Iron absorption in humans as influenced by bovine

546 Am J C/in Nuir 1989;49:546-52. Printed in USA. © 1989 American Society for Clinical Nutrition

milk proteins13

Richard F Hurrell, Sean R Lynch, Trinidad P Trinidad, Sandra A Dassenko, and James D Cook

ABSTRACT The effect ofthe two major bovine milk protein fractions on the dialyzabilityof iron in vitro under simulated gastrointestinal conditions and on the absorption of Fe byhumans was studied. Liquid-formula meals were prepared from hydrolyzed maize starch, corn

oil, and either spray-dried egg white or a milk-protein product. In meals containing egg white,3.32% ofthe Fe was dialyzable. The substitution ofcasein and whey protein products reduced

the dialyzable fraction to 0. 19-0.56% and 0.86-1.60%, respectively. Percentage Fe absorptionwas also reduced by the substitution ofcasein or whey protein for egg white. Mean absorptionvalues fell from 6.67 to 3.65% and 2.53 to 0.98%, respectively. When the intact milk-proteinproducts were replaced by enzyme- or acid-hydrolyzed preparations, the dialyzable fraction

increased markedly and in proportion to the extent ofhydrolysis. A similar but much smallereffect on absorption was observed. These studies suggest that bovine casein and whey proteinsare responsible at least in part for the poor bioavailability of the Fe in some infantformulas. Am J Clin Nuir 1989;49:546-52.

KEY WORDS Milk proteins, casein, whey proteins, iron absorption

Introduction

Foods that are important Sources ofprotein in the hu-man diet may either depress or enhance nonheme ironabsorption. In a standardized liquid-formula (SS) meal

designed to supply the major dietary constituents (pro-tein, carbohydrate, and fat) in a semipurified form, theprotein component (spray-dried egg white) had the great-est effect on Fe absorption in human volunteers (1). Itwas inhibitory. When it was omitted from the meal, ab-sorption increased 2.5 times. Doubling the quantity ofegg white (EW) further reduced absorption by almost75%. The most widely used semipurified vegetable pro-tein, soybean protein, was even more inhibitory. Whenit was substituted for EW in the SS meal, the mean Feabsorption fell from 2.5 to 0.5% (2). On the other hand,animal tissues appeared to enhance nonheme Fe absorp-tion when compared with EW. The substitution of beef,lamb, pork, liver, fish, or chicken for EW in the SS mealresulted in a two- to fourfold increase in percentage Fe

absorption (3, 4).The influence of other protein sources has not been

investigated to the same extent. Only a few studies ofmilk, milk products, or isolated milk proteins were car-ned out in human beings (2, 3, 5, 6), yet bovine caseinand whey proteins provide the protein fraction of mostinfant formulas and are widely used for their nutritionaland functional properties in many other processed foods(7). Their influence on Fe absorption in infants is partic-

ularly important. This is a critical period for Fe nutritionbecause the growing child is dependent on a constantsupply ofavailable dietary Fe.

The few published reports of the effect of milk andmilk products on Fe absorption in man gave conflictingresults. Abernathy et al (5) noted that the removal ofmilk from a mixed diet increased Fe absorption in bal-ance studies in children. On the other hand, Hallberg andRossander (6) reported that the substitution of milk forwater in a hamburger meal did not influence Fe absorp-tion. Cook and Monsen (3) and Cook et al (2) showedthat milk, cheese, or casein inhibited Fe absorption to adegree similar to that of EW. Using their SS meal theyfound that the substitution of milk, cheese, or casein forEW had no effect whereas the substitution of milk orcheese for beefin a typical American meal reduced meanpercentage Fe absorption by 70% and almost 60%, re-spectively.

I From the Division ofHematology, Department ofMedicine, Uni-versity ofKansas Medical Center, Kansas City, KS.

2 Supported by AID cooperative agreement DAN-0227-A-OO-2 104-

00, National Institutes ofHealth grant DK 39246, and Nestle ResearchCenter, Vevey, Switzerland.

3 Address reprint requests to JD Cook, Division of Hematology,University of Kansas Medical Center, 39th and Rainbow Boulevard,KansasCity, KS 66103.

Received December 7, 1987.Accepted for publication March 22, 1988.

at Tsinghua U

niversity on July 10, 2011w

ww

.ajcn.orgD

ownloaded from

MILK PROTEINS AND IRON ABSORPTION 547

TABLE 1Analytical data on test materials

Protein source Supplier Crude protein Iron Calciumt Phosphorust Amino N:total Nt

% mg/100 mg mg/lOOg mg/I 00 g

Intact proteinsEggWhite A 85.3 0.6 140 230 -

Casein (acid precipitated) D 86.7 4.6 70 310 -

Casein (acid precipitated) C 89.5 4.7 - - -

Sodium caseinate(Alenate 1 10) C 90.6 1.8 110 830 -

Whey protein (Ultrafiltered) B 79.0 0.5 780 320 -

Whey protein (Ultrafiltered) C 80.0 5.8 - - -

Whey protein (heat precipitated) C 88.4 5.9 - - -

Whey protein (heat precipitated) D 89.5 1.4 140 880 -

Hydrolyzed proteinsCasein

Alenate 140#{174} C 88.9 6.3 80 800 0.07NZ Amine E� E 78.6 13.6 20 860 0.34NZAmineA E 81.7 3.6 20 1230 0.50NZAm1neHD#{174} E 77.0 5.7 10 1170 0.70HY-CASSFf E 83.9 2.7 290 410 0.74Sigma-enzymes D 83.9 4.3 140 880 0.53Sigma-acidji D 86.7 14.8 330 10 0.76

Whey proteinAletal8l7E C 88.9 5.3 300 300 0.06Lactry#{174} B 79.0 1.5 970 620 0.13Lad#{149} B 79.0 0.9 1400 140 0.29Sigma-enzyme D 8 1.7 3.7 740 320 0.50Edamin-S E 78.4 9.9 20 290 0.56

* Supplier A-Monarch Egg Corporation, Kansas City, MO; B-Linor, Orbe, Switzerland; C-New Zealand Milk Products, Petaluma, CA;

D-Sigma Chemicals, St Louis, MO; E-Sheffield Products, Norwich, NY.

t Analytical data provided by the manufacturer.� Acid hydrolyzed.§ Sigma-enzyme C0626.

0 Sigma-acid C9386.

The present study was designed to examine the effectof casein and whey protein on Fe absorption in the SSmeal. This meal has been studied extensively and was

chosen for its well-standardized characteristics. Both invitro dialyzability using the method of Miller et a! (8)and absorption in humans were studied. Because bothprotein products were inhibitory to Fe absorption, wealso attempted to document a specific role for the pro-teins themselves by examining the extent to which acidhydrolysis or enzyme digestion before inclusion in themeal could overcome the inhibition.

Methods

Protein sources

Spray-dried egg white, three caseins, and four whey proteinswere purchased commercially (Table 1). Two of the caseinswere acid precipitated and the third was a sodium caseinatethat had been rendered water soluble by neutralization (7).Two whey proteins were produced by ultrafIltration and twoby heat precipitation (7). Seven hydrolyzed casein productsand five hydrolyzed whey protein products were also obtainedcommercially. Two ofthe whey protein products had been acid

hydrolyzed and the rest were enzyme-treated preparations. Theenzyme-treated products were all pancreatic digests preparedwith either trypsin or pancreatin.

Crude protein, Fe, calcium, and phosphorus contents of alltest materials and the ratio of a-amino nitrogen to total N ofthe protein hydrolysates are shown in Table 1. Total N wasmeasured by Kjeldahl digestion. Crude protein content (Nx 6.25 for egg white, N X 6.38 for milk proteins) ranged from77 to 91%. However, the protein products were added to thetest meals on an isonitrogenous basis for both the dialysis stud-jes and the Fe-absorption tests. Values for crude protein weretherefore calculated as N X 6.25 in all cases as is the conventionfor nutritional studies (8). Fe content was measured on ashedaliquots by atomic absorption spectroscopy and ranged from0.5 to 15 mg/i#{174} g. Ratios ofa-amino N to total N and Ca andP contents were provided by the suppliers. The approximatepercentage protein hydrolysis, assuming that an amino N:totalN ofO.84 for casein and 0.80 for whey protein is equivalent tocomplete hydrolysis, ranged from 8 to 90% (Fig 1).

Analytical methods

Dialyzable Fe was measured by the method of Miller et al(9) as modified by Hurrell et al (10). Briefly, a simulated gastro-intestinal digestion followed by the measurement ofthe dialyz-able Fe was carried out. A standard formula based on that of

at Tsinghua U

niversity on July 10, 2011w

ww

.ajcn.orgD

ownloaded from

20

10

z0a:

Ui-a

Cl)

a

w

-a

0

0

PROTEIN HYDROLYSISCs)

20 40 60 80 100

PROTEIN HYDROLYSIS

(%)

548 HURRELL ET AL

CASEIN WHEY PROTEIN

FIG I. Relationshipbetween percentagedialyzable iron and extentofhydrolysis ofcascin and whey protein. Closedcircles represent the native protein and enzyme-hydrolyzed products; open circles are acid-hydrolyzed products.

Cook and Monsen (4) was used to prepare the SS test meals.Per 100 g, the meals contained 8.5 g (N X 6.25) crude proteinderived from one of the intact proteins or hydrolyzed proteinproducts, 18 g hydrolyzed maize starch (Fro-Dex 36, Amen-can Maize Products, Hammond, IN), 9.5 g corn oil (NuggetBrand, Stockton, CA), sufficient Fe added as FeCI3 to bring thefinal nonheme Fe concentration to 2.0 mg/ 100 g, and sufficientHO to reduce the pH to 2.0. The final weight was adjusted to100 g with distilled water. Twenty�gram aliquots were digestedfirst with pepsin and then with a pancreatin and bile solution.Dialyzable Fe was measured after the second digestion wascompleted (9, 10).

Iron-absorption studies

Twelve male and three female volunteers ranging in agefrom 21 to 46 y were studied. All were in good health but theydemonstrated a wide range of Fe stores as assessed by serumferritin measurements. Written, informed consent was ob-tamed from each volunteer and all experimental procedureswere approved by the Human Subjects Committee ofthe Uni-versity ofKansas Medical Center.

Each individual participated in one of two separate Fe-ab-sorption studies. The effect ofcasein was examined in the firstand whey proteins in the second. SS meals were fed in all stud-ies. Each meal contained 67 g hydrolyzed maize starch, 35 gcorn oil, 12 mL vanilla extract (McCormick and Co, mc, Balti-more, MD), 200 mL distilled deionized water, and 30 g (Nx 6.25) crude protein. In the first study the protein was derivedfrom EW (meal A), caseinate (Alenate 1 10, meal B), or 84%enzyme hydrolyzed casein (Sheffield Products, meal C). In thesecond study the protein source for meals A-D was EW, wheyprotein (ultrafiltered), 16% enzyme-hydrolyzed whey protein(Lactry#{174}),and 36% enzyme-hydrolyzed whey protein (Lade),respectively. The whey-protein products had been producedfrom sweet whey after rennet precipitation ofthe casein.

Four separate Fe-absorption measurements were performedon each subject with double sequential radio-Fe labels. Testmeals were administered between 0700 and 0900 after an over-night fast and only water was allowed for the subsequent 3 h.All meals were labeled extrinsically by adding 37 MBq 5�FeCI3or 1 1 1 MBq “FeCI3 to a solution of �FeCI3 in I mL 0.01 mol

HO/L containing a quantity of Fe sufficient to adjust totalmeal Fe content to 4. 1 mg. Blood was drawn for serum ferritin(1 1)and background blood radioactivity determinations on thefirst day ofthe study. Each ofthe first pair oftest meals, taggedseparately with either 59Fe or 55Fe, was administered consecu-tively on days 2 and 3 ofthe study. Blood was taken on day 16to measure incorporated red-cell radioactivity. In study 1 athird tagged meal was then fed. In study 2, a second pair ofseparately tagged meals was given consecutively on days 16 and17. The final blood sample was drawn 2 wk after the last testmeal to measure the increase in circulating red-cell radioactiv-ity. All radio-Fe measurements were performed on duplicate1O-mL samples of whole blood with a modification of themethod of Eakins and Brown (12). Percentage absorption wascalculated on the basis ofblood volume estimated from heightand weight (13, 14) and an assumed red-cell incorporation forabsorbed radioactivity of8O% (15).

Statistical analysis

Percentage absorption values were converted to logarithmsfor statistical analysis and results reconverted to antilogarithmsto recover the original units (16). A paired t test was used tocompare the casein and whey-protein containing meals withthe standard EW-containing SS meal and the meals containingenzyme hydrolyzed proteins with the relevant intact-proteinmeal. This was done by determining whether mean log absorp-tion ratios differed significantly from zero, which is equivalentto determining whether the mean ratios of the percentage ab-sorptions differed from one.

Results

Dialyzable iron

Dialyzable Fe measured in the protein-free formulameal is shown in Table 2. All protein sources added tothis formula meal reduced the percentage of dialyzableFe. The most inhibitory proteins were the caseins fol-lowed by whey proteins and EW. There was a direct cor-relation between the extent ofprotein hydrolysis and the

at Tsinghua U

niversity on July 10, 2011w

ww

.ajcn.orgD

ownloaded from

MILK PROTEINS AND IRON ABSORPTION 549

S Hydrolyzed.

t Mean values for age and packed-cell volume are arithmetic; all others are geometric.

TABLE 2Influence ofdifferent protein sources contained in a liquid-formulameal on dialyzable iron in vitro

Dialyzable

Protein sources iron

%

Noprotein 5.32

Eggwhite 3.32Sodiumcaseinate 0.56Casein (acid-precipitated, New Zealand Milk Products) 0.34Casein (acid-precipitated, Sigma Chemicals) 0.19Whey protein (ultrafiltered, New Zealand Milk

Products) 1.60

Whey protein (heat precipitated, Sigma Chemicals) 1.40Whey protein (heat precipitated, New Zealand Milk

Products) 1.00Whey protein (ultraluitered, Linor) 0.86Pooled SEM 0.06

proportion ofdialyzable Fe for both the casein and whey-protein products (Fig 1). The nature of relationship ap-

peared to be different for the two types ofprotein, nonlin-ear for casein but linear for whey. Maximum dialyzable

Fe was 17.2% for 90% acid-hydrolyzed casein and 12.8%

for 70% enzyme-hydrolyzed whey protein (Fig 1).

Absorption studies

Mean Fe absorption by subjects fed the liquid-formulameal containing EW in the first study is shown in Table3. Individual absorption values for the four male andthree female subjects reflected the wide variation in Festatus. Replacing EW with casein reduced mean Fe ab-

sorption to approximately half the value with EW a!-though the difference was not statistically significant (p

TABLE 3

= 0.22). Prior 84% enzyme hydrolysis of the casein re-moved much ofthe inhibitory effect; the difference in Feabsorption between meals prepared with sodium casein-ate and enzyme-hydrolyzed casein was statistically sig-nificant (p = 0.02).

Mean Fe absorption in subjects fed the meal contain-ing EW in the second study is shown in Table 4. Thelower value of 2.53% in this study is a reflection of the

fact that eight male subjects participated, and the meanserum ferritin level for the group was higher than that ofthe subjects in the first study. Replacing EW with wheyprotein significantly reduced Fe absorption to 0.98% (p= 0.04). Prior enzyme hydrolysis ofthe whey protein toa level of 16% did not influence Fe absorption apprecia-bly but prior hydrolysis to the level of 36% increasedmean absorption. Although this increase in absorptionwas not statistically significant (p = 0. 17), the trend wassimilar to that seen with hydrolyzed casein.

Discussion

Our in vitro dialyzable Fe results with casein are in linewith those of Kane and Miller (17) and indicate that thein vitro digestion products of this protein bind Festrongly and prevent its dialysis. Acid-precipitated caseinand sodium caseinate were equally inhibitory. Our stud-ies show that whey protein also reduced the dialyzability

ofFe irrespective ofwhether it was manufactured by ul-trafiltration or by heat precipitation. Casein substitutedin SS meals gave dialyzable Fe values equal to 6-17% ofthe value obtained with the EW formula compared with26-48% for the whey proteins.

The in vitro findings were reflected in the absorptionstudy results although the degree of inhibition was lessand was statistically significant only in the case of whey

Absorption ofiron from a liquid-formula meal containing egg white, casein, or enzyme-hydrolyzed casein as the protein source (study 1)

Subject Sex AgePacked-cell

volumeSerumferritin

Iron absorption

Absorption ratioEgg white

(A)Casein

(B)Casein, 84%

hydr (C) B:A C:A C:B

y % �g/L % of dose

1 M 27 48 95 0.56 1.31 1.70 2.33 3.03 1.29

2 M 23 46 98 2.23 2.13 4.06 0.95 1.82 1.903 M 37 44 144 2.70 0.43 2.95 0.15 1.09 6.864 F 23 40 44 17.25 5.91 7.30 0.34 0.42 1.235 F 23 39 20 19.96 2.37 5.97 0.11 0.29 2.516 M 22 47 18 22.58 27.11 33.56 1.20 1.48 1.237 F 24 42 8 22.35 18.98 52.57 0.84 2.35 2.76

it 26 44 40 6.67 3.65 7.67 0.55 1.15 2.10

-1 SEM 3.81 2.09 4.76 0.36 0.83 1.66+1 SEM 1 1.68 6.38 12.36 0.83 1.60 2.66

at Tsinghua U

niversity on July 10, 2011w

ww

.ajcn.orgD

ownloaded from

550 HURRELL ET AL

TABLE 4Absorption ofiron from a liquid-formula meal containing egg white, whey protein, or enzyme-hydrolyzed whey protein as the protein source

(study 2)

Subject Sex AgePacked-cell

volumeSerumferritin

Iron absorption

Absorption ratioEggwhite

(A)

Whey protein

Intact(B) 16% hydr (C) 36% hydr(D) B:A C:A D:A C:B D:B

y % �g/L % of dose

1 M 21 46 140 1.23 0.08 1.03 0.48 0.07 0.84 0.39 12.87 6.00

2 M 23 46 66 3.88 0.47 0.75 2.47 0.12 0.19 0.64 1.59 5.25

3 M 27 46 173 1.46 0.78 0.85 0.91 0.53 0.58 0.62 1.08 1.164 M 22 44 123 1.57 0.81 0.75 3.90 0.51 0.48 2.48 0.92 4.815 M 46 45 135 0.72 1.02 0.57 0.60 1.42 0.79 0.83 0.55 0.586 M 36 43 136 3.88 1.11 1.00 0.75 0.29 0.26 0.19 0.90 0.677 M 24 44 76 2.30 1.38 0.77 2.23 0.60 0.33 0.97 0.55 1.618 M 25 44 16 23.70 23.77 7.33 17.58 1.00 0.31 0.74 0.30 0.73

it 28 45 90 2.53 0.98 1.06 1.71 0.39 0.42 0.68 1.08 1.74

-1 SEM 1.73 0.57 0.80 1.12 0.27 0.35 0.52 0.72 1.23

+1 SEM 3.69 1.71 1.41 2.62 0.56 0.51 0.88 1.60 2.46

S Hydrolyzed.

t Mean values for age and packed-cell volume are arithmetic; all others are geometric.

protein. The relative effect ofthese milk proteins can be

put into perspective by comparing several different pro-tein sources that have now been tested for their influenceon Fe absorption with the SS meal. In each case EW was

used as a reference meal. Table 5 summarizes these re-suits and relates Fe absorption in the presence of eachprotein source to that with EW (EW = 100). The EWreference meal was identical in all studies except that ofCook and Monsen (3), who used a different source of EWand added Ca and potassium phosphates to the meal.The protein-free meal gave the highest relative absorp-

tion (RA) compared with the EW meal. All proteinsources including beef muscle can be considered inhibi-tory when compared with the protein-free meal althoughthe influence of beef was marginal, which is in keepingwith its documented enhancing effect on Fe absorptionfrom inhibitory meals (18). The degree ofinhibition var-ied greatly with soy protein isolate being the most inhibi-tory.

Additional evidence that specifically implicates theprotein component of milk in the inhibition of Fe ab-

sorption was obtained from our studies ofthe hydrolyzedproducts. When casein and whey proteins were hydro-lyzed before inclusion in the meals, the level of dialyzableFe increased commensurate with the extent of hydroly-sis, eg, dialyzable Fe from the meal containing 90% acid-hydrolyzed casein was �-3O times higher than that pre-pared with intact casein. However, prior hydrolysis of thecasein and whey protein did not increase Fe absorptionin humans as markedly. The 84% enzyme-hydrolyzedcasein, which had shown a 25-fold rise in dialyzable Fe,produced only a two-fold increase in Fe absorption. Sim-ilarly, the 36% enzyme-hydrolyzed whey protein, which

had produced a five-fold increase in dialyzable Fe,showed a 70% improvement in Fe absorption. The poorquantitative agreement between the in vitro dialyzableFe method and Fe absorption in man for test meals con-taming hydrolyzed proteins is another example of thelimitations ofthe dialyzable Fe method. It correctly pre�

dicted the direction of a response but failed to predictits magnitude. Similar results were reported for bovineserum albumin (10).

The mechanism by which proteins influence Fe ab-

sorption remains obscure. Layrisse et at (19) suggestedthat the amino acids in meat and fish may be importantfor solubilizing food Fe and increasing its bioavailability.More recently they proposed a specific role for cysteine-containing peptides (20). Kane and Miller (17) used anin vitro model designed to simulate gastrointestinal con-ditions and demonstrated that beef yielded low-molecu-lar-weight digestion products that enhanced the dialysisof Fe through a membrane with a molecular-weight cut-offof6000-8000 daltons. However, dialyzability was notincreased when soy flour, gelatin, casein, soy-protein iso-late, or gluten were used. They postulated that the influ-ence ofproteins on Fe bioavailability might be related tothe properties of undigested or partially digested pro-teins. Thus, the affinity of these products for Fe and thesize ofthe compounds formed could determine both thedialyzability ofthe Fe and its availability for absorption.

Our in vitro studies suggest that the Fe-binding proper-ties of the pepsin-pancreatin digestion products of bothcasein and whey play a major role in the reduction of Fedialyzability and the inhibition ofFe absorption in man.When intact casein and whey protein were tested in vi-tro, most ofthe Fe did not cross the dialysis membrane,

at Tsinghua U

niversity on July 10, 2011w

ww

.ajcn.orgD

ownloaded from

MILK PROTEINS AND IRON ABSORPTION 551

TABLE 5

Relative iron absorption from a liquid-formula meal containing

different protein sources*

Iron Relative iron Mean serumStudy and protein source absorption absorption ferritin

% �ig/L

Cook and Monsen (3)tEggwhite 1.7 100 22Beefmuscle 5.1 300

Cooketal(2)Eggwhite 2.5 100 53

Casein 2.7 108

Soy protein isolate 0.5 20Hurrell et al (10)

Eggwhite 3.0 100 92

Bovine serum albumin 5.7 190Proteinfree 10.6 353WheatgIuten� 2.1 31 40

Present study

Study IEggwhite 6.7 100 40Casein 3.7 55

Study 2Eggwhite 2.5 100 90Whey protein 1.0 40

* Fe absorption of test protein source relative to absorption from

same meal containing egg white.

t The test meals in this study contained added Ca and potassiumphosphate.

t Unpublished result.

indicating that it was complexed in a form that renderedit insoluble or too large to pass through the pores in themembrane. With extensive acid hydrolysis or enzymepredigestion, the dialyzable fraction increased markedly,suggesting that the protein itself had played a major rolein preventing dialysis of the Fe. There are at least twoplausible explanations for this observation. Pepsin-pan-

creatin digestion of the proteins may have yielded onlycomparatively large products that retained their affinityfor Fe. Alternatively, prior hydrolysis might have led tothe release of small peptides or amino acids that boundFe preferentially in the in vitro system and allowed it tocross the membrane. Unfortunately, our studies do notallow any conclusions to be drawn about either of thesepossible mechanisms.

The finding that predigestion of casein and whey be-fore feeding did not increase Fe absorption as markedlyas it did dialyzable Fe is perhaps not totally unexpected.Cysteine is the only free amino acid that has been shownto increase Fe absorption from an inhibitory meal (21);cysteine-containing peptides have a similar effect (20).Neither casein nor whey protein has a high level of cyste-

me in its protein structure (2 1). Casein is a mixture ofphosphoproteins ofwhich the principal components area-, j3-, and K-casein. All these proteins are devoid of cyste-inc although K-casein has one disuiphide bridge per mol-ecule (22). P is present mainly as serine phosphate and

casein is known to form large phosphopeptides on in vi-tro digestion with pepsin and trypsin and on in vivo di-gestion in the rat (23). These phosphopeptides were re-ported to have strong Ca-binding properties (24) andcould possibly also bind Fe in the duodenum and upperjejunum, maintaining it in a soluble form that mightnevertheless not be available to the absorptive mecha-nisms.

Whey protein is also a mixture ofproteins. Two majorproteins, �9-lactog1obu1in and a-lactalbumin, account for80% ofthe total (22). Serum albumin, immunoglobulins,proteose-peptones, lactoferrin, and various enzymesmake up most ofthe remainder. fl-Lactoglobulin has twodisuiphide bridges and one cysteine residue per moleculeand a-lactalbumin has four disuiphide bridges but nocysteine residues per molecule (22). None of the wheyproteins are phosphoproteins although it is possible thatcommercial whey protein preparations contain some ca-scm as a contaminant. It is not clear which component ofcommercial whey protein concentrate has the inhibitoryeffect on Fe absorption. It is not likely to be lactoferrinbecause lactoferrin rapidly loses its Fe-binding propertieswhen subjected to heat treatments such as those that areused to sterilize commercial products (25).

Although our studies indicate that the protein compo-nents ofcasein and whey protein played a significant rolein limiting Fe absorption, an additional effect ofCa andP (26) cannot be excluded. The intact and hydrolyzeddairy proteins used in our study contained variable levelsof Ca and P (Table 1) reflecting different methods ofpreparation. The caseins and hydrolyzed caseins in gen-eral contained more P than Ca whereas the whey prod-ucts were usually higher in Ca. The highest levels of cal-cium (780-1400 mg/100 g) were found in the whey pro-tein and hydrolyzed wheys fed in the human study.

Finally, there is some practical importance to the dem-onstration that casein and whey protein are inhibitory toFe absorption in adult human beings. Although directcomparisons ofabsorption in adults and infants have notbeen made, the process appears to depend on the samephysiological factors at all ages. In a retrospective corn-

parison, calculated radio-Fe absorption from a corn-soya-milk infant food supplement was very similar inadults and children aged 7-21 mo if a correction wasmade for Fe status(27). The effect ofthese proteins there-fore provides a partial explanation for the lower bioavail-ability of Fe from milk-based infant formula than fromhuman milk (28, 29). Undoubtedly other factors, such asCa and P concentrations and the presence of appreciable

quantities oflactoferrin in human milk (30), are also im-portant. Unfortunately, partial hydrolysis does little toremove their inhibitory influence. More complete pro-tein hydrolysis might improve Fe bioavailability but theproducts would become organoleptically unaccept-able.

References

1. Monsen ER, Cook JD. Food iron absorption in human subjects. VEffects of the major dietary constituents of a semisynthetic meal.Am J Clin Nutr 1979; 32:804-8.

at Tsinghua U

niversity on July 10, 2011w

ww

.ajcn.orgD

ownloaded from

552 HURRELL ET AL

2. Cook JD, Morck TA, Lynch SR. The inhibitory effect ofsoy prod-

uctson nonheme iron absorption in man. Am J Clin Nutr 198 1;34:2622-9.

3. Cook JD, Monsen ER. Food iron absorption in human subjects.III Comparison of the effect of animal proteins on nonheme ironabsorption. Am J Clin Nutr 1976;29:859-67.

4. Cook JD, Monsen ER. Food iron absorption. I Use ofa semisyn-thetic diet to study absorption of nonheme iron. Am J Clin Nutr1975;28: 1289-95.

5. Abernathy RP, Miller J, Wentworth J, Spiers M. Metabolic pat-terns in preadolescent children. XII Effect ofamount and sourceofdietary protein on absorption ofiron. J Nutr 1965;85:265-70.

6. Hallberg L, Rossander L. Effect ofdifferent drinks on the absorp-

tion of non-heme iron from composite meals. Hum Nutr AppiNutr 1982;36:1 16-23.

7. Schwartz M, Wong C. Milk proteins: nutritional and functionaluses. Cereal FoodsWorld 1985;30:173-5.

8. Williams S, ed. Official methods of analysis of the Association ofOfficial Analytical Chemists. 14th ed. Arlington, VA: AOAC,1984.

9. Miller DD, Schricker BR, Rasmussen RR, Van Campen D. An invitro method for estimation ofiron availability from meals. Am JGin Nutr 198 1;34:2248-56.

10. Hurrell RE, Lynch SR. Trinidad TP, Dassenko SA, Cook JD. In-fluence of endogenous animal protein on Iron absorption in hu-mans: bovine serum albumin compared with beefmuscle and eggwhite. Am J Gin Nutr 1988;47: 102-7.

1 1 . Miles LEM, Lipschitz DA, Bieber CP, Cook JD. Measurement ofserum ferritin by a 2-site immunoradiometric assay. Anal Biochem

1974;61:209-24.

12. Bothwell TH, Chariton RW, CookJD, Finch CA. Iron metabolismin man. Oxford: Blackwell Scientific, 1979.

13. Wennesland R, Brown E, Hopper J Jr, et al. Red cell, plasma andblood volume in healthy men measured by radiochromium (Cr�1)cell tagging and hematocrit: influence ofage, somatotype and hab-

its ofphysical activity on the variance after regression of volumesto height and weight combined. J Gin Invest 1959;38:1065-77.

14. Brown E, Hopper J Jr, Hodges JL Jr, et al. Red cell, plasma, andblood volume in healthy women measuredby radiochromium cell-

labelingand hematocrit. J Clin Invest 1962;41:2182-90.

15. Hosain F, Marsaglia 0, Finch CA. BlOOd ferrokinetics in normalman. J Clin Invest 1967;46:1-9.

16. CookJD, Layrisse M, Finch CA. The measurement ofiron absorp-tion. Blood 1969;33:421-9.

17. Kane AP, Miller DD. In vitro estimation ofthe effects of selectedproteinson iron bioavailability. Am J Clin Nutr 1984; 39:393-401.

18. Bjorn-Rasmussen E, Hallberg L. Effect ofanimal proteins on theabsorption offood iron in man. Nutr Metab 1979;23:192-202.

19. Layrisse M, Martinez-Torres C, Roche M. Effect ofinteraction ofvarious foods on iron absorption. Am J Clin Nutr 1968;21:1 175-

83.20. Taylor PG, Martinez-Torres C, Romano EL, Layrisse M. The

effect of cysteine-containing peptides released during meat diges-tion on iron absorption in humans. Am J Clin Nutr 1986;43:68-71.

21. Martinez-Torres C, Layrisse M. Effect ofamino acids on iron ab-sorption from a stable vegetable food. Blood 1970;35:669-82.

22. Brunner JR. Milk proteins. In: Whittaker JR. Tannenbaum SSR,eds. Food proteins. Westport, CT: AVI Publishing Co mc, 1977:175-208.

23. Naito H, Susuki H. Further evidence for the formation in vivo ofphosphopeptide in the intestinal lumen from dietary /3-casein. AgrBiol Chem 1974;38: 1543-5.

24. Naito H, Kawakami A, Imamura T. In vivo formation of phospho-peptide with calcium-binding property in the small intestinal tractofthe rat fed on casein. Agr Biol Chem 1972;36:409-15.

25. Ford JE, Law BA, Marshall VME, Reiter B. Influence ofthe heat

treatment ofhuman milk on some ofits protective constituents. JPediatr 1977;90:29-35.

26. Monsen ER, Cook JD. Food iron absorption in human subjects.Iv The effects ofcalcium and phosphate salts on the absorption ofnonheme iron. Am J Gin Nutr 1976;29:1 142-8.

27. Morck TA, Lynch SR, Skikne BS, Cook JD. Iron availability frominfant food supplements. Am J Clin Nutr 198 1 ;34:2630-4.

28. McMillan JA, Oski FA, Lourie G, Tomarelli RM, Landaw SA.Iron absorption from human milk, simulated human milk, andproprietary formulas. Pediatrics 1977;60:896-900.

29. Saarinen UM, Siimes MA. Iron absorption from breast milk, cow’smilk and iron-supplemented formula: an opportunistic use ofchanges in total body iron determined by hemoglobin, ferritin, andbody weight in 132 infants. Pediatr Res 1979; 13:143-7.

30. Lonnerdal B. Iron and breast milk. In: Stekel A, ed. Iron nutritionin infancy and childhood. New York: Raven Press, 1984:95-118.

at Tsinghua U

niversity on July 10, 2011w

ww

.ajcn.orgD

ownloaded from