Energy metabolism in developing brain cells' JOHNEDMOND Department qf Biologkcab Cher~zkst? and the Wntcal Retardation Research Cetztei-, UCLA Schoub of Medica'rzc, Los Angela, CA 90024-1 759, U.S. A.

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Irvine on 11/24/14 For personal use only.

Received November 7. 199 1

EDMOND, J. 1992. Energy metaboIism in developing brain cells. Can. J. Physiol. Pharmacol. 70: S 1 18 - S 129. During development different energy substrates are available to cells in brain in plentiful supply. The metabolic environment, which is dictated by the milk diet rich in fat, ensures that substrates in addition to glucose are available as fuels. Some substrates serve readily as primary fuels for respiration, whereas other substrates can serve other functions in addition to serving as primary fuels. Primary fuels for respiration serve to supply acetyl CoA directly and as a result always have first priority. With this criteria in mind, a consideration of substrate priority for respiration by developing brain is presented. Many studies in the decade, 1970- 1980. in human infants and in the rat pup model show that both glucose and the ketone bodies, acetoacetate and D-(-)-3-hydroxybutyrate. are taken up by brain and used for energy production and as carbon sources for lipogenesis. Products of fat metabolism, free fatty acids, ketone bodies, and glycerol dominate metabolic pools in early development as a consequence of the milk diet. This recognition of a distinctive metabolic environment from the well-fed adult was taken into consideration within the last decade when methods becanme available to obtain and study each of the major cell populations. neurons, astrocytes, and oligodendrocytes in near homogeneous state in primary culture. Studies on these cells made it possible to examine the distinctive metabolic properties and capabilities of each cell population to oxidize the metabolites that are available in developnment. Studies by many investigators on these cell populations show that all three can rase glucose and the ketone bodies in respiration and for lipogenesis. Only one cell type, the astrocytes. can P-oxidize fatty acids such as octanoate. By comparing the production of labeled carbon dioxide from glucose labeled on carbon-1 compared with carbon-6, it is clear that all three cell populations are capable of active hexose monophosphate shunt activity. Neurons and oligodendrcxytes are capable of making good use of acetoacetate and D-(-)-3-hydroxybutyrate, whereas the best substrate for astrocytes is fatty acid. Under comparable conditions of incubation with astrocytes, fatty acids serve better than ketones, which in turn serve better than glucose in respiration. Some of the major factors that can explain the differing observations by different investigators on the capacity for substrate oxidation are presented. Over the last decade, astrocytes have captured the attention of neurobiologists because they have special attributes as metabolic support cells for the management of intermediary metabolisnm in brain. Evidence has been accumulating that astrocytes exhibit a versatility in their metabolic competency and are now regarded as metabolically multifunctional. Unlike neurons or oligodendrocytes, the astrocytes in culture exhibit metabolic versatility and substrate specialization in their management of carbohydrate and in the processing of fatty acids by @-oxidation, which also produces acetcbacetate. In this regard astrocytes process substances important for other cells. Blsd-borne ketone bodies are not mandatory substrates for growth and brain development of the infant rat. In addition, it is known that the developing brain is autonomous with respect to meeting its needs for major lipids such as cholesterol and palmitate, consequently a reliable substrate supply to support and fuel these needs is mandatory. Evidence is now available to support the conclusion that the developing brain can accommodate alternative substrates to meet its needs for respiration and cholesterogenesis. A metabolic adaptability is demonstrated in vivo when it is shown that increased glucose utilization coinpensates for the reduced availability of acetoacetate in a dietary induced hypoketonemic state in neonatal rat pups that are fed milk substitutes. This compensation is implemented without the precocious development of the key neural enzyme, pymvate dehydrogenase, which would be expected to Facilitate an increased flux of glucose-derived pyruvate for respiration and lipogenesis. Key words: developing brain, neural cells in primary culture, primary fuels, respiration, fatty acids, ketone bodies, glucose. - -

. -

EDMOND, J. 1992. Energy metabolism in developing brain cells. Can. J. Physiol. Pharmacol. 70 : S118-S129. Durant le diveloppement. une grande quantitC de substrats Cnergktiques est mise B la disposition des cellules dans le cerveau. Lqenvironnementmetabolique, qui est irnposC par la dikte lactee riche en maticres grasses, fait en sorte yue les substrats constituent des sources d'Cnergie tout comme le glucose. Certains substrats sont utilisks rapidement comme sources d9energieprimaires pour la respiration, alors que d'autres peuvent @treutiiisCs en PIUS pour d'autres fonctions. Les sources d'Cnergie primaires pour la respiration servent a approvisionner directement l'acktyl CoA et ont, ainsi, toujours priorit6 absolue. Tenant compte de ce critere, on presente une rCflexion sur la priorit6 des substrats pour la respiration dans le cerveau en dCveloppement. Au cours de la dCcennie 1970- 1980, plusieurs Ctudcs chez des nouveaux-n6s hurnains et sur le modkle du raton montrent que le glucose et les corps cetoniques. acktoacetate et D-(-)-3-hydroxybutyrate, sont captures par le cerveau et utilisCs pour la production d9Cnergieet comme sources de carbone pour la lipogenkse. Les produits du m6tabolisrne des graisse. les acides gras libres, les corps cktoniques et le glycCrol sont predominants dans les p o l s n-aCtaboliques aha debut du dCveloppement 2i cause de la diete lactke. Cette reconnaissance d'un environnement metabolique diffirent de celui de l'adulte bien nourri a CtC pris en considkration au cours de la dernikre dCcennie avec le dkveloppement de mCthodes permettant d'obtenir et d'examiner, dans un Ctat quasi homogkne dans des cultures primaires, les populations de cellules qui sont les neurones, les astrocytes et les oligodendrocytes. Les etudes sur celles-ci ont permis d'examiner les propriCaCs mCtaboliques caractCristiques et les capacitts de chaque population de cellules d'oxyder les rnktabolites qui sont disponibles drarant le ddveloppement. Des Ctudes effectukes par differents chercheurs montrent que ces trois populations de cellules peuvent utiliser lc glucose et Bes corps cCtonjiques pour la respiration et la lipogenkse. Un seul type de ccllules, Bes astrocytes, peut 0-oxyder

'This gaper was presented at the satellite symposium of the International Brain Reserch Organization meeting held August 16)- 14, 1991, University of Saskatchewan, Saskatoon, Sask., Canada, entitled Ions, Water, and Energy in Brain Cells, and has undergone the Journal's usual peer review. Printed in Canada ! Imprime ail Canada

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Irvine on 11/24/14 For personal use only.

EDMOND

les acides gras tels que 190ctanoate. En cornparant la production Be gaz carbonique marque provenant du glucose marque en carbone-l et en carbone-6, il esa clair que les trsis populations de cellules sont actives cornme voie des hexoses msnophosphate. Les neurones et les oligodendrocytes peuvent utiliser judicieusernent 19acCtoac6tateet le D-(-)-3-kydroxybutyraae, alors que le rneilleur substrat pour les astrocytes est l'acide gras. Dans des conditions soanparables d9incubation avec des astrocytes, les acides gras agissent mieux pour la respiration que les dtones, qui eux font mieux que le glucose. On presente quelques-uns des principaux facteurs qui pourraient expliquer la diversite des observations sur la capacitC d'oxydation des substrats. Au cours de la dernikre decennie, les astrocytes ont captive les neurobiologistes parce qu9ilsont des caracteristiques particulikres en tant que cellules de soutien mktabolique pour le fonctionnernent du rnCtabolisrne intermediaiare dans le cerveau. Des nombreux faits ont mis en Cvidence que let; astrocytes prtsentent des aptitudes mdtaboliques trks souples et sont maintenant considCres comme multi-fonctionnels. Contrairement aux neurones et aux oligodendrocytes, les astrocytes cultivds prksentent une souplesse mktabolique et une specificit6 en regard du substrpt dans leur gestion des glucides et dans le traitement des acides gras par une P-oxydation qui produit aussi l'adtoacktate. A cet Cgard, les astrocytes traitent des substances qui sont importantes pour les autres cellules. kes corps cktoniques diffush par voie sanguine we sont pas des substrats essentiels pour la croissance et le dCveloppenlent du cerveau chez le rat nouveau-we. Be plus, on sait que le cenreau en dCveloppement est autonome et repond 2i ses besoins en lipides, tels que le cholestkrol et le palmitate; par conskquent, un apport en substrats fiable pour soutenir et alirnenter ces besoins est essentiel. On peut maintenant presenter plusieurs faits soutenant la conclusion que le cerveau en dkveloppement peut s'adapter Zi divers substrats afin de satisfaire ses besoins au niveau de Ba respiration et de la cholestkrogenkse. Une adaptabilitk mktabolique est dkmontrke ia viva lorsqu'une utilisation accrue de glucose compense la disponibilitk reduite d'acktoacdtate lors d'une hypocktontmie induite par la dikte alirnentaire chez les ratons n6onatals nousris avec des substituts de lait. Cette compensation est rCalisCe sans le dkveloppement hatif de l'enzyrne neurale clC, pymvate dkshydrogknase, qui devrait faciliter un flux accm de pymvate dt5rivC du glucose pour la respiration et la lipogenkse. Mots C I ~ S: cerveau en dkveloppement, cellules neurales en culture primaire, sources d'Cnergie primaires, respiration, asides gras, corps c6toniques. glucose. [Traduit par la rCdaction)

Introduction During development, when different energy substrates are available in plentiful supply to cells in brain, it is worthwhile considering how they can best be used, not necessarily to the exclusion of one another, but efficiently and cooperatively. Hn addition, it is important to consider the implications that exist when different substrates are h o w n to be available and to consider if there is any information to indicate that a priority for their utilization cam be established. The metabolite environment in the newborn mammal is a reflection of the milk diet. Fat in the milks of mammals, including the milk of humans and the rat, is the dominant caloric macronutrient. Rat milk fat contributes about 65 % of total calories, human milk contains fat at about 55% of calories (Edmond et al. 1985 and references therein). In reality, these percentages are all low because that portion of the total calories ascribed to protein does not, in large part, provide energy since about 75 % s f the amino acids are reutilized in the synthesis of new protein in growth (Miller 1970). Essentially, it is the products of fat naetabolisrn, free fatty acids and the ketone bodies, which contribute in a dominant way to metabolic processes in the milk feeding period (Page et al. 1971; Edrnond et al. 1985). It should be remembered that fat provides "carbohydrate" from its glycerol moiety, much s f which is considered to be converted to hepatic glucose (Snell and Walker 1973). A consideration of substrate priority It has been recognized from the early 1960s that acetoacetate must be a ready source of mitochondrial acetyl CoA in tissues that contain the two mitochondrial enzymes responsible for its utilization (Williamson and Hems 1970; Page et al. 197I). Two points can be made. For energy production, both 3-oxoacid CoA-transferase and acetoacetyl-CoA thiolase must be present in the mitochondrion; both reactions are reversible (Scheme 1). These enzymes do not appear to be regulated in any clever way like pymvate dehydrogenase (Lai 1992), thus if acetoacetate is available it will be converted to acetoacetyl

CoA, then into two acetyl CoA to readily satisfy the acetylCoA pool; at the same time NADH and ATP pools are satisfied. Under these circumstances pyruvate, whether it be derived from glucose, lactate, or alanine, is not the primary supplier of acetyl CoA, since pyruvate dehydrogenase activity is reduced when these energetically favorable states prevail (Eai 1992). Under normal physiological conditions, regulatory processes ensure that blood glucose concentrations are remarkably constant. On the other hand, the blood concentrations of ketone bodies and free fatty acids can vary widely as they rise and fall in response to dietary conditions; there appears to be no finely tuned regulatory mechanisms to hold them in a particular steady-state concentration range like that observed for glucose. They can always be found at low basal concentrations, particularly when fat is a minor component of the diet. Consequently, these substrates contribute significantly to meeting energy needs only when circurnstances to promote their production are favored. Although glucose is probably in a class by itself as a substrate, carbon derived from glucose may not exercise first priority in fueling respiration (Lai 1982). Fuels for respiration can be assigned a priority based on how their carbon can enter the mitochondrial acetyl-CoA pool (Schemes 1 and 2). Primary fuels for respiration serve to supply mitochondrial acetyl CoA directly and primary hels always have first priority. Primary fuels are acetoacetate, free fatty acids, and glucose-derived pymvate, when the pymvate is directed to mitochondria1 acetyl CoA. Other fuels that supply acetyl CoA indirectly should not be considered primary. Substances that enter the tricarboxylic acid cycle at points other than as acetyl CoA need special consideration (Scheme 2). Such substances contain carbon that may enter the acetyl-CoA pool after the carbon has mixed with the four-carbon pool (oxalsacetate). The intricacies of this arrangement are complex and cannot be avoided because the management of carbons in the cycle is deceivingly sophisticated and results in the phenomenon recognized as "carbon exchange" (Krebs et al. 1966; Chiasson et al. 1977; Palmer and Sugden 1983 Fink et al.

%.I20

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Irvine on 11/24/14 For personal use only.

BLOOD

Glucose

CAN. J. PHYSIOL. PHARMACOL. VOL. 70, 1992

/

MITOCHONDRION

Glucose

SCHEME1. The scheme shows the potential fate of blood-borne glucose, lactate, pyruvate, and the ketone bodies, acetoacetate (AcAc), and D-(-)-3-hydroxybutyrate (3HB), in providing acetyl CoA. In the mitochondrion, three primary fuels, (i) pymvate derived frona lactate and (or) glucose, (ii) acetoacetate, or (kii) fatty acids can provide mitochondrial acetyl CoA. D-(-)-3-Hydroxybutyrate can be converted to AcAc in the mitochondrion by D-(-)-3-hydroxybutyrate dehydrogenase (step 1). Both AcAc and htty acids yield acetyl CoA directly: AcAc in two reversible steps catalyzed by 3-oxoacid CoA-transferase (step 2). then by AcAc-CoA thiolase (step 31, and by @-oxidation in four steps (5, steps not shown). In the first instance, these processes depend on the availability of the substrates. The delivery of carbon from pymvate to acetyl CoA (step 4, catalyzed by the highly regulated enzyme pyruvate dehydrogenase) depends on the need to generate acetyl CoA (ATP, NADH + H+). If acetyl CoA (ATP, NADH H+) is (are) provided by AcAc or fatty acid oxidation, carbon from pyruvate will be diverted to lactate, even when pyruvate supplies are plentiful from the metabolism of glucose. Glucose provides pymvate by the Embden-Meyerhof pathway of glycolysis, or by the hexose n-aonophosphate shunt, HMS (see also Scheme 4). AcAc from the blood p o l , or from the mttochondrion, can directly supply the cytoplasmic acetyl-CoA pool by the action of step 6, which is catalyzed by AcAe-CoA ligase. This is an ATP, co-enzyme A dependent reaction, which is not reversible. By contrast, pymvate provides cytoplasmic acetyl CsA only after its carbon passes through the mitochondrial acetyl-CoA pool; a transfer of acetyB units is mediated by citrate. Acetyk units from the oxidation of fatty acids can provide cytoplasmic acetyl CoA by two routes: (i) like that for pymvate, where transfer of the acetyl units is mediated by citrate, or (ii) mitochondrial acetyl CoA or AcAc-CoA produced in the process of /?-oxidation can be converted to mitochondria1 AcAc, then exported to the cytoplasmic compartment to be used in biosynthetic reactions or exported to the blood pc~ol.This scheme represents the array of reactions that can occur in cultured astrocytes. A similar scheme is applicable to neurons and oligodea~drocytes,except that the /?-oxidation reaction can be excluded (see text).

+

1988). "Carbon exchange" can be accounted for by several mechanisms. ( a ) The utilization of primary fuels is underestimated when labeled carbon dioxide is assayed as a measure of respiration. Both the labeled carbons in acetyl CoA, on their entry into the cycle via citrate, are retained i n oxaloacetate (scheme 3a), which may or may not be available to condense with another acetyl CoA for another turn of the cycle (Scheme 2). Thus it is not the carbons of the fuel that are always released as carbon dioxide: this means that the contribution of the labeled primary fbel directly supply acetyl CoA for respiration can be underestimated. Carbon from primary fuels such as acetoacetate or fatty acids are directed a c e t ; l - ~ o ~ production and do not make a wet contribution to the four-carbon p o l , oxdoacetate. The labeling of oxaloacetate by acetyl CsA is one example of carbon exchange. By contrast, should oxaloacetate (consider

1,4-labeled oxaloacetate, Scheme 3b) acquire label from labeled aspartate through transamination, or from labeled pyruvate by the action of pyruvate carboxylase, this labeled oxaloacetate? when processed in the cycle, can i~nmediately produce two labeled carbon dioxide molecules (Scheme 3b). Firstly, this cannot be accomplished unless a two-carbon acetyl fuel is introduced to permit the formation of citrate and the movement of carbon in the cycle to yield carbon dioxide in the next steps. In this instance it is the aeetyl unit that is introduced to provide the citrate that is fueling the cycle, albeit the labeled carbons in oxaloacetate that are accounted for as labeled carbon dioxide. Secondly, each four carbon unit of oxaloacetate introduced into citrate is recovered as oxaloacetate, thus there is no net disposal of its carbon to carbon dioxide (Scheme 3). This means that aspartate, when its carbon enters the cycle as oxaloacetate cannot provide met energy

EDMOND

Glucose

A

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Irvine on 11/24/14 For personal use only.

114

SCHEME 2. Scheme 2 serves two purposes. It shows how four carbon intermediates and substances, which supply four carbon intermediates, depend on being converted to pymvate before their carbon can supply acetyl CoA by step 1, which is catalyzed by pymvate dehydrogenase. Pymvate can be formed from the basic four carbon units rnalate or oxaloacetate (steps 4, 5 , and 6, catalyzed by malate dehydrogenase, phosphoenolpymvate carboxykinase, and pymvate kinase, respectively). Once the carbon is available in pymvate its transfer to acetyl CoA is not assured, as pyruvate carboxylase and the malic enzyme catalyzed by steps 2 and 3, respectively, can divert the pymvate from this course. The scheme indicates also the mitochondria1 intermediary metabolites and related co-metabolites that can leave the mitochondrion and the cell into a large extracellular space, as is provided in the typical culture environment (see text). Readily identifiable substances are glutamate, a-ketoglutarate, aspartate, rnalate, and citrate, leaving the cell as steps 9- 13. respectively. The convertisn of glutamate to cu-ketoglutarate can occur by either transarninase reactions or by the action of glutamate dehydrogenase, step 8. It should be noted that cells that can convert pymvate to oxaloacetate (step 2) and oxaloacetate to phosphmnolpyruvate (step 5) have the capacity for the first steps in the pathway for gluconeogenesis (14, steps not shown). m, mitochondrial; c, cytoplasmic.

directly. Pyruvate, like aspartate, if its carbon is directed to oxaloacetate or to malate, and then into the tricarbowylic acid cycle, does not contribute to net energy production. In these circumstances their contribution will be overestimated. However, pyruvate does provide net energy when its carbon is directed via acetyH CoA into the cycle. The contribution of pymvate, as a consequence s f its three routes for the flow of its carbon into the cycle is difficult to measure (Kelleher and Bryan 1985) and can be either overestimated if the flux to oxaloacetate and malate is substantial, or greatly underestimated if the flux is principally into the cycle via acetyl CoA as indicated above for fuels with only this option. The interexchange of carbon, in the micromanagement of the tricarboxylic acid cycle, "carbon exchange," as identified in this section, has been encountered by many investigators who have warned of the difficulties it brings when interpreting data. ( b ) "Acetyl-equivalent exchange" is another form of carbon exchange and is best demonstrated by transaminase reactions, for example, that of alanine transaminase, in which a potential primary acetyl unit, namely pyruvate, is exchanged to introduce other carbon blocks into the system. The effect is no net

gain in the provision s f the h e l , acetyl CoA. When pyrnvate is converted to alanine in the conversion of glutamate to a-ketoglutarate. a potential primary acetyl unit is sequestered in alanine and the a-ketoglutarate provides directly half an acetyl unit (one labeled carbon dioxide) and oxaloacetate, which enters the four carbon pool as discussed in (a) above. Oxalsacetate could be converted to acetyH CoA via pymvate, which is formed from oxaloacetate (Scheme 2), by way of the initial step of gluconeogenesis, catalyzed by phosphoenolpyruvate carboxykinase and the last committed step in glycolysis catalyzed by pyruvate kinase (Palmer and Sugden 19831, with neither a net loss nor gain in energy. Again it is not clear why the metabolic enterprise would prefer to secure pyruvate from oxaloacetate when glucose-derived pymvate or lactate-derived pyruvate provides net energy in meeting this need. Further, even if the pyruvate pool is supplied by oxaloacetate, this pymvate is worth a two-carbon unit only when it is converted to acetyl CoA by pyruvate dehydrogenase; if converted back to oxaloacetate by pymvate carboxylase, an energy-using step, there is no net gain, only a metabolic cycle involving oxaloacetate, phosphoenolpyruvate, and pyruvate (Scheme 2.).Such

CAN. J. PHYSIOL. PHARMACBPL. VOL. 70, 1992

b

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Irvine on 11/24/14 For personal use only.

C-COOH

C-COOH Elm

HO-C-COOH C-COOH

I

I0

HB- C-COOH C-COOH

CErate

LC-COOH mm -..-I---.-.----..

C-COOH SCHEME 3. The "catalytic" use of oxaloacetate in the tricarboxylic acid cycle is illustrated. Each molecule of oxaloacetate converted to citrate can be recouped as oxaloacetate ira each turn of the cycle. Scheme 3ca shows an acetyl unit labeled on both carbons entering the cycle and being retained in succinate, which is a symmetrical molecule. This means that aH four carbons in malate and oxaloacetatc contain labcl and these substances can mix into the cellular pool of these metabolites as indicated in Scheme 2. Essentially labeled carbon from acetate can pass readily to glucose, pymvate, lactate. and amino acids by this carbon exchange, yet there is no net synthesis of four carbon substances in the process. Dissipation of label to these substances in this way results in an aanderestirnate of the amount s f acetyl units that entered the cyclc in the generation of energy as measured by the production of labeled carbon dioxide. Scheme ' 3 b ~shows the loss of carbon-l and -4 of sxaloacetate in a first turn of the tricarboxylic acid cycle, yielding a maximum production in labeled carbon dioxide, yet a molecule of oxalsacetate (with no label) is returned t s the system. This cannot happen without the disposal of fuel (acetyl unit of acetyl CoA entering the system). This means the utilization of substrates such as aspartate, malate, four carbons in a-ketoglutarate or glutamate -glutamine, which flow by way s f oxalsacetats in this way to give labeled carbon dioxide, does not measure their contribution to net energy production. Net energy was produced by an unidentified fuel supplying acetyl units.

sycles are believed to confer metabolic integrity to systems (Newsholme and Stanley 1987), however, their existence creates enormous difficulties in interpreting and quantitating data for net energy production. It is worthwhile noting that when there is a need to replenish the four-carbon pool, the energetically more favorable source of carbon equivalents is provided by glutamine-glutamate utilization in the provision of half an acetyl equivalent (one C 0 2 is released in the tricarboxylic acid cycle) to gain a four carbon equivalent; by contrast the well-recognized route s f pyruvate being carboxylated to provide oxaloacetate consumes energy in meeting the requirements for the four carbon pool. Glucose and acetoacetate each can Isave distinctive k n c tions. Glucose provides pyruvate and dihydroxyacetone phosphate for Zipogenesis, however, its distinctive role is that it is oxidized in the hexose monophosphate shunt to provide ribose units and reducing equivalents, such as NADPH, which can be used in support s f biosynthetic processes in brain growth (Scheme 1). Products of fat catabolism in the milk feeding period cannot directly contribute to this function, only glucose oxidation can. In Scheme 4, the function of the hexcpse monophosphate shunt is depicted to illustrate maximum advantage for the utilization s f glucose. Essentially, in cells that have the enzymic capacity for key steps in gluconeogenesis, that is, the capacity to convert fructose 1,6-bisghosghate to fructose $-phosphate, it is possible to consume six nnolecules of glucose 6-phosphate in the hexose monophosphate shunt to generate 12 NADPH + 12H', and recoup five molecules of glucose &phosphate for further oxidation. This

is illustrated in Scheme 4. In neural cells, however, unless they contain an astive fructose l,6-bis phosphatase to convert fructose 1,6-bis phosphate to fructose 6-phosphate; any fructose 1,6-bis phosphate that is produced would be expected to follow the Embden - Meyerhof pathway s f glyccplysis to yield pyruvate. Scheme 4 shows also the three-carbon unit glyceraldehyde 3-phosphate being recouped to reform glucose; but it is equally possible that once glyceraldehyde 3-phosphate is formed, it may have no other fate in some cell types than to be processed to pyruvate. It does not matter whether glucose is oxidized by the hexose monophosphate shunt or processed by glycolysis in brain, both routes can provide pyruvate. Scheme 4 is included to illustrate that it is possible for glucose to be very efficiently oxidized and produce pyruvate. Cytoplasmic acetyl units, as in acetyl CoA for ckolestercpl, fatty acid, or acetylcholine synthesis, when derived from glucose, fatty acids, alanine, lactate, or pyruvate, are produced first in the n~itochondrialcornpartmeaat, carried in citrate to the cytosol, then released from citrate to provide acetyl CoA for cytoplasmic events. As illustrated in Scheme I , it is nnuch more direct and distinctive that acetoacetate can be activated in the cytosolic compartment and in two steps can deliver cytosolic acetyl CoA for biosynthetic processes (Buckley and Williamson 1973; Webber and Edmcpnd 1979).

Substrate utilization by neural cells in primary culture Many studies in the 1 9 7 0 ~in~ children and in the rat pup model, show that both glucose and the ketone bodies, aceto-

6 Glucose 6-43 12 NADP +

(a)

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Irvine on 11/24/14 For personal use only.

12 NADPH + H'

2 ~lyceraldehyde3-P

2 Sedoheptulose 7-P

2 Fructose 6-P

Fructose 1,6=Bispho SCHEME 4. In this scheme, the reactions of the hexose monophosphate shunt are arranged to conserve glucose for the generation of reducing equivalents for the support of biosynthetic reactions as suggested by Scheme 1 (lipogenesis, protein synthesis). Six molecules of glucose 6-phosphate are oxidized to yield 12 NADPH 12Hf at step u , two consecutive reactions that are catalyzed by glucose 6-phosphate dehydrogenase and 6-phosphogluconate dehydrsgenase. In the course of subsequent reactions in the path, five molecules of glucose $-phosphate can be recouped to be processed at step a again. To accomplish this conservation of glucose 6-phosphate, steps b-it catalyzed by epimerase (b),ketoisomerase (c), transketolase (hi), transaldolase (e), phosphotriose isomerase (f), fructose 2 ,$-bisphosphatase ( g ) , and phosphohexose isomerase (h) are required. It is not known how efficiently this recovery can be maintained. It can be anticipated that cells with little or no fmctose I,$-bisphosphatase would process any fructose 1,6-basphosphate that may form (f)by the steps of glycolysis to glyceraldehyde 3-phosphate and dihydroxyacetone phosphate, which will be converted to pyraavate ( i ) . It is of considerable interest in this respect that Fromm (1991) has reported the discrete location of fmctose 1,6-bisphosphatase in astroglia. This scheme for the hexose monophosphate shunt shows the link into the Embden-Meyerhof path s f glycolysis at glyceraldehyde 3-phosphate to pymvate (i) in steps that are not shown.

+

acetate and D-(-)-3-hydroxybutyrate are taken up by brain and used for energy production (Hawkins et al. 1971; Persson et al. 1972; Kraus et al. 1974). These findings were followed by many studies to determine how the blood-brain barrier influenced substrate supply to the neural compartment. The intricacies of the blood-brain barrier as a possible rate-limiting step in the utilization of blood-borne substrates by brain have been reviewed in considerable detail by Miller (1985). The situation is complex because the capacity to transfer certain metabolites changes with the maturation of the brain, and important substrates such as lactate and the ketone bodies seem to share a common carrier system. Further, unlike glucose, which is metabolized by brain at the same rate over a wide range of blood concentrations, the brain uptake and utilization of ketone bodies are proportional to their concentration in blood. As mentioned earlier, the supply of ketone bodies in the blood is dependent on the nature of the diet. However. the two ketone bodies are not treated in an identical manner since it appears that acetoacetate is transported into brain at a slightly faster rate than D-( -)-3-hydroxybutyrate (Miller 1985). These aspects and other considerations of this topic, particularly for lactate, are discussed by Miller (1985) and readers are

referred to this excellent presentation for the detail this subject deserves. Products of fat metabolism, free fatty acids, ketone bodies, and glycerol dominate metabolic pools in early development as a consequence of the milk diet rich in fat (Page et al. 1971; Snell and Walker 1973). For these reasons, the recognition of a metabolic environment in development, distinctive from that of the well-fed adult, was taken into consideration when examining the distinctive metabolic capabilities of neural cells in primary culture (Edmond et al. 1987). Methods became available to obtain each of the major cell populations, neurons, astrocytes and oligodendrocytes in near homogeneous state in primary culture early in the 1980s (McCarthy and de Vellis 1980; Kumar and de Vellis 1981; Yu and Hertz 1983). The competency of each cell population to oxidize each of the primary metabolites available in development has been studied by several investigators (Yu and Hertz 1983; Hertz and Schousboe 1986; Lopes-Cardozo et al. 1986; Mmond et al. 1987; Fitzpatrick et al. 1988; Hertz et a1. 1988; Sykes et al. 1986). All three cell populations use glucose and the ketone bodies in respiration and for lipogenesis. Only one cell type, the astrocytes, can @-oxidizefatty acids such as octano-

Sl24

CAN. J. PHYSIOL. PHARMACOL. VOL. 70. 1992

TABLE1. Substrate utilization for respiration by neural cells in primary culture

Astrocytes Substrate Glucose

wmol substrate h-'

42 - 84 3 -4"

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Irvine on 11/24/14 For personal use only.

Oligodendrocytes

- mg-'

4.gh

5-6"

Neurons protein

References

54 4-8"

1I 18 Acetoacetate Fatty acids

38 11-13" 6 - 30"

1.2" 24-28 C

26 30" r

Yu and Hertz 1983: Hertz and Schousboe 19186; Lopes-Cardozo et al. 1986: Fitzpatrick et al. 1988; Hertz et al. 1988; Sykes et al. 1986 Edmond et al. 19185, 1987 Lopes-Cardozo et al. 1986 Hertz and Schousboe 1986; Fitzpatrick et al. 19188 Lopes-Cardozo et al. 1986: Sykes et al. 1986 Edmond et al. 1985, 1987 Auesaad st al. 1991

NOTE: The data are taken from the references listed. "Rates of respiration were determined with cells in suspension. In addition, each substrate was tested in the presence of the other substrates under equivalent conditions of incubation. "These rates for respiration are expressed as nanomoles of substrate per hour per culture by Sykes ct al. (1986). It is not possible to compare rates for respiration in this instance because the protein content of cclls in a culture flask can vary depending on the cell density, particularly for oligodendrocytes which rarely achieve confluency. 'Neither oligodendrocytes nor neurons are capable of the /.?-oxidation of octanoic acid (Edmond et al. 1987).

ate (Edmond et al. 1987). By comparing the production of labeled carbon dioxide from glucose labeled on carbon-1 versus carbon-6, it is clear that all three cell populations are capable of active hexose monophosphate shunt activity (Edmond et al. 1987). From our data on the capacity of each of the cell populations to generate labeled carbon dioxide from glucose labeled on carbon- l versus carbon-6, we conclude that oligodendrocytes and astrocytes use glucose heavily in the hexose monophosphate shunt; at least 82 and 74% of the glucose metabolized through the hexose monophosphate shunt and through glycolysis to yield labeled carbon dioxide proceeds via the hexose monophosphate shunt (Edmond et al. 1987). By contrast, neurons under the culture conditions in use, used glucose less effectively by this pathway; at least 48% of the glucose that is metabolized to produce GO2 enters the hexose monophosphate shunt. Recently Larrabee (1989) presented a method to quantitate the utilization of glucose in the hexose monophosphate shunt and applied it to neural tissue. Neurons and oligodendrocytes are capable of making good use of acetoacetate and B-(--)-3-hydroxybutyrate, whereas the best substrate for respiration in astrocytes is fatty acid. Under comparable conditions of incubation with astrocytes, fatty acids serve better than ketones, which in turn serve better than glucose in respiration (Edmond et al. 1987). Other investigators report much higher values for the oxidation of glucose, particularly for astrocytes (Yba and Hertz 1983; Hertz and Schousboe 1986; Lopes-Cardozo et al. 1886; Fitzpatrick et id. 1986). Studies on respiration with cells in primary cultures are not simple to perform. There can be problems in obtaining comparable data and an environment that approaches the condition in viva. AS a result there can be substantial differences in outcome. Hertz and colleagues have pointed out many times how culture conditions can contribute to outcome (Hertz and Schousboe 1986; Fitzpatrick et al. 1988; Hertz et al. 1988). Probably the most obvious condition relates to whether cells are tested in the culture flask without disturbing their cellular organization in the culture, or are removed from the flask by trypsinization, or by mechanical means, then tested as cell suspensions. Another variable in the culture condition for dibutyryl cyclic AMP, astrocytes is the presence or absence 0% which is added to induce their differentiation. Further, sub-

strates may be tested alone or Bn combination, thus conditions are established to the advantage of some substrates and not others depending on their order of priority as fuels. Some comparisons on substrate utilization, though not complete by any means, are shown in Table 1 for each of the cell populations. Another consideration, discussed at this meeting, relates to the cell volume and the extracellular space that dictates its fluid environment. Unlike the condition for brain cells in v i v ~ , which have an approximately 20% extracellular space (Schneider et al. 1992), the typical cell culture condition, 10- 15 mL culture medium/0.3 -0.5 mE packed cell volume, means there is an "extracellular space9' in excess of 3080%. This culture environment means that the concentration integrity of readily diffusible intermediary metabolites is disrupted. From the perspective of mitochondria1 energy metabolism*readily diffusible intermediates of the tricarboxylic acid cycle, for example, citrate, malate, and fumarate, and substances derived from these intermediates, for example, aspartate and glutamate, are in equilibrium with the extracellular fluid volume (Scheme 2), and as a result their concentrations in the mitochondrion are suboptimal for tricarboxylic acid cycle activity. Cells in culture or incubation, in part, perform to redress the imbalance in their intracellular intermediary metabolite concentrations. Substrates such as glutamine and pyruvate would be expected to be pressed into recouping a favorable concentration of four-carbon intramitochsndrial milieu and in meeting this obligation their putative role as respiratory fi~elscan be grandly overestimated.

Metabolic versatility and specialization for astrocytes Over the last decade, astrocytes captured the attention of neurobiologists as having special attributes as metabolic support cells in the management of substances in brain. Evidence has been accumulating that astrocytes exhibit a versatility in their metabolic competency. Astrocytes are now regarded as metabolically multii%nctional, where the microenvironment determines the cell's metabolic functions. In this regard astrocytes can process substances important to other cells. The classic example of substrate management by astrocytes is their capability to produce glutamine, a function that deals with the

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Irvine on 11/24/14 For personal use only.

EDMOND

removal of free ammonium ions and glutamate (Bed and Clarke 1983; Buffy et al. 1983; Hertz et al. 1983). Several aspects sf substrate management by astrocytes are reflected in their versatility in processing primary fuels for respiration. Although astrocytes can oxidize acetoacetate in respiration, they can also produce acetoacetate, in a distinctive manner, from the 0-oxidation of fatty acids (Auestad et al. 1991). Our most recent work demonstrates that astrocytes in primary culture, as a consequence of the P-oxidation of fatty acids, also produce acetsacetate. The mechanism for ketsgenesis in astrocytes is distinctive from the mechanism for ketogenesis in liver, the 6-hydroxy-0-methyl-glutaryl (HMG)-CoA cycle of ketsgenesis. Astrocytes, in the oxidation of fatty acids have the capacity to simultaneously produce for export as much carbon as acetoacetate, as they produce in respiration through acetyl CoA (Auestad et al. 1991). The data from these experiments indicate that the P-oxidation is not a symmetrical process in which the carbon chain is broken down systematically to two carbon blocks to provide the C-2 unit acetyl CoA as the central metabolic entity through which all carbon should flow. Of the three routes by which carbon units from the 0-oxidation process can produce acetoacetate, only two are applicable, as it was demonstrated that the conventional route for the production of acetoacetate, the HMG-CoA cycle of ketogenesis plays a ~ninirnalrole, if any, in acetoacetate production (Auestad et al. 1991). Acetoacetate is produced from acetoacetyl CoA, either by the acetoactyl-CoA deacylase described by Petal and Clark (19'78), or by 3-oxoacid CoA-transferase with the formation of succinyl CoA and free acetoacetate (Robinson and Williamson 1980). The acetoacetyl CoA is either formed directly in the @-oxidation process from the omega fourcarbon unit of each fatty acid or by the condensation of two acetyl units (Auestad et al. 1991). This same capability for astrocytes in primary culture to @-oxidizefatty acids and produce acetoacetate exists in cell suspensions taken immediately from developing brain by the same manipulations applied in the preparation of mixed glial cells for primary culture (McCarthy and de Vellis 1980). The capability is an intrinsic property of the astrocyte population. As yet, there is no information on the hormonal control of this function in astrocytes . More than 50 years ago in a study in which a deuterium labeled saturated fatty acid was used in an examination of its transport across the blood-brain barrier of the adult rat, Sperry et al. (1940) showed a very minimal uptake of deuterium labeled stearic acid into brain, whereas there was a large deposition of the lipid in liver and other organs. It was concluded that the brain synthesizes the major fatty acids it needs and does not depend on an external source of supply. We have demonstrated in the artificially reared rat pup (Hall 1973; Auestad et al. 1989) in an experiment covering the period of active growth, the brain growth spurt and the onset of myelination, that the major dietary fatty acid, palmitic acid, does not enter the developing brain (Marbois et al. 1991). We conclude from this study that the brain synthesizes de nova all the palmitate it requires. Further, we propose that unsaturated fatty acids that can be produced from palmitic and stearic acid may also be biosynthesized in brain to meet its needs and speculate that a selective transport system across the blood-brain barrier may exist for the essential fatty acids and (or) the more complex fatty acids derived from them. It is our view that brain has full control of the amount and compositisn of its basic lipids by having exclusive rights for their synthesis to meet its

S125

needs; this autonomy is a protective mechanism to sustain the integrity of the brain irrespective of the extracerebral lipid environment and its needs. Hn considering the role for 0-oxidation in astrocytes, and the substrates that might supply 0-oxidation, it can be suggested the process serves a "housekeeping9? function. There is evidence from pathological conditions that short and medium chain length fatty acids, such as octanoic acid. can enter the brain (Zieve 1985), AstrogHia, by virtue of their capacity for 6-oxidation can be expected to utilize these fatty acids at physiological concentrations, but can be overwhelmed at the high concentrations known to be present in pathological situations. Although it has always been assumed that fatty acids of medium chain length are cleared and oxidized by the liver, Wells et al. (2985) presented evidence that a considerable fraction of these fatty acids can bypass the liver and be available as fuels for oxidation in extrahepatic organs such as the heart, kidney, and possibly the central nervous system now that it is known that astroglia contain the enzymic potential for their disposal by 0-oxidation. Astroglia, by virtue of their capacity to produce and process apolipoprstein E (Boyles et al. 1985) are believed to be intimately involved in the management of lipids in brain. A second function for astroglia in this role is that they remove by 0-oxidation any fatty acid no longer needed for the complex lipid milieu of membranes. Ht is well known that fatty acids in membrane lipids of brain are constantly turning over, thus it can be considered prudent that fatty acids in excess, or redundant to the needs of brain hnction, be removed in a useful way by astroglia for energy use. The alternative is that fatty acids in excess of requirements be exported to the blood, a process that would expend energy. It woulld seem, that when one considers the autonomy sf the brain, the capacity to dispose of fatty acids within the system will be an advantage. It is also clear that astrocytes contain an active glycogen pool with properties in carbohydrate metabolism that are unique to this cell population (Reinhart et al. 1990, and references therein). Glycogen is uniquely associated with astrocytes, and enzymes in glycogen metabolism, such as glycogen phosphorylase, have been shown by immunochemical techniques to be specifically associated with astrocytes (Reinhart et al. 1998). It has recently been reported that astrocytes can restore their glycogen psol, not only from glucose, but from mannose and to a lesser extent, from fructose, but not from galactose, essentially showing that certain hexoses other than glucose can be meaningfully metabolized (Dringen and Hamprecht 1991). Another fascinating aspect of carbohydrate metabolism in astrocytes is the very recent report that fructose H ,6-bis-phosphatase is specifically loca~ized in astrocytes (Fromm 1991). This was determined by irnmunmytochemical approaches (Fromm 1991). The pool sf glycogen in astrocytes is small when compared with the psol in liver at only a b u t 2 -3 prnollg wet weight brain (Hawkins 1985). It was calculated that glucose release from glycogen in brain might sustain energy metabolism for about 5 min at most. This time limitation assumes that glycogen is not being replenished and that the glucose is expended in support of energy needs in general. These assumptions are unlikely because there is no documentation that free glucose is rapidly exported from the glycogen pool in astrocytes; this would require an astrocytic glucose 6-phosphatase, if the process for glucose release is to mimic this function in liver. It is well known that in times of decreased metabolism, which is

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Irvine on 11/24/14 For personal use only.

S126

CAN. J. PHYSIOL. PMARMACOL. VOL. 70, 1992

induced by barbiturates (Gatfield et al. B966), increased glycsgen pools are readily apparent and associated with synaptic regions. It can be suggested that the glycogen pool in astrocytes is intended to serve the energy needs of specific structures such as nerve terminals and that the ready transfer of substrate is not mediated by free glucose, as has been assumed, but from pyruvate and lactate derived from the breakdown of the glycogen. Since glycogen in brain is a finite and active energy store, a highly specific rather than global role for its hnction is likely. The reason glycogen appears to be unique to astrocytes can best be explained in terms sf the unique hnction of astrocytes as a processor - s u p p r t cell that is heavily committed to the micromanagement of metabolites and substances in support of the needs of neighboring cells. There is much yet to be learned about the specific roles of astrocytes ia unique environments of an organ suck as brain. One can speculate that the astrocyte can exhibit multifunctional properties in much the same way as Sungermann and Katz (1989) describe for the "metabolic zonation" ascribed to the hepatocyte. There is every reason to suspect that the astrscyte in the multivarient regions of brain can be subject to rnetabolic zonation since each region can subject the cells to distinct metabolic environments based on the metabolic needs, the substrate supply, the function of neighboring cells, and the regulatory signals that dictate the processes of the region. In terms of function, astrocytes have the capacity to provide acetoacetate for use by neighboring cell populations (Auestad et al. 1991). Astrocytic glycogen can conceivably yield free glucose (a glucose phosphatase would need to reside in the astrocytes for this hnction to be viable and expeditious (Weinhart et al. 1990)), or more likely glucose phosphates, which then yield pymvate and lactate for use by neighboring cell populations. A role for lactate as a substrate of consequence for respiration in developing brain has been advocated by Thurston and Hauhart (1985). Recent studies by Fernandez and Medina (1986) and Dombrowski et al. (1989) illustrate the interest in lactate as an important substrate for developing brain.

Metabolic versatility of developing brain It is easy to demonstrate how important and useful is a substrate such as acetoacetate (Hawkins et al. 1971; Persson et al. 197%;Kraus et al. 1974; Edmond et al. 1987; Auestad et al. 199B), but is it mandatory that acetoacetate be supplied as an efficient source of acetyl CoA in the mitochondrion or cytoplasm in developing brain? One of our recent studies suggests that blood-borne ketone bodies are not mandatory substrates for brain development (Auestad et al. 1990). Ketone bodies might be irreplaceable because Land et al. (1977) provide significant evidence that developing brain might be hard pressed to efficiently process glucose-derived carbon to acetyl CoA. This suggestion arises because the development of pymvate dehydrogenase in brain is significantly delayed compared with the development of other enzymes of energy metabolism (Land et al. 1977), prompting the speculation that this late development sf pyruvate dehydrogenase means that ketones are mandatory for rat brain until the capacity for pymvate utilization develops sufficiently. We investigated this consideration in two ways. Hn one set of experiments the compssition of the rat milk was adjusted to produce a hypoketonernia over the milk feeding period, then the brains of animals were examined to determine if they developed normally (Auestad

et al. 1990). Studies sf this nature have been done using the artificially reared rat pup model (Hall 1973; Auestad et al. 1989). The milk substitute we formulate in the laboratory mimic's mother's milk in quality, and can be made to contain substances of interest, or specific deficiencies can be introduced by omitting one or more components (Auestad et al. 1989, 1990). In the studies where a hypoketsnernia is diet induced, the lipid component containing medium chain fatty acids in the milk fat is left out of the fat blend and replaced in an isocaloric basis in the milk, by carbohydrate, namely pslycose, which is a soluble glucose polymer. The presence of polycose does not increase the osmolarity of the milk substitute to life-threatening levels such as would be found if the substitution had been made by lactose. This medium chain triglyceride-deficient : carbohydrate-rich diet produces a metabolic environment in which blood ketone bodies are less than 18% normal (Auestad et al. 1990). Thus, a dietary metabolic environment is established to examine if more carbon from glucose can be directed to cholesterol production in the reduced acetoacetate environment such that increased glucose utilization compensates, in part, or in total. Essentially, any deficiency in acetoacetate supplied carbon to meet acetyl-CoA production for respiration and for lipogenesis may be met by glucose. In other words, the observed delay in the development of pyruvate dehydrogenase as reported by Land et al. (1977) is not a problem. In this respect, it is important to remember that carbon from glucose must pass through the rnitochondrial acetyl-CoA pool before it can supply acetyl CoA in the cytoplasm for cholesterol production (Scheme 1). Rat pups fed the hypoketonemia producing diet have the same growth, brain weight, and brain cholesterol composition as controls, indicating that other substrates can substitute for the deficiency of ketone bodies in meeting the needs of rapid development (Auestad et al. 1990). Our second approach was directed at finding out if it is glucose that substitutes for acetoacetate when the supply of acetoacetate is reduced in amount by the dietary conditions. We use an indirect approach to determine if more carbon from glucose passes through acetyl CoA in mitochondria in brain by the action of pymvate dehydrogenase in the hypoketonernic condition than in the normal condition and finds its way into the cholesterol that is produced extramitochondrially. This provision is possible because the glucose polymer replaces medium chain triglycerides in the milk when the hypoketonemic condition is promoted in the artificially reared rat pups. Further, it is h o w n that the developing brain is autonomous with respect to meeting its needs for major lipids such as cholesterol (Edmond et d.1991b); all the cholesterol that accumulates in brain is produced in brain. This property simplifies the exarnination of which substrates supply carbon to make cholesterol. Again we used our rat rearing model and the hypoketonemic condition to find out if the flow of carbon from glucose into cholesterol in brain in the hypoketonemic rat pup is greater than in brain of the normoketonemic rat pup under equivalent conditions. To do this with the artificially reared rats, carbon-13 gBucose is fed in the milk substitute at 10% of the lactose concentration in natural rats9 milk (that is, at 10% of 3.5 g % lactose in the control milk substitute and in the same amount in the medium chain triglyceride-deficient:carbohydrate-enriched milk substitute, which gives a hypoketonemic condition). The rat pups fed the milks containing the carbon-13 glucose from day 10 to 12 of age, a 48-h period centered at the important

I

EBMOND

TABLE2. BiosynGkesis of cholesterol from glucose in developing rat brain Brain sterols C-13 substrate

Milk fed (%> CCHO)

Cholesterol (5% C-I?)

Besmosterol (7% C-13)

Cholesterol made (R of total)

Blood glucose (% C-13)

Carbon from glucose in newly made sterol (%)

Glucose

3.5 10 3.5 10

0.125 0.113 0.605 0.585

0.545 0.570 2.627 2.644

22.9 19.8 23 .0 22.1

2.41 0.74

23 77

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Irvine on 11/24/14 For personal use only.

Acetate

NOTE:Milk substitutes are fed to create a normcs- and hypo-ketonemic condition in artificially reared rat pups as described in the text (Auestad et al. 1990). The data are taken from Edmorad et al. (1990). Briefly, rat pups artificially reared, as described by Auestad et al. (1990), are fed either a control milk substitute similar to rat milk in composition, 3.5 g% carbohydrate (CHO), 10 g% fat, or a test milk substitute in whish medium chain triglycerides are omitted from the fat and replaced isocalorically by CHO (10 g % CHO, 7' g% fat). Pups fed the test milk. deficient in medium chain triglyceride, arc hypketonemic; blood ketone body concentrations are