- Top of page
- Materials and methods
During the postprandial phase after ingestion of a meal, triglycerides (TGs) circulate in the plasma in the form of intestinally derived chylomicrons. These TG-rich lipoproteins distribute fatty acids to different tissues via the lipoprotein lipase system and are ultimately removed from the circulation through receptor-mediated processes in the liver . Elevated postprandial lipemia is often associated with obesity  and has been reported in patients with cardiovascular disease . Lactoferrin (Lf) is a milk protein that when injected intravenously has been demonstrated to increase postprandial lipemia by inhibiting chylomicron remnant removal from the liver .
Lf is a non-heme iron-binding glycoprotein belonging to the transferrin class. This monomeric protein of 689 residues folds into two domains of high sequence similarity, called the N- and C-lobes , each of which binds one ferric ion .
On the N-lobe, Lf contains an Arg-rich region similar to that of apolipoprotein E (apoE), which is the ligand for the uptake of chylomicron remnants . Studies on parenchymal liver cells demonstrated that Lf inhibits the association of apoE-containing lipoproteins to a receptor which is distinct from the low-density lipoprotein (LDL) receptor (LDL-R) and LDL-R-related protein (LRP) but was unidentified .
It was at this time that Bihain and Yen  reported that fibroblasts from a patient deficient in LDL-R were able to internalize significant amounts of LDL via a previously unidentified receptor. It was demonstrated that this new pathway involves a receptor that mediates the binding of apoB- and apoE-containing lipoproteins, leading to their subsequent internalization and degradation . This receptor displays highest affinity for TG-rich lipoproteins [chylomicrons and very-low-density lipoprotein (VLDL)] and is referred to as the lipolysis stimulated lipoprotein receptor (LSR) since it can only bind and internalize lipoproteins in the presence of free fatty acids (FFAs). Since LSR is activated by these lipolytic products, it is thus primarily active during the postprandial phase when the influx of chylomicrons into the circulation is high, which leads to increased lipase-mediated hydrolysis of the chylomicron TGs, leading to high FFA levels in the space of Disse of the liver [10-12]. LSR was identified as a multimeric complex consisting of three subunits derived by alternative splicing from the same gene . In vivo data now demonstrate that LSR plays an important role in the clearance of TG-rich lipoproteins during the postprandial phase . It was observed that Lf inhibited oleate-induced lipoprotein binding to rat hepatocyte plasma membranes [11, 12]. This inhibition was shown to occur without a significant decrease of the amount of membrane-associated oleate and thus differs from the inhibition caused by albumin that is achieved through oleate sequestration , suggesting that LSR is the molecular target of Lf. Even though this hypothesis was based on experiments done on liver membranes under conditions optimal for LSR activity, interaction between LSR and Lf remained to be demonstrated.
The aim of this study was to determine if the inhibition of LSR by Lf is due to direct interaction of these two proteins. Here, we show that bovine Lf and its hydrolysis products display the ability to bind to the FFA-activated form of the purified receptor and compete for binding of VLDL to LSR.
- Top of page
- Materials and methods
In order to study the direct interaction of LSR with Lf, it was necessary to use Lf with high purity. Since SDS/gel analysis revealed that commercially available Lf contained a number of contaminating proteins, including albumin (Fig. 1A), we purified Lf from raw bovine milk to > 95% purity (Fig. 1A). Since Lf inhibits postprandial lipemia when injected intravenously into rats , we next verified that the purified Lf exhibited this same biological effect. C57BL/6RJ male mice were injected intravenously with Lf and then immediately received olive oil by gavage. Plasma TGs measured 1 h after the gavage were found to be significantly increased (P = 0.04) by a factor of 4.3 in the group of mice that received purified Lf (Fig. 1B), while the 1.9-fold increase in plasma TGs in the group of mice that received commercial Lf only tended towards significance (P = 0.08). We verified that Lf did not have any effect on free glycerol levels (increase in free glycerol 1 h after gavage: controls 16 ± 4.5 μg·mL−1 and Lf group 18.8 ± 2.3 μg·mL−1). The difference between purified and commercial Lf could be due to impurities detected in the latter, including albumin and immunoglobulin (Fig. 1A). These results therefore indicate that pure Lf retains its in vivo effect on postprandial lipemia, confirming that the effect derives purely from Lf and not from other contaminants such as albumin.
Figure 1. Effect of Lf on plasma TG level during the postprandial phase. (A) SDS/PAGE showing the Lf used for this study. Lane 1, molecular weight marker (M); lane 2, bovine Lf from Sigma; lane 3, bovine Lf purified from whey proteins (Lf lab); lane 4, whey proteins obtained after casein precipitation. LC-IgG, light chain immunoglobulin; HC-IgG, heavy chain immunoglobulin; LG, lactoglobulin. (B) Plasma TG increase 1 h after Lf injection. C57BL6/RJ male mice were injected intravenously (tail vein) with 100 μL NaCl/Pi (open bars) or NaCl/Pi containing 2 mg bovine Lf (closed bars) obtained from Sigma or purified from raw milk. After gavage with 300 μL olive oil, blood samples were removed and plasma TG levels were measured as described in Materials and methods. Results are shown as mean ± SEM (n = 3 or 4 per condition) of the increase in plasma TG relative to control. Statistical significance is indicated by P values (compared to the corresponding NaCl/Pi group).
Download figure to PowerPoint
As Lf has been shown to enter the nucleus and induce gene expression [15, 16], we first sought to determine whether Lf demonstrated any effect on LSR mRNA or protein levels in the mouse hepatoma cell line, Hepa1–6. Since LSR activity as a lipoprotein receptor is dependent upon the presence of FFAs, we incubated cells for 3 h at 37 °C in the absence or presence of oleate and Lf. The concentrations of Lf used were those shown previously to inhibit lipoprotein binding to liver membranes in the presence of oleate . Incubation of cells in the presence of Lf appeared to slightly decrease cell LSR protein levels relative to actin as a loading control (Fig. 2A). However, because of the variations in expression, this difference was not statistically significant. Parallel experiments were conducted to determine whether LSR mRNA levels were affected. Only a small decrease was observed in cells incubated in the presence of 0.8 mm oleate and Lf compared with 0.8 mm oleate alone (Fig. 2B). No significant changes were observed when cells were incubated with or without Lf in the absence or presence of 0.4 mm oleate (Fig. 2B). Therefore, at least in short-term experiments, Lf did not have any detectable effect on LSR protein levels but did have a small inhibitory effect on LSR mRNA levels in the presence of oleate. We then initiated a study on the interaction between LSR and Lf both at the cellular and molecular levels. Intact cells were used, in which Alexafluor-488-labelled Lf was incubated with Hepa1–6 cells with or without oleate. Figure 3 shows strong labeling of permeabilized cells with both Lf and LSR, with little apparent colocalization in the merge picture in the absence of oleate (Fig. 3, top panels). However, when cells were incubated in the presence of oleate, colocalization of Lf and LSR labeling increased dramatically, as evidenced by the increased orange intensity in the merged picture (Fig. 3, lower right panel). This suggests that Lf interacts directly with LSR but only in the presence of oleate, thus when LSR is in its active conformation.
Figure 2. Effect of lactoferrin on LSR mRNA and protein levels in Hepa1–6 cells. Hepa1–6 cells were incubated for 3 h with 2 mg·mL−1 Lf with increasing concentrations of oleate as indicated. (A) Cell lysates were then prepared from which immunoblots were performed as described in Materials and methods to detect LSR. Actin served as loading control. A representative blot is shown and an analysis of three experiments with three different preparations of cells is shown in the bar graph as mean ± SEM ratio of LSR/actin. The values of LSR/actin for cells incubated in the absence of oleate and Lf were normalized to 100%, allowing us to compare the variations between different experiments. (B) Experiments were conducted in which mRNA was directly extracted from cells and then used in qPCR analysis. Data are shown as the relative expression of LSR versus the housekeeping gene, HPRT (mean ± SEM, n = 3).
Download figure to PowerPoint
Figure 3. Colocalization of Lf and LSR in Hepa1–6 cells. Hepa1–6 cells were incubated with 2 mg·mL−1 Lf labeled with Alexafluor 488 in the absence (top panels) or presence (bottom panels) of oleate for 20 min at 37 °C. The cells were then washed, fixed, permeabilized and labeled with anti-LSR IgG as described in Materials and methods. Visualization of Lf is shown in the left panels and LSR in the middle panels. Merged pictures are shown in the right panels.
Download figure to PowerPoint
In order to confirm this possible interaction, ligand blot experiments were performed using purified LSR and Lf. LSR was purified from rat hepatocyte plasma membranes using anion exchange chromatography. The protein appeared to be at least 95% pure, as judged by SDS/PAGE analysis (Fig. 4A). Three bands were obtained under reducing conditions, with apparent molecular mass ranging from 54 to 70 kDa, which is consistent with the masses of the subunits that compose the receptor. Under non-reducing conditions, one single band was observed at an apparent molecular weight of ~ 240 kDa, representing the LSR complex (Fig. 4A) . The identity of LSR was confirmed by immunoblot analysis (Fig. 4B).
Figure 4. SDS/PAGE and western blot of purified LSR. The purified LSR was loaded on 10% SDS/PAGE under non-reducing (NRC) or reducing (RC) conditions. (A) Silver staining. (B) Western blot. Revelation was made as described in Materials and methods. MM, molecular weight marker.
Download figure to PowerPoint
Ligand blots were then carried out in which LSR was migrated on SDS/PAGE under non-reducing conditions and, after transfer on nitrocellulose, incubated with the pure Lf. It was previously shown that FFAs were necessary for LSR binding to LDL [11, 12]. In view of the results of the colocalization experiments, the assay was conducted both in the absence and in the presence of oleate. As shown in Fig. 5A, lanes 1 and 2, Lf was revealed to bind to a band corresponding to the LSR complex (apparent molecular mass 240 kDa)  but only when oleate was present (Fig. 5A, lane 2), suggesting that FFAs are also required for Lf binding to this receptor. Control experiments in the absence of Lf demonstrate that this is not due to binding of anti-Lf IgG to LSR (data not shown).
Figure 5. Ligand blots showing the interaction between LSR and bovine Lf and its hydrolysis products. (A) Ligand blots of LSR with entire Lf or its hydrolysis products. LSR was loaded on SDS/PAGE under non-reducing conditions, and then transferred onto a nitrocellulose membrane and incubated with full-length Lf (2 mg·mL−1, lanes 1 and 2), its N-terminal lobe (2 mg·mL−1, lanes 3 and 4) and C-terminal lobe (2 mg·mL−1, lanes 5 and 6), and peptides derived from tryptic/chymotryptic hydrolysis (2 mg·mL−1, lanes 7 and 8). Strips were incubated without oleate (lanes 1, 3, 5 and 7) or with 0.8 mm oleate prior to incubation with Lf (lanes 2, 4, 6 and 8). Lf was then detected by immunoblotting with anti-Lf IgG as described in Materials and methods. (B) Ligand blots of LSR with VLDL in the presence of increasing concentrations of Lf. Purified LSR on nitrocellulose membranes was incubated with 0.8 mm oleate and 20 μg·mL−1 VLDL in the presence of the indicated concentrations of Lf. The VLDL was detected with anti-apoB IgG as described in Materials and methods. (C) SDS/PAGE of Lf and its mild hydrolysis products used for the ligand blot shown in (A). lane 1, M, molecular weight marker; lane 2, pure bovine Lf; lane 3, peptide representative of the C-domain; lane 4, peptides of the N-domain.
Download figure to PowerPoint
We have previously reported that Lf can inhibit LSR activity as a lipoprotein receptor in liver cell and membrane studies as well as ligand blots using membrane extracts [11, 12]. Ligand blots were performed by incubating immobilized purified LSR with TG-rich VLDL in the presence of oleate and increasing concentrations of Lf. VLDL binding was revealed by detection of apoB on the 240-kDa LSR complex (Fig. 5B). This binding was decreased in intensity with increasing concentrations of Lf, thus demonstrating that direct binding of Lf to LSR can prevent binding of its lipoprotein ligand.
In order to determine which part of Lf is recognized by LSR, Lf was hydrolyzed by trypsin under mild conditions to obtain the N- and C-lobes separately. Four fragments with apparent molecular masses of 24, 34, 44 and 51 kDa were obtained on SDS/PAGE profile (data not shown). N-terminal sequencing and mass spectrometry analysis (Table 1) allowed the identification of the first two fragments as deriving from the N-terminal lobe and the second two fragments as belonging to the C-lobe (with or without the peptide connecting the two lobes of Lf). Precise identification of the C-terminus of the peptides corresponding to the N-lobe (a tryptic fragment) was difficult to obtain because of the presence of two Lys residues (at positions 280 and 282) and one Arg residue (at position 284) in this region. Indeed, it has been reported that two N-lobe fragments of apparent mass 34 kDa correspond either to the 1–284 Lf region  or to the 1–280 region , showing that identification of the C-terminal extremity remains difficult to determine. Moreover, determination of the mass of the entire N- and C-lobe peptides did not allow precise identification of the C-terminus because of the microheterogeneity of the glycan structures of Lf.
Table 1. Identification of the peptides obtained after mild tryptic hydrolysis of Lf. X corresponds to a residue that was not identified by Edman sequencing
|Fragment name||N-ter sequence (Edman)||Peptide (MS-MS)||Domain|
After isolation of the fragments accounting for each lobe by liquid chromatography (Fig. 5C), ligand blots were performed once again in the same conditions as above except that incubation was performed with fragments of the N-lobe (mixture of PP24 and PP34) or of the C-lobe (P51) rather than with the whole Lf molecule. As shown in Fig. 5A (lanes 3–6), both lobes of Lf are able to bind to LSR but, again, only if oleate is present. This also suggests that there are structural motifs or sequences located not only at the N-lobe but also the C-lobe of Lf that are recognized by LSR.
We pursued this line of thought by subjecting the trypsin hydrolysate of Lf to further hydrolysis by chymotrypsin in order to determine whether small peptides derived from Lf hydrolysis possess the ability to bind to the receptor. Non-digested N- or C-lobes or intermediate peptides having apparent molecular masses > 10 kDa were removed by ultrafiltration. Mass spectrometry analysis confirmed that peptides contained in the ultrafiltrate were small (Table 2), except for four peptides displaying masses of 2726, 2564, 2402 and 2240 Da. Fragmentation by tandem mass spectrometry showed that these peptides actually correspond to four glycopeptides with the same aglycoform VKNDTVW(543–549) produced by chymotrypsin action on the C-lobe. These glycopeptides carry on the Asn545 residue an oligomannosidic-type glycan structure (data not shown). The mass spectrometry analysis showed that the glycan moiety possesses a variable number of mannose residues, up to nine units, and this result is in accordance with the N-glycan structures of Lf reported by Spik et al.  and Coddeville et al. . Furthermore, the experimental molecular mass of the deglycosylated peptide chain produced by MS2 fragmentation was 860 ± 1 Da, which is very close to the theoretical mass of the apoheptapeptide of 860.44 Da. The hydrolysate does not contain any other N-glycopeptide, as all the other molecular masses found in this fraction are lower than the usual masses of N-acetyllactosamine-type and oligomannosidic-type structures. Finally, the glycopeptides were discarded by cold ethanol precipitation in order to determine whether the interaction between the mixture of Lf peptides and LSR was due to the amino acid sequences rather than to the glycan moiety. Ligand blots performed on the mixture of these peptides (Fig. 5A) or on the mixture of peptides after elimination of the glycosylated fractions (data not shown) highlighted their ability to bind to LSR in the presence of oleate.
Table 2. Identification of the peptides obtained after hydrolysis of Lf by trypsin and chymotrypsin. The Mr values correspond to monoisotopic masses. nd, not determined; Man, mannose; GlcNAc, N-acetylglucosamine; RT, retention time
|RT (min)||Mr (experimental)||Mr (calculated)||Sequence||Residues|
|13.52a||959.5 (+57)||959.427 (+57)||CAVGPEEQK (+57)||348–356|
|18.89||2726.0||860.44 (aglycoform)||VKNDTVW (9 Man + 2 GlcNAc)||543–549|
|19.00||2564.0||860.44 (aglycoform)||VKNDTVW (8 Man + 2 GlcNAc)||543–549|
|19.27||2402.0||860.44 (aglycoform)||VKNDTVW (7 Man + 2 GlcNAc)||543–549|
|19.49||2240.0||860.44 (aglycoform)||VKNDTVW (6 Man + 2 GlcNAc)||543–550|
- Top of page
- Materials and methods
Elevated postprandial lipemia following intravenous injection of Lf was shown to be due to accumulation of chylomicron remnants in the circulation and reduced clearance of these particles by the liver . This led to numerous investigations of the molecular mechanisms involved with the objective of identifying the Lf-sensitive liver receptor(s) for these TG-rich lipoproteins. However, Lf has many other biological activities such as in host defense against infection or immunomodulation and, as a result, displays various affinities for a number of proteins on the cell surface [21, 22], which rendered the task of identifying the liver receptor difficult. Indeed, at the hepatic level, there have been studies reporting association of Lf with LRP1, the asialoglycoprotein receptor, as well as heparan sulfate proteoglycans [23, 24]. Of these different proteins, LRP1 was a potential candidate as chylomicron remnant receptor, but it was later demonstrated to play a back-up role in clearance of these particles in the absence of the LDL-R . Indeed, the presence of a third pathway other than LRP1 and LDL-R for chylomicron remnants was postulated [25, 26]. Furthermore, GST-RAP, a specific inhibitor of LRP, did not appear to modify Lf's effect on β-VLDL binding to liver cells, leading the authors to conclude that the Lf effect was not mediated through LRP nor through the LDL-R . While Lf has been shown to inhibit the binding and uptake of apoE containing lipoproteins, including chylomicron remnants, in the liver [28, 29], these studies used cell-based models or tissue extracts without identifying the actual target receptor or performing direct binding studies of Lf with the receptor in a purified state.
We identified LSR as an apoB,E receptor that plays an important role in the removal of TG-rich lipoproteins during the postprandial phase . It serves as lipoprotein receptor only in the presence of FFA, which induces a conformational change to expose a binding site on the receptor for apoB or apoE . We have previously shown that Lf can prevent the binding of lipoproteins to FFA-activated LSR . In this study, we now demonstrate that Lf colocalizes with LSR in Hepa1–6 cells and that Lf binds to purified LSR but, in both cases, only in the presence of oleate and thus only when LSR is in its active form. Furthermore, Lf is able to directly compete with the TG-rich VLDL for binding to LSR. Taken together, this suggests that the previously observed inhibitory effect of Lf on oleate-induced binding of lipoproteins to hepatocytes or hepatic membranes is mediated by the direct interaction of Lf with the receptor. Since during the postprandial phase chylomicron influx into the liver is high leading to increased FFA production by lipolysis, LSR would be in its active form. We would propose then that the elevated postprandial lipemia observed upon Lf treatment in vivo is mediated in part by its direct interaction with FFA-activated LSR, thus preventing clearance of chylomicrons and their remnants through the LSR pathway.
We tested whether Lf demonstrated any effect on LSR mRNA or protein levels. Although protein levels did not change significantly, we did observe a small but significant decrease in LSR mRNA in cells incubated with Lf and oleate. Lf has previously been reported to enter the nucleus and induce gene expression via the transcription factor AP-1 . Bioinformatics analysis revealed potential AP-1 response elements in the sequence upstream of the mouse lsr gene. Taking into account that AP-1 is involved in the control of genes related to cell survival and apoptosis [31, 32] and that a previous study has also reported that LSR expression can be induced by P53 , it may be that LSR is in some manner involved in the regulation of the cell cycle. Further investigation is required to examine this question. Nevertheless, because of the rapid effect of Lf on chylomicron remnant clearance, it seems unlikely that this was due to Lf's effect on LSR mRNA or protein levels.
Ligand blot studies using purified LSR revealed that both the C- and N-lobes of Lf were able to bind directly to LSR in the presence of oleate. Furthermore, a peptide mixture resulting from further hydrolysis of both lobes retained the capacity for recognizing the active form of LSR. Since the removal of glycopeptides did not affect the ability of Lf to interact with activated LSR, this suggests that the recognition is based purely on the amino acid sequence. The original hypothesis of the Lf effect was based on the fact that Lf possesses a number of Arg and Arg-Lys clusters on its N-terminal side that resemble the binding site of apoE, which is considered the ligand for the chylomicron remnant receptor. On the basis of chemical modification experiments, previous studies using liver parenchymal cells proposed that Arg residues were crucial for the recognition of Lf and its inhibitory effects on lipoprotein remnants [28, 34]. These studies were carried out on human Lf, which possesses a cluster of four Arg residues (positions 2–5) and an Arg/Lys-rich sequence at positions 25–31 (R-X-X-R-K-X-R), similar to that of apoE . However, conflicting results were reported when Huettinger et al.  observed that digestion of Lf by trypsin that cleaved through the Arg (positions 2–5) cluster but left the helical region near position 35 intact suppressed the inhibitory effect of Lf on the internalization of remnants. On the other hand, Ziere et al.  demonstrated that truncated Lf lacking the first 14 residues and thus the Arg (positions 2–5) cluster was a more effective competitor than Lf for liver uptake of lipoprotein remnants. Therefore, the importance of the four Arg cluster in Lf's effect on chylomicron remnant clearance remains unclear.
Here, we use purified bovine Lf which, when injected into mice, leads to elevated postprandial plasma TG levels (Fig. 1B). This is in agreement with previous reports using bovine Lf that have demonstrated inhibition of chylomicron remnant clearance during the postprandial phase . Interestingly, bioinformatics analysis revealed that bovine Lf possesses only an RK sequence instead of the Arg (2–5) cluster described in human Lf, thus clearly excluding a predominant role of this cluster in the binding to LSR. Moreover, the Arg/Lys-rich sequence at positions 25–31 is divergent in the bovine sequence (WRMKKLG) compared with that of human. This observation, along with the fact that both N- and C-terminal lobes of bovine Lf bind to the receptor, suggests that at least two sequences or structural motifs are recognized. Furthermore, given the size of the peptides present in the tryptic/chymotryptic hydrolysate that retain the ability to bind to LSR, our results suggest that this receptor recognizes one or several sequences on Lf rather than conformational features. For the N-lobe, the RMKKL sequence at positions 25–29 is found in the peptide mixture isolated from the tryptic/chymotryptic hydrolysate together with another basic peptide IVKLLSKAQEKFGK at positions 268–280, while no peptide possessing more than one basic residue and belonging to the C-lobe has been recovered in the hydrolysate, suggesting that determinants of the interaction might involve other features than basic sequences.
In conclusion, this study clearly demonstrates a direct interaction of Lf with the FFA-activated form of hepatic LSR, which prevents the binding of TG-rich lipoproteins to this receptor and therefore could explain the elevated postprandial lipemia and reduced chylomicron remnant clearance observed in rodents treated with Lf. Further experiments are needed to identify the specific peptides of Lf that bind to LSR and to determine whether these peptides, when isolated, have the ability to modulate postprandial lipemia.