Formula Feeding Skews Immune Cell
Composition toward Adaptive Immunity
Compared to Breastfeeding
This information is current as
of June 17, 2017.
Yvonne Andersson, Marie-Louise Hammarström, Bo
Lönnerdal, Gitte Graverholt, Helen Fält and Olle Hernell
J Immunol 2009; 183:4322-4328; Prepublished online 4
September 2009;
doi: 10.4049/jimmunol.0900829
http://www.jimmunol.org/content/183/7/4322
Subscription
Permissions
Email Alerts
This article cites 31 articles, 5 of which you can access for free at:
http://www.jimmunol.org/content/183/7/4322.full#ref-list-1
Information about subscribing to The Journal of Immunology is online at:
http://jimmunol.org/subscription
Submit copyright permission requests at:
http://www.aai.org/About/Publications/JI/copyright.html
Receive free email-alerts when new articles cite this article. Sign up at:
http://jimmunol.org/alerts
The Journal of Immunology is published twice each month by
The American Association of Immunologists, Inc.,
1451 Rockville Pike, Suite 650, Rockville, MD 20852
Copyright © 2009 by The American Association of
Immunologists, Inc. All rights reserved.
Print ISSN: 0022-1767 Online ISSN: 1550-6606.
Downloaded from http://www.jimmunol.org/ by guest on June 17, 2017
References
The Journal of Immunology
Formula Feeding Skews Immune Cell Composition toward
Adaptive Immunity Compared to Breastfeeding1
Yvonne Andersson,* Marie-Louise Hammarström,† Bo Lönnerdal,‡ Gitte Graverholt,§
Helen Fält,* and Olle Hernell2*
B
reastfed (BF)3 infants have fewer and milder infections
compared with formula-fed (FF) infants (1), partly because bioactive components in human milk not only contribute to high bioavailability of nutrients but also affect the maturation of innate and adaptive immunity (2). However, little is
known about the immunity of the newborn human infant and the
development of the immune system. In particular, there are no
prospective studies addressing the effect of different types of infant
feeding during the first half of infancy using the exclusively BF
infant as reference, although an increase in T cells, particularly in
Th cells (3, 4), and a decrease in NK cells (4) have been reported
for FF infants compared with BF infants at 6 mo of age.
The composition of infant formulas is continuously modified,
aiming at a composition more similar to that of human milk. When
using cow’s milk as the protein source for infant formulas, the
whey-to-casein ratio is adjusted to resemble the whey-predominant
human milk. To ensure sufficient intake of all essential amino acids, a higher protein content in formula as compared with human
*Department of Clinical Sciences, Pediatrics, and †Department of Clinical Microbiology, Immunology, Umeå University, Umeå, Sweden; ‡Department of Nutrition,
University of California, Davis, CA 95616; and §Arla Foods Ingredients, Aarhus,
Denmark
milk is required, which may have adverse effects (5, 6). By increasing ␣-lactalbumin, a major human milk whey protein, and
decreasing the content of ␤-lactoglobulin, a major bovine whey
protein, which is absent from human milk, it is possible to reduce
the total protein content in a formula and achieve a plasma amino
acid profile in FF infants more similar to that of BF infants (7).
Besides having a well-balanced amino acid composition, ␣-lactalbumin has antimicrobial and also other biological effects (8, 9).
Caseinoglycomacropeptide (CGMP) is cleaved off from ␬-casein
by pepsin digestion in the stomach and has also been shown to
have antimicrobial, immunomodulatory, and prebiotic effects (10 –
13), including promoting growth of bifidobacteria (14).
We hypothesized that the immune cell composition of FF infants would be more similar to that of BF infants if the protein
composition of the formula was made more similar to that of human milk, that is, by decreasing the ␤-lactoglobulin and increasing
the ␣-lactalbumin proportion compared with a standard formula.
The aim of this study was therefore to determine the composition
of immune cells in blood in the nearly newborn (1.5 mo of age)
and prospectively follow age-related changes up to 6 mo of age,
and to evaluate how formula feeding, depending on the ␣-lactalbumin and CGMP contents, affect this composition compared with
breastfeeding.
Received for publication March 13, 2009. Accepted for publication July 14, 2009.
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance
with 18 U.S.C. Section 1734 solely to indicate this fact.
Materials and Methods
1
Due to the very high initial breastfeeding rate (90% are exclusively BF at
2 mo), healthy term infants were recruited from well-baby clinics at 6 ⫾ 2
wk of age in Umeå, a university town in northern Sweden with 114,000
inhabitants, most of whom are middle class and well educated. After parents had given written consent for their infant to participate, 96 infants
fulfilled the following inclusion criteria: gestational age 36 – 42 wk, and
birth weight 2500 –5000 g. Thirty-four of them were recruited among
mothers who intended to exclusively breast-feed until 6 mo and served as
a BF reference group. Sixty-two of them, initially FF or weaned from the
breast at 6 ⫾ 2 wk, were randomized to either control formula (CON; 11%
This study was financially supported by Arla Foods Ingredients, Aarhus, Denmark.
2
Address correspondence and reprint to Dr. Olle Hernell, Department of Clinical
Sciences, Pediatrics, SE-901 87, Umeå University, Umeå, Sweden. E-mail address:
[email protected]
3
Abbreviations used in this paper: BF, breastfed; FF, formula-fed; CGMP, caseinoglycomacropeptide; CON, control formula; ␣-LAC, formula enriched in ␣-lactalbumin; RCGMP, formula enriched in ␣-lactalbumin and reduced in CGMP; WBC,
white blood cell.
Copyright © 2009 by The American Association of Immunologists, Inc. 0022-1767/09/$2.00
www.jimmunol.org/cgi/doi/10.4049/jimmunol.0900829
Study design
Downloaded from http://www.jimmunol.org/ by guest on June 17, 2017
The ontogeny of the immune system and the effect thereon by type of infant feeding is incompletely understood. We analyzed
frequencies and composition of immune cells in blood of breastfed (BF) and formula-fed (FF) infants at 1.5, 4, and 6 mo of age.
Three formulas with the same protein concentration but with varying levels of ␣-lactalbumin and caseinoglycomacropeptide were
compared. Twenty-nine exclusively BF infants served as reference, and 17 infants in each formula group completed the study.
Whole blood and PBMCs were analyzed by flow cytometry and immunoflow cytometry, respectively. Leukocyte count of BF
infants increased with time due to increased frequency of neutrophils. Lymphocyte count was high at 1.5 mo and was unchanged
over time, as were the relative proportions of CD4ⴙ ␣␤T cells, CD8ⴙ ␣␤T cells, B cells, NK cells, and ␥␦T cells. Most
CD45R0ⴙCD3ⴙ cells were HLA-DRⴚ and hence memory cells. Compared with breastfeeding, formula feeding resulted in a
significant decrease in proportion of NK cells, but a significant increase in naive CD4ⴙ ␣␤T cells and an elevated CD4-to-CD8
ratio, that is, 3.3 in the combined FF groups compared with 2.6 in the BF group. No significant differences were found between
the three groups of FF infants. In conclusion, blood cells of lymphoid lineage did not change significantly in frequencies or
composition from 1.5 to 6 mo of age in BF infants. In contrast, FF infants displayed an ongoing maturation of adaptive immunity
cells and a delayed recruitment of innate immunity cells as compared with BF infants. The Journal of Immunology, 2009, 183:
4322– 4328.
The Journal of Immunology
4323
Table I. Contents of ␣-lactalbumin, CGMP, and ␤-lactoglobulin in the
formulas studied
Type of Formula
Protein
CON
␣-Lactalbumin
CGMP
␤-Lactoglobulin
11
14
27
␣-LAC
RCGMP
(% of protein)
25
25
15
10
15
19
CON
␣-LAC
1.4
1.9
3.5
3.3
1.9
1.3
RCGMP
(g/L)
3.2
1.3
2.5
Statistical analysis
FIGURE 1. A, Flowchart of study subjects, that is, infants recruited to
the breastfed group (BF) and the three different formula groups (CON,
␣-LAC, and RCGMP). Values represent number of infants. B, Time schedule and study design.
␣-lactalbumin, 14% CGMP) (n ⫽ 21), formula enriched in bovine ␣-lactalbumin (␣-LAC; 25% ␣-lactalbumin, 15% CGMP) (n ⫽ 20), or formula
enriched in ␣-lactalbumin and also reduced in CGMP content (RCGMP;
25% ␣-lactalbumin, 10% CGMP) (n ⫽ 21). All formulas had at total protein concentration of 13 g/L (Fig. 1A and Table I). To prevent bias, all
investigators and parents involved in the study were blinded with respect to
type of formula given to the FF infants. Formulas were provided by Arla
Foods Ingredients and were produced as described earlier (15). After 4 mo
of age both BF and FF infants were allowed taste portions of fruit and
vegetables purées (1–2 tablespoons per day) until 6 mo when the study was
discontinued. The purées were supplied by the investigators.
Blood samples (3 ml) were taken by venopuncture into heparinized
tubes at entry 1.5 (6 ⫾ 2 wk), 4, and at 6 mo, and stored for 2– 4 h in room
temperature before analysis (Fig. 1B).
The Ethics Committee on Research Involving Human Subjects of the
Faculty of Medicine, Umeå University, approved the study.
Flow cytometry analysis of whole blood leukocyte composition
Routine hematology was performed at the Department of Clinical Chemistry, Umeå University Hospital. Concentrations of total white blood cells
(WBCs), lymphocytes, monocytes, as well as neutrophil, eosinophil, and
basophil granulocytes were assessed by forward and side scatter characteristics in flow cytometry (XE-2100; Sysmex).
Phenotyping by immunoflow cytometry
PBMCs were enriched by density gradient centrifugation on Ficoll-Paque
(GE Healthcare Biosciences) within 4 h of storage at room temperature.
PBMCs (100,000 cells) were stained with mouse mAbs directly conjugated
with PE or FITC. MAbs used were: anti-CD3 (clone SK7, IgG1), antiCD19 (clone 4G7, IgG1), anti-CD4 (clone SK3, IgG1), anti-CD8 (clone
SK1, IgG1), anti-CD16 (clone B73.1, IgG1), anti-CD56 (clones MY31,
IgG1), anti-TCR-␣␤ (clone WT31, IgG1), anti-TCR-␥␦ (clone 11F2,
IgG1), anti-HLA-DR (clone L243, IgG2a), anti-CD45R0 (clone
LUCHL-1, IgG2a), anti-CD45RA (clone L48, IgG1), anti-CD14 (clone
MøP9, IgG2b), and anti-CD45 (clone 2D1, IgG1), all from BD Biosciences. Fluorochrome, IgG subclass, and concentration-matched irrelevant mAbs served as negative controls (BD Biosciences). Fifteen thousand
ungated events were collected in two-color immunofluorescence in a FACScan (BD Biosciences) and analyzed using the CellQuest software program. The instrument was calibrated with calibrate beads. PBMCs were
distinguished by dual forward and side scatter plots and manually gated to
include all lymphocytes but exclude possible erythrocytes and granulocytes
in the analysis. The proportion of marker-positive cells is given as percentage of CD45⫹ cells in the same sample.
Results
Five of the infants in the BF group were excluded because their
mothers could not meet the criterion of exclusive breastfeeding for
6 mo. Four infants were excluded from the CON group: two because they developed cow’s milk protein allergy, and the parents
withdrew the other two from the study. In the ␣-LAC group the
corresponding numbers were two and one infant, respectively, and
in the RCGMP group one and three infants. Thus, the final numbers of infants completing the study were 29, 17, 17, and 17 in the
BF, CON, ␣-LAC, and RCGMP groups, respectively (Fig. 1A).
Time points for collection of blood samples are shown in Fig. 1B.
The mean percentages of infants from which sufficient blood samples were obtained to allow phenotyping by immunoflow cytometry were 97, 94, and 96% at 1.5, 4, and 6 mo, respectively.
We found that gradient centrifugation was an adequate way to
isolate PBMCs also from blood of very young infants. CD45 was
used as a pan-leukocyte marker in the immunophenotyping experiments, and CD45⫹ cells comprised ⬎95% of the PBMC preparations. Furthermore, the sum of the major immune cell populations in PBMCs, that is, monocytes (CD14⫹ cells), T cells (CD3⫹
cells), B cells (CD19⫹ cells), and NK cells (CD16⫹ and/or
CD56⫹CD3⫺ cells), was 97 ⫾ 3.2% (n ⫽ 76), 96 ⫾ 3.6% (n ⫽
73), and 96 ⫾ 3.1% (n ⫽ 75) at 1.5, 4, and 6 mo, respectively.
The total WBC count in BF infants increased significantly
during the first 6 mo of life, mainly explained by an increase in
circulating neutrophil granulocytes
In BF infants the total count of WBCs was 8.5 ⫾ 2.3 ⫻ 109/L at
1.5 mo of age, increased significantly with time ( p ⫽ 0.007), and
reached 9.9 ⫾ 2.6 ⫻ 109/L at 6 mo (Fig. 2A). Lymphocytes constituted the major fraction of the WBCs already at 1.5 mo, and
increased marginally during the study period (Fig. 2B). Neutrophil
granulocytes constituted the second largest immune cell population at 1.5 mo and continued to be so with a steady increase to
2.9 ⫾ 1.6 ⫻ 109/L at 6 mo ( p ⫽ 0.002) (Fig. 2B). Monocytes
constituted the third largest population at 1.5 mo, decreased significantly to 4 mo ( p ⫽ 0.01), but returned to the level at entry at
6 mo of age (Fig. 2B). Eosinophil granulocytes constituted a minor
population at 1.5 mo (0.3 ⫾ 0.1 ⫻ 109/L) and basophils were
scarce (0.03 ⫾ 0.03 ⫻ 109/L). The frequencies of the latter two
cell types remained stable throughout the study period (Fig. 2B).
Downloaded from http://www.jimmunol.org/ by guest on June 17, 2017
Comparisons between study groups were performed using linear regression
and analysis of covariance. Value at entry was used as covariate to circumvent that differences at study entry influenced outcome at 4 and 6 mo
of age. Comparisons between time points within a certain study group were
performed using general linear model with repeated measures and paired
sample t test. The software SPSS v13.0 was used for all calculations. A p
value of ⱕ0.05 was considered statistically significant. Descriptive values
are expressed as means ⫾ 1 SD.
4324
TYPE OF FEEDING AND PERIPHERAL BLOOD IMMUNE CELLS IN INFANTS
FIGURE 2. A, Frequency of WBCs
in whole blood of infants in the BF
group at the three test points, 1.5 (6 ⫾
2 wk, i.e., study entry), 4 and 6 mo of
age. Cell counts were done by flow
cytometry. B, Frequencies of lymphocytes, monocytes, neutrophils, eosinophils, and basophils in whole blood
of infants in the BF group as determined by forward/side scatter characteristics in flow cytometry at 1.5, 4,
and 6 mo of age. Bars represent
means ⫹ 1 SD from 28, 27, and 29
infants at 1.5, 4, and 6 mo,
respectively.
A panel of mAbs was used in two-color immunoflow cytometry to
characterize the lymphocytes with regard to subtypes. Fig. 3 shows
an example of analysis of PBMCs from a 1-mo-old BF infant.
B cells were defined as CD19⫹ cells (Fig. 3B), NK cells as non-T
cells (CD3⫺ cells) stained by a mixture of mAbs against the
NK cell marker CD56 and the Fc␥-receptor CD16 (Fig. 3F), and
T cells as CD3⫹ cells. For each child and time point the percentage
of CD3⫹ cells was calculated as the mean of three determinations,
that is, anti-CD3 vs anti-CD19 (Fig. 3B), anti-CD3 vs anti-CD56
plus anti-CD16 (Fig. 3F), and anti-CD3 vs anti-HLA-DR (Fig.
3E). T cells were further characterized by TCR usage, that is, ␣␤T
cells (TCR-␣␤⫹ cells) and ␥␦T cells (TCR-␥␦⫹ cells; Fig. 3C),
helper T cell phenotype (CD4⫹ cells; Fig. 3D), cytotoxic T cell
phenotype (CD8⫹ cells; Fig. 3D), activated phenotype (HLADR⫹CD3⫹ cells; Fig. 3E), memory/activated phenotype (CD45R0⫹
CD3⫹ cells; Fig. 3H), and naive phenotype (CD45RA⫹CD3⫹
cells; Fig. 3G). The experimental setup also allowed for detection
of possible circulating immature T cells as CD4/CD8 double-positive
cells (Fig. 3D) and CD3/CD56 double-positive cells (Fig. 3F).
In the BF infants T cells constituted by far the largest lymphocyte population at 1.5 mo (68.5 ⫾ 6.8%) followed by B cells
(16.6 ⫾ 4.7%) and NK cells (6.9 ⫾ 2.8%). None of the lymphocyte types changed significantly in proportion during the study
period. The vast majority of T cells expressed TCR-␣␤, while ␥␦T
cells constituted a minor population (4.4 ⫾ 2.0%) of the CD3⫹
cells. Furthermore, CD4⫹ cells (51.9 ⫾ 6.3%) dominated over
CD8⫹ cells (21.9 ⫾ 5.6%) with a CD4-to-CD8 ratio of 2.6 ⫾ 0.9.
All of these parameters remained unchanged over time.
Interestingly, although the proportion of naive T cells
(CD45RA⫹CD3⫹ cells) was high (60.9 ⫾ 8.4%), it was less
than the total population of T cells, and the proportion of
CD45R0⫹CD3⫹ cells was as high as 22.9 ⫾ 10.6% already at 1.5
mo, suggesting that circulating memory and/or activated T cells
are present in significant numbers already during the first months
of life. The frequency of HLA-DR⫹CD3⫹ cells, however, was
low, indicating that most of the CD45R0⫹CD3⫹ cells are memory
T cells rather than activated T cells. The proportions of
CD45RA⫹CD3⫹ cells, CD45R0⫹CD3⫹ cells, and HLA-DR⫹
CD3⫹ cells were approximately the same at 1.5, 4, and 6 mo (Fig.
4). The proportion of CD4/CD8 double-positive cells was initially
low (⬍2.5%) with a tendency to decrease over time. The proportion of CD3/CD56 double-positive cells was even lower (⬍1.5%).
Formula feeding had only marginal effect on WBC counts
Total WBC counts and frequencies of lymphocytes and immune
cells of the myeloid lineage in the three groups of FF infants (i.e.,
the CON, ␣-LAC, and RCGMP groups) were compared with those
of BF infants. The ␣-LAC- and RCGMP groups (experimental
formulas) were also compared with the CON group (standard formula). As expected, frequencies of WBCs were not significantly
different between children in the BF group and the FF groups at
study entry, that is, at an average age of 1.5 mo (Fig. 5A). From
entry to 4 mo there was a small decline in the ␣-LAC- and
RCGMP groups compared with the BF group and in the ␣-LAC
group also compared with the CON group. However, at 6 mo of
age there was again no significant difference in WBC counts between any of the study groups (Fig. 5A). The functional significance of this brief decrease in WBC count is not obvious but could
reflect a transient response to formula. Frequencies of lymphocytes
were the same in BF and FF infants and did not change over time
(data not shown). None of the immune cell types differed significantly in frequency in the ␣-LAC- or RCGMP groups compared
with the CON group. As in the BF group the monocytes decreased
from 1.5 to 4 mo in all FF groups ( p ⫽ 0.008, p ⫽ 0.002, and p ⫽
0.04 for the CON, ␣-LAC, and RCGMP groups, respectively) (Fig.
5B). No differences were found in monocyte count between any of
the study groups at 4 and 6 mo. Compared with the BF group, the
␣-LAC group had significantly lower frequencies of neutrophil
granulocytes at both 4 and 6 mo (Fig. 5C). Although there appeared to be a slower increase in the frequency of neutrophil granulocytes also in the CON and RCGMP groups compared with the
BF group, this tendency did not reach statistical significance (Fig.
5C). In all FF groups eosinophil granulocytes were significantly lower
at 4 mo compared with the BF group and remained significantly lower
also at 6 mo in the CON and ␣-LAC groups (Fig. 5D).
Formula feeding caused a significant increase in the proportion
of CD4⫹ ␣␤T cells
There were no significant differences between either of the three
FF groups and the BF group with regard to lymphocyte subset
composition at 1.5 mo (Fig. 6). There were, however, significant
changes following study entry. The proportion of T cells (CD3⫹
cells) in FF infants increased significantly with age (CON, p ⫽
0.005; ␣-LAC, p ⫽ 0.005; and RCGMP, p ⫽ 0.02) and was significantly higher than in the BF group at both 4 and 6 mo (Fig. 6A).
In contrast, the proportions of NK cells (CD16⫹ and/or
CD56⫹CD3⫺ cells) decreased significantly in the CON, ␣-LAC,
and RCGMP groups and were significantly lower than in the BF
Downloaded from http://www.jimmunol.org/ by guest on June 17, 2017
The proportions of different lymphocyte subsets remained
constant throughout the first 6 mo of life in BF infants
The Journal of Immunology
4325
Discussion
FIGURE 3. Two-color immunoflow cytometry analysis of PBMCs
from a 1-mo-old infant at entry into the BF group. Cells were stained with
mixtures of PE-labeled anti-CD19 and FITC-labeled anti-CD3 (B), PElabeled anti-TCR-␣␤ and FITC-labeled anti-TCR-␥␦ (C), PE-labeled antiCD8 and FITC-labeled anti-CD4 (D), PE-labeled anti-HLA-DR and FITClabeled anti-CD3 (E), PE-labeled anti-CD16, PE-labeled anti-CD56, and
FITC labeled CD3 (F), PE-labeled anti-CD3 and FITC-labeled antiCD45RA (G), and PE-labeled anti-CD45R0 and FITC-labeled anti-CD3
(H). Quadrant regions were set according to corresponding negative control
incubated with mixtures of fluorochrome-labeled, concentration- and isotype matched irrelevant mAbs, that is, PE-labeled IgG1 and FITC-labeled
IgG1 in A and not shown.
group both at 4 and 6 mo (Fig. 6H). No statistically significant
difference in the frequency of B cells (CD19⫹ cells) was seen in
any of the FF groups compared with the BF group at either 4 or 6
mo (Fig. 6G).
More detailed analysis revealed that it was the ␣␤T cells (TCR␣␤⫹ cells) that were increased in infants of the FF groups both at
4 and 6 mo (Fig. 6C), while the ␥␦T cells remained a small population of ⬃2.5% (Fig. 6D). When performing the same comparisons for CD4⫹ cells and CD8⫹ cells it became evident that the
proportion of CD4⫹ cells was significantly higher in all three FF
groups compared with the BF group at 4 and 6 mo (Fig. 6E), while
An interesting difference between BF and FF infants was that BF
infants showed a steady increase in total WBC counts, essentially
explained by an increase in circulating neutrophil granulocytes,
while FF infants did not. This potential enforcement of innate immunity by breastfeeding in the young infant with a not fully developed immune system has not been observed previously, as the
role of early feeding mode has not been studied. Since there is
individual variation in both cell counts and composition of immune cell types, changes over time may easily be missed in crosssectional studies. A stable eosinophil granulocyte count in the BF
group but a decline in the FF groups is in agreement with Cordle
et al. (16), who compared BF and FF infants at 6 mo of age, and
further indicates beneficial effects on innate immunity by
breastfeeding.
Lymphocyte counts were stable over time in all study groups,
which is in accord with previous studies on healthy, term infants
during the first half of infancy (17–19). These studies reported
age-dependent changes in proportions of lymphocyte subsets.
However, we found that lymphocyte subset composition is
strongly influenced by feeding mode, that is, stable from 1.5 to 6
mo of age in BF infants but an increase of ␣␤Th cells on the
expense of NK cells, CTLs, and B cells in FF infants. As the
impact of feeding mode was not considered in the earlier studies,
the reported changes may reflect increasing proportions of FF infants in their material. In line with this notion, neonate cord blood
had higher levels of NK cells and lower levels of T cells than did
infant peripheral blood (18, 19). Furthermore, Hawkes et al. (4, 20)
found a lower proportion of NK cells in FF compared with BF
infants at 6 mo of age, which is in agreement with our results. Our
study suggests that in some respects the immune system develops
slower in FF infants compared with BF infants because there was
a more pronounced dominance of naive Th cells over memory T
cells as well as of Th cells over CTLs and a slower recruitment of
cells with effector functions in innate immunity. The high frequency of Th cells in FF infants could reflect that their immune
Downloaded from http://www.jimmunol.org/ by guest on June 17, 2017
the proportion of CD8⫹ cells was the same or slightly lower than
in BF infants (Fig. 6F). On average the CD4-to-CD8 ratio was
higher in the FF groups compared with the BF group, and the
difference reached statistical significance at both 4 and 6 mo for the
RCGMP group (Fig. 6B).
The proportion of naive T cells (CD45RA⫹CD3⫹ cells) was
significantly higher in all three FF groups compared with the BF
group at both 4 and 6 mo of age (Fig. 4A). The increase from 1.5
to 6 mo was statistically significant in the ␣-LAC group (Fig. 4A).
In contrast, there were no significant differences between the
groups with regard to the proportion of memory T cells
(CD45R0⫹CD3⫹ cells). There was, however, a tendency to a decrease over time in all groups, and the decrease from 1.5 to 6 mo
was significant in the ␣-LAC group ( p ⫽ 0.04; Fig. 4B). The
proportion of activated T cells (HLA-DR⫹CD3⫹ cells) was low in
the BF group and even lower in all three FF groups (Fig. 4C). In
the RCGMP group this decrease reached statistical significance
from 1.5 to 6 mo ( p ⫽ 0.005) and the proportion was significantly
lower than in the BF group at both 4 and 6 mo (Fig. 4C).
Double-positive CD4/CD8 cells were low in all FF groups
(⬍2.1%) and showed a tendency to decrease over time, while the
proportion of CD3/CD56 double-positive cells was even lower
(⬍1.2% at 6 mo). All FF groups showed a tendency to lower
values of double-positive CD4/CD8 cells at 4 and 6 mo compared
with the BF group. No statistically significant differences were
found between the FF groups for any of the lymphocyte subsets
analyzed.
4326
TYPE OF FEEDING AND PERIPHERAL BLOOD IMMUNE CELLS IN INFANTS
Marker positive cells (%)
CD45RA+ T cells
CD45R0+ T cells
80
70
60
HLA-DR+ T cells
30
30
25
25
50
20
20
40
30
15
15
10
10
5
5
20
10
0
1.5
4
0
6
1.5
4
6
0
BF
CON
α -LAC
RCGMP
1.5
4
6
FIGURE 4. Mean proportions of naive T cells, memory/activated T cells, and activated T cells estimated as double-positive cells in two-color immunoflow cytometry with anti-CD3 vs anti-CD45RA (A), anti-CD3 vs anti-CD45R0 (B), and anti-CD3 vs HLA-DR (C) stained samples, respectively, in
infants of the BF group (⽧, hatched line), the CON group (f, solid line), the ␣-LAC group (Œ, solid line), and the RCGMP group (, solid line) at 1.5
(6 ⫾ 2 wk, i.e., study entry), 4, and 6 mo of age. Bars represent means ⫾ 1 SD and statistically significant differences between the BF group and a FF
group are indicated: ⴱ, p ⬍ 0.05; ⴱⴱ, p ⬍ 0.01; and ⴱⴱⴱ, p ⬍ 0.001.
FIGURE 5. Frequencies of WBCs
(A), monocytes (B), neutrophil granulocytes (C), and eosinophil granulocytes (D) in whole blood of infants in
the BF group (⽧, hatched line), the
CON group (f, solid line), the
␣-LAC group (Œ, solid line), and
the RCGMP group (, solid line) at
1.5 (6 ⫾ 2 wk, i.e., study entry), 4,
and 6 mo of age. Cell counts were
done by flow cytometry. Bars represent means ⫾ 1 SD, and statistically
significant differences between the BF
group and a FF group are indicated:
ⴱ, p ⬍ 0.05; ⴱⴱ, p ⬍ 0.01; and
ⴱⴱⴱ, p ⬍ 0.001.
Cells x 109/L
WBCs
12
11
10
9
8
7
6
5
4
3
2
1
0
Monocytes
1.25
BF
CON
α-LAC
RCGMP
1.00
0.75
0.50
0.25
1.5
4
6
0.00
1.5
6
Eosinophils
Neutrophils
4.0
0.5
3.5
0.4
3.0
2.5
0.3
2.0
0.2
1.5
1.0
0.1
0.5
0.0
4
1.5
4
6
0.0
1.5
4
6
Downloaded from http://www.jimmunol.org/ by guest on June 17, 2017
The proportion of CD45R0⫹CD3⫹ cells was on average 20% at
1.5 mo of age. No increase narrowing the gap to adult levels, that
is, ⬃55% (19), was noted in either BF or FF infants at 4 or 6 mo
of age. Both activated T cells and memory T cells express
CD45R0, while only activated T cells express the MHC class II
molecule HLA-DR. Since the frequency of CD3⫹HLA-DR⫹ cells
was ⬍3% we conclude that there is a very significant number of
memory T cells already at 1.5 mo of age.
The list of breast milk components with potential to influence
immune cell development and function is long and includes bioactive peptides and proteins, for example, cytokines, chemokines,
and growth factors, and components that promote colonization of
the infant’s gut (22). Several of these may act directly on the infant’s gastrointestinal mucosa, but may also reach the circulation.
Decreased mortality in TGF-␤-null mice fed milk from heterozygote dams as compared with milk from null dams (23) and enhanced immune tolerance to dietary Ags in mice after oral administration of TGF-␤ supports this scenario (24). When considering
system is more skewed toward adaptive immunity compared with
BF infants.
The increase in CD4⫹ ␣␤T cells resulting from formula feeding
confirms and extends the results of previous studies on infants at 6 mo
of age (3, 4, 16). In contrast, one study reported that the frequency of
T cells is the same in 6-mo-old BF and FF infants (16). This could,
however, be explained by the fact that the infants in the latter BF
group were exclusive BF only during the first 2 of the 6 mo.
The proportion of circulating naive T cells, CD45RA⫹CD3⫹
cells, in BF infants was similar to that in adults, that is, ⬃60%
(19), while the levels of these cells were significantly higher in FF
infants. Similarly, the proportion of CD45RA⫹CD3⫹ cells was
higher in preterm infants fed formula compared with those fed
human milk (21), suggesting that exposure to new food Ags in
absence of breast milk causes recruitment of naive T cells to the
blood. This might reflect propensity for an adaptive immune response rather than for induction of tolerance induction to
dietary Ags.
The Journal of Immunology
4327
5
80
70
4
60
50
3
40
30
2
20
1
10
0
1.5
4
0
6
BF
CON
α-LAC
RCGMP
1.5
4
3
2
1
1.5
4
0
6
1.5
T-helper cells
6
T-cytotoxic cells
30
70
60
25
50
20
40
15
30
10
20
5
10
0
4
4
1.5
6
0
1.5
B cells
4
6
NK cells
25
12
20
10
8
15
6
10
4
5
2
0
0
1.5
the observed differences in blood immune cell composition in BF
compared with FF infants some of these factors seem to be of
particular relevance. Lactoferrin is known to enhance NK and T
cell proliferation and function (25, 26). Immune-modulating factors such as IL-2 may provide the BF infant with important signals
during significant stages of T cell development (27) and even contribute to the larger thymus in BF compared with FF infants at 4
mo (28) and 10 mo of continued breastfeeding (29). The earlier
exposure to new dietary food Ags in FF infants may enhance T cell
release from the thymus, causing the significantly higher levels of
circulating naive helper T cells in these children. Moreover, granulocyte CSF is suggested to promote eosinophil production and
could contribute to the higher levels of eosinophils in BF infants
(30, 31).
Since there were no statistically significant differences between
the three groups of FF infants but all three FF groups differed
4
6
1.5
4
6
significantly from BF infants in several respects, it appears that the
differences in ␣-lactalbumin and CGMP concentration in the formulas studied had no or only minor effects on the distribution of
immune cells in peripheral blood during the first 6 mo of life. It is
possible that larger differences in proportions of whey proteins in
formula are needed to observe significant changes in immune cell
distribution. Alternatively, factors in human milk that affect development and distribution of leukocytes were not present in the
formulas studied.
Whether the differences found between the FF groups and the
BF group are of clinical significance cannot be concluded from this
study. However, as previously reported, there were no differences
between BF and FF infants with regard to fever episodes, number
of days with fever, and episodes of airway infections (7).
In conclusion, the most striking and consistent differences between FF and BF infants were a higher proportion of T cells
Downloaded from http://www.jimmunol.org/ by guest on June 17, 2017
Marker positive cells (%)
5
80
70
60
50
40
30
20
10
0
6
γδ Τ cells
αβ Τ cells
FIGURE 6. Mean proportions of T
cells (CD3⫹ cells; A), ␣␤T cells
(TCR-␣␤⫹ cells; C), ␥␦T cells (TCR␥␦⫹ cells; D), T cells with helper cell
phenotype (CD4⫹ cells; E), T cells
with cytotoxic phenotype (CD8⫹
cells; F), B cells (CD19⫹ cells; G),
NK cells (CD16⫹/CD56⫹/CD3⫺
cells; H), and the ratio between CD4⫹
cells and CD8⫹ cells (B) as estimated
by immunoflow cytometry in PBMCs
of infants in the BF group (⽧,
hatched line), the CON group (f,
solid line), the ␣-LAC group (Œ, solid
line), and the RCGMP group (, solid
line) at 1.5 (6 ⫾ 2 wk, i.e., study entry), 4, and 6 mo of age. Bars represent means ⫾ 1 SD, and statistically
significant differences between the BF
group and a FF group are indicated:
ⴱ, p ⬍ 0.05; ⴱⴱ, p ⬍ 0.01; and
ⴱⴱⴱ, p ⬍ 0.001.
4
4328
TYPE OF FEEDING AND PERIPHERAL BLOOD IMMUNE CELLS IN INFANTS
caused by a selective increase in naive Th cells, a lower proportion
of NK cells, and lower eosinophil granulocyte cell counts occurring already after 3 mo and persisting after 5 mo of formula feeding. Thus, it seems that the development of the immune system in
FF infants is preferentially forced toward cells in adaptive immunity, while the development of innate immunity is slower than in
BF infants.
Acknowledgments
We are grateful to the participating families. We thank the research nurse
Margaretha Bäckman for keeping contact with the families and for skilful
work when collecting the blood samples. Carina Lagerqvist is gratefully
acknowledged for invaluable assistance with analyses of blood samples.
We are grateful to Dr. Hans Stenlund for statistical advice, and to Drs. Sten
Hammarström and Lars Bläckberg for critical reading of the manuscript.
Disclosures
References
1. Hanson, L. A., and M. Korotkova. 2002. The role of breastfeeding in prevention
of neonatal infection. Semin. Neonatol. 7: 275–281.
2. Lönnerdal, B. 2004. Human milk proteins: key components for the biological
activity of human milk. Adv. Exp. Med. Biol. 554: 11–25.
3. Carver, J. D., B. Pimentel, D. A. Wiener, N. E. Lowell, and L. A. Barness. 1991.
Infant feeding effects on flow cytometric analysis of blood. J. Clin. Lab. Anal. 5:
54 –56.
4. Hawkes, J. S., M. A. Neumann, and R. A. Gibson. 1999. The effect of breast
feeding on lymphocyte subpopulations in healthy term infants at 6 months of age.
Pediatr. Res. 45: 648 – 651.
5. Karlsland Akeson, P. M., I. E. Axelsson, and N. C. Raiha. 1998. Protein and
amino acid metabolism in three- to twelve-month-old infants fed human milk or
formulas with varying protein concentrations. J. Pediatr. Gastroenterol. Nutr. 26:
297–304.
6. Schmidt, I. M., I. N. Damgaard, K. A. Boisen, C. Mau, M. Chellakooty,
K. Olgaard, and K. M. Main. 2004. Increased kidney growth in formula-fed
versus breast-fed healthy infants. Pediatr. Nephrol. 19: 1137–1144.
7. Sandström, O., B. Lönnerdal, G. Graverholt, and O. Hernell. 2008. Effects of
␣-lactalbumin-enriched formula containing different concentrations of glycomacropeptide on infant nutrition. Am. J. Clin. Nutr. 87: 921–928.
8. Bounous, G., and P. A. Kongshavn. 1985. Differential effect of dietary protein
type on the B-cell and T-cell immune responses in mice. J. Nutr. 115: 1403–1408.
9. Pellegrini, A., U. Thomas, N. Bramaz, P. Hunziker, and R. von Fellenberg. 1999.
Isolation and identification of three bactericidal domains in the bovine ␣-lactalbumin molecule. Biochim. Biophys. Acta 1426: 439 – 448.
10. Kawasaki, Y., H. Isoda, M. Tanimoto, S. Dosako, T. Idota, and K. Ahiko. 1992.
Inhibition by lactoferrin and ␬-casein glycomacropeptide of binding of cholera
toxin to its receptor. Biosci. Biotechnol. Biochem. 56: 195–198.
11. Otani, H., and I. Hata. 1995. Inhibition of proliferative responses of mouse spleen
lymphocytes and rabbit Peyer’s patch cells by bovine milk caseins and their
digests. J. Dairy Res. 62: 339 –348.
12. Otani, H., M. Monnai, Y. Kawasaki, H. Kawakami, and M. Tanimoto. 1995.
Inhibition of mitogen-induced proliferative responses of lymphocytes by bovine
␬-caseinoglycopeptides having different carbohydrate chains. J. Dairy Res. 62:
349 –357.
13. Li, E. W., and Y. Mine. 2004. Immunoenhancing effects of bovine glycomacropeptide and its derivatives on the proliferative response and phagocytic activ-
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
Downloaded from http://www.jimmunol.org/ by guest on June 17, 2017
Co-author Gitte Graverholt is an employee of Arla Foods Ingredients. She
contributed to the design of the study and product composition but had no
influence on the data compilation and analysis or on the interpretation of
the results. None of the other authors declare any conflicts of interest.
14.
ities of human macrophagelike cells, U937. J. Agric. Food Chem. 52:
2704 –2708.
Bruck, W. M., G. Graverholt, and G. R. Gibson. 2003. A two-stage continuous
culture system to study the effect of supplemental ␣-lactalbumin and glycomacropeptide on mixed cultures of human gut bacteria challenged with enteropathogenic Escherichia coli and Salmonella serotype Typhimurium. J. Appl. Microbiol.
95: 44 –53.
Bruck, W. M., M. Redgrave, K. M. Tuohy, B. Lönnerdal, G. Graverholt,
O. Hernell, and G. R. Gibson. 2006. Effects of bovine ␣-lactalbumin and casein
glycomacropeptide-enriched infant formulae on faecal microbiota in healthy term
infants. J. Pediatr. Gastroenterol. Nutr. 43: 673– 679.
Cordle, C. T., T. R. Winship, J. P. Schaller, D. J. Thomas, R. H. Buck,
K. M. Ostrom, J. R. Jacobs, M. M. Blatter, S. Cho, W. M. Gooch, 3rd, and
L. K. Pickering. 2002. Immune status of infants fed soy-based formulas with or
without added nucleotides for 1 year, Part 2: immune cell populations. J. Pediatr.
Gastroenterol. Nutr. 34: 145–153.
Shearer, W. T., H. M. Rosenblatt, R. S. Gelman, R. Oyomopito, S. Plaeger,
E. R. Stiehm, D. W. Wara, S. D. Douglas, K. Luzuriaga, E. J. McFarland, et al.
2003. Lymphocyte subsets in healthy children from birth through 18 years of age:
the Pediatric AIDS Clinical Trials Group P1009 study. J. Allergy Clin. Immunol.
112: 973–980.
Comans-Bitter, W. M., R. de Groot, R. van den Beemd, H. J. Neijens, W. C. Hop,
K. Groeneveld, H. Hooijkaas, and J. J. van Dongen. 1997. Immunophenotyping
of blood lymphocytes in childhood: reference values for lymphocyte subpopulations. J. Pediatr. 130: 388 –393.
de Vries, E., S. de Bruin-Versteeg, W. M. Comans-Bitter, R. de Groot,
W. C. Hop, G. J. Boerma, F. K. Lotgering, and J. J. van Dongen. 2000. Longitudinal survey of lymphocyte subpopulations in the first year of life. Pediatr. Res.
47: 528 –537.
Hawkes, J. S., and R. A. Gibson. 2001. Lymphocyte subpopulations in breast-fed
and formula-fed infants at six months of age. Adv. Exp. Med. Biol. 501: 497–504.
Field, C. J., C. A. Thomson, J. E. Van Aerde, A. Parrott, A. Euler, E. Lien, and
M. T. Clandinin. 2000. Lower proportion of CD45R0⫹ cells and deficient interleukin-10 production by formula-fed infants, compared with human-fed, is corrected with supplementation of long-chain polyunsaturated fatty acids. J. Pediatr.
Gastroenterol. Nutr. 31: 291–299.
Hosea Blewett, H. J., M. C. Cicalo, C. D. Holland, and C. J. Field. 2008. The
immunological components of human milk. Adv. Food Nutr. Res. 54: 45– 80.
Letterio, J. J., A. G. Geiser, A. B. Kulkarni, N. S. Roche, M. B. Sporn, and
A. B. Roberts. 1994. Maternal rescue of transforming growth factor-␤1 null mice.
Science 264: 1936 –1938.
Ando, T., K. Hatsushika, M. Wako, T. Ohba, K. Koyama, Y. Ohnuma, R. Katoh,
H. Ogawa, K. Okumura, J. Luo, T. Wyss-Coray, and A. Nakao. 2007. Orally
administered TGF-␤ is biologically active in the intestinal mucosa and enhances
oral tolerance. J. Allergy Clin. Immunol. 120: 916 –923.
Shau, H., A. Kim, and S. H. Golub. 1992. Modulation of natural killer and
lymphokine-activated killer cell cytotoxicity by lactoferrin. J. Leukocyte Biol. 51:
343–349.
Mincheva-Nilsson, L., M. L. Hammarstrom, P. Juto, and S. Hammarstrom. 1990.
Human milk contains proteins that stimulate and suppress T lymphocyte proliferation. Clin. Exp. Immunol. 79: 463– 469.
Bryan, D. L., K. D. Forsyth, R. A. Gibson, and J. S. Hawkes. 2006. Interleukin-2
in human milk: a potential modulator of lymphocyte development in the breastfed
infant. Cytokine 33: 289 –293.
Hasselbalch, H., D. L. Jeppesen, M. D. Engelmann, K. F. Michaelsen, and
M. B. Nielsen. 1996. Decreased thymus size in formula-fed infants compared
with breastfed infants. Acta Paediatr. 85: 1029 –1032.
Hasselbalch, H., M. D. Engelmann, A. K. Ersboll, D. L. Jeppesen, and
K. Fleischer-Michaelsen. 1999. Breast-feeding influences thymic size in late infancy. Eur. J. Pediatr. 158: 964 –967.
Calhoun, D. A., M. Lunoe, Y. Du, and R. D. Christensen. 2000. Granulocyte
colony-stimulating factor is present in human milk and its receptor is present in
human fetal intestine. Pediatrics 105: e7.
Calhoun, D. A., S. E. Sullivan, M. Lunoe, Y. Du, and R. D. Christensen. 2000.
Granulocyte-macrophage colony-stimulating factor and interleukin-5 concentrations in premature neonates with eosinophilia. J. Perinatol. 20: 166 –171.