Main

The nutritional and immunologic properties of human milk are unsurpassed. The protective properties of human milk against pathogenic bacteria in the intestinal tract are widely known [for a summary see Goldblum and Goldman(1)]. Breast-feeding has been shown to prevent about 70-80% of the diarrhea in young infants in a poor country (F. Jalil, A. Mahmad, R. N. Ashraf, S. Zaman, S. R. Kahn, L. Å. Hanson, J. Karlberg, and Lindblad B. S., unpublished data), as well as drastically reduce the risk of neonatal septicemia(2). The factor in milk accountable for most of these effects is secretory IgA(3–6), which comprises about 90% of the total IgA present in milk. Secretory IgA constitutes about 70-80% of the total milk Igs(7). IgG and IgM are also present in human milk, although in lower concentrations, 0.045 and 0.14 g/L, respectively, in mature milk compared with 0.3-1 g/L of secretory IgA(8). Little is known, however, regarding the activities of these two additional isotypes in milk. It has been reported that anti-HIV-1 IgG and IgM in milk may protect against postnatal infections(9). Compensatory increases of IgM and IgG have been found in milk of IgA deficient mothers (M. Hahn-Zoric, B. Carlsson, J. Bjökander, L. Mellander, and L. Å. Hanson, unpublished data).

Undernutrition is well known to impair the immune system in various ways(10). Human milk is an important source of immune factors such as IgA antibodies for the infant(11). Several studies have suggested that mild to moderate undernutrition in lactating mothers is not associated with low quality milk(12, 13), although lower milk output has been reported in undernourished Zairian women(14). In another study, a lower concentration of total secretory IgA was found in the milk of moderately undernourished mothers receiving a low calorie supplement compared with those receiving a high calorie supplement(15). Possible differences in the antibody specificities present in milk from undernourished mothers may not be revealed by antibody analysis using complex bacterial antigens in ELISA(15). Qualitative and semiquantitative analysis of antigen preparations from enteric bacteria by SDS-PAGE may increase the sensitivity concerning variations in the milk antibody repertoire.

The Gram-negative bacterial outer membrane is composed of a bilayer, of which the outer layer mainly consists of LPS, lipoprotein, and protein. The LPS contains the repeating oligosaccharide units, which constitute the O-specific antigens, the dominating somatic antigen of Gram-negative bacteria(16, 17). The outer membrane contains two types of protein molecules, a restricted amount(3–8) in high copy numbers (major Omps), which make up most of the proteins in the bacterium and 50-100 other proteins present in lower amounts (minor Omps)(17). Immunoblotting analyses of bacterial lysates by SDS-PAGE have shown that the LPS (ladder-like appearance) and major as well as minor Omps of the bacterial cell wall may induce antibody production(18, 19).

As a part of our studies in moderately malnourished mothers, we were interested in determining the immunostaining profile of IgA, IgG, and IgM antibodies in milk using antigen preparations from Escherichia coli O6, a common inhabitant of the gut, and Shigella flexneri, an enteric pathogen common in Guatemala. The significance of caloric supplementation on the antibody levels toward certain epitopes was analyzed.

METHODS

Study design and subjects. The original protocol was designed to study the effect of supplementation on milk production (T. González-Cossío, J. P. Habicht, H. Delgado, and I. L. Rasmussen, unpublished data). The women included in the study were mainly of Mayan origin, living in the poorest sections of Quetzaltenango, in the western Guatemalan highlands. Using calf circumference as the indicator of undernutrition(20, 21), women with a low calf circumference during the last trimester of pregnancy were invited to the study, and recruited after informed consent. The women were divided randomly in a double-blind design into either the low or the high calorie intake groups. Women in both groups had similar characteristics (i.e. anthropometric measures, age, breast feeding practices), and no statistical or biologic differences were found between the groups. The women were on average poorly nourished (mild to moderate malnutrition) as reflected by their height, small limb circumferences, and low weight(21).

Supplement. The supplement consisted of high calorie cookies (250 kcal each, two per day) or low calorie cookies (70 kcal each, two per day) given during a period of 20 wk postpartum. The supplement was given in addition to the women's normal diet. Their food consumption, as well as supplement intake, was evaluated daily (except Sundays) by field workers. The cookies'composition was based on maize, wheat, and soy flour, sweetened with sugar and flavored with chocolate, vanilla, or licorice. The caloric difference was achieved by controlling the amount of vegetable lard, sugar, and sesame seeds. Ten to twelve percent of the energy content in both cookies was derived from protein, in agreement with protein recommendations for adults(21).

Milk samples. We analyzed two samples from each mother, one collected before the start of the supplementation (5 wk after parturition, d 0) and the other 20 wk later (end of the supplementation period). Milk from healthy Swedish mothers (n = 20) was obtained and used as control for the SDS-PAGE gels. Out of 102 mothers who were included in the study (49 in the high calorie and 53 in the low calorie groups), we analyzed the milk of 21 mothers (20 paired samples) in the high calorie supplemented group and 24 (22 paired samples) in the low calorie group.

Bacterial antigens. Bacterial cells of E. coli O6:K13:H1 (WHO designation Su4344/41) and S. flexneri type 6 (Culture Collection, University of Göteborg No. 9106) were prepared. Briefly, bacteria were grown on BHI plates over night at 37 °C and harvested the next day. After washing (Shigella were first killed by heating at 63 °C for 50 min), the pellet was suspended in distilled water, diluted to an amount corresponding to approximately 1 × 109 bacteria/mL, and stored frozen in aliquots.

E. coli O6 LPS was obtained by the hot phenol-water method(22). All samples were diluted 1:4 in sample buffer when used in the SDS-PAGE. The amount of each bacterial antigen was calculated to correspond to approximate 2.5 × 108 bacteria/mL for the whole cell preparations and 400 μg/mL of LPS for the O6 LPS preparation.

SDS-PAGE. Whole bacterial cell preparations or LPS were pretreated and separated according to Laemmli(23). Electrophoresis was performed at constant 200 V for 42-45 min on a Miniprotean II vertical electrophoretic cell (Bio-Rad Laboratories, Richmond, CA), with discontinuous stacking and separating gels containing 4 and 12% acrylamide, respectively. A low molecular weight standard (Bio-Rad) together with milk obtained from healthy Swedish mothers was included in each run as controls. Samples from both groups and each time point were included in each run.

The transfer of antigens was performed with a Mini Transblot transfer cell using a nitrocellulose membrane at 100 V, for 70-80 min (Bio-Rad). After the transfer, the nitrocellulose membrane was submerged in blocking solution (3% gelatin in Tris-buffered saline, TBS). The membrane was washed twice in TBS with 0.05% Tween 20 (TTBS) and placed in a Mini-protean II Multiscreen apparatus (Bio-Rad). Milk samples, diluted 1:20, were added to the wells (600 μL). After overnight incubation and washing, the membrane was cut into three pieces, each containing the same series of milk samples. The nitrocellulose pieces were incubated in either of the affinity-purified rabbit anti-human IgG, IgM, or IgA antibodies (Jackson ImmunoResearch Labs. Inc., West Grove, PA) diluted 1/1000 in TTBS. After 60-min incubation on a shaker followed by washing, a second antibody (biotinylated goat anti-rabbit IgG diluted 1/6000, Bio-Rad) was added. After another 60 min and washing, streptavidin-biotinylated alkaline phosphatase (diluted 1/3000, Bio-Rad) was added to the membranes and incubated as above. After washing, alkaline phosphatase substrate 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium (Bio-Rad) were used for staining according to the manufacturer's manual.

The antigen bands used for comparison were chosen by their molecular distribution and the ease by which they were identified. The color intensity of the bands was graded subjectively using a scale from 0 to 3, where 0 described no immunoreactivity and 3, the highest intensity. The results were based on the values assigned to each band. The bands studied for each antigen were the same for the three isotypes (IgA, IgM, and IgG). Nine bands were selected for the E. coli bacterial cell preparation and LPS and seven bands for the S. flexneri bacterial preparation (see Figs. 1,3,and 5). For each Ig class, some milk samples or pairs were excluded from the band pattern analysis due to a high background staining.

Figure 1
figure 1

The immunostaining pattern of E. coli O6 LPS. The molecular masses of the standard bands are shown to the left. The bands chosen for comparison between samples are shown next to the standard. The approximate molecular mass of bands 1, 2, 3, 4, 5, 6, 7, 8, and 9 is 108, 101, 79, 74, 63, 39, 31, 14, and 13.5 kD, respectively. The IgA, IgM, and IgG antibody patterns are shown for milk samples from two mothers (nos. 37 and 39) at the onset of supplementation (a) and at wk 20 (b). Sample nos. 315 and 321 represent control milk samples from two Swedish mothers.

Full size image
Figure 3
figure 3

Immunostaining pattern of E. coli O6 whole cell preparation. The molecular masses of the standard bands are shown to the left. The bands chosen for comparison between samples are shown next to the standard. The approximate molecular mass of bands no. 1, 2, 3, 4, 5, 6, 7, 8 and 9 is 109, 101, 84, 68, 63, 50, 36, 17 and 14 kD, respectively. The IgA, IgG and IgM antibody patterns are shown for milk samples from two mothers (38 and 39) at the onset of supplementation (a) and at wk 20 (b). Numbers 305 and 311 represent control milk samples from two Swedish mothers.

Full size image
Figure 5
figure 5

Immunostaining pattern of S. flexneri whole cell preparation. The molecular masses of the standards are shown to the left. The bands chosen for comparison between samples are shown next to the standard. The approximate molecular mass of bands no. 1, 2, 3, 4, 5, 6, and 7 is 109, 81.4, 60.3, 42.8, 33.3, 15, and 13.7, respectively. The IgA, IgG, and IgM antibody patterns are shown for milk samples from two mothers (nos. 6 and 7) at the onset of supplementation (a) and at wk 20 (b). Numbers 280 and 284 represent control milk samples from two Swedish mothers.

Full size image

Statistical analysis. The difference in the intensity of the bands between the two groups was analyzed by the Mann-Whitney U test. The Wilcoxon test for paired differences was used for analyzing the paired samples from d 0 and wk 20 within each group.

RESULTS

Effects of high calorie versus low calorie supplementation. The immunostaining analysis of the milk samples revealed no statistically significant differences in the antibody IgA, IgG, or IgM pattern between the two groups for any of the three antigens studied at the onset of supplementation (d 0).

E. coli O6 LPS. The typical ladder pattern of LPS can be observed in Figure 1. Antibodies of all three isotypes were observed, although IgA antibodies by far gave the strongest reactions (Fig. 1). IgM or IgG antibodies were not seen in all milk samples. IgM antibodies were, however, more frequent than IgG. About 85% of the samples expressed IgM antibody activity, whereas only 43% expressed IgG. Comparing the two groups after the 20-wk treatment, a significant difference was observed only for one band in the low molecular range (14 kD) concerning IgA (Fig. 2). No differences were observed regarding IgM or IgG antibodies.

Figure 2
figure 2

Immunostaining intensity of milk IgA, IgM, and IgG antibodies against the nine bands of E. coli O6 LPS as separated by SDS-PAGE and transferred by electroblotting to a nitrocellulose membrane. The bands chosen for comparison are presented below the bars showing their approximate molecular mass. Bands 1, 2, 3, 4, 5, 6, 7, 8, and 9 correspond to 108, 101, 79, 74, 63, 39, 31, 14, and 13.5 kD, respectively. Hatched bars represent the high calorie-supplemented group and filled bars the low calorie supplemented group. The immunostaining results are compared at wk 20 (end of the supplementation period). SEM of the means are indicated. Statistical analysis was performed by Mann-Whitney U test (*p < 0.05).

Full size image

E. coli O6 whole cell preparation. The band pattern of the E. coli lysate was different from that of LPS (Fig. 3). Antibodies of all three Ig isotypes were observed also for this complex antigen preparation. A weaker immunostaining was seen for the low calorie intake group compared with the high calorie one, especially regarding the IgM isotype (Fig. 4). Statistically significant differences were obtained with IgM antibodies directed against four high molecular mass bands (109, 101, 68, and 63 kD; Fig. 4). The IgG antibodies were comparatively higher against this whole cell preparation than to the extracted LPS antigen (Figs. 2 and 4). IgA antibodies seemed also slightly higher in staining intensity for the whole cell preparation, whereas IgM antibodies were similar for both antigen preparations (Figs. 2 and 4).

Figure 4
figure 4

Immunostaining intensity of milk IgA, IgM, and IgG antibodies against the nine bands of E. coli O6 whole cell preparation separated by SDS-PAGE and transferred by electroblotting to a nitrocellulose membrane. The bands chosen for comparison are presented below the bars showing their approximate molecular mass. Bands no. 1, 2, 3, 4, 5, 6, 7, 8, and 9 correspond to 109, 101, 84, 68, 63, 50, 36, 17, and 14 kD, respectively. Hatched bars represent the high calorie supplemented group; filled bars represent the low calorie-supplemeted group. The immunostaining results are compared at wk 20 (end of the supplementation period). SEM of the means are indicated. Statistical analysis performed by Mann-Whitney U test (*p < 0.05; **p < 0.01).

Full size image

S. flexneri whole cell preparation. A typical band pattern of S. flexneri is shown in Fig. 5. In total, seven bands were studied. No differences were obtained between the groups for any of the isotypes concerning this antigen (not shown). As observed for the E. coli lysate preparation, the IgG antibody activity seemed to be higher than that of IgM. The immunostaining pattern was similar to that of IgA (not shown).

Effects of the supplementation within each group. When analyzing the paired samples collected at the two time points (d 0 and wk 20), the high calorie intake group showed similar immunostaining intensity for all three isotypes in all the bands for each antigen (Fig. 6; only IgM antibodies to O6 LPS are shown). The low calorie intake group presented more variations, with an overall lower intensity of the bands at wk 20 concerning IgM antibodies to O6 LPS (Fig. 6). A statistically significant difference was seen regarding the low molecular mass bands no. 7 and 8 of E. coli O6 LPS (31 and 14 kD, respectively). No differences were observed in any of the other Igs or antigens tested.

Figure 6
figure 6

Immunostaining intensity of milk IgM antibodies against nine of the bands of E. coli O6 LPS separated by SDS-PAGE and transferred by electroblotting to a nitrocellulose membrane. The bands chosen for comparison are presented below the bars showing their approximate molecular mass (kD). The first panel shows the antibody reactivity of the treatment group (high calorie intake) at d 0 (onset of the supplementation, hatched bars) and wk 20 (end of the supplementation, filled bars). The second panel shows the control group (low calorie intake) at d 0 and wk 20. SEM of the means are indicated. Statistical analysis was performed using the Wilcoxon test for paired differences, where **p < 0.01.

Full size image

DISCUSSION

In human field studies it is very difficult to establish a control group in which all the conditions are controlled, especially when nutritional aspects are considered. For that reason the low calorie intake group was created. In this group the low calorie supplement was given more as a placebo, because it was significantly lower than that given to the treatment group. Community workers who supervised the study confirmed that the difference in the diet between the two groups consisted only in the calorie intake. The net difference intake in the high calorie supplemented group was calculated to be 269 kcal/d, which represented a 12% increase of the normal calorie intake for the women in that community (T. González-Cossío, J. P. Habicht, H. Delgado, and I. L. Rasmussen, unpublished data). At the onset of the supplementation period, no statistically significant differences were observed in the antibody pattern between the two groups regarding the different antigen preparations. Thus, both groups were considered to be equal concerning their antibody levels at the start of the calorie supplementation treatment.

As expected, milk IgA antibodies were more abundant and diverse, as judged by comparing the intensity, frequency, and distribution of the bands for each Ig. In accordance with our earlier report(15), IgA-specific antibodies seemed not to be negatively affected by a low calorie intake in relation to a high calorie intake for those mild to moderate undernourished mothers. Even a significant increase toward one low molecular mass band of E. coli O6 LPS was observed in the low calorie-supplemented group (Fig. 2). Previous comparison of milk IgA antibody titers against E. coli O6 and a pool of 10 common E. coli O antigens between the low calorie- and high calorie-treated groups revealed no differences(15). A study involving undernourished women in Zaire showed no decreases in IgA antibody titers to several bacterial and viral antigens when compared with normal Zairian women, although milk output was lower, resulting in correspondingly less Ig ingestion by the children(14).

According to the IgA subclass distribution in colostrum, IgA1 has been found directed preferentially against protein antigens and IgA2 mainly directed against LPS, with a slightly higher proportion of IgA1 compared with IgA2(24). In our study, we observed higher immunostaining intensity of IgA antibodies against the whole cell preparation in comparison with the IgA antibodies directed to O6 LPS, which seems to indirectly correlate with the study mentioned above.

In our present study, both IgG and IgM antibodies were present in milk, although only IgM antibodies to certain antigens appeared to be affected in the low supplemented group. Our immunostaining results suggest that a part of the IgM anti-E. coli O6 antibody production may be affected by a low calorie intake. This was seen when comparing the staining intensity of the bands between the two groups at wk 20 (Fig. 4), and between paired samples of the two sampling periods in the low calorie intake group (Fig. 6). Because IgM antibodies are associated with a primary response and have a short half-life of around 5 d(25), the low calorie intake may have influenced newly synthesized IgM antibodies directed against E. coli O6. The contribution of milk IgM antibodies to the immunity of infants is not well known. Yet, milk IgM anti-HIV-1 has been suggested to protect against postnatal transmission of the HIV-1 virus(9). They could as well play a role in protection against bacterial pathogens. This shows that it is important to separate the antigens in a complex antigen mixture and to study all isotypes to reveal differences. Although the IgM antibody intensity in the immunoblots was similar against both LPS and whole cell preparations, IgG antibodies had a much stronger reaction with the whole bacterial cell preparation compared with the O6 LPS. In one study, especially IgG3 and IgG4 antibodies against E. coli were demonstrated in human milk (I. Mattsby-Baltzer, personal communication). It has been reported in a study on serum Igs, that IgG1 is mainly produced against protein antigens, whereas IgG2 is predominant regarding carbohydrate antigens(26, 27). The IgG subclass distribution in milk may be different, however, from that seen in serum. In a study of human colostrum and milk, the percentage distribution of total IgG1 was significantly higher than IgG2, whereas IgG3 and IgG4 were similar compared with sera from normal healthy adults and mothers. In addition, the concentration of the IgG subclasses was lower in mature milk than in colostrum, but the distribution of each subclass was similar(28). IgG4 has been reported to be locally produced and present in higher concentrations in human colostrum than in serum(29), although the role of IgG in milk remains unknown.

Four bands out of the nine studied in the E. coli cell preparation (corresponding to the molecular masses of 63, 68, 101, and 109 kD) were significantly lower in the immunostaining intensity for IgM in the low calorie intake group (Fig. 4). The 63-kD band is close to Btub (60 kD), which regulates the uptake of vitamin B12(16). Minor Omps in the molecular range of 74 to 83 kD have been reported as receptors for ferric siderophores, essential for bacterial iron uptake. A 68-kD molecule has also been described in E. coli O18, which probably has the same function(30), the molecular mass of this protein could probably correspond to our 68-kD band studied. The importance of antibodies directed to these regions in bacteria is yet inconclusive. In one report, passive immunization with anti-iron-regulated Omps protected turkeys from E. coli septicemia(31). It cannot be excluded, however, that the bands could be related to internal bacterial structures.

Only one of our selected bands (band no. 7, 36 kD) seemed to coincide with the major Omps described for E. coli. Presumably it corresponds to OmpC (36 kD)(16), although the immunostaining intensity for this band was similar in the two groups. Antibodies directed against OmpC(16) have been shown to induce protective immunity in diverse Gram-negative infections(18). In another study in mice, partial cross-reactive protection against two Salmonella species was obtained after immunization with Omps(32).

Two out of the nine bands studied for E. coli O6 LPS were significantly reduced in the low calorie group after 20 wk (Fig. 6). These bands were in the low molecular mass region (31 and 14 kD). The low molecular mass bands, no. 8 and 9 of the O6 LPS (14 and 13.5 kD) could probably correspond to the core LPS containing one or no oligosaccharide units, respectively. In some instances a double band was found, mainly in the low molecular area. These double bands could be explained by a different antibody specificity in relation to core heterogeneity, as reported by Palva and Mäkelä(33). The immunoblotting patterns of O6 LPS are similar to those reported for Salmonella typhimurium LT2 and E. coli O111 LPS, in which two major distributions are obtained, one corresponding to long and the other one to short LPS side chains(34). The ladder-like bands obtained for O6 LPS correspond to the different sizes of the LPS molecules with varying numbers of repeating oligosaccharide units, composing the side chains of the LPS molecule(33).

No differences were obtained for any of the bands against S. flexneri antigen, which is a true intestinal pathogen. One explanation for this may be that there was no exposure to this bacterium during the study period.

The women in this study were mild to moderate undernourished as judged by anthropometrical features(21). The leaner women in the high calorie group had significantly larger milk outputs than the leaner women in the low calorie group, whereas the less lean women in the high calorie group experienced no benefit. In consequence, no differences in the milk output were observed comparing all the women in both groups(21, 35). In addition, the total kilocalories ingested by the children in the high calorie group were slightly higher during wk 10-12, when is the period of peak milk production (T. González-Cossío, J. P. Habicht, H. Delgado, and I. L. Rasmussen, unpublished data)(35). González-Cossío, et al. also found that a higher proportion of the children in the high calorie treated group were exclusively breast-fed during this period (T. González-Cossío, J. P. Habicht, H. Delgado, and I. L. Rasmussen, unpublished data)(35). A possible increased milk consumption by the child would also mean an increased intake of antibodies. Thus, caloric supplementation in moderately undernourished mothers may only have consequences for subgroups regarding milk output. A greater positive effect of caloric supplementation could be expected in women with a more severe undernutrition.

In summary, IgA antibodies to E. coli and S. flexneri were not affected by a low calorie intake. IgM and IgG antibodies were also present in milk. IgG antibodies seemed to be mainly directed against E. coli protein antigens rather than against the LPS. A decrease in the antibody activity against certain E. coli epitopes was demonstrated only for IgM when comparing low and high calorie intake. The significance of a decreased IgM antibody activity against E. coli needs to be further investigated. Our study supports previous findings reporting that mild to moderate undernutrition may not significantly impair the transfer of antibodies to the breast-fed infant.