope of the linear regression line.

Efflux after i.c.v. injection

Brain-to-blood studies were performed in mice (n=5/group) anesthetized with urethane. The scalp of the mouse was removed, and the right lateral ventricle was localized at 1 mm lateral and 0.2 mm posterior to the bregma and 2.5 mm deep. Each mouse received 1 l of ¹²⁵I-exendin-4 and ^99mTc-albumin (about 25 000 cpm each) in lactated Ringer's solution with 1% BSA (pH 6.5). The intracerebroventricular (i.c.v) injection was performed with a Hamilton syringe (7000 series; Reno, NV, USA). Mice were decapitated at 2, 5, 10, and 20 min after i.c.v. injection. Their brains were removed immediately after decapitation and radioactivity measured in the -counter. The logarithm of brain radioactivity was plotted against time, and the slope of the regression line was obtained. The half-time disappearance (t_1/2) from the brain for each compound was calculated: t_1/2=0.301/slope.⁸

Statistical analysis

Means are expressed with their standard errors. Groups were compared by analysis of variance followed by Tukey's multiple range test. Regression analysis was performed by the least-squares method. GraphPad Prism statistical software (GraphPad, San Diego, CA, USA) was used for all analyses.

Octanol/buffer partition coefficient

To a mixture of ¹²⁵I-exendin-4 and 1 ml of octanol was added 1 ml of a 0.25 M phosphate buffer solution at pH 7.4. After being vigorously mixed for 1 min and gently mixed for an additional 10 min, the two phases were separated by centrifugation at 4000 g for 10 min. Aliquots were counted for radioactivity and the partition coefficient was expressed as the ratio of cpm in the octanol phase to cpm in the buffer phase. The octanol/buffer partition coefficient measures lipophilicity.

Hydrogen bonding

The standard method for the determination of hydrogen bonding (desolvation potential) involves measurement of the difference between the logarithms of the partition coefficients in a hydrogen bonding solvent (octanol) and a nonhydrogen bonding lipophilic organic solvent (isooctane).^9,10 The procedure we used for the determination of the isooctane/buffer partition coefficient was identical to that described above for the octanol/buffer partition coefficient except that isooctane was substituted for the octanol.

Results

Stability of ¹²⁵I-exendin-4 in blood and brain

In serum, the percent intact ¹²⁵I-exendin-4, corrected for processing, was 100% at 10 min; at 30 min it was 97.7%. The corrected percent intact ¹²⁵I-exendin-4 in brain, correlating with the elution position of the ¹²⁵I-exendin-4 stock solution, was 82.8% at 10 min (the latest time studied), and 60.0% at 30 min. This shows that exendin-4 injected i.v. reached the brain intact during the study.

Compartmental distribution of ¹²⁵I-exendin-4 in brain

At 10 min after i.v. injection, the brain/serum ratio of ¹²⁵I-exendin-4 in the nonperfused cerebral cortex was 17.82.19 l/g. This was composed mainly of the radioactivity in three compartments: that loosely associated with the capillary lumen and removable by intracardial perfusion (-0.701.33 l/g), that taken up by the capillaries (2.901.24 l/g) which constitute the BBB, and that crossing the BBB completely to enter brain parenchyma (15.61.21 l/g). Significantly (P<0.001) more ¹²⁵I-exendin-4 reached the brain parenchyma than was bound to the capillaries or was loosely adherent to the vasculature (Figure 1). This shows that almost 90% of the exendin-4 injected i.v. reaches the brain parenchyma.

Figure 1.

Capillary depletion: tissue to serum ratios for mice injected with ¹²⁵I-exendin-4 corrected for simultaneously injected ^99mTc-albumin. Values for the cortex are composed of three components: parenchyma (remaining after washout), capillaries (remaining after washout), and reversible vascular association (removed by washout). Significantly more ¹²⁵I-exendin-4 was found in the brain parenchyma than was bound or loosely adherent to the endothelial cells of the capillaries composing the BBB (n=4 mice/group). ***P<0.001 compared with either cortex or parenchyma.

Full figure and legend (13K)

Entry of ¹²⁵I-exendin-4 into brain after i.v. injection

In each of four individual experiments (n=7/group) considered separately, excess (5 g/mouse) unlabeled exendin-4 did not decrease the entry of ¹²⁵I-exendin-4 to a statistically significant extent. When the results of these four experiments were combined (n=28/group), however, the results became significant at a P-value of 0.048. For these combined experiments (Figure 2), the K_i of ¹²⁵I-exendin-4 was 4.6210.557 10^-4 ml/g min (r=0.85). Addition of 5 g/mouse unlabeled exendin-4 decreased the K_i to 2.6310.818 10^-4 ml/g min (r=0.54). In a preliminary experiment, 1 g exendin-4 was ineffective. This shows that exendin-4 readily enters the brain from blood, but that the entry rate may be limited when high doses are administered.

Figure 2.

Blood-to-brain entry of ¹²⁵I-exendin-4 (I-Exendin-4) with (+Exendin-4) and without the addition of 5 g/mouse of unlabeled exendin-4. The self-inhibition characteristic of a saturable transport system was only significant (P<0.05) when the results of four individual experiments (n=7/group/experiment) were combined. The lack of effect of the excess exendin-4 on the influx of the simultaneously administered ^99mTc-albumin shows the absence of any disruption of the BBB.

Full figure and legend (22K)

In each experiment, influx of ¹²⁵I-exendin-4 (with or without unlabeled exendin-4) was significantly (P<0.0001) faster than the K_i of ^99mTc-albumin. The combined values for the ^99mTc-albumin were -1.6881.118 10^-4 ml/g min when coinjected with ¹²⁵I-exendin-4 alone and -0.1161.009 10^-4 ml/g min when the injectate consisted of ¹²⁵I-exendin-4+5 g unlabeled exendin-4. These values for ^99mTc-albumin were not significantly different from zero, indicating that an excess (5 g) of the exendin-4 did not disrupt the BBB (Figure 2).

Exendin-4 shows the same affinity as GLP-1 to binding sites that are present in the CVOs.⁴ Figure 3 shows the similar rates of entry of ¹²⁵I-exendin-4 into whole brain, cortex alone (devoid of CVOs and choroid plexuses), and the remainder of the brain (containing CVOs and choroid plexuses). This shows that most of the exendin-4 injected i.v. reaches the brain through the BBB rather than through CVOs.

Figure 3.

Blood-to-brain entry of ¹²⁵I-exendin-4 into whole brain, cortex (devoid of CVOs), and the remainder of the brain (containing CVOs). There were no significant differences, indicating that most of the ¹²⁵I-exendin-4 entered the brain by crossing the BBB.

Full figure and legend (11K)

The entry of a few ingestive peptides with readily saturable transport systems into the brain is affected by food deprivation.^11,12 Food deprivation for 24 h was without significant effect on the influx of ¹²⁵I-exendin-4.

Entry of ¹²⁵I-exendin-4 during in situ brain perfusion in blood-free buffer

In the blood-free in situ brain perfusion system, ¹²⁵I-exendin-4 was delivered at a constant rate together with ^99mTc-albumin. There was no significant inhibition in any of the three separate experiments, so the results were combined. As illustrated in Figure 4, the combined group with only ¹²⁵I-exendin-4 had a high influx rate of 18.953.70 10^-3 ml/g min (r=0.730). However, even when the results were pooled, the decrease (10.502.60 10^-3 ml/g min, r=0.690) in the rate of entry of ¹²⁵I-exendin-4 in the group receiving the excess unlabeled exendin-4 only approached statistical significance (P=0.077). This shows that protein binding or other blood processes do not explain the weak nature of any self-inhibition of exendin-4 influx seen after i.v. injection. Constant exposure during perfusion to the high dose does not show the almost complete inhibition of transport typical of a high-capacity saturable system when assessed by in situ perfusion.

Figure 4.

Influx (K_i) of ¹²⁵I-exendin-4 (I-Exendin-4) into brain by in situ perfusion in blood-free buffer plotted against time. These results from three individual experiments (n=7/group/experiment) show that the trend toward self-inhibition with excess unlabeled exendin-4 (+Exendin-4) did not reach statistical significance (P=0.077).

Full figure and legend (14K)

Brain-to-blood efflux of ¹²⁵I-exendin-4

There was a linear relationship between the logarithm of brain radioactivity and time for both ¹²⁵I-exendin-4 (r=0.94) and ^99mTc-albumin (r=0.91). The half-time disappearance was 14.6 min for ¹²⁵I-exendin-4 and 15.7 min for ^99mTc-albumin. The difference between the two regression lines was not statistically significant. These results are shown in Figure 5. They fail to provide any evidence for a rapid transport system out of the brain for exendin-4 that might have confounded the determination of the rate of influx.

Figure 5.

Efflux of i.c.v. ¹²⁵I-exendin-4 (I-Exendin-4) from brain. The ¹²⁵I-exendin-4 did not exit significantly faster than the usual reabsorption of cerebrospinal fluid as measured with ^99mTc-albumin (Albumin) (n=5/group).

Full figure and legend (15K)

Octanol/buffer partition coefficient

The log D octanol coefficient, calculated as the cpm in the octanol phase divided by the cpm in the buffered saline phase, was 0.0990.006 for ¹²⁵I-exendin-4. This shows that exendin-4 is highly lipophilic.

Hydrogen bonding

The isooctane/buffer partition coefficient for ¹²⁵I-exendin-4 was 0.0470.006. The difference between the logarithms of the octanol and isooctane partition coefficients, representing hydrogen bonding, was 0.344.

Discussion

Although there are several indirect ways by which a substance in blood can affect the brain,¹³ we showed that exendin-4 is stable in blood and can directly penetrate the BBB in intact form. This was not dependent upon CVOs. Coadministration of albumin, used as the vascular control, showed that this did not occur by leakage. Entry, therefore, occurred either by transmembrane passive diffusion or by a saturable transport mechanism. For most ingestive peptides and polypeptides, one of these processes clearly predominates within the usual test range. For exendin-4, entry appeared to be passive except that a high dose of the peptide showed weak self-inhibition. This unusual situation is compatible with a transport system of limited capacity. It has practical implications for a compound with therapeutic potential since it might partially limit the effectiveness of very high doses.

Although GLP-1 is rapidly metabolized,¹ addition of a hydroxyl group to its Ala⁸ changes it to Ser,⁸ which provides a more stable form of GLP-1 that retains strong biological activity and receptor specificity.¹⁴ This Ser⁸ analogue of GLP-1 has an even more rapid rate of entry into the brain than exendin-4, and its influx is not saturable even at a high dose (5 g/mouse).¹⁵ However, [Ser⁸]GLP-1 shows a high degree of loose association within brain capillaries so that it may be less effective in reaching the parenchyma of the brain.

Substances with saturable transport systems across the BBB are of particular interest because their rates of influx are more susceptible to manipulation by pharmacological and physiological factors like blood concentrations and food deprivation. If the system has limited capacity, as seems to occur with exendin-4, it should be less susceptible to these influences which also would not affect passive diffusion that is mainly dependent on physico-chemical factors such as lipophilicity and hydrogen bonding.

The lipophilicity of ¹²⁵I-exendin-4, as measured by its octanol/buffer partition coefficient, was greater than that found for many other ingestive peptides. These include [Ser⁸]GLP-1,¹⁵ adrenomedulin,¹⁶ agouti-related protein (83–132) (AgRP),¹⁷ cocaine- and amphetamine-related transcript (55–102) (CART),¹⁸ galanin-like peptide (GALP),¹¹ insulin,¹⁹ mahogany (1377–1428),²⁰ and neuropeptide Y (NPY).²¹

The octanol/buffer partition coefficient for exendin-4 was about the same as that for Phe¹³,Tyr¹⁹-melanin-concentrating hormone (MCH)²² and less than that for the urocortins (UCN),²³ orexin A,²⁴ and cyloHis-Pro.²⁵ None of these peptides has a saturable transport system into the brain.

Another physico-chemical determinant of penetration of the BBB is hydrogen bonding. It has been reported that hydrogen bonding potential—the energy required to desolvate polar amide bonds—rather than lipophilicity has the greatest influence on the passive diffusion of small D-phenylalanine containing model peptides in which the amide bond was sequentially N-methylated to alter the hydrogen bonding.²⁶ Using the same methods, we found that the hydrogen bonding value for ¹²⁵I-exendin-4 was less than half of that found for the three UCN peptides (the only ones measured by our group so far). Low hydrogen bonding potential might be advantageous for translocation to the highly apolar interior region of the cell membrane.

The lipophilic and hydrogen bonding properties of exendin-4 are consistent with the ability of exendin-4 to readily cross the BBB. Its rate of entry is almost the same as that for insulin²⁷ and leptin.⁶ Yet injection of only 1 g/mouse of excess insulin²⁸ or leptin^6,29 almost totally inhibited their entry into brain. For exendin-4, the larger dose of 5 g/mouse only showed statistically significant inhibition of entry when the results of four experiments were combined, and even then the inhibition was much less than the inhibition seen for insulin and leptin with smaller doses.

Together, our results show that intact exendin-4 readily enters the brain, but that there may be some limit to the extent of this entry at high doses.

References

1. Kieffer TJ, McIntosh CHS, Pederson RA. Degradation of glucose-dependent insulinotropic polypeptide and truncated glucagon-like peptide 1 in vitro and in vivo by dipeptidy1 peptidase IV. Endocrinology 1995; 136: 3585–3596. | Article | PubMed | ISI | ChemPort |
2. Young AA, Gedulin BR, Bhavsar S, Bodkin N, Jodka C, Hansen B, Denaro M. Glucose-lowering and insulin-sensitizing actions of exendin-4: studies in obese diabetic (Ob/Ob, Db/Db) mice, diabetic fatty zucker rats and diabetic rhesus monkeys (Macaca mulatta). Diabetes 1999; 48: 1026–1034. | Article | PubMed | ISI | ChemPort |
3. Rodriquez de Fonseca F, Navarro M, Alvarez E, Roncero I, Chowen JA, Maestre O, Gomez R, Mundoz RM, Eng J, Blazquez E. Peripheral versus central effects of glucagon-like peptide-1 receptor agonists on satiety and body weight loss in zucker obese rats. Metabolism 2000; 49: 709–717. | Article | PubMed | ChemPort |
4. Goke R, Larsen PJ, Mikkelsen JD, Sheikh SP. Distribution of GLP-1 binding sites in the rat brain: evidence that exendin-4 is a ligand of brain GLP-1 binding sites. Eur J Neurosci 1995; 7: 2294–2300.
5. Banks WA, Kastin AJ. Opposite direction of transport across the blood–brain barrier for Tyr-MIF-1 and MIF-1: comparison with morphine. Peptides 1994; 15: 23–29.
6. Kastin AJ, Akerstrom V, Pan W. Validity of multiple-time regression analysis in measurement of tritiated and iodinated leptin crossing the blood-brain barrier: meaningful controls. Peptides 2001; 22: 2127–2136. | Article | PubMed | ChemPort |
7. Banks WA, Kastin AJ. Measurement of transport of cytokines across the blood–brain barrier. In: Conn PM, De Souza EB (eds). Neurobiology of cytokines, Part A. Academic Press Inc.: San Diego; 1993. pp 67–77.
8. Banks WA, Fasold MB, Kastin AJ. Measurement of efflux rates from brain to blood. In: Irvine GB, Williams CH (eds). Methods in molecular biology, neuropeptide protocols, Vol. 73. Humana Press Inc.: Totowa, NJ; 1997. pp 353–360.
9. Burton PS, Conradi RA, Hilgers AR, Ho NFH, Maggiora LL. The relationship between peptide structure and transport across epithelial cell monolayers. J Control Rel 1992; 19: 87–98.
10. Burton PS, Conradi RA, Hilgers AR. Mechanisms of peptide and protein absorption, (2) transcellular mechanism of peptide and protein absorption: passive aspects. Adv Drug Deliv Rev 1991; 7: 365–386.
11. Kastin AJ, Akerstrom V, Hackler L. Food deprivation decreases blood galanin-like peptide and its rapid entry into the brain. Neuroendocrinology 2001; 74: 423–432.
12. Kastin AJ, Akerstrom V. Fasting, but not adrenalectomy, reduces transport of leptin into the brain. Peptides 2000; 21: 679–682.
13. Kastin AJ, Wade LA, Coy DH, Schally AV, Olson RD. Peptides and the blood–brain barrier. In: Wuttke W, Weindl A, Voigt KH, Dries RR (eds). Brain and pituitary peptides. Karger: Munich; 1980. pp 71–78.
14. Ritzel U, Leonhardt U, Ottleben M, Ruhmann A, Eckart K, Spiess J, Ramadori G. A synthetic glucagon-like peptide-1 analog with improved plasma stability. J Endocrinol 1998;159: 93–102.
15. Kastin AJ, Akerstrom V, Pan W. Interactions of glucagon-like peptide-1 (GLP-1) With the blood–brain barrier. J Molec Neurosci 2002; 18: 7–14.
16. Kastin AJ, Akerstrom V, Hackler L, Pan W. Adrenomedullin and the blood–brain barrier. Hormones Metab Res 2001; 33: 19–25.
17. Kastin AJ, Akerstrom V, Hackler L. Agouti-related protein (83–132) aggregates and crosses the blood–brain barrier slowly. Metabolism 2000; 49: 1444–1448. | Article | PubMed | ChemPort |
18. Kastin AJ, Akerstrom V. Entry of CART into brain is rapid but not inhibited by excess CART or leptin. Am J Physiol 1999;277: E901–E904.
19. Banks WA, Jaspan JB, Kastin AJ. Effect of diabetes mellitus on the permeability of the blood–brain barrier to insulin. Peptides 1997; 18: 1577–1584.
20. Kastin AJ, Akerstrom V. Mahogany (1377–1428) enters brain by a saturable transport system. J Pharmacol Exp Ther 2000;294: 633–636. | PubMed | ChemPort |
21. Kastin AJ, Akerstrom V. Nonsaturable entry of neuropeptide Y into the brain. Am J Physiol 1999; 276: E479–E482.
22. Kastin AJ, Akerstrom V, Hackler L, Zadina JE. Phe¹³,Tyr¹⁹-melanin-concentrating hormone and the blood-brain barrier: role of protein binding. J Neurochem 2000; 74: 385–391.
23. Kastin AJ, Akerstrom V, Pan W. Activation of urocortin transport into brain by leptin. Peptides 2000; 21: 1811–1818.
24. Kastin AJ, Akerstrom V. Orexin A but not orexin B rapidly enters brain from blood by simple diffusion. J Pharmacol Exp Ther 1999; 289: 219–223. | PubMed | ChemPort |
25. Banks WA, Kastin AJ, Akerstrom V, Jaspan JB. Radioactively iodinated cyclo(His-Pro) crosses the blood–brain barrier and reverses ethanol-induced narcosis. Am J Physiol 1993; 267: E723–E729.
26. Chikhale EG, Burton PS, Borchardt RT. The effect of verapamil on the transport of peptides across the blood–brain barrier in rats: kinetic evidence for an apically polarized efflux mechanism. J Pharmacol Exp Ther 1995; 273: 298–303. | PubMed | ChemPort |
27. Banks WA, Kastin AJ. Differential permeability of the blood–brain barrier to two pancreatic peptides: insulin and amylin. Peptides 1998; 19: 883–889.
28. Banks WA, Jaspan JB, Huang W, Kastin AJ. Transport of insulin across the blood–brain barrier: saturability at euglycemic doses of insulin. Peptides 1997; 18: 1423–1429. | Article | PubMed | ISI | ChemPort |
29. Banks WA, Kastin AJ, Huang W, Jaspan JB, Maness LM. Leptin enters the brain by a saturable system independent of insulin. Peptides 1996; 17: 305–311. | Article | PubMed | ISI | ChemPort |

Acknowledgements

Supported by NIH (DK 54880) and the Department of Veterans Affairs.