Evidence that [125I]N-Tyr-delta sleep-inducing peptide crosses the blood-brain barrier by a non-competitive mechanism

Brain Research, 301 (1984) 201-207 Elsevier

201

BRE 10016

Research Reports

Evidence that [125I]N-Tyr-delta sleep-inducing peptide crosses the blood-brain barrier by a non-competitive mechanism WILLIAM A. BANKS*, ABBA J. KASTIN and DAVID H. COY

VA Medical Center and Tulane University School of Medicine, New Orleans, LA 70146 (U.S.A.) (Accepted October llth, 1983)

Key words: blood-brain barrier - - peptide - - sleep - - brain - - cerebrospinal fluid - - plasma binding - age - - delta sleep-inducing peptide

Delta sleep-inducing peptide (DSIP), a nonapeptide, has previously been shown to cross the blood-brain barrier (BBB) in rats4,8,9 and the blood-CSF barrier in dogs2. New experiments were conducted to determine if this crossing was competitive. Neither DSIP nor several analogs, including non-radioactive [127I]N-Tyr-DSIP,injected by the jugular vein or carotid artery, inhibited passage of radioactive [12SI]N-Tyr-DSIPacross the rat BBB. Column chromatography of brain samples confirmed that the radioactivity in the brain represented intact [125I]N-Tyr-DSIPand that the non-radioactive competing materials did not interfere with the degradation or binding of [125I]N-Tyr-DSIP. In addition, N-Tyr-DSIP was unable to inhibit the appearance of radioactive [125I]N-Tyr-DSIPin the CSF of dogs. In conclusion, the evidence from these experiments suggests that [125I]N-Tyr-DSIPcrosses the rat BBB and dog blood-CSF barrier by a non-competitive mechanism. INTRODUCTION It is now widely accepted that peptides administered peripherally can affect brain function and behavior6,7. In addition, levels in the C S F of some of the g u t - b r a i n p e p t i d e s have been r e p o r t e d to correlate with their b l o o d levels1-3,1],15. These findings have raised the possibility that p e p t i d e s m a y cross the b l o o d - b r a i n or b l o o d - C S F barriers (BBB). Although this is currently disputed for m a n y peptides, we have shown that D S I P crosses the B B B in rats4,8,9 and dogs 2. In this p a p e r , we e x a m i n e d the question of whether this crossing could be inhibited in a competitive fashion. MATERIALS AND METHODS D S I P (delta sleep-inducing p e p t i d e ) and its analogs were synthesized by solid phase methods. NT y r - D S I P was iodinated with chloramine-T and pu-

d r i e d by partition c h r o m a t o g r a p h y on a column of Sephadex G-25 (fine) with a b u t a n o l - a c e t i c acid-water solution in a ratio of 4:1:5. Purity was checked by chromatoelectrophoresis and a n t i b o d y binding.

Study I: competition using D S I P Male rats o b t a i n e d from Blue Spruce F a r m s (Altamont, NY) and weighing 200-250 g were anesthetized with i.p. p e n t o b a r b i t a l (65 mg/kg). The jugular veins were then exposed and 0.5 ml of I of 5 solutions ( p r e p a r e d with 0.9% NaC1) injected. F o u r of the solutions varied only in their concentration of D S I P , so that each animal received 2 x 105 cpm of [125I]NT y r - D S I P and 0, 0.01, 0.1, or 1.0 mg of DSIP. The fifth solution contained only r a d i o i o d i n a t e d serum albumin ( R I S A ) and was used as an intravascular marker. A d d i t i o n a l concentrations of D S I P tested were 5 ng, 50 ng, 500 ng, 5/~g, 50/~g, or 500 pg/animal (n = 3/group). Two min after injection, the other jugular vein was incised and about 0.5 ml of blood

Correspondence: W. A. Banks, VA Medical Center and Tulane University School of Medicine, New Orleans, LA 70146, U.S.A. 0006-8993/84/$03.00 (~) 1984 Elsevier Science Publishers B .V.

202 was collected in a chilled 12 mm × 75 mm borosilicate tube containing 50 U of heparin sulfate (1000 U/ml). The rat was then immediately decapitated and the whole brain (except the pineal and pituitary) removed, rinsed in normal saline, and counted in a gamma counter for 3 rain. Blood was centrifuged at 4 °C for 10 min at 1300 g and 0.1 ml of plasma counted for 3 rain. Additional animals were injected via the carotid artery and decapitated 5 s later. Blood samples (collected from the trunk) and brains were treated as above. Only 0 and 1 mg doses were tested in this part. The brain to plasma ratio for each animal was determined by dividing the cpm for whole brain by the cpm for 1 ml of plasma and multiplying by 100. Means are presented with their Standard errors. Intact, bound, and damaged [1251]N-Tyr-DSIP were separated on a 60 cm × 1 cm column of G-15 Sephadex (fine). Distilled water containing 4% Trasylol was added to 17 mm x 100 mm chilled polypropylene tubes containing whole brains from animals injected via the jugular vein and this was homogenized with a polytron at a setting of 4.5 for 30 s. The homogenate was centrifuged at 4500 g for 30 min and the supernatant chromatographed. Plasma samples were added directly to the column with no pretreatment.

Study H: effect of size Male Blue Spruce rats ranging in size from 70 to 870 g were injected via the jugular vein. The animals formed 3 size groups: small (70-100 g), medium (275-376 g) and large (650-870 g). Weight determined the amount of pentobarbital (65 mg/kg), [125I]N-Tyr-DSIP (1.6 × 106 cpm/kg), and DSIP (1 mg/kg) given. Blood collection, circulation time, techniques of decapitation, brain removal and processing of samples were the same as in study I.

Study III: competition using non-iodinated DSIP analogs Male Blue Spruce rats (200-250 g) were injected via the jugular vein with 1 of 4 solutions so that, in addition to 2 x 105 cpm of [I25I]N-Tyr-DSIP, they received 0.5 mg of either DSIP, N-Tyr-DSIP, D-Ala 4DSIP-NH2, or no unlabeled peptide. The procedure was otherwise identical to study I.

Study IV: competition using non-radioactive [I:71]NTyr-DSIP Non-radioactive [le71]N-Tyr-DSIP was prepared by the same technique used to produce radioactive [1zSI]N-Tyr-DSIP. Both mixtures were dried with a stream of nitrogen gas. These samples were then reconstituted with 0.9% NaCI so that the final solution for injection contained either [1251]N-Tyr-DSIP or a 10:1 ratio of non-radioactive [127I]N-Tyr-DSIP to radioactive [125I]N-Tyr-DSIP. Male Blue Spruce rats weighing about 250 g were injected with about 2 x 105 cpm under one of the following protocols. (1) Jugular vein injection. This protocol was identical to the jugular vein injections of study I, except that [127I]N-Tyr-DSIP was used as the competing material. (2) Carotid artery injection: no washout. This protocol was identical to the study I carotid artery injections, except that [Ie71]N-Tyr-DSIP was used as the competing material. (3) Carotid artery injection: 60 s washout. Five seconds after carotid artery injection a blood sample was taken from the contralateral jugular vein. A 60-s washout with 0.9% NaC1 was then begun through a large bore needle directed toward the brain that occluded the jugular vein proximally. The volume of wash, as measured by collection from the contralateral jugular vein, was about 30 ml and near the end of the 60 s tended to appear clear. After completion of washout, the animal was decapitated, and brain and blood samples treated as above.

Study V: competition in the dog blood-CSF barrier Two mongrel dogs were anesthetized with i.v. sodium pentobarbital (30 mg/kg). They were injected in a hindleg vein with about 108 cpm of [125I]N-TyrDSIP either with or without 0.1 mg/kg of N-TyrDSIP. Blood samples were collected from a foreleg vein and CSF (0.5 ml) from the posterior fossa before ('0' minute sample) and at 10, 20, 30, 45 and 60 min after injection. Both CSF and blood samples were centrifuged at 4 °C for 10 min at 1300 g and 0.2 ml counted on a gamma counter. RESULTS

Study I: competition using DSIP Analysis of variance followed by Duncan's mul-

203 TABLE I

DSIP competition for [1251]N-Tyr-DSIP: dose and injection route No statistical differences were found with either route or amount of injection. The brain/plasma ratios show that, based on RISA, less than 10% of the [125I]N-Tyr-DSIP was intravascular.

DSIP/animal

Route of injection

n

Plasma (cpm/ml)

0 mg 0.01 mg 0.1 mg 1 mg RISA

jugular vein jugular vein jugular vein jugular vein jugular vein

10 9 10 9 8

1687 1577 1717 1489 8712

0 mg 1 mg

carotid artery carotid artery

6 6

+ + + + +

Whole brain (cpm)

75 70 56 183 611

389 384 410 394 195

1840 + 200 2180 _+ 266

+ + + + +

(Brain~plasma)× 100

18 13 18 17 35

23.6 24.6 24.1 24.8 2.19

997 + 128 1054 + 83

+ + + + +

1.7 1.0 1.4 1.7 0.297

54.8 + 5.7 51.5 + 7.0

TABLE II

DSIP competition: column chromatography Most of the brain radioactivity represented intact [125I]N-Tyr-DSIP. The presence of exogenous DSIP had little effect on degradation and binding of [125I]N-Tyr-DSIP.

Source

Dose (rag)

% Bound

% Free iodine

% [1251]N-Tyr-DSIP

Plasma Plasma Brain Brain

0 1 0 1

1.5 2.1 1.5 2.6

43.7 46.8 22.6 18.9

49.8 42.8 74.5 65.1

tiple range test showed that no significant differences

activity in t h e b r a i n t h a n in t h e b l o o d . T h e s i m u l t a -

occurred among the varying concentrations injected

n e o u s i n j e c t i o n of 1 m g o f D S I P w i t h [125I]N-Tyr-

by the jugular vein for plasma, whole brain, or brain

D S I P a p p e a r e d t o h a v e little e f f e c t o n t h e e l u t i o n p a t -

to p l a s m a v a l u e s ( T a b l e I). T h e b r a i n / p l a s m a r a t i o

t e r n s of b r a i n o r b l o o d ( T a b l e II).

f o r R I S A w a s less t h a n o n e - t e n t h t h a t o f [1251]N-TyrD S I P , d e m o n s t r a t i n g t h a t v e r y little o f t h e [I25I]NT y r - D S I P in t h e b r a i n w a s i n t r a v a s c u l a r . L i k e w i s e ,

Study II: effect o f size The mean

weight (+

S.E.)

for small rats was

no differences occurred between the 0 mg and 1 mg

88.2 + 5.2 g, f o r m e d i u m r a t s w a s 323 + 11.0 g, a n d

doses injected by the carotid artery.

f o r l a r g e r a t s w a s 747 + 38.4 g. T h e c p m in b o t h

Column chromatography

showed that intact pep-

b r a i n a n d p l a s m a i n c r e a s e d w i t h t h e size o f t h e ani-

tide accounted for a higher percentage of the radio-

mal, suggesting that the volume of distribution for

TABLE III

DSIP competition: effect of size of rat Competition between DSIP and [125I]N-Tyr-DSIP was not observed in any size rat.

DSIP (1 mg/kg)

Small

wt(g)

+

Plasma (cpm/ml)

+

Whole brain (cpm)

+

(Brain/Plasma) × 100

+

93 85 2320 2620 731 647 32.4 25.4

- 3 --- 9 + 220 + 410 + 152 + 62 + 9.6 + 2.5

Medium

Large

328 320 3560 4660 626 826 17.3 18.5

770 700 5470 4990 1247 1056 23.5 22.0

+ + + + + + +

20 17 150 66 47 111 1.4 2.3

+ 47 + 75 + 480 + 910 + 204 + 51 + 4.5 -_+ 5.0

204 TABLE IV

Effect of DS1P analogs Neither DSIP nor the analogs inhibited BBB passage of [12SI]N-Tyr-DSIP.

Peptide

n

Plasma/ml (cpm/ml)

Whole brain (cpm)

(Brain/Plasma) x 100

Control DSIP N-Tyr-DSIP o-Ala4-DSIP-NH2

8 10 9 10

1558 + 81 1616 _+ 72 1558 + 75 1620 + 55

365 361 377 368

24.0 + 1.5 22.5 _+ 0.94 24.8 + 1.9 23.1 + 1.6

+ + + +

9.9 15.4 15.2 18.5

DSIP does not increase linearly with weight (Table-

collected from the jugular vein representing blood

III). Injection of unlabeled DSIP produced no statistically significant differences within a group size in

returning directly from the brain rather than from the trunk representing mixed venous and arterial blood.

cpm for plasma or whole brain, or in the brain/plasma ratio. Regression analysis showed no statistically significant correlation between rat size and brain/blood

For any given method, however, no significant differences in values occurred between animals injected with or without non-radioactive [lZ7I]N-Tyr-DSIP for

ratios.

plasma, brain, or their ratio values (Table V).

Study III: competition using non-iodinated D S I P analogs

Study V: competition in the dog b l o o d - C S F barrier

No significant differences occurred among the 4 solutions for values in plasma, brain, or the brain/ plasma ratio (Table IV). The brain/plasma ratio for DSIP (22.5 _+ 1.5) was lower than for D-Ala4-DSIP NH 2 (23.1 _+ 1.6) or for N-Tyr-DSIP (24.8 _+ 1.9), but these differences were not statistically signifi-

Peaks in radioactivity in the CSF occurred between 30 and 45 min after injection. The peak brain/ plasma ratio was 1.8 for the dog injected with [t25I]N-Tyr-DSIP and 2.04 for the dog for the dog that also received unlabeled N-Tyr-DSIP (TableVI). DISCUSSION

cant.

[127I]N-

We have previously shown that small amounts of DSIP can cross the BBB in the rat 4,8,9 and the

Higher brain levels were achieved after carotid ar-

b l o o d - C S F barrier in the dog 2. In the rat, injection of

tery injection without washout than by either of the other two methods. Blood levels after carotid injections were higher in the rats after washout than in the rats not washed-out because the blood samples were

[125I]N-Tyr-DSIP resulted in brain levels of radioactivity about seven times higher than after injection of RISA used as a marker for contamination of tissues

Study IV: competition using non-radioactive Tyr-DSIP

by blood 4. Some analogs of DSIP crossed better than

TABLE V

[1251]N-Tyr-DSIP vs [1271]N-Tyr-DSIP Non-radioactive [127I]N-Tyr-DSIPdid not inhibit BBB passage of [lZSl]N-Tyr-DSIPby either route of injection or with washout.

Route of inlection

Material injected

n

Jugular vein

diluent peptide

5 5

Plasma/ml (cpm/ml) 1091 + 65 1117 + 53

Carotid artery - - no washout

diluent peptide

6 6

2991 + 429 3044 + 462

Carotid artery - - washout

diluent peptide

6 5

16472 + 1708 15826 +__1990

Whole brain (cpm) 323 + 14.6 315 + 17.1 1067 + 96 1272 + 169 297 + 31.1 377 + 54.5

(Brain~plasma)×100 29.8 + 1.49 28.4 + 2.26 39.6 + 5.52 45.4 + 6.54 1.83 + 0.145 2.40 + 0.281

205 TABLE VI Competition in the dog blood-CSF barrier

N-Tyr-DSIP did not compete with the penetrance of [I:SI]N-Tyr-DSIPacross the blood-CSF barrier in the dog. Time

0 10 20 30 45 60

[t251]N- Tyr-D SIP only

[1251]N-Tyr-DSIP and N-Tyr-DSIP

CSF(ml)

Plasma (ml)

CSF (lO0)/Plasma

CSF(ml)

Plasma (ml)

CSF (lO0)/Plasma

0 280 370 395 400 345

0 26525 24530 24055 22120 22195

-1.06 1.51 1.64 1.81 1.55

0 305 345 380 260 270

0 22090 18780 18665 17235 17365

-1.38 1.85 2.04 1.51 1.55

others 4,8. Size (and, therefore, perhaps age) also appeared to be a factor, with more crossing occurring in large rats. Because of the greater crossing of [125I]NTyr-DSIP as compared to RISA, and the varying degrees of passage of analogs, a specific mechanism seemed possible. Despite these previous studies, however, the mechanism by which DSIP crosses the rat BBB remained undetermined. The present studies confirm that [125I]N-Tyr-DSIP crosses the BBB regardless of the method of administration. Results from study I show that, as assessed by RISA, less than 10% of the [125I]N-Tyr-DSIP in the brain is intravascular. Likewise we have shown that [125I]N-Tyr-DSIP brain levels are 5-7 times greater than RISA levels when the intracarotid injection washout technique is used 4 and two times greater when washout is not performed (unpublished data). Those counts that are not intravascular are presumed to have crossed the BBB. Since easily detectable radioactive compounds were used, passage (and hence competition) should be demonstrable even if the transport mechanism were saturated at physiologic levels, since radioactive peptide would freely intermix with endogenous material. Also, because large amounts of unlabeled peptide were injected, even a relatively high capacity transport system should have been saturated. Furthermore, if a saturable system existed only for transporting peptide out of the brain, the brain levels of radioactivity should increase with administration of unlabeled materia113. The results presented here do not support a competitive transport mechanism for [125I]N-Tyr-DSIP. Inhibition of the passage of [125I]N-Tyr-DSIP could not be demonstrated with various amounts of unla-

beled peptide (0.01 mg/animal, 0.1 rag/animal, and 1 mg/animal), analogs (N-Tyr-DSIP, D-Ala4-DSIPN H 2 and [127I]N-Tyr-DSIP), time intervals between injection and decapitation (5 s, 65 s, 2 min), techniques (washout and no washout) and sizes of rat (70-870 g). Brain levels of exogenous radioactivity increased with increasing animal size. This is consistent with previous studies4, 8 in which it was shown that bigger animals had larger increases in brain levels of DSIPlike immunoreactivity than did smaller animals after the intracarotid injection of DSIP on a mg/kg basis. In this study, however, the blood levels of radioactivity that were also higher in the larger animals suggested that the volume of distribution for [125I]N-TyrDSIP did not change proportionately with weight. 'No clear correlation existed between brain/blood ratios and weight. In previous studies, DAlaa-DSIP-NH2 was shown to be one of the analogs that crossed the rat BBB best 4. By comparison, smaller amounts of N-TyrDSIP crossed, but because of its similarity to [125I]NTyr-DSIP, this decapeptide was used in these experiments. Neither analog nor DSIP inhibited the crossing of [125I]N-Tyr-DSIP. Non-radioactive [127I]NTyr-DSIP did not inhibit crossing either, regardless of route of injection (jugular vein vs carotid artery) or use of washout. In dogs, we have shown that DSIP crosses the blood-CSF barrier 2. In that study, a correlation was observed between plasma and CSF levels in the basal state and after injection of [125I]N-Tyr-DSIP, but not after the injection of 0.1 mg/kg. DSIP. Since the [125I]N-Tyr-DSIP injection contained only about 0.1% of the amount of peptide present in the injec-

206 tion of 0.1 mg/kg of unlabeled peptide, this is compatible with saturation. Free DSIP seemed to cross more than bound DSIP. Analogs showed varying CSF/ plasma ratios. Taken together, these results suggested a relatively specific mechanism for BBB passage of DSIP in the dog. Injection of N-Tyr-DSIP, however, did not inhibit the appearance of [125I]NTyr-DSIP in dog CSF in the present studies. Chromatography of samples from the rat showed most of the radioactivity in the brain to coelute with [125I]N-Tyr-DSIP demonstrating that intact peptide can enter the brain. Adjustment of the levels to reflect only free, intact peptide would increase brain/ plasma ratios by about 50% both for animals that received peptide and for those that did not receive peptide. The small differences in levels of intact [125I]NTyr-DSIP between animals that received unlabeled peptide and those that did not rules out the possibility that DSIP might have protected [125I]N-Tyr-DSIP from degradation or binding in the plasma. In such a case, more of the radioactivity in the plasma would have been intact or free [125I]N-Tyr-DSIP and available for crossing. Thus, the results of the present experiments indicate that DSIP, and perhaps other peptides as well, cross the BBB by a non-competitive and presumably non-specific mechanism. It is also conceivable that [125I]N-Tyr-DSIP could cross independently of DSIP, N-Tyr-DSIP, D-Ala4-DSIP-NH2, and [127I]N-TyrDSIP or that higher levels of DSIP than those tested are needed to block passage of [125I]N-Tyr-DSIP. These results provide additional information important in understanding the actions of peptides on

the brain. When administered peripherally, peptides do affect brain function and behavior and do cross the BBB in small amounts. While some believe that this crossing accounts for the behavioral effects seen after peripheral administrationS,9,10, ~2, others continue to deny that crossing occurs 14. Also, the CSF and plasma levels have been shown to correlate for several peptidesl-3,11,LL Although this is a growing list, it is still unclear if the correlation is due to direct BBB passage of the peptides or to some unknown mechanism that coordinates both levels. Furthermore, previous studies in rats4, 9 and dogs 2 show that some analogs of DSIP cross better than others. The current results would suggest that these differences in crossing may be due to variations in such factors as degree of protein binding, distribution, degradation, etc. rather than specificity of the crossing mechanism. The amount of peptide that crosses the BBB seems to depend on physiochemical parameters other than just molecular size, suggesting that the BBB may exhibit selectivity and specificity towards peptides not based on classic competitive mechanisms. Further elucidation of BBB/peptide interactions should be important for the eventual therapeutic uses of peptides.

REFERENCES

5 Kastin, A. J., Banks, W. A., Marks, N. and Olson, R. D., Dissociative effects of CNS peptides with some unusual applications of RIA. In Ch. A. Bizollon (Ed.), Physiological Peptides and New Trends in Radioimmunoassay, Elsevier/ North-Holland Biomedical Press, Amsterdam, 1981, pp. 89-99. 6 Kastin, A. J., Banks, W. A., Zadina, J. E. and Graf, M., Brain peptides: the dangers of constricted nomenclatures, Life Sci., 32 (1983) 295-301. 7 Kastin, A. J., Coy, D. H., Schally, A. V. and Miller, L. H., Peripheral administration of hypothalamic peptides results in CNS changes, Pharmacol. Res. Comrnun., 10 (1978) 293-312. 8 Kastin, A. J., Nissen, C. and Coy, D. H., Permeability of blood-brain barrier to DSIP peptides, Pharmacol. Biochem. Behav., 15 (1981) 955-959. 9 Kastin, A. J., Nissen, C., Schally, A. V. and Coy, D. H.,

1 Banks, W. A., Evidence for a cholecystokinin gut-brain axis with modulation by bombesin, Peptides, 1 (1980) 347-351. 2 Banks, W. A., Kastin, A. J. and Coy, D. H., Delta sleepinducing peptide crosses the blood-brain barrier in dogs: some correlations with protein binding, Pharmacol. Biochem. Behav., 17 (1982) 1009-1014. 3 Hammer, M., Sorensen, P. S., Gjerris, F. and Larsen, K., Vasopressin in the cerebrospinal fluid of patients with normal pressure hydrocephalus and benign intracranial hypertension, Acta endocr., 100 (1982) 211-215. 4 Kastin, A. J., Banks, W. A., Castellanos, P. F., Nissen, C. and Coy, D. H., Differential penetration of DSIP peptides into rat brain, Pharmacol. Biochem. Behav., 17 (1982) 1187-1191.

ACKNOWLEDGEMENTS This work was supported in part by the Veterans Administration and ONR. We also thank Cheryl Nissen and Paul Castellanos for their technical assistance and Linda A. May for secretarial assistance and Dr. James E. Zadina for assistance with statistical analysis.

207 Additional evidence that small amounts of a peptide can cross the blood-brain barrier, Pharmacol. Biochem. Behav., 11 (1979) 717-719. 10 Kastin, A. J., Olson, R. D., Schally, A. V. and Coy, D. H., CNS effects of peripherally administered brain peptides, Life Sci., 25 (1979) 401-414. 11 Luersson, T. G. and Robertson, G. L., Cerebrospinal fluid vasopressin and vasotocin in health and disease. In J. Wood (Ed.), Neurobiology of Cerebrospinal Fluid, Plenum Press, New York, 1980, pp. 613-623. 12 Mens, W. B. J., Witter, A. and Van Wimersma Greidanus, T. B., Penetration of neurohypophyseal hormones from plasma into cerebrospinal fluid (CSF): half-times of disap-

pearance of these neuropeptides from CSF, Brain Re-

search, 262 (1983) 143-149. 13 Reed, D. J., Woodbury, D. M., Jacobs, L. and Squires, R., Factors affecting distribution of iodide in brain and cerebrospinal fluid, Amer. J. Physiol., 209 (1965) 757-764. 14 Verheugen, C., Laufer, L. R., DeFazio, J. L., Pardridge, W. B., Lu, J. K. H., and Judd, H. L., Impermeability of the rat blood-brain barrier to exogenously administered gonadotropin-releasing hormone, Neuroendocrinology, 36 (1983) 102-104. 15 Woods, S. C. and Porte, D., Relationship between plasma and cerebrospinal fluid insulin levels of dogs, Amer. ]. Physiol., 233 (1977) E331-E334.