Transepithelial Transport and Metabolism of Boronated Dipeptides Across Caco-2 and HCT-8 Cell Monolayers

Oral delivery of proteins and peptides as therapeutic agents is problematic due to their low bioavailability. This study examined the effect of boronation on the transepithelial transport and metabolism of three glycine-phenylalanine dipeptides in Caco-2 and HCT-8 cell monolayers. The three dipeptides exhibited passive transport characteristics in the monolayer systems. However, metabolism of the boronated dipeptides did occur, but to a lesser extent than the non-boronated glycine-phenylalanine dipeptide. The same metabolic scheme was seen in both cell monolayer system, but greater metabolism was seen in the HCT-8 cell monolayers.


INTRODUCTION
Boron derivatives of amino acids and di-and tripeptides have been synthesized xvith a boron atom incorporated in place of the carbon atom in the alkyl chain. A number of these compounds have been screened for potential activity and have been found to exhibit antineoplastic, anti-inflammatory, antiosteoporosis, and hypolipidemic properties. One of these compounds, N-[(trimethylamine-boryl)carbonyl]-L-phenylalanine methyl ester, 1, exhibited potent c,vtotoxicity against various murine and human tissue cultured cancer cells (1). In L_o cells, the boron derivative blocked DNA, RNA, and protein synthesis in a concentration dependent manner achieving greater than 50/'0 inhibition at 100 (1). From these results, it is evident that I is able to penetrate cellular barriers to inhibit cell growth. Oral delivery of proteins and peptides as therapeutic agents is problematic due to their low bioavailability. This is mainly due to both their poor permeability across cellular barriers and to enzymatic degradation that occurs in the gut lumen, brush border, and enterocyte cytoplasm (2). Substitution of a boron atom for an amino or-carbon in I demonstrated an increased in vitro stability (3). Such as a substitution may also have the potential to alter the in vivo bioavailability of dipeptides. These characteristics could have beneficial therapeutic applications in the development of peptidyl drug therapies. This study was undertaken to determine the effect of boronation on the transepithelial transport and metabolism of three glycine-phenylalanine dipeptides. The transport and metabolism of two boron substituted dipeptides, 1 and N-[(dimethylamine-boryl)-carbonyl]-L-phenylalanine methyl ester, H, were compared to a non-boronated dipeptide, N,N-dimethylglycyl-L-phenylalanine methyl ester, 111, using Caco-2 and HCT-8 cell monolayers as model systems for the small intestine epithelium (see Figure 1). Caco-2 cells are human colon adenocarcinoma cells first isolated from a primary colonic tumor (4). The cell line undergoes spontaneous enterortic differentiation in culture after reaching cotfflucnce, and morphologically resembles small intestinal epitheliun after three xveeks in culture (5). These morphological characteristics include the development of a brush border region expressing microvilli, occluding junctions, dome formations, and the formation of tight junctions leading to cell polarity. In addition, the Caco-2 cells exhibit small intestine biochemical markers such as brash border enzymes, and enzymes are secreted into the growth medium from both the apical and basal regions (6)(7)(8). The cell line expresses a dipeptide carrier transport system in both the apical and basal membranes (9)(10). Caco-2 cells also exhibit transport pathways similar to the distal ileum, and the electrical properties are closely related to colonic epithelium. HCT-8 cells are human adenocarcinoma cells from the ileocecal region; as such, these cells may give a truer reflection of gut absorption compared to Caco-2 cells which are derived from the colonic region. The cells produce a polarized differentiated monolayer consisting of enterocyte-like columnar cells that are interconnected by mature tight junctions and desmosomes and exposed microvilla (11). The structural and functional integrity of the tight junctions of cells grown on polyester filter cups (Falcon, BD) and membrane filter inserts (Transwell(R), Costar) have been confirmed by transmission electron microscopy, transepithelial electrical resistance measurements, and transport of impermeant markers such as mannitol and polyethylene glycol (11,12).

Sources of Compounds
The synthesis of/has been reported (13). Compound//was donated by Boron Biologicals, Inc., Raleigh, NC. Compound 111 was synthesized as follows: L-phenylalanine methyl ester hydrochloride (2.09 g, 9.7 mmol) and dimethylglycine (1 g, 9.7 retool) xvere suspended in 30 ml dry tetrahydrofuran with 10 mmol of triethylamine, dicyclohexyl-carbodiinfide and hydrox3'benzotriazole. The reaction was stirred at room temperature for 48 hours under nitrogen and the solvent was removed under reduced pressure. The product was purified on silica gel, two void volumes of hexane, two void volumes hexane/ethanol (8'1), and then hexane/ethanol (5:1). The combined fractions of purified product yielded a clear liquid (421.9 mg, 16.7%), and had a Rf 0.35 using 1:1 hexane/ethanol. H NMR (CDCI3) 7.55 (d, IH, NH), 5 7.26 (m, 5H, C6I-I5), 4.91 (m, IH, CH), 5 3.73 (s, 3H, CH3), 3  The Caco-2 cell line was obtained from the Lineberger Cancer Center at the University of North Carolina at Chapel Hill at different serial passage numbers (13)(14)(15)(16). For the transport studies, cells were diluted with growth medium to give a total of 7.5 x 10 cells/cm in 1.5 mls. The 1.5 ntis was placed in the apical compartment of a Transwcll (R) (Costar). The Transxvell (R) system is a 6-well plate with each well having an apical compartment capacity of 1.5 ntis, a basal compartment capacity of 2.5 ntis, and a 4.71 cm polycarbonate microporous (0.45 gin) membrane separating the two compartments. Cell monolayers in the Transwell (R) were allowed to grow for 21 days while the medium was replaced every 3 days. The cells were grown in Dulbecco's Modified Eagle Medium (D-MEM) containing L-glutamine and sodium pyruvate at 37C under 5% CO-. and saturated humidity. Additives to the medium included 10% heatinactivated fetal bovine serum, 10 mM nonessential amino acids, 100 units of penicillin and 100 mg of streptomycin per ml of medium.
On the day of the transport study, the culture was placed in a medium of Hamk's Balanced Salt Solution (I-IBSS) (Gibco) containing 0.35 g sodium bicarbonate and 1.25 g N-(2-hydroxyethyl)-piperazine-N'-2ethanesulfonic acid (HEPES) (Sigma Chemical Co.) and allowed to equilibrate for hour. The apical-tobasal transport experiment was initiated by removing the buffer solution from the apical compartment and adding 1.5 ml of fresh HBSS/HEPES containing the compound. At specified time intervals (0.25, 0.50, 1.0, 1.5, 2.0, 3.0, and 4.0 hours), the monolayer insert (the apical compartment with membrane) was moved to a new well with the basal compartment containing 2.5 ml fresh HBSS/HEPES buffer to maintain sink conditions. Basolateral solutions at each time interval were collected and analyzed by HPLC (see below). After 4 hours, the remaining apical solution was collected for HPLC analysis. All experiments at 37C were performed in an incubator at 5% CO. and 95% atmosphere. Experiments conducted at 4C were performed in a cold room. All experiments were performed in duplicate; when agitation was desired, the Transwell (R) system was placed on a cell Nutator (Clay Adams, Model 1105). The monolayer integrity was tested both before and after transport experiments by monitoring the passage of the fluid-phase marker Lucifer yellow CH from the apical to the basal compartment. The integrity was further tested by measuring the potential difference across the monolayer using a transepithelial electrical resistance device described elsewhere (14) in which 50 A of current is applied across the monolayer and the electrical resistance across tight junctions is determined from Ohm's Law in units of f) cm2. After completing the transport experiments, the monolayers were immersed in a 37% formaldehyde solution for 30 minutes. The monolayers were then rinsed in distilled water and stained for 2 minutes with Mayer's hematoxylin which has an affinity for negatively charged particles and reveals DNA, RNA, and acidic proteins. After a second rinsing with distilled xvater, the monolayers xvere covered with an eosinphloxine solution (87% ethanol, 11% eosin, 1% phloxine, and 0.4% glacial acetic acid) which stained cellular proteins. Areas devoid of growth where leakage may have occurred were not stained and such data were not included in analyses.

HCT-8
The HCT-8 cell line was obtained from the Lineberger Cancer Center at the University of North Carolina at Chapel Hill at serial passage number 190. The transport experiment protocol was the same as described in the Caco-2 cell experiments.

Partition Coefficient
The apparent partition coefficient (PC') was obtained by initially preparing 2 mM solutions of compounds 1-111 in I-IBSS/HEPES buffer. One milliliter of water saturated n-octanol and ml of the buffer were added to 5 ml vials for duplicate determinations. Each vial was mixed for 2 minutes with a vortex mixer at room temperature, and then centrifuged at 2000 rpm for 10 minutes. The n-octanol layer was removed using a disposable pipet, and the remaining n-octanol film xvas removed by vacuum suction. HPLC analysis (see below) was used to determine the concentration of the compounds.

HPLC Analysis
For preliminary transport studies (Tables 8 in the Results section), HPLC analyses for the compounds were performed with an isocratic system consisting of a LC pump (Spectra-Physics), an autosampler (SSI, model 280D), a Dynamax UV/VIS absorbance detector (Rainin) at 203 nm, an integrator (Shimadzu, model CR601), and an Econosil Cts 101.t (Alltech) 4.5 x 250 mm column. The mobile phase was methanol and disodium phosphate (60:40) at pH 7.0. However, this initial mobile phase was altered for later studies (Table 9) to a more acidic pH (2.7) in order to lengthen the elution time of the degradation products. These later studies were performed with an isocratic system consisting of a Model 6000A pump (Waters), and Model 210B autosampler (Waters), a Dynamax UV/VIS absorbance detector (Rainin) set at 210 nm, a Econosil Cts 10m column (Alltech), and a model CR601 integrator (Shimadzu). The flow rate was 1.5 ml/min and the k' values for compounds 1-111 were 2.3, 2.5, and 2.8 respectively. The following mobile phases were used for each compound: I and 11, methanol/0.25% H3POa in a 60/40 ratio at pH 2.7; and 111, methanol/0.1N NaH2PO in a 60/40 ratio at pH 7.0. The respective free acids of the three compounds (see Results) were detectable without modifying the appropriate mobile phases. All quantitations were accomplished using standard calibration curves of the individual compounds.

RESULTS
The stability of the three compounds was determined in buffer and media without the Transwell (R) system.
Each of the three compounds were found to not degrade in HBSS/HEPES buffer solution at pH 7.4 and 37C, or the apical and basal media for 10 hours. When the study was extended, degradation was noted only in the last sample taken at 25 hours. At 25 hours, I contained 96% of the original ester,//contained 90%, and 111 contained only 55%. These results show that the three compounds undergo some degradation between 10 and 25 hours. However, the transport experiment was conducted for only 4 hours, and therefore hydrolysis in buffer solution at pH 7.4 and 37C should not be a factor. To determine if transport of I was concentration dependent, varying concentrations of the compound was studied. Table 1 shows that the mass of I transported across Caco-2 cell monolayers was proportional to the initial concentration over the range of 0.75 to 7.50 nuM. Essentially the same percentage of mass was transported at each concentration. These results suggest that 1 was transported across the Caco-2 cell monolayer by passive diffusion. A plot of flux versus initial concentration was linear (plot not shown), again indicating that passive transport vas occurring. The transport of/followed first-order kinetics, with mass transported increasing with time. The following equation was fitted to the experimental data using nonlinear least-squares analysis where 13 and 13o equal the amount of I in the basal compartment at times and oo respectively, and k equals the apparent firstorder rate constant.
The apparent permeability coefficient (Pe) can be calculated from the slope of a plot of the mass transported divided by the product of the Transvell (R) surface area (4.71 cm2) and the initial concentration versus time (14). Alternatively, Pe can be calculated from the derivative of the first order equation at time zero, e.g., k x 13oo. The rate constants (k) and permeability coefficients (Pc) for the transport experiments using I at different initial concentrations are given in Table 2. The values of k and Pe for the various initial concentrations were not significantly different, again indicating that passive diffusion was occurring.  Tables 3 and 4). More of ! remained in the apical compartment under static conditions than in rocking conditions (16% compared to 2%). At 4C, rocking conditions, both the amount and rate of 1 transported was decreased compared to 37C, rocking conditions. a initial concentration ofI was 2.5 raM.
b mmol colle'tel in the basal medium after 4 hours. Compound H was found to transport through the cell monolayers by passive diffusion using the same criteria applied to I (data not shown). At 37C, rocking conditions, the rate constant (k) and permeability (Pe) of II were significantly less than the values for 1 and 111 (Tables 5 8). However,//degrades at a faster rate than I, but not as quickly as 111 when exposed to Caco-2 cell monolayers. Degradation was reduced at 4C but not completely inhibited. The permeability values at 4C were not significantly different between rocking and static conditions.     Compound III also was found to transport through the monolayers via passive diffusion (data not shown). At 37C, rocking conditions, the rate constant (k) and permeability (Pe) of 111 were significantly greater than the values for I and// (Tables 7 and 8). Hoxvever, 111 degrades at a faster rate than the other two compounds when exposed to the cell monolayers. Degradation was reduced at 4C but not completely inhibited; however, the permeability values at 4C xvere significantl.v different between rocking and static conditions (Student's t-test, 95% confidence). Approximately 70% of Compound I xvas recovered after 4 hours (see Table 1). During those experiments. one additional peak was seen to elute to earlier than I. To determine if this degradation product was the result of ester hydrolysis, was incubated xvith 0.0IN NaOH at 37C for 6 hours. HPLC conditions xvere modified as described in the Methods section and the analysis of the incubated solution showed a peak that occurred at the same retention time as the degradation product observed in the transport experiments.
A time study of the incubation solution was done by injecting numerous samples over a 6 hour period, and the ester peak continued to decrease with a simultaneous increase in the degradation peak. After 6 hours, all of the ester was degraded to what was thought to be the free acid of 1. To firther confirm the identib' of the degradation peak,//and phenylalanine methyl ester vere analyzed by HPLC; both compounds had a different retention time than both I and the presumed free acid peaks. The identity xvas finally confirmed as the free acid of I by NMR. Identical studies were conducted with compounds H and 111; and their free acids confirmed.
Initial experiments with 1 showed that transport across Caco-2 cell monolayers occurred by passive diffusion. Further, the initial experiments showed that the Caco-2 cell monolayers had some metabolic specificity for all three compounds. A set of experiments xvere designed to compare the effects of boronation on transport and metabolism. Side-by-side experiments were conducted with Caco-2 cell monolayers to ensure that the enzyme concentration was as comparable as possible for each compound studied. Experiments were performed in duplicate using nlM of Compounds 1-111. At 37C, static conditions, 44% of I xvas transported over four hours (see Table 9), and none of I remained in the apical compartment; only the free acid of I was found in the apical compartment. In the side-by-side experiment, 44% of I was found as the free acid in the basal compartment, comqrming that was being substantially degraded as it transported through the cell monolayer. At 4C, at both static and rocking conditions, degradation of 1 to its free acid was inhibited since none of the free acid was found in the basal compartment or remained in the apical compartment. The transport profile of 11 was similar to 1. After 4 hours of transport at 37C, static conditions. 49%, of 11 xvas recovered in the basal compartment with no remaining ester present in the apical compartment. In the side-by-side experiment, 52% of H was degraded to the corresponding free acid as H transported through the monolayer. At 4C, at either static or rocking conditions, the degradation of H to its respective free acid was inhibited, again, similar to Compound I. The transport profile of compound III was similar to I and II; however, III differed from I and//in that degradation was not completely inhibited at 4C. But it was markedly decreased compared to the 37C, static conditions.    In the HCT-8 cell monolayer system, compounds I and 111 transport was studied at only two concentrations, 0.75 mM and 1.5 mM (see Table 10). With two data points per experiment, it is more difficult to conclusively determine that Compounds 1 and 111 are transported through HCT-8 cell monolayers by passive diffusion Using the same criteria applied to the Caco-2 experiments. The apparent permeability coefficients of I showed little change as the initial concentration changed (see Table 11), again supporting the hypothesis that 1 transports by passive diffusion through HCT-8 cell monolayers.
However, Pe was greater for the higher initial concentration of Compound 111. Approximately 95% of 111 and 45% of I were unaccounted for as the ester or their respective free acid. These data suggest that in addition to a ester hydrolysis pathway, other metabolic pathways were active in the HCT-8 cell monolayers. One possibility is that the compounds under,vent amide hydrolysis; however, no L-phenylalanine methyl ester was detected by HPLC. Apparent partition coefficients (PC') were determined for all three compounds and are presented in Table  12. Compound III had the highest PC', meaning that compared to the other txvo compounds, a greater amount of 111 was found in n-octanol than in buffer. Compound 111, the non-boronated compound, was therefore more lipophilic and the boronated compounds xvere more hydrophilic.

DISCUSSION
The first aim of the study was to characterize the transepithel.ial transport of two boronated dipeptides and compare their transport properties xvith a structurally related, non-boronated, dipeptide analog. Compound I is a glycine-phenylalanine derivative xvith boron substituted at the (x-carbon position of glycine. The Nterminal position contains three methyl groups resulting in an increase in lipophilicity. Likexvise, the Cterminal end is blocked by a methyl ester. The plrtial positive charge surrounding the N-terminal nitrogen is offset by the partial negative charge from boron so that the compound should remain uncharged at physiological pH. Compound 11 is another boron substituted glycine-phenylalanine analog with the boron substitution occurring at the x-carbon position of glycine as in I. Txvo methyl groups surround the Nterminal nitrogen and the C-terminal position is again blocked by a methyl ester. Compound 111 is a dipeptide analog of I without boron substitution. In this compound, the N-terminal is blocked by two methyl groups to afford enhanced lipophilicity and stability xvithout introducing a charge on the nitrogen. The C-terminal position is blocked by a methyl ester as in 1. The conclusion that Compounds 1-111 undergo transepithelial transport in Caco-2 cell monolayers via a passive diffusional process was supported by the folloving evidence. Mass transported across the cell monolayer is proportional to the initial concentration in the apical compartment over a concentration range of 0.75 to 7.50 nflVl (Compound 1, see Table 1" Compounds 11 and 111, data not shown). The apparent permeability coefficients obtained from the transport data for each initial concentration are not significantly different except Table 2 instead of Table 1. The transport of ! and I11 across HCT-8 cell monolayers may be via passive transport, it is difficult to substantiate using only txvo concentrations (see Tables 10 and 11). According to Fick's Law, decreasing the contribution of the aqueous boundary layer to transport by agitation should increase the rate of transport for lipophilic compounds for which the aqueous boundary layer is a rate determining step. In accordance with this expectation, I transport was significantly increased as judged by percent mass transported and the apparent permeability coefficient when the system was agitated at 37C (see Table 3 and 4). Similar experiments were not conducted with Compounds H and 111 at 37C, rocking conditions. Temperature dependent studies were designed to show significant differences in transport rates between experiments performed at 37C and 4C. Active proccsses that contribute to transport should be inhibited at 4C and therefore the rate and permeability values should be much higher at 37C compared to 4C. Both transport rates and permeability values were higher at 37C for all three compounds as expected. The second aim of the study was to compare the metabolism of the three peptides during transepithelial transport. Dipeptides with glycine as N-terminal residues have been shown to have a slower rate of hydrolysis than other dipeptides due to the decreased enzymatic affinity towards glycine. However, various studies performed with glycine-phenylalanine have shown that the dipeptide is extensively metabolized by enzymes in vivo. Two human perfusion studies, perfusing glycine-phenylalanine into the jejunum, have shown a 68% and 79% disappearance rate respectively (16,17). Glycine-phenylalanine has also been shown to be a substrate for brush border aminopeptidases in the rat with an 85% substrate specificity (18). Possible sites for enzymatic degradation in the three compounds chosen for the present study include the amide bond and the methyl ester on the C-terminal portion of the molecule. However, the addition of blocking groups on both ends of the peptide derivatives should afford increased stability towards enzymatic degradation during transport studies. When 1 transported through the Caco-2 cell monolayers, approximately 70% of the compound was recovered (see Tables 1 and 3). Propyl paraben was used as a marker to test for recovery since it undergoes passive diffusion without chemical degradation. Propyl paraben transported across the Caco-2 cell monolayers by passive diffusion and xvith 100% recovery. Therefore, 1 xvas being metabolized at some point during transport. Stability experiments shoxved that no degradation was occurring over the 4 hour period in the medium, indicating that hydrolysis must be occurring xvithin the cell monolayer. A series of basal-to-apical transport experiments showed that both I and prop.vl paraben were 100% recovered. Thus, the metabolism must occur in the bnsh border region where the I contacts the cell monolayer in the apical-to-basal experiments.
Similar results were found with Compounds//and III with 58% of//recovered (see Table 5) and 42% of III recovered (see Table 7). The stability studies in media also showed that Compounds//and 111 were stable throughout the time frame of the experiment. Basal-to-apical studies were not conducted with Compounds // and llI; however, it was anticipated that similar results would be obtained as with Compound I. Thus the metabolism of//and III is likewise assumed to occur in the brush border region of the cell monolayer. A number of hydrolases are present in the Caco-2 cell nonolayer brush border including sucrose-isomaltase, aminopeptidase N, dipeptidyl peptidase IV, alkaline phosphatase, gglutarnyl transpeptidase, lactase, trehalase, maltase, and ornithine decarbox3'lase (6)(7)(8).
The transport of Compounds I and 11I shoxved similar re;ults in HCT-8 cell monolayers compared to Caco-2 cell monolayers in that I had much higher percent mass balances and apparent permeability coefficients compared to III (see Tables 1, 2, 7, 10, and 11). Although the rocking conditions, temperature conditions, propyl paraben and basal-to-apical transport studies were not conducted in HCT-8 cell monolayers, it is likely that Compounds I and III are being metabolized by tle HCT-8 cell monolayer brash border enzymes. Alkaline phosphatase, leucine aminopeptidase and Zn'-+-resistant a-glucosidase have been identified in the cell line (19).
Methyl ester degradation of III was significantly higher than I in both Caco-2 and HCT-8 cell monolayers and in the buffer hydrolysis studies. The difference in hydrolysis is likely a consequence of intramolecular catalysis by the secondary amino group in the dipeptide which is absent in I. The protonated secondary amino group may stabilize the negative charge being developed in the transition state leading to the methyl ester hydrolysis. Such a mechanism would be absent in the boron containing N-terminus. The data further suggest that metabolism may be partially blocked by the presence of the boron atom. Whether the boron atom inhibits enzymatic activity or is just more stable to enzyme attack remains to be elucidated. Compound 111 has only two methyl groups on the N terminal end of glycine whereas 1 has three. Steric interference may therefore be a simple explanation of Compound/'s increased stability.. Partition coefficients can be used to estimate the lipophilicity difference between different compounds and indicate the ability of the compounds to partition into a cell membrane. In this case, the structural differences between the three analogs are reflected in their partition coefficients which may explain the differences in transport rate and permeability (see Table 12). For example, the boron substituted analogs have a lower partition coefficient or are more hydrophilic than the non-boronated 111. The only difference between the compounds is the additional methyl group at the N-terminal position of I which attributes a slight increase in partition coefficient in I compared to//. The three glycine-phenylalanine derivatives in this study were all found to undergo passive diffusional transport in the apical-to-basal direction in Caco-2 cell monolayers. The rate of permeability appeared to follow the order of partition coefficient for the three compounds. For example, 111, with the largest partition coefficient, transported across the Caco-2 monolayer at the fastest rate. In addition, with the partition coefficient of 1.1, the permeability of 111 is probably independent of the partition coefficient and the rate determining step for diffusion is the aqueous boundary layer adjacent to the cell surface (20). In fact, experiments performed under both static and rocking conditions show a pronounced increase in transport rote with agitation compared to static conditions. However, for the boron substituted derivatives, their partition coefficient values are very similar (0.66 and 0.61). Therefore, due to their more hydrophilic nature, partitioning into and diffusion across the cell membrane appears to be the rate determining step for transport. Examination of transport under rocking and static conditions at 4C shows that there is virtually no difference between transport rate and permeability at the two conditions. In summary, the boron-containing dipeptides I and//undergo transepithelial transport across the two cell lines via a passive diffusional process. Since the N-terminus is no longer a nucleophile, the dipeptides 1 and H were much more stable in terms of alkyl ester hydrolysis at the C-terminus than 111. At the conclusion of transport experiments, 100% recovery of I and//xvas achieved after quantitating their free acids but 100% recovery of 111 was not achieved, indicating that firther metabolism may be occurring at the amide bond. Even at 4C, 111 degraded to some extent to its corresponding free acid, while 1 and// showed no metabolism.
It is still unclear why boron substitution at the c-carbon position in glycine should affect the stability of the phenylalanine methyl ester. This increased stability could result from a decreased affinity, for the hydrolases in the brash border, or from the lower rate of permeability into the cell membrane leading to decreased exposure to the hydrolases. ACKNOWLEDGMENTS The studies on the Caco-2 cells were a part of Amy Elkins' thesis work under the direction of Dr. Moo Cho. The authors wish to thank Dr. Robert P. Schrcwsburs.' for his valuable assistance with the writing of the manuscript.