Main

Oxidative stress is present in very-low-birth-weight preterm newborns (1) and is suspected to play a key role in several complications of prematurity. Although the contribution of oxygen supplementation is recognized (1), total parenteral nutrition (TPN) is also an important source of oxidants. Peroxides in TPN are derived mainly from the reduction of dissolved oxygen by electron donors such as vitamin C, amino acids, and lipids (2). These reactions are catalyzed by ambient light through photosensitive riboflavin (2). Complete photoprotection reduced the generation of peroxides by half (1); however, this is not practical in the neonatal unit. During reactions occurring in TPN, several reactive components are produced, including hydrogen peroxide (3), lipid hydroperoxides (4), and 4-hydroxynonenal (4-HNE; ref. 5).

In vivo, 4-HNE is detoxified by conjugation with glutathione (6). Preterm infants, who have an immature antioxidant system (7) and low glutathione levels (8), are at risk to develop several pathological complications, such as bronchopulmonary dysplasia, retinopathy of prematurity, and hepatocellular injury (9,10,11). Therefore, reactive molecules, such as 4-HNE, can interfere with proteins (12), lipids (13), and DNA (14); disrupt the cellular redox balance (15); act on cell signaling; and induce inflammatory response by activating the nuclear factor-κB pathway (16).

It is well known that human milk (HM) promotes antioxidant capacity in the newborn as compared with artificial milk (AM) (17). This property is associated with a greater free radical–scavenging capacity (18,19). However, at the time when very-low-birth-weight premature newborns are nourished with TPN, their gastrointestinal tract is too immature to receive significant amounts of HM. Recently, Tsopmo et al. (20) isolated two hexapeptides from enzymatic hydrolysates of HM possessing a high free radical–scavenging capacity in vitro (20). We hypothesized that the addition of those peptides to TPN could be a valuable alternative to light protection. Therefore, the aims of this study were to assess the scavenging capacity of these peptides in the prevention of oxidation of TPN components and to test the ability of peptides added to TPN to (i) reduce the hepatic level of oxidative stress markers, especially 4-HNE, and (ii) to prevent the induction of two proinflammatory cytokines: tumor necrosis factor-α (TNFα) and interleukin-1 (IL-1). Since these peptides were derived from HM, animals were nourished or not with HM in order to generate reference values.

Results

The levels of peroxides generated in TPN solutions ( Figure 1 ) were lower (F(1,12) = 10.2; P < 0.01) in the presence of peptide-A (YGYTGA) without any difference (F(1,12) = 0.5) at varying concentrations (10–40 mmol/l). Peptide-B (ISELGW) was without effect (F(1,12) < 1.4).

Figure 1
figure 1

Concentration of peroxides in TPN solutions in the presence of peptide-A or peptide-B. (a) The concentration of peroxides was lower in TPN solution in the presence of peptide-A (YGYTGA) independent of the concentration of peptide. (b) Peptide-B (ISELGW) was without effect on the generation of peroxides in TPN solution; mean ± SEM, n = 4; **P < 0.01. TPN, total parenteral nutrition.

Full size image

An interaction (F(1,24) > 8.9; P < 0.01) between the parameters influencing the hepatic redox potential of glutathione ( Figure 2a ) allowed us to reanalyze the data in the presence or absence of intravenous H2O2. In the absence of peroxide, redox potential was less oxidized (F(1,24) = 18.5; P < 0.01) in the HM than the AM group, but no difference was observed (F(1,24) = 0.01) between the HM and the control (C) groups. In animals infused with H2O2, there was no difference between the HM and AM groups (F(1,24) = 1.1). However, redox potential was more oxidized in the HM and AM groups as compared with the C group (F(1,24) = 16.9; P < 0.01).

Figure 2
figure 2

Redox potential of glutathione and levels of F2α-isoprostanes in liver of animals infused with ± H2O2 and fed regular chow, human milk, or artificial milk. (a) In the absence of peroxide, redox potential was more reduced in the HM group than in the AM group. In animals infused with peroxides, there was no difference in redox potential between the HM and AM groups; both were more oxidized than the C group. (b) In the absence of peroxide, there was no difference between the groups. In animals infused with peroxides, F2α-isoprostane level was higher in the AM group than in the HM and C groups; mean ± SEM, n = 4–6; **P < 0.01. AM, animals fed artificial milk; C, (control) animals fed regular chow; HM, animals fed human milk; ± H2O2, animals infused (open bars) or not infused (filled bars) with solution containing H2O2 at similar concentration as measured in TPN solutions; TPN, total parenteral nutrition.

Full size image

The diet effect on the redox potential seems to be linked to changes in the levels of the reduced form of glutathione (GSH) ( Table 1 ), as the results of statistical analyses were similar: (i) interaction between parameters (F(1,24) > 9.1; P < 0.01); (ii) in the absence of H2O2, GSH levels were higher in the HM group than they were in the AM group (F(1,24) = 19.0; P < 0.01), without difference (F(1,24) = 0.04) with the C group; and (iii) in the presence of H2O2, GSH levels did not differ between the HM and AM groups (F(1,24) = 0.2), but were lower than those of the C group (F(1,24) = 16.7; P < 0.01). Conversely, there was no interaction (F(1,24) < 2.6) between the parameters for the disulfide form of glutathione (GSSG). Its level, similar in the HM and AM groups (F(1,24) = 0.1), was lower with the C group (F(1,24) = 5.0; P < 0.05). There was no effect of infused peroxide (F(1,24) = 2.4).

Table 2 Glutathione levels in liver from animals infused with TPN with or without peptide-A or peptide-B
Full size table

The interaction (F(1,24) = 6.2; P < 0.05) between the parameters influencing F2α-isoprostane levels ( Figure 2b ) suggests that the impact of diet was dependent on the presence of H2O2. In the absence of H2O2, there was no difference between the HM, AM, and C groups (F(1,24) < 0.2). In contrast, in the presence of infused H2O2, the F2α-isoprostane level was higher (F(1,24) = 15.7; P < 0.01) in the AM group than in the HM and C groups. There was no difference between the HM + H2O2 and the C + H2O2 groups (F(1,24) = 0.2). The hepatic levels of glutathionyl 1,4-dihydroxynonenal (GS-HNE) were similar (F(1,22) < 3.3) in these six groups ( Figure 2 ) with a mean value of 2.4 ± 0.2 pmol/mg protein (n = 28).

As compared with the control TPN (without light protection), peptide supplementations reduced (F(1,14) = 12.2; P < 0.01) the hepatic redox potential ( Figure 3a ). There was no difference between peptides (F(1,14) = 0.2). The difference in redox potential is linked with a higher level (F(1,14) = 7.6; P < 0.05) of GSH ( Table 2 ) in animals infused with TPN enriched with peptides. GSSG levels were not influenced by the presence of peptides (F(1,14) < 1.1).

Figure 3
figure 3

Oxidative stress markers in liver from animals infused with TPN with or without peptide-A or peptide-B. (a) Hepatic redox potential was more reduced if animals were infused with TPN containing peptide-A or peptide-B than in animals on TPN without peptide. The two peptides had similar effects. (b) There was no statistically significant effect of peptides on the hepatic levels of F2α-isoprostane. (c) Hepatic GS-HNE (glutathione-4-hydroxynonenal adducts) levels were lower in animals infused with TPN containing peptide-A or peptide-B than in animals receiving TPN without peptides. The two peptides had similar effects; mean ± SEM, n = 5–6. **P < 0.01. pept., peptide; pept.-A, YGYTGA; pept.-B, ISELGW; TPN, total parenteral nutrition.

Full size image
Table 3 5′–3′ Sequences of primers used for TNFα, IL-1, and housekeeping gene 18S
Full size table

Levels of F2α-isoprostane ( Figure 3b ) did not vary between groups (F(1,14) < 2.0), whereas GS-HNE levels ( Figure 3c ) were lower (F(1,14) = 9.2; P < 0.01) in liver from animals infused with TPN enriched with peptide-A or peptide-B as compared with liver from animals receiving TPN without peptide. There was no difference in GS-HNE levels between animals infused with peptide-A and those infused with peptide-B (F(1,14) = 0.2).

Hepatic mRNA of TNFα and of IL-1 ( Figure 4 ) were lower (F(1,14) > 4.9; P < 0.05) when TPN was supplemented with peptides. The effect of peptide-A was similar to that of peptide-B on both cytokines (F(1,14) < 1.4).

Figure 4
figure 4

mRNA levels of TNFα and IL-1 in liver from animals infused with TPN with or without peptide-A or peptide-B. The mRNA levels of (a) TNFα and (b) IL-1 were lower in liver from animals infused with TPN containing peptide-A or peptide-B than in animals on TPN without peptides. The two peptides exerted similar effect; mean ± SEM, n = 5–6. *P < 0.05. IL, interleukin; pept., peptide; pept.-A, YGYTGA; pept.-B, ISELGW; TNF, tumor necrosis factor; TPN, total parenteral nutrition.

Full size image

Discussion

The main finding of the study is that a nutritional alternative for the prevention of oxidative stress caused by peroxide from TPN in premature newborns is possible. The results demonstrate that the peptides from HM have properties that may be useful for this population.

This study also provides evidence that enteral HM is better that AM in counteracting free radical injuries (lower F2α-isoprostane levels) induced by infusion of peroxides at concentrations measured in TPN solutions. In fact, AM possessed low antioxidant capacity whereas HM had a similar effect to that of regular chow. As compared with a previous report with the same animal model infused with H2O2 (21), measured hepatic levels of F2α-isoprostane suggest that AM does not possess antioxidant effect. Our results are in agreement with the observation that urinary levels of F2α-isoprostanes are lower in newborn infants fed HM as compared with those nourished with formula (18). However, as compared with regular chow, there was no advantage of HM as suggested by the absence of a beneficial effect on hepatic redox potential in animals infused with peroxides. Further comparisons of chemical compounds between chow and the milks used here could provide new information to improve the antioxidant properties of HM or AM.

If the neonates cannot be orally fed, they require TPN. When peptide-A was added to TPN, it reduced the generation of peroxides by half, similar to the results obtained with full light protection (2). This finding was not reproduced with peptide-B, suggesting that free radical–scavenging properties are peptide specific. Peptide-A (YGYTGA) contains two phenolic groups from tyrosyl residues (Y) that can serve as radical scavengers (22) whereas the free radical–scavenging property of the peptide-B (ISELGW) was reported to be linked with the tryptophan residue (W) (23). As the generation of H2O2 in TPN is secondary to the reduction of superoxide anion (2), a further study would challenge the superoxide anion–scavenging capacity of these peptides.

The in vivo part of this study demonstrates that peptide-A and peptide-B have significant effects on markers of oxidative stress and on mRNA of proinflammatory cytokines. Furthermore, these effects are different from those observed in animals fed HM. The route of administration (parenteral vs. enteral), the dose, or intestinal absorption can explain the difference. For instance, the hepatic redox potentials calculated in animals infused with H2O2 fed with AM ( Figure 2a ), or with TPN without light protection and additional peptides ( Figure 3a ) were similarly oxidized. Supplementation of TPN with peptides protected this redox shift through an oxidized status whereas HM was without effect. The high GSH level in liver from the TPN + peptides groups, explaining their more reduced redox potential, suggests that peptides could favor glutathione synthesis. This provides a basis for future studies that will be important for the premature newborn population in whom the level of glutathione is low (7,8) although they have a normal capacity of synthesis (24,25). The fact that an early administration of amino acids in premature infants favored glutathione synthesis (24) supports the notion that the availability of amino acids may be a limiting step for this synthesis. Because both peptides contain glycine, one of the glutathione constituents, the impact of peptides on GSH levels may be derived from the hydrolysis of their amino acids. Because the maximum release of glycine from peptide is 1 mmol/l as compared with the 3 mmol/l from TPN, and because glycine is not known to be the limiting step for GSH synthesis, the probability of a greater amino acid availability explaining the higher GSH levels is not very probable. However, the hydrolysis hypothesis is nevertheless interesting. As mentioned earlier, the free radical–scavenging abilities of peptides may be related to their tyrosine or tryptophan content. The complete hydrolysis of peptides could provide 2 mmol/l tyrosine and 1 mmol/l tryptophan, which are considerably higher than the amount usually present in TPN solution: ~0.04 mmol/l for tyrosine and ~0.2 mmol/l for tryptophan.

On the other hand, the in vivo free radical–scavenging property of HM has been reproduced with peptides. Even if F2α-isoprostane levels did not reach statistically significant difference between groups, the hepatic levels of GS-HNE were lower in animals infused with TPN containing peptides as compared with those animals receiving only TPN. Both F2α-isoprostanes and HNE are derived from lipid peroxidation via free radical attack. In liver, HNE is detoxified during glutathionylation, in which the product, GS-HNE, is actively exported in bile (26). The fact that GS-HNE was very low (2.4 ± 0.2 pmol/mg protein) in liver from animals enterally nourished suggests that GS-HNE measured in animals on TPN (58.0 ± 9.9 pmol/mg protein) could be associated with a lower bile flow, a well-known complication of TPN administration such as cholestasis. Thus, by their antioxidant properties, the tested peptides could contribute to the reduction of the incidence of cholestasis in newborns on TPN. The observation of lower mRNA levels of proinflammatory cytokines—TNFα and IL-1—supports the notion that these peptides exert an in vivo antioxidant property. Because the presence of these proinflammatory cytokines is cited as part of hepatic TPN-related liver diseases such as cholestasis (27), the relation between HNE and bile flow is reinforced. However, as amino acids have a main impact on the reduction of bile flow (28), HNE might have only a partial impact on cholestasis, which may be related to its inflammatory properties. Indeed, cholestasis has been shown to be associated with bacterial infection (29), a condition associated with oxidative stress (30) and generation of lipid aldehydes such as HNE (31,32).

The aim of this study was to test the capacity of new hexapeptides to reduce oxidative stress associated with TPN. Our results show that the hexapeptides are involved in liver antioxidant actions that are accompanied by lower levels of expression of the two cytokines. Although the beneficial effect of these peptides on hepatic cholestasis is not yet proved, this study brings hope to neonatal units where the incidence of several complications of prematurity associated with oxidative stress is high. The results also demonstrate that the enrichment of TPN solutions with these isolated peptides from HM seems more prone to improve the antioxidant defenses of the neonates than whole HM. However, before the extrapolation of these data to the clinic, further studies should delineate the optimal peptide concentrations as well as the mechanisms involved.

Methods

Experimental Designs

Experiment 1: to test the peptides’ capacity to prevent the generation of peroxides in TPN. Peptide-A and peptide-B were identified in HM as described previously (20). The peptides (95% purity) were synthesized by GenWay Biotech (San Diego, CA). TPN solutions were prepared with 0, 10, 20, or 40 mmol/l peptide-A or peptide-B (peptides were added first) + 2.2% (w/v) amino acids (Travasol; Baxter, Mississauga, ON, Canada) + 8.7% dextrose + 1% multivitamins (Multi-12 Pediatrics, Sandoz, Boucherville, QC, Canada) + 1.5% (v/v) lipids (Intralipid, Fresenius Kabi, Toronto, ON, Canada). After 3-h incubation at room temperature under 66 foot-candles fluorescent light exposure, peroxide concentrations were measured.

Experiment 2: to test the effect of HM vs. AM on the prevention of the induction of oxidative stress by infused peroxides. Three-d-old male Hartley guinea pigs (Charles River, St Constant, QC, Canada; body weight = 104 ± 1 g, n = 26) received intravenous solutions (described in the “Intravenous protocol”) via a catheter secured in the right jugular vein (21). Infusions of 200–500 μmol/l H2O2 daily in this animal model induces hepatic lipid peroxidation as indicated by increased F2α-isoprostane levels (21). This amount of infused peroxides was comparable to the ~400 μmol/l peroxides (1) measured in TPN as previously reported (4) and measured in this study, of which 80% are H2O2 (4). At the infusion rate of 240 ml/kg/d, the animals received 40–120 μmol H2O2/kg/d (21). In this study ( Figure 5 ), H2O2 was administered in this range.

Figure 5
figure 5

Experimental design for the first in vivo protocol generating data for Figure 2. AM, animals fed artificial milk; HM, animals fed human milk.

Full size image

Intravenous protocol

  1. 1

    During the first 18 h following catheter placement, animals received intravenous solutions containing: 1 U/ml heparin + 0.45% (w/v) NaCl + 8.7% (w/v) dextrose ± 500 μmol/l H2O2 (infusion rate = 240 ml/kg/d). Thus, animals on H2O2 received 120 μmol H2O2/kg/d.

  2. 2

    From day 2 (d2) to d4, during the day (6:00 AM to 6:00 PM), animals received intravenous solutions containing: 1 U/ml heparin + 0.9% NaCl ± 755 μmol/l H2O2 (infusion rate = 150 ml/kg/d). Animals on H2O2 received 113 μmol H2O2/kg/d and were fed regular chow, HM, or AM (see the following).

  3. 3

    From d2–d4, during the night (6:00 PM to 6:00 AM), animals received intravenous solutions containing: 1 U/ml heparin + 0.45% NaCl + 5% dextrose ± 755 μmol/l H2O2 (infusion rate = 150 ml/kg/d). Animals on H2O2 received 113 μmol H2O2/kg/d.

Experimental groups

  1. 1

    Group C: from d2–d4, animals were fed regular guinea pig chow (2041-Teklad Global High Fiber Guinea Pig Diet; Harlan-Canada, Montreal, QC, Canada) and had free access to chow and water. Their intravenous solution was devoid of added H2O2.

  2. 2

    Group C + H2O2: animals were fed the same diet as group C except that the intravenous solutions contained H2O2.

  3. 3

    Group HM (provided from pooled human donors and lyophilized by J.K.F.): From d2–d4, every 2 h (five times a day), the mouth of the animal was filled with 1.5 ml of reconstituted milk (20 g in 100 ml water). Animals had free access to water. Their intravenous solution was devoid of H2O2.

  4. 4

    Group HM + H2O2: animals were treated as in the HM group, except that their intravenous solution contained H2O2.

  5. 5

    Group AM (prepared from Carnation Good Start; Nestle Canada, North York, ON, Canada): From d2–d4, every 2 h (five times a day), the mouth of the animal was filled with 1.5 ml of AM (20 g in 100 ml water). Animals had free access to water. Their intravenous solution was devoid of H2O2.

  6. 6

    Group AM + H2O2: animals were treated as in the AM group, except that their intravenous solutions contained H2O2.

Experiment 3: to analyze the in vivo effect of TPN enriched with HM-derived specific peptides. Three-d-old guinea pig pups (body weight 99 ± 2 g, n = 17) were infused with TPN solutions as previously described (21) for 4 d via a catheter secured in the right jugular vein. TPN was administered in a binary mode, with lipid emulsion (1.5%) separated from other components (1% multivitamins + 2.2% amino acid + 8.7% dextrose ± 1 mmol/l peptides (a concentration close to those of other amino acids in TPN)). TPN solutions were infused without light protection. The effects of peptides were compared in three groups of animals:

  • TPN: animals receiving TPN without addition of peptide.

  • TPN + A: animals receiving TPN containing 1 mmol/l peptide-A.

  • TPN + B: animals receiving TPN containing 1 mmol/l peptide-B.

Daily-prepared TPN solutions were infused continuously at a rate of 220 ml/kg/d (animals receiving 4.8 g/kg/d of amino acids and 3.2 g/kg/d of lipids).

For both in vivo protocols, after 4 d of treatments, the liver was collected, rinsed in 0.9% NaCl, rapidly minced, and aliquoted; one aliquot was immediately treated for glutathione determination and the others were stored at −80 °C.

Guinea pigs were housed in an animal facility on a 12-h light/dark cycle at constant temperature. The study was approved by the institutional review board for the care and handling of animals (CHU Sainte-Justine), in accordance with guidelines of the Canadian Council of Animal Care.

Methods

GSH and GSSG were determined as previously described (1) by using the capillary electrophoresis P/ACE MDQ system (Beckman Coulter, Mississauga, Ontario, Canada). The redox potential was calculated using the Nernst equation (25 °C, pH 7.0; 33).

F2α-isoprostanes were measured (21) using a commercial enzyme immunoassay kit (Cayman Chemical, Ann Arbor, MI).

Peroxide concentrations were determined in TPN solutions by using the ferrous oxidation–xylenol orange assay as previously described (34), using H2O2 for the standard curve.

Hepatic levels of GS-HNE were measured by liquid chromatography electrospray ionization mass spectrometry using an Eclipse Plus C18 column coupled with an Agilent 1100 single quadrupole (Agilent, Montreal, QC, Canada). The approach used (35) is based on a differentiation of the aldehyde from the alcohol by reduction with 1 mol/l NaB2H4 (Cambridge Isotope Laboratories, Andover, MA), generating stable [2H1]DHN. Briefly, liver sample homogenized in buffer (N-2-hydroxyethylpiperazine-N9-2-ethanesulfonic acid: 39 mmol/l, EDTA: 0.4 mmol/l, butylated hydroxytoluene: 0.9 mmol/l, pH 7.4) was treated with NaB2H4 and spiked with the internal standard GS-[2H4]DHN, made from [9,9,9-2H3]trans-4-hydroxy-2-nonenal (Cayman Chemical), before proteins were precipitated by 5-sulfosalicylic acid. GS-[2H]DHN and the internal standard GS-[2H4]DHN were directly detected in the supernatant at m/z = 467 and 470, respectively. Operating parameters were a cone voltage of 60 V, source block temperature of 350 °C, nitrogen nebulizer gas flow at 12 l/min, and acquisition time between 8 and 18 min.

Total RNA was isolated from 100 mg of liver after homogenization in 1 ml of Trizol (Invitrogen, Carlsbad, CA) + 200 μl of chloroform and centrifuged at 31,200g for 15 min at 4 °C. The RNA from supernatant was precipitated in 500 μl isopropanol for 10 min. After 10 min of centrifugation at 31,200g at 4 °C, the pellet was washed in 500 μl ethanol 70% and centrifuged at 31,200g for 5 min at 4 °C. The dried pellet (37 °C, 1 h) was solubilized in RNase-free water. RNA concentration was determined by the ratio of absorbance read at 260/280 nm. A volume of 100 ng was reverse-transcribed by using the Quantitect Reverse transcription Kit (Qiagen, Montréal, QC, Canada). Levels of mRNA were quantified by real-time PCR using the SYBR green I labeling method. Sequences of primer sets are shown in Table 3 . The amplification protocol was denaturation at 95 °C for 5 min followed by 40 cycles of 95 °C/10 s, 64 °C/30 s. mRNA levels of target genes are reported as ΔCT values (cycle number at which PCR target plots cross threshold, normalized to the expression levels of 18S rRNA), and converted to the relative expression where ΔCT values of target genes are normalized to a control value, as described by Livak and Schmittgen (36).

Table 1 Glutathione levels in liver from animals infused with or without H2O2 and fed regular chow, human milk, or artificial milk
Full size table

Statistical Analyses

Data, expressed as mean ± SEM, were compared by ANOVA after verification of the homoscedasticity by Bartlett’s χ2-test. All comparisons were orthogonal and are described in the figures. In the first protocol ( Figure 2a ), the impact on redox potential of HM was compared with that of animals receiving AM. The effects of milk used as food were compared with a control group fed with regular guinea pig chow. For F2α-isoprostane determinations ( Figure 2b ), the comparisons were (C vs. HM) vs. AM. The impacts of infusion of H2O2 at similar levels to those measured in TPN, according to the enteral food, were compared in a factorial ANOVA. In the second protocol ( Figure 1 ), the following comparisons were made: ((40 mmol/l vs. 20 mmol/l) vs. 10 mmol/l) vs. 0 mmol/l. In the third protocol ( Figures 3 and 4 ), TPN containing peptide-A was compared with TPN + peptide-B, and both were compared with TPN without any peptide addition. Statistical analysis of TNFα and IL-1 mRNA levels was achieved with ΔCT values, and graphically represented as relative expression. The threshold of significance was assigned to P < 0.05.

Statement of Financial Support

This work was supported by a grant from the Advanced Foods and Materials Network (Project 11F) and the Canadian Institutes for Health Research (MOP 77637).