Main

Both human and animal breast milk contain hormones and growth factors(1–3). The IGF in milk are thought to contribute to the maturation process of the gastrointestinal (GI) tract of the newborn(4). This is indicated by expression of the IGF type 1 receptor and by expression of IGF-II and IGFBP in the epithelia of the GI tract of newborn rodents(5–7). However, the physiologic role of the IGFBP in milk is not fully understood. Adsorption of 125I-labeled IGF-I from milk was detected in the GI tract of newborn animals. As a consequence of IGF-I uptake, increased growth and DNA synthesis and increased intestinal adsorption of nutrients were observed(8).

A striking feature was the higher mitogenic potential of the milk from humans and other mammals derived from the initial phase of lactation compared with that of mature milk(4,9). One mechanism to enhance the mitogenic potency of IGF in biologic fluids includes the cleavage of IGF-IGFBP complexes by specific proteases(10). It increases the availability of the IGF to the IGF-I receptor. IGFBP-specific proteases are found in human plasma, e.g. during pregnancy or severe illness, and produce defined IGFBP fragments(11,12). Recently, defined fragments of IGFBP-2 with a low affinity to IGF, possibly resulting from specific proteolysis, were detected in early human milk(13). However, no corresponding enzyme activity could be measured.

To search for a putative functional role of IGFBP in milk, we studied the composition of breast milk from 99 donors during lactation at regular intervals for a period of 6 mo. In particular, we measured the concentration of IGF-I, IGF-II, and of IGFBP-1, -2, and -3, the most important IGFBP in milk, by means of RIA. In a parallel study, we sought longitudinal evidence for a likely specific proteolysis of IGFBP-2 in milk. For this purpose, we used ligand blots and immunoblots and developed a PA for IGFBP-2. Further aims included the biochemical analysis of the IGFBP-2 protease in detail as well as a comparison of the proteolytic pattern with that of other IGFBP proteases detectable in breast milk.

METHODS

Milk samples. Breast milk samples of 2 mL were donated by 99 healthy lactating women with no clinical or laboratory sign of inflammation of the breast. We chose the following intervals after delivery: 3 d, 1, 2, 4, and 6 wk, 3 and 6 mo. The milk was centrifuged at 8000 × g for 10 min to remove fat and cellular components and was stored at -20°C until further analyses. The study was approved by the local ethics commission.

Quantitative measurement of IGF and IGFBP. IGF-I, IGF-II, and IGFBP-3 were measured by commercial RIA (Mediagnost, Tuebingen, Germany). IGFBP-1 and IGFBP-2(14) were measured by validated in-house RIA, standardized upon preparations of purified human IGFBP.

WLB and WIB. WLB was performed according to the method described by Hossenlopp et al.(15) to achieve a semiquantitative analysis of the IGFBP and their fragments in milk with respect to their ability to bind 125I-IGF-II. IGF-II was used because it is the major IGF in milk and can be bound by all six IGFBP. Milk samples (20 µL) were diluted at a ratio of 1:10 in sample buffer [50 mM Tris/HCl pH 8.5, 35% (vol/vol) glycerol, 1% SDS (wt/vol)] before electrophoresis. An activity level of 2 × 106 cpm 125I-h-IGF-II per milliliter was used as a tracer.

WIB was done with the objective of examining protein bands that are recognizable by a specific rabbit anti-hIGFBP-2 antiserum (diluted 1:5000)(14). For WIB, the blotted membrane used in WLB was washed twice and reprobed with IGFBP-2 antiserum. IGFBP-2 immunoreactive bands were visualized by a chemiluminescence detection system (ECL, Amersham-Pharmacia Biotech, Freiburg, Germany).

IGFBP-2 PA. To assess IGFBP-2 protease activity in milk, we produced 125I-labeled (1000 cpm/µL) IGFBP-2 from purified human IGFBP-2. For IGFBP-3 PA, commercially available 125I-IGFBP-3 (1000 cpm/µL; DSL, Webster, TX) was used. We modified a protocol to measure serum IGFBP-3 protease activity(11): 20 µL 125I-IGFBP-2 or 125I-IGFBP-3 was incubated with 20 µL from the milk sample for 6-12 h at 37°C at the pH of milk. The mixture was diluted at a ratio of 1:2 with sample buffer (Tris/HCl pH 8.5, 1% SDS, 35% glycerol) and then heat-denatured. Following this procedure, 40 µL was then separated by SDS-PAGE (12.5% polyacrylamide, 450 V, 4 h). The gel was dried and autoradiographed. The bands were scanned and quantified by means of a specific computer software package for densitometric analysis (Win-Cam 2.1, Cybertech, Berlin, Germany). The decrease of the 31-kD IGFBP-2 band, in percent of the nonproteolytic control, was used as a measure for fragmentation, i.e. for its proteolytic activity.

Biochemical analysis of the IGFBP-2 protease. The biochemical nature of the IGFBP-2 protease was examined by performing PA in the presence of protease inhibitors and under different pH conditions(10). The following protease inhibitors were used: the serine protease inhibitors PMSF (10 mM), aprotinin (4000 U/mL), and antithrombin-3 (2 IU/mL), and, in addition, EDTA (10 mM) as a nonspecific metalloproteinase inhibitor, E-64 (1.4 µM) as a cysteine protease inhibitor, pepstatin-A (0.15 µM) as an aspartic acid protease inhibitor, and ZnCl2 (20 mM). All inhibitors were purchased from Boehringer, Mannheim, Germany. The amounts of all compounds used conformed with the recommendations of the manufacturer. Alternatively, PA were performed with milk samples (pH 7.2) adjusted stepwise to pH 3-10 by using 0.1 M HCl and 0.1 M NaOH, respectively. The percent fragmentation of IGFBP-2 served to assess the protease activity. Furthermore, IGFBP-2 PA were performed in the presence of compounds, known as IGFBP proteases, detectable in human milk(4,16): PSA (purified from seminal plasma; generously provided by DPC Biermann GmbH, Bad Nauheim, Germany), the 7S (Svedberg) fraction of the γ-NGF(16,17), and a polypeptide of the EGF-R (Chemicon Inc., Temecula, CA)(4).

PSA concentration in milk. The concentration of total, i.e. free and complexed, PSA in milk(16) was measured by using a highly sensitive (detection limit 0.003 ng/mL; range 0.01 to 20 ng/mL; imprecision at 1 ng/mL, 6.8% CV) automated chemiluminescent immunoassay (PSA-Immulite, third generation, DPC, Los Angeles, CA). PSA was found to be highly stable in milk against storage up to 6 wk and against freezing and thawing.

Mathematical analysis. Data were analyzed using standard statistical methods, including a paired t test. Results are reported as mean ± SEM of at least 4 or up to 99 independent parallels, as indicated in the Figure and Table legends. Significance was assigned a value of p < 0.05 for the differences between the times of sample collection.

RESULTS

Longitudinal analysis of IGF, IGFBP, and IGFBP-2 fragmentation in milk. Table 1 shows the mean concentrations of IGF-I, IGF-II, IGFBP-2, and IGFBP-3 in breast milk derived from the 99 subjects studied. The last row indicates the results of IGFBP-2 PA of four representative milk samples. With the exception of IGFBP-3, the concentrations of each factor measured in the first week of lactation showed a slight increase (p < 0.02), most strikingly (25%) IGFBP-2. This was partially attributed to immunoreactive IGFBP-2 fragments, because the measurable IGFBP-2 concentration increased by up to 95% after incubation of early milk at 37°C (Fig. 1). Further evidence of this was found in the immunoreactive IGFBP-2 bands that were visualized by means of immunoblots and ligand blots (Fig. 2). At the end of 6 mo of lactation, the composition of milk had changed markedly. The values for IGF-I decreased to 47%, IGF-II to 44%, IGFBP-2 to 37%, and, in the case of IGFBP-3, the values decreased to as low as 30% of those for early milk. At no stage was IGFBP-1 detectable in breast milk (data not shown). In contrast with human neonatal plasma, IGFBP-2 was the most prominent IGF component in breast milk (Fig. 3). Values as high as 4100 ng/mL (mean 2163 ± 108 ng/mL) on d 3 of lactation were measured (Table 1). The values for IGFBP-3, at 100.0 ± 5.0 ng/mL, were about 1 order of magnitude lower than that of IGFBP-2 and decreased right from the start of lactation. At the same point of time, IGF-II, at 10.7 ± 0.4 ng/mL, was the most predominant IGF. Based on these measurements, therefore, a 50-fold molar excess could be calculated for IGFBP-2, whereas only a 2.5-fold molar excess in the IGFBP-3 concentration was found compared with the measured IGF.

Table 1 Parameters of the IGF system in human breast milk during 6 mo of lactation
Full size table
Figure 1
figure 1

The effect on the recovery (%) of IGFBP-2, measured by RIA after incubation of milk samples at 37°C for 0, 6, and 12 h. Results are presented as mean ± SEM of four early (d 3; striped bars) and four late (6 mo; filled bars) milk samples before and after incubation.

Full size image
Figure 2
figure 2

Analysis of the IGFBP-2 proteolysis in early (d 3) and mature (6 mo) milk samples by three electrophoretic techniques (representative autoradiographs are shown). Samples II were obtained from the same mother. Left, WLB shows IGFBP-2 forms binding 125I-IGF-II. Fifty nanograms of purified human IGFBP-2 (BP-2), which was not incubated with milk, was used as standard. Middle, WIB shows immunoreactive IGFBP-2 fragments, using the same standard (BP-2). Right, PA shows cleavage of introduced 125I-IGFBP-2 by the milk protease at 37°C after 12 h.

Full size image
Figure 3
figure 3

WLB using 125I-IGF-I and 125I-IGF-II to compare the IGFBP pattern of plasma of a newborn child after 1 d and of breast milk from 3 d and 3 mo after beginning of lactation.

Full size image

The fragmentation of the 31-kD IGFBP-2 in the PA (Table 1) revealed higher variance than those observed for the factor concentrations, ranging from 25 to 92% (mean value ± SEM = 72 ± 26%), because of limited random sampling. Despite this, however, we observed a strong slackening of proteolytic activity during lactation, in particular during the first 4 wk. After 6 wk, the IGFBP-2 protease activity evidently ceased completely. No proteolysis of IGFBP-3 was detectable (not shown).

Detection of IGFBP-2-specific protease activity in early milk samples. Figure 2 represents the comparative results of three methods used to examine IGFBP-2 proteolysis in breast milk. Tests of two representative samples for early (3 d) and one for mature (6 mo of lactation) breast milk showed high proteolytic activity that leads to the cleavage of IGFBP-2. This was demonstrated by using the WLB, WIB, and PA. A proportional decrease in the 31-kD band of IGFBP-2 occurs during this process. By means of WIB, IGFBP-2 fragments reactive toward IGFBP-2 antibodies could be visualized in breast milk. We assumed that IGFBP-2 proteolysis would lead to false high IGFBP-2 values (Fig. 1), as the same antibody was used in the RIA (cf. Table 1). The WLB, however, showed that the fragments of the same blot obviously lacked any significant affinity to 125I-labeled IGF-II. An almost identical banding pattern, compared with the WIB, was observed in autoradiographs of the PA (Fig. 2) after the introduction of 125I-labeled IGFBP-2 into the early (d 3) milk samples. This pattern was not evident in the later milk samples. Two doublets of fragments were particularly striking. They had a molecular mass of 14 and 16 kD as well as 23 and 25 kD. Because of its low concentration in breast milk, IGFBP-3 could only be weakly visualized in WLB (Fig. 3). Protein staining after electrophoretic separation of the milk sample from d 3, with high IGFBP-2 protease activity, revealed an unchanged pattern of mass proteins in the milk after incubation (data not shown).

Biochemical examination of the IGFBP-2 protease activity. The use of suitable concentrations of specific protease inhibitors, such as PMSF, aprotinin, and antithrombin (Table 2) during PA resulted in almost total inhibition (up to 96%) of IGFBP-2 fragmentation. Zn2+ ions and heat exceeding 95°C over 10 min also led to protease inhibition by >80%. EDTA, which binds bivalent cations of alkaline earth metals such as Ca2+, had a weak inhibitory effect (10%, p < 0.02) on the fragmentation of IGFBP-2. The cysteine protease inhibitor E-64 and the aspartate protease inhibitor pepstatin had no inhibitory effect. The slight autolysis of IGFBP-2 (Fig. 2, right panel) could neither be halted by the inhibitors nor was pH dependent. It was taken into account during the calculation of the zero-valency. Despite this, however, the tracer was used for a maximum of 28 d after iodination. The natural pH value of the milk samples was 7.2 to 7.3. The highest IGFBP-2 fragmentation, i.e. protease activity, was evident between pH 7 and 8 (Fig. 4). The activity soon reached values in the region of 0 toward higher (up to 10) or lower(3) pH values.

Table 2 Effect of protease inhibitors on the IGFBP-2 protease activity
Full size table
Figure 4
figure 4

Dependence of the IGFBP-2 fragmentation in milk on the pH. The maximum fragmentation of two representative milk samples (3 d postpartum; I, II), adjusted to different pH values, was set to 100%. Fragmentation was measured as the decrease of intact IGFBP-2 by way of densitometric analysis of the autoradiograms (see "Methods").

Full size image

Comparison of fragment patterns by means of known IGFBP proteases. Figure 5 depicts the banding pattern of IGFBP-2 fragments that emerge after the introduction of PSA from seminal plasma, PSA plus milk (4 wk after beginning of lactation), 7S γ-NGF, and EGF-R in PA. Figure 5 illustrates tests in which noncomplexed IGFBP-2 was incubated with the factors listed above. A comparison with the intact tracer (*BP-2) also showed that the incubation of the IGFBP-2 tracer with PSA or EGF-R did not cause any significant fragmentation. As anticipated, neither PSA plus breast milk of 4 wk nor the milk alone led to a significant increase in the fragment bands (Fig. 2). 7S γ-NGF proved to be a potent IGFBP-2 protease and produced a 100% cleaving of 31-kD IGFBP-2. The banding pattern, however, is different from that of the IGFBP-2 protease modified by milk. Discrete bands of approximately 19, 15, 11, and 4 kD were found (Fig. 5).

Figure 5
figure 5

PA after introduction of 125I-IGFBP-2 in the presence of proteases putatively cleaving IGFBP-2. PSA, 7S γ-NGF, and EGF-R are known as IGFBP-specific proteases that are detectable in human milk. Here, a nonincubated tracer (*BP-2) served as control; PSA was incubated solely or mixed with a milk sample (M, 4 wk postpartum). A representative banding pattern of the obtained fragments is shown after a 6-h incubation at 37°C.

Full size image

Comparison of total PSA content and IGFBP-2 proteolysis in milk. In a parallel study, we examined PSA concentrations in breast milk (Fig. 6) by means of an ultrasensitive automated assay, inasmuch as PSA is an IGFBP-3 protease. The comparison of several longitudinal series of milk samples derived from 17 women revealed higher initial concentrations, as seen during IGFBP-2 fragmentation, followed by a sharp decline after the fourth week of lactation (our unpublished data). A strong correlation between PSA concentrations and IGFBP-2 fragmentation (r = 0.84; p < 0.00003) in early (d 3) milk samples from 17 women is shown in Figure 6.

Figure 6
figure 6

Correlation between the PSA concentration (ng/mL) measured by an automated chemiluminescent immunoassay in 17 early (d 3) milk samples and the degree of IGFBP-2 fragmentation (decrease of 31-kD IGFBP-2) measured densitometrically after the IGFBP-2 PA. Each value represents the mean of 2-3 double measurements; the SD was <10%.

Full size image

DISCUSSION

The present investigations into the composition of breast milk in 99 mothers provide evidence that the growth factors contained in breast milk stimulate development of the GI tract in newborns and possibly also that of the maternal breast epithelium(4,5,7,18). It is known that the IGF and their binding proteins control growth and differentiation processes in many tissues(19–21). The ratios of IGF and IGFBP found in milk are completely different, however, from the ratios found in serum. IGF-I and IGF-II concentrations in milk are up to 2 orders of magnitude lower than in serum and the molar ratio of IGF-I plus IGF-II to IGFBP-2 plus IGFBP-3 is approximately 50. In addition, a longitudinal decrease of the factors in the milk during the monitored 6-mo lactation period was observed. The decrease was most prominent during the first 4 wk, leading to the assumption that growth factors in the milk are most likely of high physiologic importance for newborns during the initial phase after birth. IGFBP-2 was the most abundant of all IGF compounds in milk, in contrast with IGFBP-3, which is the most abundant IGFBP in blood.

A second important finding was the existence of an IGFBP-2-specific protease in milk. Although IGFBP-2 fragments with similar molecular mass were recently identified by Ho and Baxter(13) in milk, these authors did not succeed in identifying a specific enzyme. With the PA developed in the present study, we succeeded in showing an enzyme activity cleaving IGFBP-2 for the first time. The protease responsible is a metal2+-ion-dependent serine protease, most likely a kallikrein with a pH optimum of 7-8. The physiologic pH range of milk is also 7.2-7.3. The activity of the IGFBP-2 protease shows a longitudinal decrease similar to the concentrations of IGF and IGFBP. Its presence can only be detected during the first 4 wk of lactation and its concentration decreases dramatically after 2 wk. Surprisingly, however, none of the known IGFBP proteases, the proteins of which are also found in milk(4,10,11,15,17,22), revealed a fragment pattern comparable to that of the IGFBP-2 protease. The same is also true for the serine protease, i.e. PSA, that can cleave IGFBP-3 in seminal plasma(23). Although PSA concentrations in milk vary to a large degree from individual to individual, a clear correlation of PSA with IGFBP-2 protease activity was observed. PSA concentration in milk, therefore, also decreases rapidly during the first few weeks.

IGFBP-3, the only other IGFBP that can be in detected in milk, and other milk proteins were not cleaved. The protease is, therefore, highly specific. As shown in Figures 2 and 3, the IGFBP-2 fragments, which are observed both in the PA and with the use of a specific IGFBP-2 antibody, no longer bind iodinated IGF-I and IGF-II in the ligand blot. Ho and Baxter(13) also found that the IGFBP-2 fragments had an extremely low affinity to the IGF, which is in accordance with the assumption that IGFBP proteases cleave IGF/IGFBP complexes to bring about an increase of the biologic availability of IGF(10–12). Apparently, this leads to a temporary increase in the mitogenic effects of IGF concentrations(9), which are relatively low in human milk compared with blood. A different mechanism that also leads to these effects was found in cow's milk. In cows, the mammary glands produce truncated IGF peptides of high mitogenic activity that only form weak complexes with the IGFBP(6,9).

The significance of IGFBP-2 in milk gives rise to speculations regarding its biologic role and possible IGF-independent effects(19–21). The possible binding of IGFBP-2 via its RGD (ArgGlyAsp) amino acid motif to the integrin receptors of the cells in the GI tract of newborns or the maternal breast epithelium could have an influence on cell adhesion and growth and apoptosis of these cells. Independent effects of IGFBP-2 fragments cannot be ruled out, and PSA could be part of an activating mechanism of the IGFBP-2 protease(24).