The protein digestibility-corrected amino acid score (PDCAAS)² method has been considered to be a simple and scientifically sound approach for routine assessment of dietary protein quality for humans (FAO/WHO 1991). FAO/WHO (1991) recommended use of the FAO/WHO/UNU (1985) amino acid requirement pattern for children of preschool age for the evaluation of dietary protein quality for all age groups except infants. The PDCAAS is now a federally approved alternative method to the protein efficiency ratio (PER) rat bioassay procedure still official in Canada and U.S.A. Major questions have, however, been raised about the validity of the PDCAAS relative to its inability to credit the extra nutritional values of proteins having scores higher than that of the reference protein, its failure to fully account for the possible adverse effects of antinutritional factors, and its assumption about complete biological efficiency of supplemental amino acids in improving quality of proteins, which may not be true in the case of poorly digestible, low quality proteins (Food Chemical News 1991). These concerns about the PDCAAS, however, require proper documentation (Young 1995).

Antinutritional factors may occur naturally or may be formed during heat processing. Some examples of naturally occurring antinutritional factors include glucosinolates in mustard and rapeseed protein products (Fenwick et al. 1982), trypsin inhibitors and hemagglutinins in legumes (Rackis and Gumbmann 1981), phytates in cereals and oilseeds (Sandberg 1991), and gossypol in cottonseed protein preparations (Martinez and Hopkins 1975), which could adversely affect nutrient utilization and may contribute to growth depression in animals.

During processing of foods, protein sources are treated with heat, oxidizing agents (such as hydrogen peroxide), organic solvents, alkalis, and acids for a variety of reasons such as to sterilize/pasteurize, to improve flavor, texture, and other functional properties, to deactivate antinutritional factors, and to prepare concentrated protein products (Cheftel 1979, Friedman et al. 1984, Schwass and Finley 1984). These processing treatments may cause the formation of Maillard compounds, oxidized forms of sulfur amino acids, D-amino acids, and crosslinked peptide chains (such as lysinoalanine and lanthionine), resulting in lower amino acid bioavailability and protein quality.

The purpose of this study was to assess the validity of the PDCAAS method in evaluating the quality of six protein sources containing antinutritional factors, and of one poorly digestible protein supplemented with limiting amino acids. A rat growth and balance study was conducted to determine protein digestibility and quality of fourteen test diets. Amino acid profiles of test proteins were also determined to calculate scores by the PDCAAS method, using a rat and a human pattern of amino acid requirements. The PDCAAS values were compared to the protein quality indices based on rat growth, such as protein efficiency ratio (PER), net protein ratio (NPR), relative PER (RPER) and relative NPR (RNPR).

MATERIALS AND METHODS

Total nitrogen was determined (in duplicate) by the microKjeldahl method using a Kjeltec Auto 1030 Analyzer (Tecator, Herndon, VA). Protein was calculated using a nitrogen-to-protein factor of 6.25. Protein sources were hydrolyzed (in duplicate) with 6 mol HCl/L at 110°C for 22 h for the determination of total amino acids except tryptophan and sulfur amino acids (AOAC 1990). Hydrolysis with NaOH (4.2 mol/L) was used for quantitative recovery of tryptophan (AOAC 1990). Hydrolysates for determination of methionine as methionine sulfone and cyste(i)ne as cysteic acid were prepared by performic acid oxidation of protein followed by hydrolysis with HCl (6 mol/L) (AOAC 1990). All amino acids, except tryptophan, in hydrolysates were determined by liquid chromatography of precolumn phenylisothiocyanate derivatives (Sarwar et al. 1988). Tryptophan in alkaline hydrolysates was determined by a simple liquid chromatography method requiring no derivatization (Sarwar et al. 1988). Amino acid analysis by liquid chromatography of precolumn phenylisothiocyanate derivatives has been successfully validated by conventional ion-exchange methods (Beaver et al. 1987; Sarwar et al. 1988; White et al. 1986).

The casein + L-methionine (0.2% of diet) control and test products (fourteen) were fed as the sole source of protein in diets containing 8 g protein/100 g diet (Table 1). A protein-free diet (Table 1) was also included in the feeding study to enable calculation of NPR, and for providing an estimate of metabolic fecal protein (MFP) needed for calculating true protein digestibility. The protein diets were made isonitrogenous by the addition of L-glutamic acid (Table 1). The actual analysis revealed that the casein + Met, casein, lactalbumin, skim milk, treated lactalbumin, heated skim milk, SPI, treated SPI, raw soybean meal, heated soybean meal, raw black beans, heated black beans, mustard, zein and zein + amino acids diets contained 9.00, 9.01, 8.80, 8.78, 8.90, 8.92, 8.94, 8.86, 8.88, 9.05, 8.90, 8.76, 9.00, 8.78 and 8.97 g protein/100 g diet (N × 6.25), respectively. The determined protein values were used in calculating protein intake data, required for the determination of protein digestibility, PER and NPR.

Table 1. Composition of experimental diets fed to rats [View Table]

Table 2. A comparison of essential amino acid requirements for rat growth, as recommended by NRC (1978) and NRC (1995) [View Table]

Table 3. Growth of rats and protein efficiency ratios (PER), net protein ratios (NPR) and true protein digestibility values of experimental diets1 [View Table]

True protein digestibility values were calculated using the following equation (Sarwar and Peace 1986): where PI = protein intake, FP = fecal protein and MFP = metabolic fecal protein. The amount of protein in the feces of rats fed the protein-free diet was used as the estimate for MFP. Relative protein efficiency ratio (RPER) and relative NPR (RNPR) values (2-wk) were calculated using the following equations (Sarwar and Peace 1994): where PER = weight gain of test rat/protein consumed by test rat, and NPR = (weight gain of test rat + weight loss of nonprotein rat)/(protein consumed by test rat).

The PDCAAS was calculated as follows (Sarwar and McDonough 1990): Amino acid ratios [(mg of an essential amino acid in 1.0 g of test protein/mg of the same amino acid in 1.0 g of reference protein) × 100] for nine essential amino acids (His, Ile, Leu, Lys, Met + Cys, Phe + Tyr, Thr, Trp and Val) were calculated using a rat growth pattern of amino acid requirements (NRC 1995, Sarwar et al. 1985) and a human pattern of amino acid requirements (FAO/WHO/UNU 1985 suggested pattern of amino acid requirements for preschool children, 2-5 y) as the reference proteins.

Two rat growth-based patterns of essential amino acid requirements (NRC, 1978 and 1995) are compared in Table 2. The values are expressed as g amino acid/100 g dietary protein since the requirements remain constant or decrease slightly when expressed a percentage of protein (NRC 1995). The values for all essential amino acids except arginine and methionine + cystine are similar in both the patterns. The arginine value was considerably lower in the 1995 report compared to the 1978 report (2.87 vs. 5.00 g/100 g protein), while the methionine + cystine value was higher in the 1995 report than in the 1978 report (6.53 vs. 5.00 g/100 g protein, respectively). Since the ASNS-recommended methionine- or cystine-supplemented casein-based control diet for laboratory rodents (providing 5.0 g of methionine + cystine/100 g protein) meets or exceeds rat growth requirements for essential amino acids (AIN 1977, Reeves et al. 1993), the NRC 1995 rat requirement value for methionine + cystine is too high. Moreover, the methionine-supplemented casein (containing 4.74 g of methionine + cystine/100 g protein) was reported to meet or exceed the requirements of rats for essential amino acids in terms of RPER and RNPR (Sarwar 1984, Sarwar et al. 1984). Pick and Meade (1971) reported that the sulfur amino acid requirement of growing rats was 4.26 g/100 g dietary protein. Peace et al. (1986) and Sarwar et al. (1985) estimated the methionine + cystine requirement of growing rats to be 4.1 and 4.0 g/100 g dietary protein, respectively, using RNPR as the response criterion. The NRC-specified requirement may be higher to make allowance for the active role of methionine as a methylating agent in addition to the minimum needed for protein synthesis. Since the vitamin mixture used in the studies of Peace et al. (1986) and Sarwar et al. (1985) provided an adequate level of dietary choline, the requirement for methionine as a methyl donor may have been reduced. Based on the findings of Peace et al. (1986) and Sarwar et al. (1985), a value of 4.0 g/100 g dietary protein for growth requirements of rats for sulfur amino acids was used in calculating amino acid scores. For all other essential amino acids, the values reported by NRC (1995) were used in the calculation of the scores.

RESULTS

The rat growth and food intake data needed to determine PER and NPR of the experimental diets are also shown in Table 3. The PER and NPR values of mustard flour were zero. The alkaline treatment had a drastic negative effect on the PER and NPR values of lactalbumin (4.29 vs. 0%; 5.18 vs. 0%) and SPI (2.69 vs. 0%; 3.66 vs. 0%). As noted in the case of protein digestibility, moderate heat processing improved the PER and NPR values of black beans (0 vs. 0.87%; 0 vs. 1.97%) and SBM (1.30 vs. 3.04%; 2.51 vs. 4.00%), while overheating caused a drastic reduction in the PER and NPR values of skim milk powder (3.72 vs. 0%; 4.65 vs. 0%). The PER and NPR values of the zein diet were 0% and 1.45%, respectively. Supplementation of the zein diet with limiting amino acids (lysine, methionine, threonine + tryptophan) caused significant improvement in the NPR value (2.53%).

The calculations of amino acid ratios (the lowest amino acid ratio being the amino acid score) for the four most limiting amino acids (lysine, methionine + cystine, threonine and tryptophan) using the rat growth pattern for amino acid requirements are shown in Table 4. Only casein + methionine and untreated lactalbumin met or exceeded all the essential amino acids requirements for rat growth (Table 4, data for all essential amino acids not shown). Casein, skim milk powder, untreated SPI, alkaline-treated SPI, raw SBM, heated SBM, raw black beans, heated black beans, mustard flour and amino acid-supplemented zein were all first-limiting in sulfur amino acids with amino acid scores of 86, 79, 65, 45, 72, 70, 63, 61, 91, and 86%, respectively. Alkaline-treated lactalbumin was first-limiting in threonine (amino acid score of 76%), while heated skim milk and zein were first-limiting in lysine with amino acid scores of 38 and 2%, respectively.

Table 4. Lysine, methionine + cystine, threonine and tryptophan in protein sources, expressed as a proportion of the respective amino acids for rat growth requirements1 [View Table]

The calculations of amino acid ratios (the lowest amino acid ratio being the amino acid score) for four most limiting amino acids (lysine, methionine + cystine, threonine and tryptophan) using the human pattern for amino acid requirements are shown in Table 5. Casein + methionine, casein, untreated lactalbumin, skim milk powder, untreated SPI, raw SBM, heated SBM, raw black beans, heated black beans and mustard flour met or exceeded all the essential amino acids requirements for humans using the FAO/WHO (1991) amino acid reference pattern (Table 5, data for all essential amino acids not shown). The overheated skim milk powder was first-limiting in lysine (score of 40%), while the alkaline-treated SPI and lactalbumin were limiting in methionine + cystine (score of 72%) and threonine (score of 92%), respectively. Zein was first-limiting in lysine (score of 2%), and the supplemented zein was marginally limiting in lysine (score of 99%).

Table 5. Lysine, methionine + cystine, threonine and tryptophan in protein sources, expressed as a proportion of the respective amino acids in human reference amino acid pattern (FAO-WHO 1991) [View Table]

The protein digestibility-corrected scores, RPER and RNPR are shown in Table 6. Due to the higher amino acid requirements for rat growth compared to that of human (especially sulfur amino acids), the PDCAAS of most products based on rat requirements were considerably lower compared to those based on human requirements.

Table 6. Protein digestibility-corrected amino acid scores (PDCAAS), relative protein efficiency ratio (RRER), and relative net protein ratio (RNPR) values for several protein products [View Table]

The differences between the PDCAAS based on rat requirements and RNPR or RPER of casein, lactalbumin, skim milk powder, SPI, heated SBM were small (about 10%) (Table 6). But, the PDCAAS based on rat requirements for alkaline-treated lactalbumin, heated skim milk powder, alkaline-treated SPI, raw SBM, raw black beans, heated black beans, mustard flour and supplemented zein were up to 84% higher compared to the RPER or RNPR values for these products.

The PDCAAS based on human requirements for casein, untreated lactalbumin, untreated skim milk, untreated SPI and mustard flour were high (Table 6). The PDCAAS for the alkaline-treated lactalbumin and treated SPI were 67 and 49% of the values for untreated lactalbumin and SPI, respectively. The PDCAAS for overheated skim milk powder was 31% of that for the untreated powder. The difference between the PDCAAS of raw and heated SBM was small (80 vs. 83%), but the PDCAAS of heated black beans was higher than that of raw black beans (84 vs. 72%). The PDCAAS of amino acid-supplemented zein was 70% higher than that of unsupplemented zein.

DISCUSSION

The RNPR values of all the test protein sources were higher than RPER values (Table 6) because unlike the PER method, the NPR method credits protein used for both growth and maintenance. The protein required to prevent weight loss of rats fed the protein-free diet is assumed to be equivalent to the protein needed for maintenance.

A comparison of the PDCAAS based on rat requirements and RPER or RNPR indicated that the scoring method clearly overestimated protein quality of those protein sources which contained antinutritional factors such as mustard flour (PDCAAS of 84% vs. RPER or RNPR of 0), raw black beans (PDCAAS of 45% vs. RPER or RNPR of 0), alkaline-treated lactalbumin (PDCAAS of 55% vs. RPER or RNPR of 0), alkaline-treated SPI (PDCAAS of 44% vs. RPER or RNPR of 0), and overheated skim milk powder (PDCAAS of 29% vs. RPER/RNPR of 0-5%) (Table 6). The overestimation of protein quality of these protein sources by the PDCAAS method was even more marked when the scores were calculated based on human requirements.

Some of these protein sources contain naturally occurring growth-depressing factors such as glucosinolates and isothiocyanates (breakdown products of glucosinolates) in mustard flour (Fenwick et al. 1982) and trypsin inhibitors and hemagglutinins in raw black beans and raw SBM (Rackis and Gumbmann 1981). Some other protein products contained antinutritional factors formed during alkaline and heat treatments such as lysinoalanine and lanthionine (Friedman et al. 1984, Robbins et al. 1980) in the alkaline-treated lactalbumin (containing 4.42 g lysinoalanine/100 g protein; G. Sarwar, unpublished data) or SPI (containing 1.94 g lysinoalanine/100 g protein; G. Sarwar, unpublished data), and Maillard compounds in the overheated skim milk powder.

Lysinoalanine is an unnatural amino acid derivative formed (during alkaline treatment of proteins) mainly by the addition of the -amino group of lysine residue to the double bond of a dehydroalanine residue, that has been generated by the -elimination reaction of cystine, phosphoserine or glycoserine residues (Friedman et al. 1984, Mega 1984). The formation of lysinoalanine adversely affects the nutritional value of proteins because bioavailable lysine, cystine and phosphoserine are lost. Moreover, sufficiently large doses of lysinoalanine, especially free lysinoalanine released from the protein-bound form during the digestion process, can induce a reversible form of renal toxicity known as nephrocytomegaly in rats. It has been proposed that lysinoalanine, which chelates certain metal ions, may exert its toxic effect by metal-binding in renal tubule cells (Hayashi 1982). Since extracts from human kidney were less effective in metabolizing lysinoalanine than corresponding extracts from several animal species, human kidney cells may be more susceptible to lysinoalanine toxicity than those from rats or other animals tested (Kawamura and Hayashi 1987). The formation of lanthionine during alkaline and heat processing of foods also results in significant loss of bioavailable cystine (Robbins et al. 1980).

The Maillard reaction between proteins and reducing sugars is primarily responsible for the loss in nutritional value of food proteins during heat processing (Hurrell 1984), such as overheating of skim milk powder in this study. The overheating of skim milk and the alkaline treatment of lactalbumin and SPI may also have resulted in racemization of amino acid residues (Friedman et al. 1981). Protein-bound D-amino acids formed during processing may have adverse effects on biological value and safety of processed foods (Friedman et al. 1984). Since the routine amino acid methodology used in the determination of PDCAAS does not distinguish between D- and L-forms of amino acids, the scoring method does not take into account the presence and nutritional implications of D-amino acids in processed foods. Special analyses for D-amino acid compositions of processed foods are needed to assess the importance of this problem to human nutrition.

The PDCAAS method also does not take into account bioavailability of individual amino acids, which may be up to 44% lower than the overall digestibility of protein in the same food product (Eggum et al. 1989, Sarwar and Peace 1986). Therefore, the PDCAAS method may give misleading results about the quality of proteins co-limiting in more than one essential amino acid (Sarwar and Peace 1994). In the present study, it was difficult to identify the true first-limiting amino acid in SPI, alkaline-treated SPI, raw SBM, heated SBM, raw black beans, heated black beans and AA-supplemented zein using the PDCAAS method based on human requirements (Table 5). Depending upon the bioavailability of individual amino acids, these protein sources could be first-limiting in methionine + cystine, lysine, threonine or tryptophan which would affect the true score. Moreover, biological experiments would be required to document beneficial effects of supplementation with limiting amino acids. Emmert and Baker (1995) illustrated the benefit of using amino acid analysis and bioavailability data in correctly predicting the protein quality of several processed soybean protein products. In chick assays, SBM was found to be limiting in methionine + cystine, while soybean protein concentrate and SPI were limiting in methionine + cystine as well as threonine (Emmert and Baker 1995). Between the two samples of SPI (functional and edible), the functional SPI was superior to the edible SPI in terms of protein quality.

The amino acid scoring method also overestimated the protein quality of AA-supplemented zein (PDCAAS of 71% vs. RPER and RNPR values of 3-44%) (Table 6). The PDCAAS assumes complete biological efficiency of supplemental amino acids in improving protein quality. This was, however, not true in the case of AA supplementation of zein, a protein of low digestibility and poor quality. A marked difference between the PDCAAS and RPER or RNPR of amino acid-supplemented zein would suggest incomplete biological efficiency of the supplemental amino acids. The poor biological response to amino acid supplemented zein may have been due to the poor bioavailability of essential amino acid(s) other than those supplemented.

Since scores > 100 are considered to be 100, the PDCAAS does not credit the extra nutritional value of a protein having a score higher than that of the reference protein such as egg, fish, milk and most meat protein products. There is a need to revise the calculation of PDCAAS to permit values > 100 for high quality proteins, especially if they are intended to be used as supplements to other low quality proteins. However, this revision may not be required in calculating the PDCAAS for combined proteins, where the supplemental effect of excess amino acids is included in the calculation.

In the amino acid scoring method, methionine and cyst(e)ine concentrations are combined. Due to lack of knowledge of the proportion of the total sulfur amino acid requirement which can be met by cystine, expression of protein values based on the sum total of methionine and cyst(e)ine has limitations (FAO-WHO 1991). Moreover, dietary methionine may be less effective in meeting the cellular requirements for cysteine as is a preformed dietary supply of cyst(e)ine (Baker and Han 1993, Young 1995). This may be especially applicable to protein sources with low cysteine:methionine ratios (0.2) such as casein, where some methionine would be needed for cysteine synthesis which is only about 80% efficient on a weight basis (Baker and Han 1993). Therefore, there may be a need for the inclusion of a desirable cysteine:methionine ratio in the scoring pattern. There is a large variation in cysteine:methionine ratios of dietary protein sources. For example, the cysteine:methionine ratios (molar basis) in beef, egg white, soy protein isolate, rapeseed protein concentrate, wheat flour and pea flour have been reported to be 0.5, 0.9, 1.2, 1.5, 1.6 and 1.8, respectively (Sarwar et al. 1983). Another protein source with one of the highest cysteine:methionine ratios (1.9) is mature human milk (Räihä 1985). The cysteine:methionine ratios (2.32 to 2.38) were especially high in preterm and term transitional human milks (Sarwar et al. 1996).

Since the introduction of the PDCAAS method by FAO/WHO (1991), several questions about the validity of the method have been raised (Food Chemical News 1991, Sarwar and Peace 1994). Some of these concerns were also documented by this investigation. Therefore, there is a need to address these issues, and to suggest proper revisions to the scoring method. Meanwhile, the PDCAAS remains the preferred method for routine prediction of protein quality of properly processed (containing minimal amounts of residual antinutritional factors) and highly digestible (where the overall digestibility of protein is a good approximation of bioavailability of individual amino acids) food products for human consumption.

ACKNOWLEDGMENTS