Developmental Changes of Hematoside of Rat Small Intestine POSTNATAL HYDROXYLATION OF FATTY ACIDS AND SIALIC ACID*

The hematoside of rat intestine is analyzed from 1 day to 60 days of age. During the first 3 weeks of life, GMa (N-acetylneuraminylgalactosylglucosylceramide) contains only nonhydroxylated fatty acids and ac- counts for 80-90% of the ganglioside sialic acid. Its concentration is maximum at 6 days (315 pg of NeuAc/ g of intestine) and falls abruptly over the next 2 weeks. It reaches 45 pg of NeuAc/g of intestine at 60 days. Between 28 and 60 days, G M ~ accounts for 72% of the total intestinal gangliosides. From 21 days on, structural modifications of GMJ are observed. acids. Both components days.

of weaning in order to acquire its morphologic, metabolic, and kinetic properties of adult age (7). In a previous communication, we have reported on the major changes affecting the neutral glycolipid composition after birth (8). The present study is devoted to the ganglioside composition of rat intestine during postnatal development. As it appears that changes affect mainly GMM3 hematoside, we have focused our investigation on the quantitative and the qualitative evolutions of this ganglioside.

EXPERIMENTAL PROCEDURES
Animals-Wistar rats were obtained from the "Institut Merieux" (L'Arbresle, France). Day 1 was the day following birth. The entire smdl intestine was excised from several rats, and intestines were pooled before lipid extraction. At 1 day and at 3 days, intestines were taken from one litter (12 to 17 rats). From 6 days on, all the rats were male. At 6, 13, 21, 24, and 28 days, intestines were pooled from 10 rats. At 38 and 60 days, they were pooled from 4 rats. In addition, at 38 and 60 days, epithelial cells were isolated from the jejunum-ileum of 6 rats by the method of Weiser (9) with 10 successive steps (4). Cells isolated during the four f i s t steps were villus cells. Cells collected during the three last steps were crypt cells. The intestinal tube from which epithelium had been removed was kept and studied as nonepithelial residue.
Lipid Extractiun-Intestines were cut into small pieces and they were homogenized in methanol (7 ml/g, wet tissue) with an Ultra Turrax homogenizer for 30 s at medium speed. Chloroform was added (7 ml/g, wet tissue) and the mixture was homogenized again for 10 S. Lipids were extracted overnight under gentle stirring. The suspension was centrifuged. The supernatant was pipetted out and the pellet was homogenized in chloroform/methanol (1:l) (5 ml/g, wet tissue). After 2 h at room temperature, the suspension was centrifuged and the pellet was extracted again with chloroform/methanol (1:2) (5 ml/g, wet tissue). Combined supernatants were evaporated to dryness with a rotary evaporator. They were kept in chloroform/methanol (2:l).
Ganglioside Purification-Gangliosides were purified from 20 mg of lipids. Lipids were dissolved in cNoroform/methanol/water (60: 30:4.5) and desalted by chromatography on a column of Sephadex G-25 (I g) (10). Purified lipids were dried down and dissolved in chloroform/methanol/water (30:60:8). Acidic lipids were separated from neutral lipids by chromatography on a column of DEAE-Sephadex (A-25, acetate form, 0.5 g) according to the method of Ueno et al. (11). After saponification of the alkali-labile phospholipids, they were sonicated in 2 ml of water and the solution was desalted by Uppsala, Sweden). chromatography on a Sephadex G-50 column (K15-30, Pharmacia, Chemical Assay-The sialic acid content of lipid mixtures was determined by the method of Svennerholm (12) modified by Miettinen and Takki-Lukkainen (13).
Analytical studies were done by high performance thin layer chromatography. Ganglioside mixtures containing 0.5-1.5 pg of sialic acid 299 were laid on the plate as 2-mm streaks separated by 5-mm spaces. The plate was run on a IO-cm height in solvent A. Gangliosides were visualized by spraying the plate with the resorcinol/HCl reagent (12) and by heating on a heating block at 95 "C for 30 min as described by Ando et al. (14). The chromatograms were scanned with a Vernon densitometer (Vernon, Paris, France). The intensity of colored spots was measured in the transmittance mode with a slit (0.3 x 4 mm) with a Wratten Kodak 16 A filter on the light path.
For preparative purpose, gangliosides were isolated by thin layer chromatography in solvent A. After chromatography, individual gangliosides were localized on the plate by their fluorescence in ultraviolet light after spraying a solution of Primuline (ICN Pharmaceutical, Inc., Plainview, NY) (0.1 pg/ml of acetone/water (8:2)). They were scraped from the plate and eluted from silica gel with solvent B.
Neuraminidase Treatment-100 nmol of purified gangliosides were dried under nitrogen at the bottom of a test tube. They were dissolved in 0.1 ml of a solution of neuraminidase (Vibrio cholerae, Behringwerke AG, Marburg, West Germany) (1 unit/ml). The tube was capped under nitrogen and incubated a t 37 "C for 16 h. The reaction was stopped by addition of 2 ml of chloroform/methanol (21). The mixture was partitioned according to Folch et al. (15) and the lower phase was washed once with 0.5 ml of methanol/water (1:l). The lower phase, containing the asialogangliosides, was dried under nitrogen and analyzed by high performance thin layer chromatography in solvent B. The combined upper phases were lyophilized.
Free Sialic Acid Analysis-The lyophilized free sialic acids were dissolved in 0.5 ml of water and the solution was decationized. The ion exchange column was made of Dowex 50W-X8 resin (100-200 mesh, 0.5 g) packed in a Pasteur pipette. The resin was fist rinsed with 3 bed volumes of water. The sample was eluted through the column which was rinsed with 1.5 ml of water. The combined eluants were lyophilized. Purified sialic acids (1-2 pg), dissolved in water, were analyzed by high performance thin layer chromatography in solvent C. The spots of sialic acids were revealed with the resorcinol/ HCI reagent (12) and quantified by scanning the chromatograms as it is described for gangliosides.
Fatty Acid and Carbohydrate Analysis-Asialogangliosides were methanolyzed in 0.8 N anhydrous methanolic HCI a t 80 "C for 16 h. Fatty acid methyl esters were extracted in n-hexane, and the methanolic phase was kept for carbohydrate analysis. Fatty acid methyl esters were separated into hydroxy-and nonhydroxymethyl esters by chromatography on a Florisil column and analyzed by gas-liquid chromatography as already described (16). An aliquot of C2, fatty acid methyl ester was added to both samples as an internal standard to calculate the ratio of hydroxy to nonhydroxy fatty acids. Hydroxy fatty acid methyl esters were trimethylsilylated before analysis. Fatty acid methyl esters were injected on a glass column (2 mm X 1.5 m) packed with Anachrom ABS coated with 3% OV-1 (w/w) (Analabs, North Haven, CT).
The methanolic phase was evaporated at room temperature under the vacuum generated by a water pump. Sugars were kept dry on P205 under vacuum until analyzed as N,O-trifluoroacetyl derivatives.
Samples were derivatized in the presence of dichloromethane and trifluoroacetic anhydride at 150 "C as described by Zanetta et al. (17). They were analyzed by gas-liquid chromatography on a glass column (2 mm X 2 m) packed with Supelcoport coated with 3% SP-2401 (w/ w) (Supelco, Inc., Bellefonte, PA).
Sphingoid Base Analysis-Gangliosides were hydrolyzed in methanol/concentrated HCl/water (83:8.6:9.4) at 80 "C for 18 h. Long chain bases were extracted and submitted to periodate oxidation as described by Sweeley and Moscatelli (18). Aldehydes were separated by gas-liquid chromatography on a glass column (2 mm X 2 m) packed with Gas-chrom P coated with 10% EGSS-X (Applied Science Laboratories, Inc., State College, PA), according to a published procedure (19). Native G M~ was also submitted to periodate oxidation and analyzed under the same conditions as above.
An alternate method was used to analyze undegraded long chain bases. Free bases were N-acetylated and 0-trimethylsilylated with the conditions given by Carter and Gaver (20). They were analyzed by gas-liquid chromatography on a OV-1 column. The oven temperature was 220 "C a t the time of injection and it was programmed a t 3 "C/min immediately after injection. Injector and detector temperatures were set, respectively, a t 250 and 300 "C.
Gas-liquid chromatographic analyses were performed on a Hewlett-Packard 5710 A gas chromatograph equipped with a double flame ionization detector. The detector signal was recorded on a Hewlett-Packard 3980 A integrator.

RESULTS
In this study, the ganglioside content of rat intestine is analyzed during the postnatal development which spans the suckling period (from birth to 21 days), the weaning period (from 21 to 28 days), and the period of puberty around 35 days.
Dramatic changes affect both the ganglioside concentration and the ganglioside pattern. Quantitative changes begin first. The ganglioside concentration, which is 4 to 5 times higher than in adult intestine during the first 2 weeks of life, decreases sharply after 6 days and it reaches the adult level at the end of the fourth week (Fig. 1). The ganglioside profie does not change at first and then it appears of increasing complexity from 21 days on (Fig. 2). A t all ages, GM3 hematoside is by far the major ganglioside of rat intestine. It accounts for 85% of the ganglioside content at 1 day, 90% at 6 days, 82% at 21 days, and 72% at 28 days. Thus, the fall of the ganglioside concentration is due essentially to the fall of the G M ;~ concentration and to a minor extent to a decreasing concentration of other gangliosides (Fig. 1). Minor gangliosides are scarcely detectable during the neonatal period. At 6 days, as GD:~ alone accounts for about 50% of the minor gangliosides, it is the only one which is identifiable. During the weaning period, the contribution of minor gangliosides to sialic acid content increases and they all become clearly individualized on chromatograms at 28 days. These changes explain partly the increasing complexity of the ganglioside composition which is observed on thin layer chromatograms (Fig. 2).
The are calculated from the mean values of total gangliosides, using the percentage distribution obtained for each point by densitometer scanning of thin layer chromatograms (Fig. 2). intestinal &3. One can notice that, in epithelial cells, G M~ is distributed on thin layer chromatograms among more polar spots at 60 days than at 38 days.
All these observations have led us to conclude that the differentiation of intestinal tissue which is initiated at the time of weaning brings about an increasing polarity of G M~ hematoside. Furthermore, the increased polarity of G M~ is likely to occur chiefly in the epithelium. In order to elucidate the molecular basis of this phenomenon, our study has been focused on the structural alterations which affect intestinal G M~ after 21 days.
Postnatal Hydroxylatwn of Sialic Acid of Intestinal GM~-( 3~3 has been quantitatively cleaved into sialic acid and lactosylceramide by V. cholerae neuraminidase. From 1 day to 13 days, the only sialic acid is N-acetylneuraminic acid (Fig.  3). From 21 days on, a second spot of lower mobility appears in increasing proportion on the chromatogram. This spot has the same mobility as the prominent sialic acid of epithelial cells of adult intestine that we have already characterized as N-glycolylneuraminic acid (4). As N-glycolylneuraminic acid is formed by hydroxylation of the acetyl group of N-acetylneuraminic acid (21) and as the percentage of N-glycolylneuraminic acid increases rapidly during the 4th week of life (Table I), one can conclude that the mechanisms of hydroxylation of N-acetylneuraminic acid are activated at this period of the intestinal development. Furthermore, the comparative analysis of the &3 sialic acids of epithelial cells and of nonepithelial residue in the oldest animals demonstrates that the hydroxylating process takes place mainly, but not exclusively, in the epithelium and that the percentage of N-glycolylneuraminic acid in G M~ sialic acids has not yet reached at 38 days the level found at 60 days. These findings seems to give a coherent explanation to the heterogeneity of G M~ that is observed after weaning, as the appearance of more polar spots of G M~ (Fig. 2) is paralleled by an increasing content of N-glycolylneuraminic acid (Fig. 3). This is not, however, the only explanation of the evolution of the chromatographic pattern of G M~, because the ceramide part of G M~ becomes also more hydroxylated after 21 days.

Postnatal Hydroxylatwn of the Ceramide of Intestinal
GM3-Upon high performance thin layer chromatography, the lactosylceramide of G M~ hematoside displays a novel pattern of evolution between 13 and 60 days (Fig. 4). This evolution does not originate in the carbohydrate portion of the molecule which is made of glucose and galactose in equimolar amount. Consequently, the increasing proportion of molecular species with a lower mobility must reflect the transformation of the ceramide constituents, fatty acids and sphingoid bases.
Gas-liquid chromatography of sphingoid bases as their aldehyde derivatives gives a major peak of pentadecanal (Fig.   5, left). This aldehyde is likely to come from 4D-hydroxysphinganine but it can also be generated by the periodic oxidation of C17-sphinganine. In order to assess the exact nature of the major base of intestinal Gun at birth, two other experiments were conducted. The genuine G M~ molecules were submitted to periodate oxidation. We have reasoned after Crossman and Hirschberg (22) that, in this case, trihydroxy bases bearing free hydroxyl groups on C-3 and C-4 can be oxidized and yield aldehydes with 3 carbons less than the original bases while dihydroxy bases cannot be degraded. We have found that the native G M~ of 1-day-old and 38-day-old rat intestine yields pentadecanal. Furthermore, upon gas-liquid chromatography of the N-trimethylsilyl derivatives of the free bases, the most prominent peak has the same retention time as standard 4D-hydroxysphinganine (Fig. 5, right). From these analyses, we conclude that 4D-hydroxysphinganine is the major base of rat intestinal G M~ throughout the developmental period. Among minor bases, sphingosine, detected as hexadecenal upon periodate oxidation, can be found in significant amount only after the 4th week of life. It occurs mainly in nonepithelial tissue, analyzed at 38 and 60 days. At these ages, the G M~ of nonepithelial residue contains an equal amount of 4D-hydroxysphinganine and of sphingosine.

TABLE I N-Glycolylneuraminic acid appearance in intestinal hematoside
during development N-Glycolylneuraminic acid is expressed as percentage of the total response recorded by scanning of the thin layer chromatograms (Fig.  3). At 13 days and before, sialic acid is exclusively N-acetylneuraminic acid.  Fig. 2). S, lactosylceramide from bovine milk fat globule membrane. cantly between 21 and 28 days, a t a time of major alterations of the chromatographic pattern of lactosylceramide, the origin of the new spots has been searched in a modification of fatty acids. The fatty acid analysis has c o n f i e d t h a t a n evolution in the fatty acid composition of G M~ does exist. From birth to 13 days, all the fatty acids are nonhydroxylated. At 21 days, 26% of the G M~ fatty acids of whole intestine is already ahydroxylated (Table 11). The percentage of hydroxylation doubles at the end of the 4th week but then, it decreases and remains stable after 38 days. However, the degree of hydroxylation goes on increasing in epithelial cells. In 38-day-old rats, two-thirds of the GMD fatty acids of epithelial cells are a-hydroxylated and this proportion is even higher a t 60 days.

Total intestine
In nonepithelial residue, G M~ contains a low percentage of ahydroxylated fatty acids which increases from 5 to 9% between 38 and 60 days. The decreased percentage of a-hydroxylated fatty acids in the G M~ of whole intestine after 28 days can be explained by a modification of the respective contribution of epithelium and of nonepithelial residue. Nonepithelial residue, which is a minor contributor before 28 days, becomes an important one after, and a t 60 days, it provides 52% of intestinal G M~. This finding may explain also why sphingosine, which is a long chain base of nonepithelial residue, appears only in small amounts in intestinal G M~ before 28 days and in higher amounts at 38 and 60 days. Thus, the developmental pattern of lactosylceramide which is illustrated by lactosylceramide. It is likely that N-glycolylneuraminic acid is associated to less polar ceramide and that N-acetylneuraminic acid is associated to more polar ceramide in such a way that both associations give G M n of similar mobility. When N-acetylneuraminic acid is the only sialic acid, its removal yields a lactosylceramide which is also resolved into more spots than the original GM3. This is the case for the A spot of G M~ at 24 days which gives two spots of lactosylceramide of similar intensity and this is the case in neonatal intestine where G M~ gives two spots (Fig. 2) and the related lactosylceramide gives three spots (Fig. 4). The degree of hydroxylation of the bases and of the fatty acids as well as their chain length determine the resultant polarity, and thus the mobility of the different molecular species of lactosylceramide (23,24).  These findings show that the analysis of GM:) by high performance thin layer chromatography resolves partially the different molecular species occurring in rat intestine. It has been sufficient to detect that major changes of composition occur during the developmental period. However, in order to characterize these changes, it has been necessary to analyze separately the sialic acids and the lactosylceramide.

DISCUSSION
We have already discovered that, in adult rat, G M~ hematoside containing N-glycolylneuraminic acid is the prominent ganglioside of epithelial cells of small intestine, that it is at least five times more concentrated in villus cells (mature cells) than in crypt cells (proliferative cells) (4,5), and that, in both cell types, its fatty acids are highly a-hydroxylated (25). These findings have been confirmed by Breimer et al. (26,27). Besides GMa which accounts for 17% of the glycosphingolipids of epithelium, the other major sphingolipids, except sphingomyelin, namely free ceramide, glucosylceramide, globotriaosylceramide (5,26), and tetrahexosylceramide (27) are also affected by the cellular differentiation occurring in adult rat intestine. It is not known at the present time whether the numerous, but quantitatively minor, complex glycolipids that have been recently identified in the intestinal epithelium of the white rat by Breimer et al. (28) are also affected by this type of differentiation. The regeneration of the intestinal epithelium of adult rat provides a good system for studying a normal type of differentiation in vivo, thanks to the method of Weiser (9) which allows a separation of the cells according to their stage of differentiation. But, up to now, it was not known whether the differentiation of epithelial cells in adult rat intestine mimics the differentiation occurring during the development of intestine.
The major discoveries of the present study are that: (i) G M 3 hematoside is the major ganglioside of neonatal intestine; (ii) it is 7 times more concentrated than in adult intestine; (iii) its fatty acids and its sialic acid, nonhydroxylated a t birth, are progressively hydroxylated, starting at the beginning of the 4th week of life.
In a previous study, we have already demonstrated that glucosylceramide is by far the major neutral glycolipid of neonatal intestine and that globotriaosylceramide, which is a major glycolipid of crypt cells of adult intestine, is detected in noticeable amount only after 21 days (8).
Therefore, under all these aspects, namely G M .~ concentration and structure and neutral glycolipid composition, the cell differentiation taking place during the postnatal growth of rat intestine is different from the crypt to villus cell differentiation taking place during the regeneration of the epithelium of adult rat intestine.
The present study was conducted on the entire intestine of 1-day-old to 28-day-old rats because the fragility of the intestinal wall prevented us from using the method of Weiser (9) in order to separate epithelial cells from nonepithelial residue. Epithelium and nonepithelial residue were separated in 38and 60-day-old rats. This procedure has led us to observe that the hydroxylation of the fatty acids and sialic acid of GMJ takes place mainly in epithelial cells and that the adult degree of hydroxylation of both components is not reached by the end of puberty which occurs around 35 days in male Wistar rat. At 60 days, nonepithelial residue contains half the intestinal GM:3 and aU the gangliosides more complex than G M :~ whereas epithelium contains the other half of G M~. Thus, the GM:j of nonepithelial residue is a more important contributor to the intestinal GM3 of adult rat than it is suggested by Angstrom et al. (29) but, like these authors, we find that nonepithelial GMs is less hydroxylated in its fatty acids and in its sialic acids than epithelial GM:~.
The a-hydroxylation of fatty acids appears to follow a similar evolution in GM: c and in glucosylceramide which is the most abundant glycolipid of epithelial cells (5,8,28). However, at all ages of the developmental period, the degree of hydroxylation is lower in GM3 than in glucosylceramide. This difference may originate in the fact that intestinal G M~ comes from the epithelium as well as from the nonepithelial residue while glucosylceramide comes almost exclusively from the epithelium. It is also possible that the difference in the degree of hydroxylation of the fatty acids of GMa and of glucosylceramide expresses the delay between the completion of GM:I and the synthesis of its distant precursor glucosylceramide.
An age-dependent increase of the degree of hydroxylation of fatty acids has been found in the galactosylceramide of rat brain by Kishimoto and Radin (30). This change depends on an a-hydroxylating system which appears in brain after 9 days, reaches a maximum activity around 21 days, and then declines rapidly (31). We have demonstrated that, in intestinal epithelium also, a large part (up to 75%) of the fatty acids of the glycolipids is a-hydroxylated ( 2 5 ) , and in the present study we show that this a-hydroxylation is a process under developmental regulation. However, intestine and brain are different on two accounts. First, the a-hydroxylating activity of intestine is not likely to decline after 21 days since the percentage of a-hydroxylated fatty acids increases greatly in GM:) after this age and since intestinal epithelial cells have a mean lifetime which does not exceed 48 h (32). Consequently, in intestinal epithelial cells, an a-hydroxylating activity must be maintained during the whole life. The second difference between intestine and brain is that intestinal GM:~ contains a high percentage of a-hydroxylated fatty acids whereas brain gangliosides contain only nonhydroxylated fatty acids (33).
This study on G M~ hematoside shows that the development of rat intestine is characterized by a hydroxylation of fatty acids and sialic acid after weaning, while bases do not change. 4D-Hydroxysphinganine, which is a trihydroxy base distinctive of intestinal epithelium in many species including human (25, 34, 35), is already present at birth in the GMn of rat intestine, as in its glucosylceramide (8). It is also present in G M~ and in NeuAca2-+6neolactotetraosylceramide (36), as in neutral glycolipids (37,38) of human fetal intestine analyzed at birth in meconium. But, unlike what happens in rat intestine, human 4D-hydroxysphinganine is already associated to a-hydroxylated fatty acids at birth. One must keep in mind that rat is an animal with a short gestational period and that of Hematoside of Rat Small Intestine it undergoes only after birth some of the stages of development which are completed before birth in species with a long gestational period such as man (39). Therefore, the study of the glycolipids of the developing rat intestine gives evidence that the synthesis of 4D-hydroxysphinganine, which requires the addition of an hydroxyl group on C-4 of sphinganine (40), occurs at an earlier stage of development than the a-hydroxylation of the fatty acids of glycolipids.
N-glycolylneuraminic acid derives from N-acetylneuraminic acid by the action of a N-acetyl hydroxylase (21). It is the prominent sialic acid of the G M~ of the intestinal epithelium of adult rat (4,29). The present study shows that, at birth, GM3 contains only N-acetylneuraminic acid and that the progressive hydroxylation of the molecule is concomitant with the cy-hydroxylation of fatty acids, beginning at the time of weaning. N-Glycolylneuraminic acid is a common sialic acid of glycolipids and glycoproteins but, up to now, it has never been suspected that its occurrence may be under developmental control. Further investigations are needed to know whether this control is specific of the development of rat intestine or whether it is a general phenomenon occurring in other tissues and other species.
As it is known that rat intestine undergoes important transformations in its structure as well as in its way of absorbing the nutrients during postnatal development, it is likely that the modifications of the GM3 content are an expression, at the molecular level, of the changes affecting the plasma membranes of intestinal cells. After birth, the only nutrient, milk, is absorbed by intestine mostly by pinocytosis. Pinocytosis is replaced gradually by an absorption of the adult type, first in the proximal intestine, then in the ileum, and it is terminated at 21 days (39). Simultaneously, the activity of lactase, a brush border enzyme, which is maximum during the days following birth, declines during the 2nd and the 3rd week of life and reaches its low adult level at 21 days (6). Both processes are likely to involve important transformations of the plasma membranes of epithelial cells. Our results show that G M~ is a t its highest level a t 6 days and that it falls abruptly afterward. This evolution suggests that a correlation may exist between the Gun content of plasma membranes, the pinocytotic capability, and/or the lactase activity.
By 20 days begins what has been called a "redifferentiation" period because of extensive functional changes in rat intestine (7); the cell turnover is accelerated (41) and new proteins such as sucrase-isomaltase (6) and the adult form of alkaline phosphatase (42,43) are inserted into the brush border membrane. It is likely that the increased hydroxylation of G M~ together with the changes in the neutral glycolipid composition (8) are parts of a reorganization of the structure of the plasma membrane. In adult rat, 20% a t least of the lipids of the brush border membrane are glycolipids (44). This particular composition may be responsible for the low fluidity that characterizes this membrane (45). The additional hydroxyl groups that appear during the "redifferentiation" period in the carbohydrate as well as in the ceramide part of G M~ may contribute significantly to the exceptional cohesion of the membrane components (46, 47).