Human Milk Fortification Increases Bnip3 Expression Associated With Intestinal Cell Death In Vitro : Journal of Pediatric Gastroenterology and Nutrition

Secondary Logo

Journal Logo

Original Articles: Nutrition

Human Milk Fortification Increases Bnip3 Expression Associated With Intestinal Cell Death In Vitro

Diehl-Jones, William*; Archibald, Alyssa; Gordon, Joseph W.§; Mughal, Wajihah; Hossain, Zakir||; Friel, James K.||

Author Information
Journal of Pediatric Gastroenterology and Nutrition: November 2015 - Volume 61 - Issue 5 - p 583-590
doi: 10.1097/MPG.0000000000000876
  • Free


What Is Known

  • Breast milk feeds given to preterm infants must be fortified for optimal growth and maturation of premature infants.
  • Fortification of breast milk with bovine milk–based supplements (HMF) is associated with higher rates of necrotizing enterocolitis in extremely-low-birth-weight infants.
  • Increased expression of BCL2/adenovirus E1B 19 kDa protein-interacting protein increases apoptosis.

What Is New

  • Digested breast milk + HMF increases intestinal cell damage and death in vitro.
  • Digested breast milk + HMF increases BCL2/adenovirus E1B 19 kDa protein-interacting protein gene and protein expression in vitro.
  • This effect is rescued by the prostaglandin analogue misoprostol.

The benefits of human breast milk (BM) for premature infants are well known; lower rates of infection and a lower incidence of necrotizing enterocolitis (NEC) are but 2 of the advantages conferred upon infants who are exclusively breast-fed (1,2). The caloric and nutrient needs of premature infants, however, often cannot be met by mother's milk alone, necessitating supplementation of BM with multicomponent human milk fortifiers (HMFs), most of which are predominantly bovine milk based (3). The addition of these HMFs to BM is a standard of care in all of the North American neonatal intensive care units (NICUs), and it is generally regarded as both safe and efficacious (4–7). More recent evidence from our laboratory, however, indicates that HMF is associated with increased oxidative stress in preterm infants (8), which in turn is implicated in several diseases of prematurity, including NEC (9), an inflammatory bowel disease that afflicts up to 14% of preterm infants born at <1000 g (10).

Several recent clinical studies on either extremely premature (>28 weeks) or very-low-birth-weight infants (VLBW) indicate that this more fragile population of infants are at greater risk of feeding intolerances and/or NEC when fed bovine-based HMFs. Higher rates of NEC occur in VLBW infants fed bovine-based HMF, suggesting that human milk–based fortifiers compared with bovine-based fortifiers may reduce the incidence of NEC (11). In addition, preterm infants fed an exclusive human milk–based diet have lower rates of NEC than those fed human milk and bovine milk–based products (12). Extremely premature infants receiving bovine-based fortifier diets also have significantly greater durations of enteral nutrition and higher rates of surgical NEC compared with those receiving human milk–based diets (13).

Little is known of the physiological and nutrigenomic effects of HMFs. We recently demonstrated that HMFs increase oxidative stress in vivo (8), leading us to postulate that feeds supplemented with a bovine-based HMF may elicit signaling pathways, which in turn trigger cell damage and cell death in enterocytes. One key cell death pathway involves BCL2/adenovirus E1B 19 kDa protein-interacting protein (Bnip3). Bnip3 is a member of the BCL2 family, which is activated by oxidative stress and is capable of triggering either apoptotic or necrotic cell death in enterocytes (14). Furthermore, Bnip3 has been implicated in the pathogenesis of NEC because the protein product of Bnip3 is found in higher concentrations in the colon of infants in whom NEC has been diagnosed (15). We therefore hypothesized that bovine milk–based HMFs increase Bnip3 in vitro and tested this with a validated in vitro digestion assay.


Cell Culture

The effects of digested BM and BM/HMFs admixtures were assayed using either FHs 74 Int cells (American Type Culture Collection number CCL-241) or HT-29MTX clone E12 cells (a kind gift from Dr Per Artursson). FHs 74 Int cells are primary human epithelioid cells from the human fetal small intestine and were used for oxidative stress assays and genomic analysis. All other cellular assays were conducted with HT-29MTX, which is a secretory human adenocarcinoma cell line that constitutively expresses mucins. FHs 74 Int cells were used in genomic analyses because of the fact that these are nontransformed cells, fetal, and would therefore be more comparable with enterocytes in human preterm infants. HT-29MTX cells were used in other assays because of their ability to remain adherent for longer periods of time in assay conditions and for the higher transfection efficiency compared with the primary cell line. Cells were incubated at 37°C under 5% CO2 and 95% air at 85% relative humidity. Dulbecco Modified Eagle Medium (DMEM) (Gibco Chemicals, Paisley, Scotland, UK) with additives was used as the medium for growth of the cells. Additives included insulin 1 mg/mL (FHs 74 Int only), human epidermal growth factor 20 μg/mL, penicillin (10,000 U/mL)/streptomycin (10 mg/mL), 10% of fetal bovine serum (Sigma-Aldrich, Oakville, ON, Canada), L-glutamine, and 2.5 mmol/L of sodium pyruvate (Sigma-Aldrich). For exposure to experimental treatments, culture media was aspirated, and cells growing in either 25-cm2 sterile Petri dishes or polycarbonate 96-well test plates (Corning Life Sciences, Corning, NY) were overlain with either phosphate-buffered saline (PBS) or in vitro–digested HM ± HMF (Ross Laboratories [Columbus, OH], prepared as per manufacturer's directions before the digestion procedure) diluted 1:100 with PBS. This dilution was empirically chosen as a result of our observation that, in our assay conditions, cultured cells lose adhesion with the substrate at higher concentrations.

After exposure at 37°C under 5% of CO2, 95% of air, and 85% of relative humidity, test solutions were aspirated and cells washed in PBS at pH 7.4 and processed for endpoint assays.

BM Samples

BM samples were collected from 21 mothers of premature infants (born between 29 and 37 weeks of gestation) during the second week of lactation. Consent and ethical approval according to the University of Manitoba human research ethics board were obtained from women who participated in the study. BM samples were collected from mothers by using a breast pump or hand expression, transported on ice to the laboratory, frozen immediately, and stored at −80°C until analysis. To minimize individual variance, BM was pooled before any analysis.

In Vitro Digestion

An in vitro digestion model mimicking premature infant digestion was utilized to digest BM ± HMF as described earlier (16). Briefly, in the gastric phase, 1.73 mL of 140 mmol/L NaCl and 5 mmol/L KCl solution was added to 3.27 mL of BM to mimic the in vivo dilution of ingested BM by digestive secretions. After adding 0.25 mL of pepsin (Sigma-Aldrich, St Louis, MO) (0.2-g pepsin dissolved in 5 mL of 0.1 N HCl), HM samples were adjusted to pH 4.0 by HCl and placed in a 37°C shaking incubator for 30 minutes. HM samples were then brought to pH 6.0 by NaHCO3 and incubated for further 30 minutes. In the intestinal phase, 1.25 mL of pancreatin (Sigma-Aldrich)/bile (Sigma-Aldrich) solution (0.05 g of pancreatin and 0.3 g of bile extract dissolved in 25 mL of 0.1 M NaHCO3) and 17.2 mU of lactase (Sigma-Aldrich) were added to HM samples. HM samples were next adjusted to pH 7.0 by 1 mol/L NaHCO3 and to a final volume of 10 mL by 140 mmol/L NaCl and 5 mmol/L KCl solution. After being shaken in a 37°C incubator for 2 hours, intestinal phase end products were heat-inactivated at 95°C for 2 minutes, then stored in a −20°C freezer until further analysis.

Intracellular Oxidation

Measurement of intracellular oxidation is based on the methods described on the basis of oxidation of the cell-permeant fluorochrome 2′,7′-dichlorodihydrofluorescein diacetate (H2DCFDA) (Life Technologies [Burlington, Canada], catalog no. D-399) (17). FHS 74 Int cells were grown in 96-well polystyrene culture plates. Cells in each well were exposed to either 100 μL DMEM (Sigma-Aldrich, catalog no. D0572), which constituted the basal cell culture medium, or 100 μL of indicator loading mix (50-μg H2DCFDA in 50-μL DMSO + 950-μL DMEM to reach a concentration of 10 μM). The samples were then incubated in the dark for 60 minutes under 5% CO2 and 95% air at 37°C. All fluids were then aspirated from wells and cells exposed to 200 μL of BM ± supplements at a dilution of 1:100 in PBS, except for wells to which only PBS was added. All of the wells were again aspirated, and 200 μL of DMEM was added. Fluorescence readings were taken at 0-, 1-, and 2-hour intervals on a Fluoroskan Ascent fluorometer (ThermoLab Systems, Helsinki, Finland) with an excitation/emission filter pair of 485/527 nm. As per standard protocol in our laboratory, which has been validated in earlier published results using the plate reader assays described herein, each treatment was based on a replicate of 8 wells in 96-well culture plates.

Lactate Dehydrogenase (LDH) Cytotoxicity Assay

HT-29MTX cells were grown to confluency in Falcon polystyrene tissue culture–coated 96-well plates (VWR International, Burlington, Canada). After aspiration of culture medium, 100 μL of PBS was transferred to each well, and 30 μL of test media (PBS ± BM ± HMF) was added to a final diluted concentration of 1:100. After 3 hours of incubation at 37°C, test media were aspirated and 50 μL of this media was pipetted into wells on a 96-well plate (8 wells per condition), followed by 50 μL of LDH reaction mixture (Sigma-Aldrich MAK066), as per manufacturer's directions. Cell culture plates were incubated at room temperature for 30 minutes, protected from light, after which 50 μL of Stop Solution was added to each well. Absorbance was measured at 490 and 630 nm (for background) on an Opsys multiwell plate reader (Dynex Technologies, Chantilly, VA). LDH activity was expressed as a ratio of absorbances at 490/680 nm.

Live/Dead Epi-Fluorescent Imaging

Numbers of live versus dead cells were assessed 3 hours after exposure to test media described above using the fluorochromes calcein AM and ethidium D homodimer 1. Briefly, HT-29MTX cells grown in 6-well polystyrene Petri dishes were exposed for 3 hours to test media at dilutions of 1:100, after which test media were aspirated and 1 mL of cell culture medium containing 2 μmol/L of Calcein AM and 2 μmol/L of Eth-D-1 was added to the cells. Both ethidium and calcein were prepared from powdered stocks reconstituted in dimethylsulfoxide (DMSO; Sigma catalog no. D2650). The final concentration of DMSO was 2/1000 μL of PBS. Samples were kept in the dark at 37°C for 30 minutes, after which media were aspirated and replaced with fresh PBS. Live (green emission) and dead (red emission) cells were imaged on a Zeiss IM35 inverted microscope using a 20× Neofluar objective (NA = 0.5) using the appropriate excitation and emission filters.

Quantitative Gene Expression

The genomic effects of in vitro–digested BM ± HMF on FHs 74 Int cells were quantified using the SABiosciences Human Oxidative Stress RT2 Profiler (PAHS 065Z). In addition to including Bnip3, this kit utilizes quantitative real-time PCR (qRT-PCR) to profile the expression of 84 genes related to oxidative stress including peroxidases and genes involved in reactive oxygen species metabolism, such as oxidative stress responsive genes, and genes involved in superoxide metabolism such as superoxide dismutases. Cells grown in 25-cm2 flasks were exposed for 3 hours to either BM or BM + HMF, digested as described. After washing 3 times with PBS, messenger RNA was collected using the RT2 qPCR-Grade RNA Isolation Kit (catalog no. PA-001) and, following spectrophotometric checks for purity and quantity, first strand synthesis was done with the SABiosciences RT2 First Strand Kit (catalog no. C-03). Real-time PCR was performed in triplicate for each treatment, and plates were read on a BioRad MiyQ thermalcycler. Fluorescence signals from qRT-PCR reactions are reported as the mean ± standard error of 3 reactions. If after 37 cycles a sample's signal did not rise above the threshold value, it was considered undetectable. Genomic DNA contamination controls, positive PCR controls, and standard housekeeping genes were all included in each array plate.

Western Blot Analysis

Cells were grown in 10 cm of Petri plates and exposed to test media for 3, 6, or 12 hours, after which they were washed 3 times in PBS. For detecting total protein, HT29-MTX cells were harvested in NP-40 lysis buffer containing protease inhibitors. Protein cell lysates (20–25 μg) were denatured for 5 minutes at 100°C and resolved on a 4% to 12% sodium dodecyl sulfate polyacrylamide gel at 80 V for 20 minutes followed by 100 V for 1 hour. The protein lysates were electrophoretically transferred to a nitrocellulose (Amersham Biosciences, Baie-d’Urfé, QC, Canada) membrane or a polyvinylidene difluoride membrane at 100 V for 1 hour at 4°C or at 20 V overnight at room temperature. Membranes were subjected to Ponceau S staining to visualize protein bands and blocked for 1 hour in 5% powdered skim milk in tris-buffered saline-TWEEN (150-mmol/L NaCl, 50-mmol/L tris–HCl, 0.3% Tween-20, and pH 7.4) at room temperature. Membranes were incubated with primary Bnip3 immunoglobulin G (IgG) (PA1-46218; Thermo Scientific, Waltham, MA) at a 1:200 dilution overnight at 4°C. Following incubation, membranes were washed 3 times with 1× tris-buffered saline-TWEEN for 10 minutes each and incubated with specific secondary antibodies for 1 hour at room temperature. A chemi-36 luminescence reaction using horseradish peroxidase–conjugated antibody with enhanced chemiluminescence reagents (Amersham Biosciences) was used to detect bound proteins. To detect equal protein loading, membranes were probed with α-actin IgG (Actin [1-19]:sc-1616; Santa Cruz Biotechnology, Inc) at a dilution of 1:1000 in 2.5% BSA.

Bnip3 Expression and Inhibition

To confirm the role of Bnip3 in HT29-MTX cell death, we overexpressed full-length Bnip3 or an endogenous inhibitor of Bnip3 function, Bnip3ΔEx3. This is an alternative splice variant, which is missing the third exon, and has been earlier shown to completely abrogate cell death and mitochondrial perturbations induced by Bnip3 (18). Full-length and ΔEx3 Bnip3 were PCR amplified from complementary DNA (cDNA) generated from total RNA isolated from NIH 3T3 mouse embryonic fibroblasts. Restriction enzyme sites were integrated into the amplicon using the following primers: forward 5′-GCCGGAATTCATGTCGCAGAGCGGGAG-3′ (EcoRI), reverse full-length 5′-CGGCGCTCGAGTCAGAAGGTGCTAGTGGAAGTTGTC-3′ (XhoI), and reverse Bnip3ΔEx3 5′-CCGGCGCTCGAGTCAGGATACTTTCAACTTCTCTTCTTCTCTC-3′ (XhoI). Following digestion, the amplicon was ligated into pcDNA3 (EcoRI-XhoI) containing an in-frame N-terminus Myc-tag for full-length Bnip3 or an in-frame N-terminus HA-tag for Bnip3ΔEx3.

Statistical Analysis

Data were expressed as mean SD of different independent experiments. Differences between groups in 96-well plate assays (LDH, oxidative stress assays) were analyzed statistically with 2-way analysis of variance (ANOVA) followed by the Turkey HDS post hoc test for multiple comparisons. The level of P ≤ 0.05 was considered statistically significant. Eight biological replicates were used for all of the microplate assays, as validated in other published studies from our laboratory (16). For Live/Dead assays, representative microscope field were digitally captured, and a replicate of 4 fields were chosen for manual cell counting of live versus dead cells based on similar studies. qRT-PCR macroarrays were analyzed using ΔΔCt method with normalization of the raw data to housekeeping genes. Expression of each gene was normalized using the mean expression of 5 housekeeping genes. Raw data was analyzed using the integrated web-based automated software for RT2 Profiler PCR Array Data Analysis available through SABiosciences ( The relative gene expression was calculated by comparative threshold cycle (ΔΔCt) method from uploaded raw threshold cycle data using average cycle threshold values; fold changes were expressed as the difference in average gene expression of control (BM treated) cells. For data analysis, the RT2 Profiler PCR Array software package was used and statistical analyses performed (n = 3 based on software analytical requirements). This package uses ΔΔCT-based fold change calculations and the Student t test to calculate 2-tail, equal variance P values.


Intracellular Oxidation

Two hours after exposure to BM + HMF, intracellular oxidation significantly increased in FHS 74 Int cells compared with the BM control, as evidenced by increased fluorescence intensity (Fig. 1). Fluorescence levels progressively increased after 2 hours in all of the cell cultures, an effect attributable to basal oxidation of the fluorochrome probe.

Effects of BM ± HMF on intracellular oxidation in FHS 74 Int cells at 0, 1, and 2 hours after exposure to in vitro–digested feeds. BM and BM + HMF admixtures diluted 1:5 in PBS and were loaded with 1 μg/μL H2DCFDA. Bars with different letters are significantly different from each other (P < 0.05). AAPH, a free radical–generating azo compound, was used as a positive control to induce intracellular oxidation. Blank, Cells without H2DCFDA. AAPH = 2,2′-azobis(2-amidinopropane) dihydrochloride; BM = breast milk; H2DCFDA = 2′,7′-dichlorodihydrofluorescein diacetate; HMF = human milk fortifier; PBS = phosphate-buffered saline.

Cell Damage/Cell Death

In vitro–digested BM + HMF significantly increased cell damage, as measured via LDH release (Fig. 2). This effect was first detected after 3 hours of incubation, and exposure for 6 and 24 hours to BM + HMF showed progressively more cell damage. Incubation of HT-29MTX cells with BM alone also showed increasing LDH release at 3 hours of incubation, although this was significantly less than the corresponding BM + HMF treatments. This is consistent with the effects of BM + HMF on cell death (Fig. 3A and B). After 3 hours of incubation, BM + HMF significantly reduced cell viability by ∼15%. Evidence of cell detachment from the culture plate surface was noted in both the BM and BM + HMF treatments.

LDH assay at 3, 6, and 24 hours of incubation shows cell damage increases with HMF exposure and with longer exposure times: a, b, c indicates significant changes between BM and PBS control; A, B, C indicates significant changes between BM + HMF and BM. Two-way ANOVA (P < 0.05, n = 8 wells per replicate). ANOVA = analysis of variance; BM = breast milk; Ctrl = control; HMF = human milk fortifier; LDH = lactate dehydrogenase; PBS = phosphate-buffered saline.
A and B, HT-29MTX Live/Dead assay. Cells were treated with PBS or BM ± HMF. A, Green fluorescence indicates live cells; red fluorescence indicates cell death. B, Statistical differences were determined by 2-way ANOVA (P < 0.05, n = 4 microscope fields). Cell viability is significantly decreased after exposure to BM + HMF after 3 hours of incubation (star). ANOVA = analysis of variance; BM = breast milk; HMF = human milk fortifier; PBS = phosphate-buffered saline.

Bnip3 Expression

In our in vitro digestion/enterocyte model, HMF induced significant fold changes in the expression of several antioxidant genes (Fig. 4), including a zinc transporter metallothionein 3 (MT3) and Bnip3. Bnip3 is a member of the BCL2 family, which is activated by oxidative stress and is capable of triggering either apoptotic or necrotic cell death. The identities of other genes significantly upregulated or downregulated in this assay are specified in Table 1.

Bnip3 protein levels increase after exposure to either BM or BM + HMF. BM = breast milk; Bnip3 = BCL2/adenovirus E1B 19 kDa protein-interacting protein; HMF = human milk fortifier; PBS = phosphate-buffered saline.
ARE-driven genes significantly upregulated by in vitro–digested HMF + BM

Consistent with the effects of HMF on gene expression, Bnip3 proteins levels also increased after exposure to either BM or to BM + HMF, compared with the saline controls. BM increased Bnip3 expression ∼2-fold compared with the saline control, whereas BN + HMF induced an ∼4-fold increase in protein expression (Fig. 5).

qRT-PCR array illustrating significant (P < 0.05, n = 3) fold differences in ARE-driven gene expression in FHs-74 intestinal cells treated with BM + HMF. Bnip3 circled in red. BM = breast milk; Bnip3 = BCL2/adenovirus E1B 19 kDa protein-interacting protein; qRT-PCR = quantitative real-time PCR.

Bnip3 Inhibition and Misoprostol Rescue

HT-29-MTX cells transfected with Bnip3 had significantly lower viability compared with nontransfected cells (Fig. 6A and B). This effect was abolished by cotransfection with the Bnip3 exon 3 splice variant (Bnip3ΔEx3). Transfection with the splice variant alone had no significant effects. Misoprostol, a prostaglandin analogue, used clinically in treating inflammatory bowel disorders in adults, significantly reduced HMF-induced cell death by ∼55%, compared with the saline control (Fig. 7A and B). This effect was observed at 3 and at 6 hours (data not shown), time points at which Bnip3 protein expression was elevated.

A and B, Effect of Bnip3 and Bnip3ΔEx3, an endogenous inhibitor of Bnip3, on viability of HT29-MTX cells. A, Cells were treated with Calcein AM and Ethidium Homodimer. B, Cell viability as a percentage change from control (mean ± SD). Statistical differences were determined by 2-way ANOVA (P < 0.05; n = 4 microscope fields). Star indicates significant difference (n = 3 replicates, P < 0.05). ANOVA = analysis of variance; Bnip3 = BCL2/adenovirus E1B 19 kDa protein-interacting protein.
A and B, Misoprostol restores cell viability when incubated with BM + HMF. A, Cells were treated with Calcein AM and Ethidium Homodimer. B, Cell viability as a percentage change from control (mean ± SD). Statistical differences were determined by 1-way ANOVA (P < 0.05; n = 4 microscope fields). Star indicates significant difference (P < 0.05). ANOVA = analysis of variance; BM = breast milk; EtOH = ethanol; HMF = human milk fortifier; Miso = misoprostol; PBS = phosphate-buffered saline.


Our data support the hypothesis that HMF increases intracellular oxidation, cell death, and Bnip3 expression in vitro. Increased LDH release, which is taken to be indicative of cell injury or plasma membrane damage, was consistent with HMF-induced cell death. We cannot state that the observed increase in Bnip3 expression is causally associated with HMF-induced cell death, although the links between Bnip3 expression and either apoptotic or necrotic cell death are well established. Bnip3 is a member of the Bcl-2 gene family and has been implicated in virtually all forms of programmed cell death (19,20). The prototypical family member, Bcl-2, plays a survival role and is localized to both the mitochondria and the endoplasmic reticulum, where it gates the inositol trisphosphate receptor (IP3-receptor) and prevents excessive calcium transfer to the mitochondria (21). We interpret our findings that Bnip3 overexpression increased cell death and that this effect was abolished by the Bnip3 exon 3 splice variant, to be confirmatory of the putative effect of the Bnip3 gene product on cell death, although we cannot say whether this is via apoptotic or necrotic pathways. Furthermore, this is consistent with the established association between Bnip3 and NEC in preterm infants (15).

Although Bnip3 gene expression has been an endpoint of interest in our study, we have also shown other ARE-driven genes, which are upregulated by HMF. Most notable among these is MT3, which was upregulated almost 40-fold by HMF. MT3 is a gene whose product is a cysteine-rich protein that is involved in zinc transport, metal detoxification, and protection against oxidative stress (21,22). It is beyond the scope of the current paper to further analyze or discuss the other genes identified through our macroarray screen, although it is interesting to note that, in a recent study in which piglets were fed cow-milk formula, the expression of metalothioneins was increased in liver and intestinal tissue of piglets fed cow's milk (23). Although the elevation of MT3 gene expression in our study is consistent with its established role in oxidative stress, it may also be as a result of differences in trace element composition and bioavailability between BM and BM + HMF. Conversely, upregulation of zinc transporters may confer a survival advantage to enterocytes: zinc prevents iron-induced oxidative damage in rats, and it also enhances the Bcl-2/Bax ratios and reduces caspase-3 activity, which can decrease apoptosis in Caco-2 cells (23).

Our findings are congruent with another study, which demonstrated that digested cow's-milk formula, but not digested BM, causes death of intestinal cells in vitro (24). In this study by Penn et al (24), primary rat intestinal epithelial cells, human neutrophils, and bovine aortic endothelial cells were incubated with a wide range of term and preterm formulae digested with either lipase and/or protease. Interestingly, these investigators found that free fatty acid (FFA) concentration was significantly increased in formula versus HM and that pretreatment with lipase inhibitor significantly decreased FFA and cell death. Penn et al (24) speculate that the observed cytotoxicity may be a result of the creation and transport of unbound FFAs. In our study, we chose a specific fortifier commonly used in our local NICUs, and we also focused on one of the likely cell death pathways. Taken together, the findings from the 2 studies are congruent. We cannot state which of the many components of HMF are responsible for the observed effects in our study, and we are currently isolating fractions of digested HMF + BM to determine which of these is the bioactive component resulting in either increased oxidative stress or cell death, or in the genomic effects observed herein.

Our additional finding that misoprostol rescued cells from the effects of BM + HMF is consistent with other studies showing a protective effect of this prostaglandin E1 (PGE1) analogue in intestinal inflammation (25). Furthermore, other investigators have shown that a combination of melatonin and PGE1 has cytoprotective and healing effects in a rodent model of NEC (26). The presumed mechanisms of action of PGE1 are diverse, although a plausible mechanism of action of misoprostol would be that it phosphorylates with protein kinase C, which in turn can prevent mitochondrial PTP opening that is concomitant with some apoptotic pathways.

Although the results reported herein do not definitively link HMF, Bnip3, and NEC, they do suggest a possible association between enteral feeds and dysregulated intestinal function. Furthermore, it may be useful in evaluating potential nutritional interventions aimed at mitigating intestinal cell death. We also have not identified specific components in digested BM + HMF, which elicit the observed responses, nor have we investigated the wide variety of fortifiers and admixtures currently in clinical use. The work of Penn et al (24) certainly suggest that FFAs may be a causal factor, but a thorough analysis of BM + HMF fractions and/or assessment of the digested milk metabolome may yield clues as to the identity of other bioactive molecules. It is worthwhile to note that both cell death and Bnip3 protein were also elevated after BM treatments, although not to the same extent as BM + HMF. One causal factor may be endogenous lipid peroxides found in BM (27,28), which in turn may trigger Bnip3 production. Alternative explanations could include the effects of other in vitro conditions in our assays, which may induce the formation of other free radicals in native BM.

As with any in vitro experiment, caution needs to be exercised when generalizing results to the in vivo state. Furthermore, the current study focused on a commercially available HMF and thus may not be reflective of all the fortifiers. In addition, one limitation of these experiments is that we utilized BM samples, which had been frozen and refrozen. It is possible that repeat freezing may compromise results, although BM + HMF was always compared with HMF, which was similarly refrozen. Finally, the current work does not elucidate the specific organic and/or inorganic moieties responsible for the observed effects. These caveats notwithstanding, our findings are consistent with recent clinical studies reporting increased risk of intestinal complications in VLBW preterm infants receiving higher levels of fortified feeds, and at the very least suggest a need to further explore the physiologic and genomic effects of BM fortification. Moreover, we anticipate that these findings will add to the current discussion on the relative benefits of bovine-based HMFs versus humanized fortifiers and supplementation with pasteurized donor milk (29,13,30).

In future work, we anticipate screening fractionated HMF-fortified BM to identify specific components involved in the effects reported herein. We speculate that FFAs may also be elevated in BM + HMF compared with BM alone, and we are currently investigating this and other components of digested feeds. Equally important will be testing of HMFs, humanized fortifiers, and donor milk for their relative impacts on other ARE-driven genes and components of the innate immune system, such as the Toll-like receptors. We suggest that these will be worthwhile targets for further investigation of the nutrigenomic effects of preterm infant feeds in general and will permit a more empirical approach to formulating preterm infant feeds.

The American Academy of Pediatrics (31) supports the use of HMFs in preterm infant feeding (30). Indeed, the dangers of not fortifying preterm infant feeds are well known (32). Our data, however, suggest that there is a need to further optimize HMFs, based on their cellular and genomic effects. In this manner, we should be better able to minimize any potentially harmful effects, especially in the VLBW preterm infant who is at greater risk of diseases of prematurity such as NEC.


1. Ahrabi AF1, Schanler RJ. Human milk is the only milk for preemies in the NICU!. Early Hum Dev 2013; 89 (suppl 2): 2013; S51–S53.
2. Johnson TJ1, Patel AL, Bigger HR, et al. Economic benefits and costs of human milk feedings: a strategy to reduce the risk of prematurity-related morbidities in very-low-birth-weight infants. Adv Nutr 2014; 5:207–212.
3. Ziegler EE. Human milk and human milk fortifiers. World Rev Nutr Diet 2014; 110:215–227.
4. Lucas A, Fewtrell MS, Morley R, et al. Randomized outcome trial of human milk fortification and developmental outcome in preterm infants. Am J Clin Nutr 1996; 64:142–151.
5. Amione A, Rovet J, Ward W, et al. Growth and body composition of human milk-fed premature infants provided with extra energy and nutrients early after hospital discharge: 1-year follow-up. J Pediatr Gastroenterol Nutr 2009; 49:456–466.
6. Kuschel CA, Harding JE. Multicomponent fortified human milk for promoting growth in preterm infants (Review). Cochrane Database Syst Rev 2009; 1:1–25.
7. Tillman S, Brandon DH, Silva SG. Evaluation of human milk fortification from the time of the first feeding: effects on infants of less than 31 weeks gestational age. J Perinatol 2012; 32:525–531.
8. Friel JK1, Diehl-Jones B, Cockell KA, et al. Evidence of oxidative stress in relation to feeding type during early life in premature infants. Pediatr Res 2011; 69:160–164.
9. Perrone S, Tataranno ML, Santacroce A, et al. The role of oxidative stress on necrotizing enterocolitis in very low birth weight infants. Curr Pediatr Rev 2014; 10:202–207.
10. Rees CM, Eaton S, Pierro A. National prospective surveillance study of necrotizing enterocolitis in neonatal intensive care units. J Pediatr Surg 2010; 45:1391–1397.
11. Ramani M, Ambalavanan N. Feeding practices and necrotizing enterocolitis. Clin Perinatol 2014; 40:1–10.
12. Sullivan S, Schanler RJ, Kim JH, et al. An exclusively human milk-based diet is associated with a lower rate of necrotizing enterocolitis than a diet of human milk and bovine milk–based products. J Pediatr 2010; 156:562–567.
13. Cristofalo EA, Schanler RJ, Blanco CL, et al. Randomized trial of exclusive human milk versus preterm formula diets in extremely premature infants. J Pediatr 2013; 163:1592–1595.
14. Chernikov VP. Role of hypoxia-induced transcriptional factor HIF-1 in the regulation of enterocytic metabolism. Arkh Patol 2008; 70:6–9.
15. Zamora R1, Vodovotz Y, Betten B, et al. Intestinal and hepatic expression of BNIP3 in necrotizing enterocolitis: regulation by nitric oxide and peroxynitrite. Am J Physiol Gastrointest Liver Physiol 2005; 289:G822–G830.
16. Yao L, Friel JK, Suh M, et al. Antioxidant properties of breast milk in a novel in vitro digestion/enterocyte model. J Pediatr Gastroenterol Nutr 2010; 50:670–676.
17. Lautraite S, Bigot-Lasserre D, Bars R, et al. Optimisation of cell-based assays for medium throughput screening of oxidative stress. Toxicol In Vitro 2003; 17:207–220.
18. Gang H, Hai Y, Dhingra R, et al. A novel hypoxia-inducible spliced variant of mitochondrial death gene Bnip3 promotes survival of ventricular myocytes. Circ Res 2011; 108:1084–1092.
19. Heath-Engel HM, Chang NC, Shore GC. The endoplasmic reticulum in apoptosis and autophagy: role of the BCL-2 protein family. Oncogene 2008; 27:6419–6433.
20. Imazu T, Shimizu S, Tagami S, et al. Bcl-2/E1B 19 kDa-interacting protein 3-like protein (Bnip3L) interacts with bcl-2/Bcl-xL and induces apoptosis by altering mitochondrial membrane permeability. Oncogene 1999; 18:4523–4529.
21. Schulkens IA, Castricun KCM, Wiejers EM, et al. Expression, regulation and function of human metallothioneins in endothelial cells. J Vasc 2014; 51:231–238.
22. Hanson CJ1, Bootman MD, Roderick HL. Cell signalling: IP3 receptors channel calcium into cell death. Curr Biol 2004; 14:R933–R935.
23. Kilari S, Pullakhandam R, Nair KM. Zinc inhibits oxidative stress-induced iron signaling and apoptosis in Caco-2 cells. Free Radic Biol Med 2010; 48:961–968.
24. Penn AH, Alshuler AE, Small JW, et al. Digested formula but not digested fresh human milk causes death of intestinal cells in vitro: implications for necrotizing enterocolitis. Pediatr Res 2012; 72:560–567.
25. Takeuchi K1, Tanaka A, Kato S, et al. Roles of COX inhibition in pathogenesis of NSAID-induced small intestinal damage. Clin Chim Acta 2010; 411:459–466.
26. Cekmez F, Cetinkaya M, Tayman C, et al. Evaluation of melatonin and prostaglandin E1 combination on necrotizing enterocolitis model in neonatal rats. Regul Pept 2013; 184:121–125.
27. Turoli D, Testolin G, Zanini R, et al. Determination of oxidative status in breast and formula milk. Acta Paediatr 2004; 93:1569–1574.
28. Bertino E, Giribaldi M, Baro C, et al. Effect of prolonged refrigeration on the lipid profile, lipase activity, and oxidative status of human milk. J Pediatr Gastroenterol Nutr 2013; 56:390–396.
29. Quigley MA1, Henderson G, Anthony MY, et al. Formula milk versus donor breast milk for feeding preterm or low birth weight infants. Cochrane Database Syst Rev 2007; 4:CD002971.
30. Unger S, Gibbins S, Zupancic J, et al. DoMINO: Donor milk for improved neurodevelopmental outcomes. BMC Pediatr 2014; 14:123.
31. Section on Breastfeeding. Breastfeeding and the use of human milk. Pediatrics 2012; 129:e827–e841.
32. Embleton NE, Pang N, Cooke RJ, et al. Postnatal malnutrition and growth retardation: an inevitable consequence of current recommendations in preterm infants? Pediatrics 2001; 107:270–273.

Bnip3; breast milk; cell death; human milk fortifier; intestinal cells

© 2015 by European Society for Pediatric Gastroenterology, Hepatology, and Nutrition and North American Society for Pediatric Gastroenterology,