Protein Region Important for Regulation of Lipid Metabolism in Angiopoietin-like 3 (ANGPTL3) ANGPTL3 IS CLEAVED AND ACTIVATED IN VIVO *

Angiopoietin-like 3 (ANGPTL3) is a secreted protein that is mainly expressed in the liver and regulates lipid metabolism by inhibiting the lipolysis of triglyceride-rich lipoproteins. Using deletion mutants of human ANGPTL3, we demonstrated that the N-terminal coiled-coil domain-containing fragment-(17–207) and not the C-terminal fibrinogen-like domain-containing fragment-(207–460) increased the plasma triglyceride levels in mice. We also found that the N-terminal region 17–165 was required to increase plasma triglyceride levels in mice and that a substitution of basic amino acid residues in the region 61–66 of the fragment showed no increase in the plasma triglyceride levels and no inhibition of lipolysis by lipoprotein lipase. In addition, when we analyzed ANGPTL3 in human plasma, we detected cleaved fragments of ANGPTL3. By analyzing recombinant ANGPTL3 in mouse plasma, we found that it was cleaved mutagenesis. Construction Recombinant Adenoviruses— Recombinant adenoviruses were generated as previously described (12) using an adenovirus expression vector kit (Takara Shuzo). Eco RI/ Xba I fragments of recombinant pME18Ss generated above were blunt-ended and inserted into the Swa I site of a cosmid vector pAxCAwt. Co-transfection of these constructs and adenoviral DNA generated recombinant adenoviruses. Recombinant genes were under the control of a CAG promoter. Recombinant adenoviruses encoding His/ANGPTL3, ANGPTL3-(17–165), ANGPTL3-(17–207), ANGPTL3-(79–207), ANGPTL3-(207–460), ANGPTL3-[17–207mu], and ANGPTL3-[5Rmu] were designated as Ad/ His/ANGPTL3, Ad/ANGPTL3-(17–165), Ad/ANGPTL3-(17–207), Ad/ ANGPTL3-(79–207), Ad/ANGPTL3-(207–460), Ad/ANGPTL3-[17– 207mu], and Ad/ANGPTL3-[5Rmu], respectively. Recombinant adenoviruses were cultured and purified as described previously (13). We confirmed the presence of the secreted proteins in the culture supernatant of all recombinant adenovirus-infected HeLa cells as fol- lows. We infected 5 (cid:2) 10 7 pfu of recombinant adenovirus on a 25-cm 2 sheet of confluent HeLa cells. At 3 days after infection, we collected the supernatant of the cells. After adding trichloroacetic acid of 1/10 the volume, proteins were obtained from the supernatant by centrifugation and analyzed by Western blotting using anti-ANGPTL3 polyclonal antibody as a probe.

Angiopoietin-like 3 (ANGPTL3) 1 is a secreted protein that is expressed predominantly in the liver (1). ANGPTL3 is structurally similar to angiopoietins (2)(3)(4), which are vascular endothelial growth factors. It also has a signal peptide for secretion, a helical domain predicted to form coiled-coils (CCD), and a C-terminal globular fibrinogen-like domain (FLD). Angiopoietins are ligands that bind to the receptor tyrosine kinase Tie2. The FLDs of angiopoietins bind to Tie2, and the CCDs of angiopoietins mediate ligand homo-oligomerization (5). Al-though ANGPTL3 does not bind to Tie2, it may also induce angiogenesis by binding to integrin ␣ V ␤ 3 (6).
We reported that KK/Snk mice (previously, KK/San) had abnormally low plasma lipid levels and identified a mutation in the Angptl3 gene as the cause of the hypolipidemic phenotype of KK/Snk mice (7,8). Injection of recombinant ANGPTL3 or adenoviruses encoding ANGPTL3 resulted in an increase of plasma lipid levels in mice. It has been speculated that ANGPTL3 was capable of inhibiting lipolysis of triglycerides, because an injection of recombinant ANGPTL3 induced a rapid increase in mouse plasma triglyceride levels (9,10). We further reported that very low density lipoprotein clearance was enhanced in KK/Snk mice and that recombinant ANGPTL3 directly inhibited lipoprotein lipase (LPL) activity in vitro (11). These observations suggest that ANGPTL3 could be involved in the regulation of lipid metabolism by inhibiting LPL activity.
Thus, ANGPTL3 seems to be a multifunctional factor that may regulate both lipid metabolism and angiogenesis. The C-terminal FLD of ANGPTL3 was reported to be involved in angiogenesis via binding to integrin ␣ V ␤ 3 (6), but the domain of ANGPTL3 that is involved in the regulation of lipid metabolism has not been elucidated.
In this study, we constructed various mutants of ANGPTL3 and analyzed the region of ANGPTL3 required for its function in lipid metabolism. Further, as it was revealed that ANGPTL3 was cleaved in the linker region between CCD and FLD, we estimated the importance of this cleavage to the activity of ANGPTL3 in vivo.
Construction of Recombinant Adenoviruses-Recombinant adenoviruses were generated as previously described (12) using an adenovirus expression vector kit (Takara Shuzo). EcoRI/XbaI fragments of recombinant pME18Ss generated above were blunt-ended and inserted into the SwaI site of a cosmid vector pAxCAwt. Co-transfection of these constructs and adenoviral DNA generated recombinant adenoviruses.  [17-207mu], and Ad/ANGPTL3-[5Rmu], respectively. Recombinant adenoviruses were cultured and purified as described previously (13). We confirmed the presence of the secreted proteins in the culture supernatant of all recombinant adenovirus-infected HeLa cells as follows. We infected 5 ϫ 10 7 pfu of recombinant adenovirus on a 25-cm 2 sheet of confluent HeLa cells. At 3 days after infection, we collected the supernatant of the cells. After adding trichloroacetic acid of 1/10 the volume, proteins were obtained from the supernatant by centrifugation and analyzed by Western blotting using anti-ANGPTL3 polyclonal antibody as a probe.
Animal Assay-KK/Snk mice were established as previously described (7,8). We injected a recombinant adenovirus intravenously to KK/Snk mice via the tail vein. We obtained blood from the retro-orbital plexus at several time points after injection. We measured the plasma triglyceride concentration with a commercial kit (WAKO Pure Chemical Co., Osaka, Japan).
Protein Preparation-We transfected mammalian gene expression constructs into COS-1 cells using reagent LipofectAMINE 2000 (Invitrogen). At 3-4 days after transfection, we collected the supernatants of the transfected cells and purified the recombinant proteins using the His Trap kit (Amersham Biosciences). The purified recombinant proteins were further dialyzed with phosphate-buffered saline (pH 7.4) (Sigma). Thereafter, the inhibitory activity on lipolysis was assessed by an LPL assay.
LPL Assay-The LPL activity assays were conducted using rat LPL obtained from rat preadipocytes as described previously (11).
Detection of Human Plasma ANGPTL3-We prepared a mouse monoclonal antibody and a rabbit polyclonal antibody that recognized the recombinant ANGPTL3 prepared as described previously (8). The monoclonal antibody used in this study was confirmed to recognize the N-terminal portion 17-207 of ANGPTL3. We purified human plasma (Sigma) by affinity chromatography that was prepared by immobilizing the anti-ANTPTL3 monoclonal antibody to an N-hydroxy-succinimideactivated HiTrap column (Amersham Biosciences) at a concentration of 5 mg/ml packed gel. We eluted the proteins that bound to the column with 0.5 M NaCl and 0.1 M glycine-HCl (pH 2.7) and immediately neutralized them with 1/10 volume of Tris-HCl, pH 8.0. The purified proteins were analyzed by Western blotting using an anti-ANGPTL3 polyclonal antibody as a probe.
Characterization of the Cleavage Site of ANGPTL3-KK/Snk mice were intravenously injected with 2 ϫ 10 9 pfu of the recombinant adenovirus, Ad/His/ANGPTL3. We collected the plasma from the recombinant adenovirus-injected KK/Snk mice at 2 days after injection and then purified the proteins that contained a hexahistidine tag from the plasma using the His Trap kit (Amersham Biosciences). The purified proteins were deoxidized in 10 mM dithiothreitol (Sigma) and alkylated in 55 mM iodoacetamide (Wako Pure Chemical Co.). The deoxidized and alkylated proteins were digested with lysyl endopeptidase (Wako Pure Chemical Co.) or deglycosylated with PNGase F (New England Biolabs, Beverly, MA) followed by digestion with sequencing grade endoproteinase Glu-C (Promega, Madison, WI). Then, formic acid was added to the peptides (final concentration 0.05%), and the digested proteins were analyzed by liquid chromatography electrospray ionization mass spectrometry: column, original fused silica electrospray needle packed with Inertsil ODS-2 (GL Science, Tokyo, Japan); solvent delivery, CapLC System (Waters, Milford, MA); solvent A, 0.05% formic acid/H 2 O; solvent B, 0.05% formic acid/acetonitrile; linear gradient elution, 0 -30% solvent B over 1 h; mass spectrometer, Q-Tof2 (Micromass, Manchester, UK) with an electrospray ion source. The resulting spectra were processed with ProteinLynx (Micromass, Manchester, UK) software.

N-terminal Fragment of ANGPTL3 Increases Plasma Triglyceride Levels in Mice: Determination of the Region Essential
for Activity-To determine the region of ANGPTL3 involved in lipid metabolism, we first tested whether the N-terminal CCDcontaining region or the C-terminal FLD-containing region, alone, could increase plasma triglyceride levels in mice. We constructed deletion mutants of ANGPTL3, ANGPTL3-(17-207) and ANGPTL3-(207-460) (Fig. 1A) and generated recombinant adenoviruses, Ad/ANGPTL3-  and Ad/ ANGPTL3-(207-460). After confirming that recombinant proteins were released in the supernatant of the adenovirusinfected Hela cells (Fig. 1B), the adenoviruses were injected into KK/Snk mice. The recombinant proteins were also detected in the plasma of mice injected with Ad/ANGPTL3-  and Ad/ANGPTL3-(207-460) (Fig. 1C). The plasma triglyceride levels of Ad/ANGPTL3-(17-207)-injected mice were dramatically increased, whereas those of Ad/ANGPTL3-(207-460)-injected mice were not (Fig. 1D). To narrow down the essential region of the functional domain of ANGPTL3, we generated recombinant adenoviruses, Ad/ANGPTL3-  and Ad/ANGPTL3-(79 -207) and injected them into KK/Snk mice. Injection of Ad/ANGPTL3-(17-165) resulted in an increase in plasma triglyceride levels, whereas injection of Ad/ ANGPTL3-(79 -207) did not (Fig. 1E). Although the amount of recombinant proteins secreted from Ad/ANGPTL3-(79 -207)infected HeLa cells was greater than that from Ad/ANGPTL3-(17-165)-infected HeLa cells (Fig. 1B), we could not detect any recombinant proteins in the plasma of mice injected with Ad/ ANGPTL3-(79 -207) (Fig. 1C, lane 4). These data show that the region 17-165 elicits activity and suggest that the region 17-78 could be essential for protein stability in vivo. To evaluate whether a shorter fragment consisting of the first 78 N-terminal amino acid residues and not including the coiled-coil domain was active or not, we constructed an expression vector that encoded 1-78 of ANGPTL3 and transfected it into COS-1 cells. However, we could not detect any recombinant protein in the supernatant of the COS-1 cells (data not shown). Therefore, we could not elucidate whether region 17-78 was sufficient for the activity or not. On the other hand, we found that there was a putative heparin-binding motif (14), 61 VHKTKG 66 , in the region 17-78. To analyze the importance of this motif, we constructed a mutant in which basic amino residues of the motif VHKTKG were substituted with VINTNG (ANGPTL3- [17-207mu]) ( Fig. 2A) and generated an adenovirus Ad/ANGPTL3- [17-207mu]. Injection of Ad/ANGPTL3- [17-207mu] into KK/Snk mice did not increase plasma triglyceride levels (Fig. 2B) even though recombinant proteins were detected in the plasma of Ad/ANGPTL3- [17-207mu]-infected mice as much as in the plasma of Ad/ANGPTL3-(17-207)-injected mice (Fig. 2C), where plasma triglyceride levels were dramatically increased (Fig. 2B).
ANGPTL3 Is Cleaved in Vivo in Human and Mouse-As mentioned above, it was shown that the N-terminal region of ANGPTL3 was involved in lipid metabolism, whereas the Cterminal region of ANGPTL3 was involved in angiogenesis. We could not detect any ANGPTL3 protein in human plasma by Western blotting using either anti-ANGPTL3 polyclonal or monoclonal antibodies (data not shown). To detect ANGPTL3 protein in human plasma and determine whether ANGPTL3 can be processed into smaller fragments, we purified ANGPTL3 from human plasma by affinity chromatography using anti-ANGPTL3 monoclonal antibody that recognizes the N-terminal 17-207 of ANGPTL3. Purified ANGPTL3 was visualized by Western blotting using anti-ANGPTL3 polyclonal antibody. We detected cleaved ANGPTL3 as well as full-length ANGPTL3 (Fig. 3A).
In addition, we detected cleaved proteins, whose molecular masses were ϳ30 -35 kDa, as well as full-length proteins in the plasma of Ad/His/ANGPTL3-injected mice (Fig. 3B). Then, we purified the N-terminal fragment of ANGPTL3 from the plasma of Ad/His/ANGPTL3-injected mice by nickel affinity chromatography. We digested the purified protein sample with lysyl endopeptidase or endoproteinase Glu-C and analyzed the digested fragments by liquid chromatography electrospray ionization mass spectrometry. As a result, we detected peptides 198 EIENQLRRTSIQEPTEISLSSKPR 221 and 198 EIENQLRRT-SIQEPTEISLSSKPRAPR 224 in the sample digested with lysyl endopeptidase, and peptides 214 ISLSSKPR 221 and 214 ISLSSK-PRAPR 224 in the sample digested with endoproteinase Glu-C. Thus, it was found that ANGPTL3 was cleaved at two sites, Arg 221 2Ala 222 and Arg 224 2Thr 225 , in the linker region between CCD and FLD (Fig. 3C).
Cleavage of ANGPTL3 Is Important for Its Activation in Vivo-We showed that ANGPTL3 was cleaved in mice and that the N-terminal fragment of ANGPTL3 increased plasma triglyceride levels. Although full-length ANGPTL3 inhibited lipolysis of triglycerides by LPL in vitro to the same extent as the N-terminal fragment of ANGPTL3, we considered that cleavage of ANGPTL3 could be required to increase plasma triglyceride levels. To evaluate the effect of the cleavage, we generated a recombinant adenovirus (Ad/His/ANGPTL3-[5Rmu]) that encoded a mutant ANGPTL3 in which all five arginine residues located in and near the linker region between the CCD and the FLD were substituted with other amino acids (R203T, R204T, R221S, R224S, and R235S) (Fig. 4A). We injected Ad/ His/ANGPTL3-[5Rmu] into KK/Snk mice. We confirmed that cleaved recombinant proteins were not detected in the plasma of Ad/His/ANGPTL3-[5Rmu]-injected mice (Fig. 4B). This result suggests that the substitution of these amino acids alters the susceptibility to cleavage in vivo. At 1 day after injection of the recombinant adenovirus, the plasma triglyceride concentration in Ad/His/ANGPTL3-[5Rmu]-injected mice was 186.4 Ϯ 11.2 mg/deciliter (n ϭ 3), whereas that of Ad/His/ANGPTL3and Ad/ANGPTL3-(17-207)-injected mice was dramatically increased and reached 1243 Ϯ 184 and 2564 Ϯ 67 mg/deciliter, respectively (Fig. 4C). Recombinant proteins were detected in the plasma of Ad/His/ANGPTL3-[5Rmu]-injected mice as much as in the plasma of Ad/His/ANGPTL3-and Ad/ANGPTL3-(17-207)-injected mice (Fig. 4B). These results indicate that His/ ANGPTL3-[5Rmu] is less active than His/ANGPTL3 in mice. Next, we prepared recombinant His/ANGPTL3 and His/ ANGPTL3-[5Rmu] and confirmed that recombinant His/ ANGPTL3-[5Rmu] inhibited lipolysis of triglycerides by acting on LPL as potently as recombinant His/ANGPTL3 (Fig. 4D). These results indicate that the substitution of the five amino acids did not result in the loss of lipolysis inhibition in vitro and that cleavage of ANGPTL3 is an important process for the activation of ANGPTL3 in vivo.

DISCUSSION
ANGPTL3 consists of the N-terminal CCD, C-terminal FLD, and a linker region. We have shown that ANGPTL3 is cleaved in the linker region between CCD and FLD in vivo and that the cleaved N-terminal portion of ANGPTL3 inhibits LPL activity and increases the level of circulating plasma triglycerides in mice. Furthermore, a cleavage-resistant mutant of ANGPTL3 was found to be less active than the cleavable wild-type ANGPTL3 in increasing mouse plasma triglyceride levels, which suggests that cleavage of ANGPTL3 is an important   FIG. 2. Importance of a putative heparin-binding motif for the activity. A, a putative heparin-binding motif of ANGPTL3 and the three substituted amino acid residues (arrow) in ANGPTL3- [17-207mu]. KK/Snk mice were each intravenously injected with 200 l of 2.5 ϫ 10 10 pfu/ml recombinant adenoviruses, Ad/ANGPTL3-(17-207) (n ϭ 2) and Ad/ANGPTL3- [17-207mu] (n ϭ 3). B, plasma triglyceride levels before (Pre) and at 1, 2, 4, 7, 11, and 14 days after injection. Error bars indicate S.E. C, recombinant proteins in 1 l of plasma at 1 day after injection were detected by Western blot analysis using polyclonal anti-ANGPTL3 antibody. Culture medium supernatant of Ad/ ANGPTL3-(17-207)-infected HeLa cells (lane 1), phosphate-buffered saline-injected mouse plasma (lane 2), Ad/ANGPTL3-(17-207)-injected mouse plasma (lanes 3 and 4), and Ad/ANGPTL3-[17-207mu]-injected mouse plasma (lanes 5-7). D, effect of recombinant proteins on LPL activity. Error bars indicate S. E. (n ϭ 3). process for the activation of ANGPTL3 in vivo. We have also shown that the putative heparin-binding motif located at 61-66 is important for the activity.
We showed that the N-terminal 17-207 of ANGPTL3 has the ability to regulate lipid metabolism. It was reported that the C-terminal portion of ANGPTL3 had the ability to induce angiogenesis (6). Therefore, the regions of ANGPTL3 involved in angiogenesis and lipid metabolism were determined to be different from each other. Furthermore, we were able to narrow the region involved in lipid metabolism down to the region 17-165. The plasma triglyceride levels of mice injected with Ad/ANGPTL3-(17-165) (1304 Ϯ 262 mg/deciliter at 1 day postinjection) were lower than those injected with Ad/ ANGPTL3-(17-207) (5056 Ϯ 93 mg/deciliter at 1 day postinjection) (Fig. 1, D and E). However, these plasma triglyceride levels seemed to be correlated with the plasma levels of recombinant protein (Fig. 1C). These results suggest that the region 17-165 could increase the plasma triglyceride levels as well as the region 17-207. We also showed that a substitution of basic amino residues in the putative heparin-binding motif located at 61-66 of the N-terminal resulted in a dramatic decrease in its inhibitory activity on LPL in vitro. Although it was reported that heparin modulated LPL activity in vitro (15), only adding heparin did not affect the inhibitory effect of ANGPTL3 on LPL activity (data not shown). We reported that ANGPTL4 could also inhibit LPL activity (16), but we could not find a putative heparin-binding motif in the active region of ANGPTL4. These findings suggest that a region containing the putative heparinbinding motif may interact with a factor involved in lipolysis other than heparin, and the binding of this factor to the putative heparin-binding motif may be important for the activity. To elucidate how ANGPTL3 inhibits the lipolysis of triglyceride, it must be determined whether ANGPTL3 directly inhibits LPL or whether another factor is involved. A putative heparinbinding motif-substituted mutant protein would be useful as a negative control in further studies.
With regard to the cleaved N-terminal fragment, we detected cleaved ANGPTL3 in human and ANGPTL3-overexpressed mice. Although we did not use any protein inhibitors when collecting plasma, we did confirm that the recombinant fulllength ANGPTL3 protein was not cleaved during incubation with mouse plasma at 37°C for an hour. We also considered that the cleavage of ANGPTL3 that we observed did not occur after we collected the plasma. We determined that the main cleavage sites were at Arg 221 2Ala 222 and Arg 224 2Thr 225 , but other small bands of cleaved proteins were also detected ( Fig.  3B), suggesting that ANGPTL3 may have been cleaved at other sites albeit only a little. To examine this further, we substituted five arginine residues located in and near the linker region between CCD and FLD. We confirmed that the substitutions of the five amino acid residues altered the susceptibility to the cleavage in vivo (Fig. 4B) and that the substitution of these amino acids did not alter the LPL inhibitory activity of these recombinant proteins in vitro (Fig. 4D). We demonstrated that the increase of plasma triglyceride levels in the cleavageresistant mutant ANGPTL3-overexpressed mice was far less than that in the wild-type ANGPTL3-overexpressed mice (Fig.  4C), even though the amounts of recombinant protein detected in the plasma of the cleavage-resistant mutant ANGPTL3overexpressed mice were greater than those of the wild-type ANGPTL3-overexpressed mice (Fig. 4B). These results indicate that non-cleavable, full-length ANGPTL3 is less active in increasing triglyceride levels in vivo despite its high LPL inhibitory activity in vitro.
There could be two explanations behind this inconsistency between the in vitro and in vivo activities of full-length ANGPTL3. One is the possibility that ANGPTL3 inhibition of LPL is not the cause of the increase in triglyceride levels in vivo but that the increase in triglyceride levels is due to ANGPTL3 affecting hepatic lipase. The other is the possibility that fulllength ANGPTL3 is not able to inhibit LPL in vivo unless it is cleaved.
With regard to the first hypothesis, we had found that the inhibition of hepatic lipase by ANGPTL3-(17-207) was very weak compared with that of LPL (data not shown). Moreover, when the N-terminal fragments of ANGPTL3 (ANGPTL3- ) and its mutant (ANGPTL3- [17-207mu]) were tested, the inhibitory activity on LPL in vitro corresponded well to the triglyceride-increasing activity in vivo. Therefore, we considered that the triglyceride increase was because of the inhibition of LPL by ANGPTL3.
With regard to the second hypothesis, it is known that some growth factors and cytokines are associated with other proteins that can suppress their actions and that in order to activate these factors and cytokines, they must first be liberated from these proteins by proteolytic processing (17)(18)(19). Natural fulllength ANGPTL3 may also be in complex with a protein, and this protein may be bound to the active site of ANGPTL3. It could be considered that the cleavage of ANGPTL3 dislodges the protein and releases the active ANGPTL3 fragment. We previously found that ANGPTL3 was able to bind to adipocytes and heparin was partially able to inhibit this binding (20). These observations suggest that ANGPTL3 may bind to proteoglycans. Therefore, proteoglycans could be the proteins binding to ANGPTL3 and suppressing its activity.
ANGPTL3 is mainly expressed in the liver (1). In plasma, full-length ANGPTL3 was also detected (Fig. 3, A and B). These findings let us speculate that after being secreted from the liver, ANGPTL3 circulates in the blood in the full-length form and carries out its role in lipogenesis at the target site upon its cleavage.