Biological Inactivation of Proteins by the Maillard Reaction

The effects of the Maillard reaction on the biological activity of insulin as a hormone, and on the utilization of egg albumin as a source of amino acids have been studied. In order to know the number, the position, and the effect of the glucose residues that bind a protein, crystalline zinc insulin was stored with 14C-glucose at 37° and 68% R.H. for four months. The unreacted sugar was separated from insulin in a Bio-Gel column and the bound radioactivity counted. The first sugar residue reacted within five days, and additional 2.7 residues reacted through the end of the storage. Binding of 1.4 residues increased the a~id solubility of insulin ten times suggesting impairment of hexamer formation. The ability of this Maillard insulin in controlling the blood glucose and tryptophan levels suffered little or no change. After four months the s9lubility decreased by a factor of

the intestine of duodenum/portal vein cannulated rats, indicating that the residue adjacent to the N-terminal amino acid of this Maillard dipeptide was not released and transported even if it did not directly participate in the Maillard reaction. From the portal blood levels of free tryptophan,it was learned that the various N-substituted dipeptides present in an L~lysyl-L-tryptophan premelanoidin mixture may be hydrolyzable and serve as a source of tryptophan without interfering with the abso rption of the normal dipeptide.
The exclusion of small, soluble peptides from absorption suggested that they may be detected in the feces. The soluble fraction of the feces of rats fed Maillard egg albumin was analyzed, after extraction and filtration with solutions of (i) sodium chloride, (ii) ammonium acetate, and (iii) pyridine acetate. The NaCl extracts showed the accumulation of ninhydrin-positive bands of about (I) 1300, (II) 1000, and (III) 800 moiecular weight, all exhibiting browning characteristics. The firt two bands contained high proportions of lysine, plus other essential and non-essential amino acids. Band III yielded 90% NH3 upon hydrolysis. (ii) Portions from the ammonium acetate extracts, containing bands I and II, were resolved into six p eaks by cation exchange and analyzed for amino acids. (iii) Samples treated with pyridine acetate gave a profile different from that in (1).
Cation exchange chromatography reveal ed some fifteen peaks , which were neither peptides nor glycosidic residues. Maillard ovalbumin tagged with either 14 c-lysine or -tryptophan was fed to rats to assess 'the fecal recovery. Feeding the Maillard ovalbumin raised the recovery of radioactivity from 1 to 7% for the former, and from .2 to 3% for the latter. Such loss was not considered significant, and thus other possibilities are suggested to account for the drastic reduction of nutritive value of mildly b~own egg a lbumin. xiii Second, we he.e_ d to know the i~portance of the sugarsubsti tuted «-amino terminus in the hydrolysis and absorption of, not only the terminal amino acid, but also those residues adjacent to it. Something is known on the fate of a few amino acid-sugar compounds when they reach the intestine, but extension of these results to proteins would be premature.
Therefore, I chose to study the absorption and hydrolysis of glycyl-L-leucine and L-lysyl-L-tryptophan after allowing them to react with glucose.
Lastly, a proof is needed for the accumulation of small natural peptides in the feces of rats fed Maillard egg albumin. If the hypothesis is correct that Maillard proteins are hydrolyzed down to fragments of a certain size and that some of these fragments are not absorbed because of the lack of free -amino groups, then such fragments may be found excreted in the feces.
There is a second part to this thesis which is a series of three appendices designed to provide supplementary inf ormation to the reader. Appendix A contains the chemical principles of the Maillard reaction. It also points out some of the early conceptions and misconceptions that came with the development of this field of research. Appendix B contains a detailed description of a surgical technique for the in vivo study of intesti nal absorption in rats. The major advantage of this technique is that it allows the researcher to use every animal as its own control and employ small amounts of substrate. The procedure is the result of xiv both the first-hand knowledge acquired from its initiator, Dr. Hugo Gallo-Torres of Hoffman-LaRoche Laboratories, and a modification that is a result of my own experience. Appendix B also describes some buffer gradients which were useful in the chromatographic separation of the fecal peptides.
The conditions for a modification of the p-dimethylaminobenzaldehyde in the determination of tryptophan and phenol are also recorded in this appendix. Appendix C is the summary and significance of the conclusions as they would appear in the classical thesis format.

EFFECT OF THE MAILLARD REACTION ON THE BIOLOGICAL ACTIVITY OF INSULIN
ABSTRACT Crystalline zinc insulin was stored with 14c-glucose (1:55 mole ratio) at 37° and 68% R.H. to undergo the 1 Maillard reaction for four months. The unreacted sugar was separated from insulin in a Bio-Gel column and the bound radioactivity counted. On the average, one carbohydrate residue reacted with the polypeptide within the first five days of storage. Throughout the rest of the incubation period additional 2.7 residues bound the hormone. The binding of an average of 1.4 hexose residues, at fifteen days of storage, brought about a ten-fold incr~ase in the acid solubility of insulin, thus suggesting that the modification impairs hexamer formation. By a simultaneous, yet slower, process the solubility was observed to decrease by a . factor of about 1000 after four months of storage. The fifteen-day Maillard insulin had lost 22% of its potency while its ability to raise the level ot, blood tryptophan in young rats had not changed.

INTRODUCTION
In the Maillard reaction, reducing sugars form condensation products with the amino groups of amino acids, peptides. and proteins. At the initial stages, ~-N-l-deoxy-2-ketohexose derivatives have been isolated and characterized from the reaction of glucose with amino acids  and peptides . The Maillard reaction with proteins has been mostly studied on proteins used for food, which appear to lose considerable solubility, digestibility and biological value along the course of the reaction.
!As a result of working with undefined protein systems, it has not been possible to ascertain (a) which amino acid residues besides the terminal amino and lysine in the polypeptide chain directly participate in the reaction; (b) to what extent do the physiochemical changes impede the functionality of a Maillard protein, and (c) how many hexose residues react per molecule of protein. The only attempt to use insulin as a model protein for the Maillard reaction was made by , who estimated the radioactivity of the ocand f,-amino groups with glucose but did not evaluate the product in terms of biological function.
The objective of this work was to modify crystalline insulin by the Maillard reaction, measure the average number of hexose residues bound to the protein, and estimate the effect of any physiocochemical changes on two different biochemical functions.

MATERIALS AND METHODS
The Browning of Insulin. Reaction mixtures were prepared in 20-ml glass vials suspending 50 mg of crystalline insulin (bovine pancreas, .5% zinc, Sigma Chemical Co.) in 3 10 ml of 7.75% glucose. The mixtures were lyophilized using minimum heat and time. The vials were then stored uncapped in sealed chambers at 37° and 68% R.H. for up to four months.
For the glucose binding studies, each vial also contained from 25 to 50 ~Ci of UL 14c-D-glucose (International Chemical & Nuclear Co., Irvine, California).
Glucose Binding. For every time point, the contents of a vial were dissolved in 10 ml of an acetic acid solution (pH 2.6), and .3 ml loaded onto a Bio-Gel P-6 (200-400 mesh, Bio-Rad Laboratories, Richmond, California) column, and eluted with the same acetic acid solution . . Other conditions for the gel filtration were: Column size, 1 x 50 cm; flow rate, 22 ml/hr; void volume, 18 ml; fraction volume, 1.5 ml.
Fifty microliters of each fraction were counted in a liquid scintillation counter (Nuclear Chicago, Mark I) using a fluorescent solution made by mixing 2 1 of toluene, 2 1 of 1,4-dioxane, 1.2 1 of ethanol, 260 g of napthalene, 26 g of PPO, and .5 g of POPOP. The ninhydrin pattern was obtained after hydrolysis of .15 ml of each fraction with 1 ml of 2 .5 N KOH at 1 21° for 40 minute s, followed by n eutralization with .6 ml of 30% a cetic a cid. Unreacted gl ucose was dete rmi n ed in the fractions by the Glucostat method (Worthington Biochemicals, Freehold, N~w Jersey).

4
The glucose-free Maillard insulin was dansylated and analyzed according to the procedure of , except that (a) 10 microliters containing 10-15 nmole of either Maillard or normal insulin were used; (b) the urea was Sequanal grade from Pierce Chemicals (Rockford, Illinois); (c) ~he TLC plates were spotted, developed and s craped promptly to avoid loss of fluorescence. The material collected from the spots was read in a Perkin-Elmer-Hitachi MPF-4A fluorescence spectrophotometer.
Biological Assays of Maillard Insulin. Batches ten times the size of those described above were stored for fifteen days. The unreacted sugar was eliminated by washing with distilled water and c entrifuging three times. This procedure afforded 95-98% recovery of the insulin as monitored by the alkaline hydrolysis-ninhydrin test. The biological activity of the resulting insulin was determined by the bloodsugar rabbi t (Glochest er Rabbitry, Glochester, Rhdde I sland) assay (US Pharmacopeia), and it was further tested in its ability to raise the blood tryptophan levels in young rats (Charle s River Breeding Laboratories, Wilmingt on, Massachusetts) a s described by Fernstrom and Wurtman (1972 ).
The stability of the Maillard insulin in the presence of deacylases was tested in vitro. Fifteen-day Maillard ins ulin of high specific 1 4c-ac tivity (180 m Ci/m mol e ) was 4-a incubated with a rat kidney homogenate (Price and Greenstein, 1948) at 37° for O, 1, 2 and 4 hours. The reaction was stopped with 1.5 volume of 10% TCA, the precipitate centrifuged and the supernatant counted for radioactivity.

RESULTS
In the dry state (16% water), glucose slowly binds the reactive amino groups of insulin by the Maillard reaction at 37° and 68% R.H. The average number of hexose residues bound per molecule was determined from the radioactivity recovered in the insulin band of the gel filtration pattern shown in Figure 1. This pattern and the amount of bound radioactivity were the same whether the brown sample was dissolved and eluted with bicarbonate buffer (pH 7.5) or acetic acid (pH 2.6) thus demonstrating the stability of Maillard insulin at low pH.
The binding of glucose to insulin as a function of storage time is presented in Figure 2. At five days after storage, an average of .98 glucose residues had bound the hormone with a resulting increase in solubility of about ten times. After fifteen days of storage, the acid solubility of Maillard insulin was observed to decrease. At four months, when an average of 3.7 glucose residues had reacted, more acid (pH 2.4) was needed to bring the Maillard insulin into solution.
Insulin ~hat had been stored with glucose for fifteen days was separated from the unreacted glucose and then assayed for available amino groups and biological activity.
Analysis of the hydrolysis products of the dansylation reaction indicated that 16.8, 33.3 and 47 per cent remained available of the initially reactive amino groups in phenylalanine Bl, glycine Al, and lysine B 29 + arginine B22, respectively. The extent of the reaction of arginine B22 with glucose could not be determined with the TLC system used I for the experiment. The dansylation reaction appeared to give higher values of glucose-reacted -amino groups but this was most likely due to lower recoveries of fluores c~nce from the TLC spots containing small amounts of dansyl phenylalanine and dansyl glycine. The biological assays (Table 1) showed that the blood-glucose depressing activity of the Maillard insulin was reduced by about 20 per cent while the bloodtryptophan elevating activity remained unchanged.
When Maillard insulin of high specific 14 c-activity was incubated with a rat kidney homogenate at 370, there was no release of TCA-soluble counts before two hours. At four hours of incubation, however, some radioactive supernatant was already evident but this could have been greatly promoted by proteolytic action.   The binding of hexose residues and its effect on the acid solubility of insulin versus the time of storage with glucose. The initial solubility (app. lmg/ml) rose about ten times but then decreased to the level of ug/ml after four months of storage. ' .

DISCUSSION
The temperature-dependent reaction of D-glucose with the primary amino groups of insulin, which yields the moderately stable N-1-deoxy-D-fructosyl derivatives, may prevent the formation of the hexamer in solution. The relative ease with which glucose attacks the amino groups of insulin Bl Phe was suggested by , who reported that, at four days of reaction, 75 per cent of the Bl Phe residue had reacted compared to 37 per cent of the Al Gly or 20 per cent of the B29 Lys residues. By the dansylation reaction we found that the total number of amino groups reacted after fifteen days with glucose was 2.02. Atlhough this was in disagreement with the value of 1.4 obtained from the bindirfg of 14c-glucose, dansylation of Maillard insulin corroborated the observation by Schwartz and Lea that the order of reactivity is Bl)' Al';> B2:9. In the crystal, neither of these reactive residues is known to be involved in monomer-monomer contacts. The Phe Bl residue is known, however, to partipate in five dimer-dimer contacts which can be essential in maintaining the hexamer organization (Blundell and Assoc., 1972 ). Considering that the fast increase in acid solubility noticed after five days of incubation coincided with the binding of about one glucose residue, it is conceivable that formation of N-1-deoxyfructosyl Phe Bl is principally responsible for the failure of dime rs to associate into soluble polymers at pH 2.6. Marcker (1960) showed that hexamer formation was impaired with large substituents such as phenylcarbamyl at Bl. The extent of the reaction of B22 Arg with glucose was considered unimporiB.nt at fifteen days because of the unreactivity of the guanidinium group at neutral pH, and the participation of B22 in a salt bridge with the A 21 carboxylate forming an essential bond in the functional structure of the monomer.
While the solubility of the hormone at pH 2.6 was markedly increased by mild browning, the biological activity was decreased only slightly. By examining the biological activity of acetylated insulin, it was proposed that unsubstituted dand ~-amino groups were not required for the bloodglucose lowering effect of insulin ). The introduction of bulkier substituents at the amino groups resulted in reduced activity perhaps due to considerable distortion of the tertiary structure·rather than to the blocking of these groups . Despite its relatively large size, glucose can approach and readily react with the Bl terminal amino group. The «-N-l-deoxy-2-ketohexose substi tuent, though bulky enough to cause the dislocation of dimer-dimer contacts, did not appear to produce any major distortions in the tertiary structure of the monomer as inferred from the results of the potency assay. Furthermore, the fifteen-day Maillard insulin exhibited the same activity of normal insulin in raising the blood-tryptophan level in rats suggesting again that the Bl and, perhaps, the Al and B29 primary amino groups are not directly involved in this function.
The N-deoxyfructosylinsulin was stable at 4° in acidic pH for at least a week but its stability was not tested in ~· It is known, however, that radioactive tryptophan is not available for protein synthesis if it is injected intravenously in the form of N-deoxyfructosyltryptophan (Sgarbieri and Assoc., 1973). Since the in vitro test showed that Maillard insulin is apparently stable for two hours in the presence of kidney deacylases, it was assumed that the Nglycosidic bond was resistant to enzymatic hydrolysis in vivo -for the duration of the biological assays.
When insulin is mixed with glucose and stored in the dry state at mild temperatures, approximately four sugar residues will bind the hormone in a period of four months suggesting that monosubsti±ution of the reactive amino groups may take place at the initial stages of the reaction. The acid solubility, which increased about ten times with the binding of an average of 1.4 residues, was observed to progressively decrease upon further reaction with hexose. Impairment of hexamer association in solution was the probable reason for the initial increase of solubility, while significant monomer distortion could have accounted for the subsequent loss of solubility. The insulin-mediated sugar metabolism in rabbits was somewhat decreased but no change was detected in the ability to raise blood tryptophan in rats as a result of  ). N-acetyl-Ltryptophan that was fed to or injected into rats was demonstrated to produce normal growth after deacetylation in the body (du Vigneaud et al., 1932).  showed that o(-N-acetyllysine fed to rats had no nutritional value whereas the f-N-acetyl derivative did to some extent. Substitution of the O'-amino function of amino acids with larger groups such as l-deoxy-2-ketohexose, as it occurs in the Maillard reaction of amino acids with glucose, makes the amino acid generally unavailable to rats (Sgarbieri et al., 1973a). About 85% of the N-1-deoxyfructosyl derivative of leucine was shown to remain unabsorbed in the intestine of rats three hours after stomach-intubating the compound. A similar fate was suggested for N-1-deoxy-fructosyl-L-tryptophan .
While in bacteria the structural requirements for oligop eptide absorption have been worked out to some d~tail, virtually nothi ng is known about the rol e of the p eptide functional groups or structure in intestinal absorption. In bacteria, the terminal amino group of an oligopeptide must b e unacylated for abs orption. For instanc e , §. coli cannot absorb either OC-N-acetyldilysine or <X'-N-acetyltetralysine, and Q'-N-acetyltrilysine is only partially absorbed . Although lysine is the only residue within a polypeptide chain which has been proven to react with glucose (or a reducing sugar) during the mild Maillard reaction of proteins, resulting in a critical loss of nutritive quality, the terminal 0(-amino groups appear to react more readily.
Therefore, it is of interest to learn if the N-terminal oligopeptides produced in the animal digestion of Maillard proteins can be degraded and utilized down to the terminal, N-substituted residue.

Preparation of Q'-N-fructosyl glycylleucine
All materials were reagent grade. The procedure used was an adaptation of that employed by  for the synthesis of fructose-L-leucine. Glycyl-L-leucine (.9137 g; Schwarz-Mann, Orangeburg, New York) and glucose Sinc e the amino-sugar bond is labile at 100°, the fractions were tested for both a ketose compound by the ferricyanide reaction , and an amino acid by the ninhydrin reaction . Three bands were eluted between fractions 45 and 80 including a major ferricyani de/ninhydrin-positive band about fraction 76. Although weakly po s itive t o the ninhydrin test, the fir s t band was n egative in the amino acid analyzer. The second band was also discarded on the basis of its low ninhydrin color.
The fractions of the large peaks which accounted for about 80-90% of the t otal ninhydrin positive ma t erial, wer e pooled into two aliquots and each extracted seven times with ether; the ether was removed in a flask evaporator, and the product lypohilized. Each aliquot was dissolved in 50 ml of water, followed by repetition of the ether extraction evaporation and lyophilization steps. In the amino acid analyzer (Technicon Auto Analyzer) the compound was eluted in a single band near methionine while the unreacted dipeptide chromatographed between tryosine and ammonia. Upon acid hydrolysis, the product showed equimolar amounts of glycine and leucine whose proportion to total input suggested the compound was disubstituted. There was no detectable TCA residue as monitored in a GLC equipped with a free fatty acid column.

Preparation of Maillard lysyltryptophan
The dipeptide was browned in two different ways. Two Fifty microliters of freshly activated LPA were added to the substrate and the decrease in absorbance at 238 nm was recorded every five minutes. Since Maillard compounds absorb rather strongly in the UV range, it was necessary to adjust the absorbance of the Maillard substrate prior to addition of the enzyme by adding drops of brown glucose or fructose (browned on a hot plate for five minutes) to the reference cuvet. When the Maillard substrate was exposed to the UV light, the absorbance was occasionally observed to rise slightly (first five minutes). In such cases, the substrate was allowed to stand in the spectrophotometer without the enzyme until the rising trend ceased.
In vivo absorption of Maillard peptides

RESULTS
The major Maillard product of reacting glycyl-Lleucine with an excess of glucose chromatographed as a single peak near methionine in the amino acid analyzer.

23
Upon acid hydrolysis, equimolar amounts of glycine and leucine were obtained whose ratio to the total weight suggested the compound was d-N,N-di-(l-deoxy-2-ketohexosyi)~g~ycyl-L leucine .
The premelanoidins formed during the reaction of Llysine-L-tryptophan with glucose in the dry state were found to contain five ninhydrin-positive peaks plus unreacted dipeptide when chromatographed in an amino acid analyzer u s ing a special gradient ( Figure 3). Peaks A through D could be selectively adsorbed onto activated carbon leaving equal amounts of product E and unreacted dipeptide. Passing the mixture through a cell~lose column yielded a mixture containing products C and E plus the unreacted dipeptide.
The products from the methanolic reaction present in the major band of the Dowex-pyridine acetate columns were peak C (75%) and unreacted dipeptide (27%), which we r e also present among the products of the reaction in the dry state .
The Maillard gycylleucine was not hydrolyzable by leucine amino peptidase (LAP in vitro. Figure 4 shows that while no hydrolysis of the brown dipeptide was observed, the enzyme action on n ormal dipeptide was apparently inhibited  in the presence of the brown substrate. Addition of about 10% brown dipeptide to normal dipeptide lowered the rate of hydrolysis by mo r e than 35%. The hydrolysis was arrested as the inhibitor-to-substrate ratio approached 1:2, respectively.
Maillard glycylleucine was not absorbed in vivo but no effect on the absorption of normal dipeptide was apparent. The absorption patterns of the hydrolyzed amino acids is shown in both Figure 5 and Table 2. The sharp increase in both glycine and leucine in the portal blood minutes after duodenal injection of the normal dipeptide was contrasted by the almost total absence of either the free amino acids or the dipeptide after injection of the Maillard peptide.
When the brown dipeptide was injected together with regular glycylleucine in a 1:2 molar ratio, respectively, the concentration pattern of free glycine and leucine in the portal blood resembled that of normal absorption. This indicated that although the brown dipeptide was not a substrate in the small intestine, its presence did not impair the hydrolysis and absorption of the regular dipeptide. The absence of either dipeptide from all the chromatograms also suggested that, at the concentration used, the intact dipeptides were not absorbed.
The in vivo absorption of L-lysyl-L-tryptophan is compared in Figure~6 and Table 3 with the absorption of a mixture containing peak E (73%) and unreacted peptide (27%) which had been prepared in methanol and partially purified in pyridine acetate buffer. Twenty-five minutes after     Blood was sampled from cannulated portal vein, deproteinized and analyzed in a Technicon Auto Analyzer. No dipeptide in either form was detected in blood. One fasted female rate (160 ~) was injected with 1 ml of saline containing .125 u mole of neutralized peptide (pH 8).   1. Blood wa s s ampled from cannul a ted portal veins and anal y~ed for free trypto~han by the fluorescent method of . Both the normal and a form of the Maillard dipeptide were chromatographically detected in the plasma. 3. This material, first reacted in methanol and then chromatographed with pyridine acetate, was found to be a mixture of peak C (73%) and reacted dipeptide (27%). Each of the two fasted female rats (165-200 g) were injected with 1 ml of saline containing about .125 m mole of peptidyl residue at pH 8.

35
infusion of L-lysyl-L-tryptophan, the level of free tryptophan in the plasma had risen about five times. The increase was hardly noticeable, however, when the methanolic preparation was infused.
A different result was obtained when all the premelanoidins from the reaction in the dry state were infused (Table 3). The level of free tryptophan increased considerably even when the Maillard and the normal dipeptides were close to a 1:1 ratio. Unlike the neutral dipeptide glycylleucine, the basic lysyltryptophan and its Maillard derivative in peak E were observed to difuse unhdryolyzed into the blood.
To test if the lack of a free tryptophan in the serosal side after infusion of pyridine-acetate preparation was totally or in part due to inhibition caused by pyridine, According to the currently accepted mechanism for the absorption of protein hydrolyzates, a protein is not broken down to amino acids by intralumen digestion before the products are taken up and transported across the intestine.
Instead, the oligopeptides produced in the liquid-medium digestion are more e f fi.ciently adsorbed to class-specific sites, transported inside the mucosal cells, hydrolyzed I intracellularly, and the products transported to the serosal side .  has provided a unified concept for amino acid transport which may also explain the uptake, hydrolysis and transport of oligopeptides The massive influx of free tryptophan into the portal blood after infusion of the premelanoidins (Table 3), formed during the browing of L-lysyl-L-tryptophan with limited amounts of glucose, was an indication that the various 1-deoxy-fructosyl derivatives of lysine may not interfere with the transfer and hydrolysis of adjacent residues, regardless of the nutritional availability of lysine itself.
The transfer and hydrolysis of the normal dipeptide was neither considered to be impaired by the presence of the premelanoidins.
The role and relative importance of the functional groups of L-lysyl-L-tryptophan could not be conclusively interpre ted given that (1 )  The first two bands contained high proportions of lysine, plus other essential and non-essentional amino acids. Band III yielded 90% NH3 upon acid hydrolysis.
(ii) Portions from the ammonium acetate extracts, which were presumed to contain bands I and II, could be resolved into six peaks by cation exchange. Each peak was analyzed for amino acids.
(iii) Samples extracted and gel-filtered in pyridine acetate gave a ninhydrin-positive profile similar to that in (i) but in different proportions. Cation exchange chromatography revealed at least fifteen peaks, which were positive to the alkaline-hydrolysis test but were neither peptides nor glycosidic residues.
Similarly prepared Maillard ovalbumin (2x crystalline) tagged with either 14 c-lysine or-tryptophan was also fed to rats to assess the fecal recovery. Feeding the Maillard 40 ovalbumin raised the recovery of radioactivity from 1 to 7% for the 'lysine' label, and from .2 to 3% for the 'tryptophan' label. Refeeding the 'lysine'-labeled extract increased the recovery to 15%.

INTRODUCTION
In digestibility and nutritional studies of proteins treated by the Maillard reaction (for a review see , several authors have suggested that the binding of sug~r residues makes certain regions of the protein undigestable, unabsorable and/or unavailable for animal growth (Clark and Tannenbaum, 1970;Valle-Riestra and Barnes, 1970;. However : feasible, this hypothesis has not been tested, or its importance evaluated as a possible mechanism for the loss of nutritive value of mildly brown proteins.
Two types of Maillard browning reactions should be distinguished according to the heating conditions. One that is carried out at miltt , below denaturing, temperatures,l and another at high, processing or sterilizing temperatures.
Under mild conditions, . the reaction due to the presence of sugars, could be observed without the intervention of heat denaturation (insolubilization, condensation; oxidation).
The duration of the treatment, of course, is an important factor.
The type of physiocochemical or conformational changes introduced to a protein within one month of mild heat treatment could be minimal, so that any digestive and lAs the term shall be used throughout this report.

42
nutritive changes observed when exposed to reducing sugars can be mostly ascribed to the chemical addition of sugar moieties to the protein.   Besides the terminal amino groups of a protein, the only other amino groups likely to react with a reducing sugar is the t-amino group of lysine (pK 10.51) and the guanidinium group of arginine (pK 12.48). Their high pK values raise the energies of activation making the reaction rate relatively low for the t-amino and negligible for the guanidinium, compared to the ~-amino group. (The guandinium group of insulin is known not to react with glyoxal at pH 7; .) It would seem, therefore, that for mildly brown proteins, the entire mechanism of 'unavailable 43 peptides' would have to rest on either or both of the following assumptions: (a) ~-N-ketosylated polypeptides, which contain essential amino acids, must be excluded from uptake and surface hydrolysis in the mucosal cells ; (b) f-N-substituted lysine must significantly delay endopeptidase action, hindering the release of neighboring residues.
Related to the first assumption, we have recently reported that the major product of the reaction of glycyl-Lleucine with glucose is not taken up by the small intestine of the rat; in fact it was observed that neither the whole Maillard dipeptide, nor any of its parts were able to cross the wall of the small intestine . Regarding the second assumption, not much more is known either.  claimed that trypsin-mediated hydrolysis of a lysine-containing peptide was signficantly lowered after allowing the (-amino group to react with glucose. It is also known that l-N-fructosyl-lysine can slowly diffuse through the intestine though it is hardly available for growth (Erbersdobler, 1973 Kekwick and Cannan (1936). The radioactivity recovered throughout crystallization and dialysis was 81.2% for the lysine-labeled and 86.1% for the tryptophan-labeled product.

Feeding the Maillard protein
A homogeneous mixture of egg albumin (Nutritional Biochemicals Corporation, Cleveland, Ohio) and glucose (3:2, w/w) was prepared and incubated for 30 d9ys at 37° and 68% R.H. After brown, the protein was dialyzed and diets con-t~ining 10% of either untreated (U) or brown (M) protein were made. Between the third and fifth weeks of feeding the diets to groups of five growing rats, the feces were collected and analyzed. All the steps involved fr0m the browning of the protein to the preparation and feeding of the diets were already described in detail in the nutritional evaluation of the proteins by .

46
The radioactive and 'cold' ovalbumins were incorporated into diets as explained above and fed to pairs of male (Sprague-Dawley, Charles River Breeding Laboratories, Wilmington, Massachusetts) rats, weighing about 100 g each.
The rats were trained for five days to eat two meals a day Vitamin fortification mixture (see  Corn oil 10% 20% 54% 5%

10%
The water-soluble extracts of the feces containing the lysine-labeled material were added to the ,abgve diet and fed in 6-g meals. Coprophagy was prevented and the feces were collected daily for three days.

Analysis of the feces
The water-soluble fraction of the feces was extracted and passed through a polyacrylamide gel column using either The following tests were performed on .2 -.5 ml of sample: Protein Lowry , ninhydrin , and ferricyanide  for the typical ketosyl residue of glucose-generated Maillard compounds.
Amino acid analysis was also performed on each one of the bands obtained from the NaCl/Bio-Gel patterns. For this, the pooled lyophil ±zed fractions were efficiently desalted by extracting twice with 95% ethanol at room temperature, and then centrifuging at 10,000 x g for ten minutes.
After evaporation of the 85% ethanol, the residue was hy-

49
Each of the peaks of the cation ex change column was subject to the following test: amino acid analysis (Technicon Auto Analyzer), rapid alkaline hydrolysis (see Appendix B), the ketose ferricyanide reaction , and the fluorescamine reaction for amines (Undenfriend et al., 1972; Fluram T.M., Hoffmann-LaRoche, Inc., Nutley, New Jersey).
t  a. The lyophilized, water-soluble material was incorporated to an amino acid diet, in which the starch was replaced by cellulose (see experimental section). · b. The rats were trained as in Table ~A except that a synthetic amino acid mixture was used instead of ovalburnin.  -52 a. Samples of .2 to .6 g of dry, ground feces were extracted and sedimented three times in 10 ml of .05 M NaCl, the r esulting fractions lyophilized and the N determined in a . s erili·~rili. t:rokj eldahl apparatus. Values were corrected for NaCl and moisture content. b. Number of replicates.

Feeding the Radioactive Ovalburnin
Six growing rats were fed a standard diet containing Maillard ovalburnin which was radioactively labeled in either the lysyl or the tryptophyl residues. Table 4Ei compare1 s the fecal recovery of radioactivity from these rats with that of four rats that received unreacted ovalburnin. On the average, close to 90% of the radioactivity was found in the water-soluble fraction of the feces, and, about 7% in the low-molecular weight region. The storage-like browning produced a five-fold increase in the re~overy of fecal radioactivity for the 14c-lysine ovalburnin. Browning of the l4c-tryptophan ovalburnin brought about an increase greater than ten-fold over the background of the unreacted ovalburnin (Table 4A).
When the water-s.oluble fraction of the feces from the 14c-lysine ovalburnin was refed, to two rats, the recovery of radioactivity was raised from 6.7 to 14.7%. The results of this experiment are recorded in Table 4B.

Analysis of the Water-Soluble Fraction
The following anlysis was carried out on feces collected during the last two weeks of a five-week feeding trial, in which non-crystalline egg albumin was used instead of ovalburnin. Table 5 shows the distribution of nitrogen, according to its water (NaCl) solubility, in the feces,   Figure 1. b. Ve is elution volume. Vo is void volume. The column bed was Bio-Gel· P-2 (200-400 mesh) and the eluant .05 M NaCl. The average of two trials is reported. c. BSA is bovine serum albumin. generated from both Maillard and unreacted egg albumin. It was app~rent that the total nitrogen of the water-soluble fractio n of the feces generated by Maillard egg albumin rose 470% while the insoluble nitrogen was only 16% aqove the control.
Extraction and gel-filtr ation of the water-soluble portions was accomplished by using three different solutions. The material present in these bands ranged between 800 and 1300 molecular weight ( Table 6). The ketose/ ferricyanid e test was positive for the thre _ e bands, but most prominently so for band II (Figure 8). Amino acid  analysis was also performed on each band after HCl hydrolysis. Four major _ featu~es stand out from the amino acid data in Table 7 Us ing thi s gel filtration system, the water-soluble extract of radioactive feces from Maillard 14c-Lys ovalbumin was eluted. Figure 9 indicated that about 80% of this radioa ctivity co-chr omatographed with peaks I and II of a 'cold'egg albumin fecal carrier.
(ii) The dried, · gr ound feces were alternatively extracted and eluted with .05 M ammonium acetate. Although this s y s t em was c apabl e of s eparating a diffuse band which appearedl brown, uv-fluorescent, and low in molecular weight, it did not provide a reproducible ninhydrin profile after removal of exc ess ammonium ion. The 'band' thus prep a r ed, however, was used to demonstr ate the presence of several peptide-like bands as resolved by cation ex change. Figure   10 shows a t ypi cal ninhydrin profile on a Dowex 50W X-4 column. Tabl e 8 gi v es the inte rnal ami no acid c ompo s i t ion   Internal amino acid composition (mole%) of five peaks from the Do:wex 50W X-4 column · in Figure 10 Amino b. Peak D-3 was lost accidentally. of each band after HCl hydrolysis.
(iii) In addition to the above separation procedures, the feces were also extracted and eluted through 63 a Bio-Gel column with .2 -.5 M pyridine acetate (pH 4.1).
The ninhydrin profile (Figure 11) was clearly different from that obtained in (i).
Moreover, when bands II and III were re-chromatographed in a Dowex column ( Figures 12A, 12B, the ninhydrin profile differ ed from that of the ammoniwn acetate ex traction. After HCl hydrolysi s , each peak in Figure TABLE 9 Summary of the chemical tests performed on the Dowex fractions of the Bio-Gel/pyridine acetate peaks II and III whose profile is shown in Figure 1 2A ( 100% ) NH 3 Gly Gly a. Dowex 50W X-4 separation of bands II, III from the EiG-GeL-.PT4 /pyridine acetate system. Feces were from the rats fed Maillard egg albumin.
b. If ketoses as in the Amadori compound--are present, the ferricyanide ion is reduced and forms Prussian blue (660 nm).
c. Samples of .5 ml are hydrolyzed in 1 ml of 2.5 N KOH in the autoclave for 50 min., then neutralized with .6 ml of 30% AcOH, and the ninhydrin test is compared against the unhydrolyzed sample. d. Is a hexosamine determination with the Earlich's reagent (Johnson, 1971). e. Is a fluorescent reaction for the determination of primary amines . It was carried out in borate buffer pH 8 . 0 (Undenfriend et al., 1972). Values are ratios of.'fluorescence over ninhydrin color. The value for n-leucine was 2.2.

DISCUSSION
From the feeding and refeeding experiments it wa s appa rent tha t mild Maill a rd browning of ovalbumin made certain portions of the polypeptide ch~in, bearing a 14c-tagging essenti al amino acid, from four to ten times less likely to be absorbed by the intestine of the rat (Table 4 ). The concentration of lysine and arginine were indisputably eleva ted in the fecal water-soluble fr a ction a fter browning (Table ?. ). It wa s also observed tha t, besides a spa rtic and glutamic acids, the neutral amino acids glycine, alanine, valine, isoleucine and leucine were found in high proportions in some of the isolated Dowex peaks (Tgble 8). Since neutral amino acids participate in the Maill a rd reaction only if they are N-terminal residues, it would not be surprising if a major egg protein, for instance, has either glycine or leucine or isoleucine as its N-terminal residue (Table 6,  Although the level of fecal excretion was below 10% of the ingested radioactivity, this loss was considered important be cause ( a ) the radioactivity was concentrated between the 1000 and 1300 molecular weight fragment ( Figure   9); (b) the relative nutritive value a ssociated with such protein was estimated to be 7%  the absence of radioactivity from the 800-molecular weight band warranted investigation of the nature of such band and its role in the loss of n~tritive value induced by mild Maillard bvowning.
Aqueous solutions of sodium chloride and ammonium acetate extracted compounds which, upon acid hydrolysis, yielded various amino acids plus ammonia or ammonia alone.
Meanwhile, pyridine acetate appeared to selectively extract the ammonia-yielding compounds characteristic of the lowmolecular weight band. Bands III (Figure 7), II and III ( Figure 11) were major contributors to the total soluble nitrogen of the feces. It was shown by alkaline hydrolysis + that this nitrogen was not ionically bound NH4, but rather in the form of primary or secondary non-peptide amines.
Their excellent solubility in ethanol and pyridine also suggested the compounds were not peptides.
One remote source of non-protein nitrogen with Maillard characteristics in the feces was the possibly undigested glycosidic moieties of the albumin's glycoproteins. This possibility was discarded after the negative results of the Elson-Morgan reaction for hexosamines.
Another source considered was that of bacterial or intestinal origin, but this was unlikely given the molecular weight range and the very Maillard characteristics . of the band.
So far as the author is aware, the presence of polyamines in egg white proteins has not been reported. Nevertheless, it is possible that such important growth factors   Our data suggest that mild Maillard browning is not sufficient treatment to produc e. high lysine losses (Table 4.) as is known to occur with autoclaved Maillard proteins (Valle- Riestra and Barnes, 1970). This contention is in agreement with investigations on the glucose-to-lysine binding in proteins (Schwartz . ana.Lea, 1950;, and also with the higher energy of activation required for the nucleophylic attack on the carbonyl carbon of the sugar by the E-amino group.
In contrast with the apparently IlLt.file amino acid damage,  have reported that as much as 50% of the nutritive value is lost after ten days of browning. These authors also suggested that growing rats could be more sensitive to the inferior quality of Maillard protein than are adult rats.  came to conelude that even a thorough amino acid supplementation did not completely restore the lost nutritive value of a mildly browned Maillard protein.
Maillard proteins reacted at high temperatures should, therefore, be distinguished from those reacted in mild conditions. In the first case, recoveries of up to 70% of the ingested lysine radioactivity have been reported (Valle- Riestra and Barnes, 1970), whereas less than 7% is recovered in the second case despite the low nutritive value of the protein. Whether the mild treatment is to result in an increased absorption of still unutil izable lysine (because of ' the less cross linking and higher distibility) is not known.
For certain, the low l~c radioactivity recovered in the feces when labeled ovalbumin was fed does not seem to be a nutritionally important loss. It must be emphasized that, to the present, no conclusive evidence has been found on the possible toxicity of mildly browned Maillard proteins.
The recent finding that free radical products can form early in sugar-amine reactions (Namiki et al., 1973) should be given ample consideration, not only from the chemical but also from the nutritional standpoint. Free radicals formed at 1000 could add to and cross link amino acid residues in a protein thu s contributing to its insolubility and tmdigestibility. It would be of interest to study the importance and stability of the Maillard reaction free radicals at mild temperatures.
In summary, evidence has been presented for the accumulation of (a) oligopeptide and (b) non-peptide nitrogen in the feces fed mildly browned egg albumin. Both types of residues exhibited the characteristics of Maillard products.
The Maillard oligopeptides, three to five residues long, could have escaped absorption and surface hydrolysis in the intestine because of the lack of free a-amino groups.
From crystalline ovalbumin only 6% 14 c-lysine radioactivity was recbvered associ.ated with the small peptides, a loss products. The mechanism for fission appears to be the rever se of aldol condensation (dealdolization).
Another reaction that has been observed i n the intermediate stage of browning is the Strecker degradation. In this degradation, an «-amino acid decarboxylates if heated in the presence of certain polycarbonyl compounds to give an aldehyde containing one carbon less. The nitrogen atom is transferred to the keto compound --just like in the ninhydrin reaction --although there is no production of color.
In fact, it is believed that the Strecker degradation accounts for about 80% of the C02 formed and that it slows down the development of brown color.
In the third and final stage, the intermediates polymerize by aldol condensation, aldehyde-amine polymerization and heterocyclization. The final product, called melanoidins, are fluorescent, unsaturated compounds that exhibit a brown color without an absorption maximum in the visible range.
Revaluation of the mechafilism for the reaction may be needed in the near future due to the recent detection of free radicals even at the initiation steps. Nakimi et al. (1973) suggested that an early amino-sugar intermediate A BRIEF HISTORICAL ACCOUNT The Maillard reaction was regarded as little mor e than a chemical curiosity for quite some time after its discovery.
During the two-and-one-half decades that followed L.C.
Maillard's report in 1912 on the dramatic reaction that could occur between amino acids and glucose, no serious consideration was given to the finding by any field of science. It was not until the middle of the 19~0's when the need arose to ship large quantities of dehydrated foods for troops stationed overseas that research on the reaction actually began to flourish. It is somewhat surprising to note that, even at that point, all interest in the reaction was triggered by the food processor's concern over the development of undesirable flavors and the low acceptance by soldiers of products such as dehydrated eggs.
Although Although the British investigators adopted c a sein, and the American counterparts egg albumin as the protein "model system," it soon became obvious that despite their practical relevance, such proteins were highly undefined and complex chemi-  (Reynolds, 1969).
A couple of misconceptions remained from all that work in the decade of the 1950's. The idea was put forth by Lea and Hannan (1950) that, even the peptide nitrogen of a protein is reactive towards the carbonyl carbon of a reducing sugar. This concept was neither proved nor disproved. It is now known that the reactivity of the peptide nitrogen is negligible given that its lone pair of electrons participate in the resonant structure of the peptide bond, and are not readily available for a nucleophilic attack. Such an idea, however, must have mislead Horn et al. (1968 ), who concluded that all the peptide nitrogen atoms of a protein must react with the carbohydrate.
Another idea, which was more like a passing fancy, was based on the claim that ketose-amino acid (Amadori ) co~pounds could stimulate the synthesis of protein in vitro . Since this research was not followed up, and recent in vivo nutritional experiments attest to the contrary, the report is no longer considered of biochemical significance. These authors, however, did start a new age of research in the Maillard reaction; namely, the biochemical nutrition.
It should be added at this point that somewhat of a con-   9. The animal is then secured in a restraining cage where the experiment is initiated one hour after it has regained consciousness. Acceptance of water or saline water is considered a sign of good physical condition.
The duodenal cannula should be occiuded with a pin or a hemostat immediately after infusion of the substrate.
This modification involves the same ba sic steps already described except that a syringe is used instead of a cannula, and the experiment is conducted in the unclosed animal under Penthrane anesthesia. Since the operation can be shortened to about fifteen minutes prior to the duodenal infusion, this procedure is recommended for absorption studies of no more than one and one-half hours long.

Procedure
The rat is prepared in the same manner as indicated Gradient for Fecal Peptides. This gradient was devised to separate a number of fecal peptides using a Pharmacia 1.6 x 90 cm column (Pharmacia, Piscataway, New Jersey) at room temperature. The bed gel used was the Dowex 50W X-4 resin, which was extern3i vely cleaned before use. The cleaning of the resin started by washing and decanting twice in one volume of 2N NaOH, followed by several rinses with H20, and finally washing twice in one volume of 2N HCl. After extensive rinsing with H~O, the gel was equilibrated with starting buffer. In an a t tempt to monitor the decrease of tryptophan availability by che~ical methods, we encountered difficulties in obtaining reproducibility of the color intensity and maximum of absorption of the p-dimethylaminobenzaldehyde (p~DMA:~) r eaction. Obviously, there was interference in the formation of the blue chromophore complex by the presence of browning pigment (probably the polyketo compounds). As a result of this the color often shifted towards a golden or red hue.
· Since not much is known about the chemistry of concentrated sulfuric acid-mediated color reactions and I was also iIBterested in shortening the reaction time of the classical reaction, I developed a modification which is about five times more sensitive than the classical one, and could have application in preliminary protein screening.
Because the activation of tryptophan requires the use of s trong HCl, the procedure has two limitations: the color procedure has relatively short stability and corrosive fumes may be harmful to the optics of instruments.

Procedure
The procedure I later discovered, has some similarities with an earlier, less developed procedure (Sullivan, Milone and Everett, 1938). Read color at 585 NM within 10 min.

Making the p-DMAB Solution
To 15 ml H20 add 50 ml of Baker A.R. H2S04. Cool in water bath. Dissolve 1.5 g of p-DMAB while the acid is still warm. Store protected from light.

Example
Analysis of "Carnation Instant Breakfast." The food was extracted once with ether· and once with CHC1 3 .
Later it was found that this step was not necessary. One gram of food was suspended in 10 ml of H20 and the reaction was carried out as described above. APPENDIX C

SUMMARY OF CONCLUSIONS
The problem under study was the loss of the biological properties of a protein when stored with reducing sugars at temperatures below 4o 0 c. The process was referred to as the 'Maillard reaction at mild temperatures,' when there appears to be little or no development of brown color.
Emphasis was placed on the temperature factor because marked differences have emerged between the type of damage caused to a Maillard protein depending on the temperature and the stage of the reaction. The conclusions drawn from the foregoing observations were as follows: 1. When insulin was allowed to undergo the Maillard reaction at mild temperature for fifteen days, little or no denaturation took place as evidenced by the high biological activity of the deoxyfructosyl-substituted hormone. Although this conclusion cannot automatically be extended to egg albumin or other food proteins, other pieces of research cited below appear to support the generalization.
2. The small number of sugar residues bound to insulin indicated that, even after four months of storage with glucose, the peptide nitrogen had no participation in the reaction. If that was the case, we would have a situation in which (a) up to forty-nine glucose residues could react with the peptide linkage alone, and (b) reaction of the amide nitrogen would imply perturbation of a resonant structure wfuich could, in turn, result in the disruption of the tertiary, seconda~y and primary structures of a polypeptide.
Although the presence of free radicals in the Maillard reaction at higher temperatures (Nakimi et al., 1973) could conceivably perturbate the peptide bond or induce polymerization, this did not seem to be the case in the mild browning of insulin. On the assumption that polymeric melanoidins are stable at pH 2.4-2.6 (slightly higher than the gastric pH), one could add with good certainty that no polymeric, reductone-like compounds had yet developed bound to the protein. Clark and T~nnenbaum (1974) have lately reported the isolation of limit peptide pigments (LPP) containing at least 8 sugar residues per amino group. Clark and T9nnenbaum' s LLP's were obtained from an insulin-14 C-glucose system reacted at 55° for 37 days.  .
A clear distinction, therefore, must be made between the Maillard reaction catalyzed by high tempe~atures (55°-1210; processing and autoclaving) and that catalyzed by mild temperatures (ambient -37°; storage). It is not known if the results eventually are the same or not, but in the latter case it is possible to observe the initial stages of the reaction. For the case of high-temp enature browning it has been reported that up to 70°/o of the "lysine" radioactivity fed to rats is recoverable in the feces (VRlle-Ri e stra and B~~nes, 1970). Also, Clark and Tannenbaum (1974) have reported that pronase digestion of 55°c-