An anti-ANGPTL3/8 antibody decreases circulating triglycerides by binding to a LPL-inhibitory leucine zipper-like motif

Abstract

Triglycerides (TG) are required for fatty acid transport and storage and are essential for human health. Angiopoietin-like-protein 8 (ANGPTL8) has previously been shown to form a complex with ANGPTL3 that increases circulating TG by potently inhibiting LPL. We also recently showed that the TG-lowering apolipoprotein A5 (ApoA5) decreases TG levels by suppressing ANGPTL3/8-mediated LPL inhibition. To understand how LPL binds ANGPTL3/8 and ApoA5 blocks this interaction, we used hydrogen-deuterium exchange mass-spectrometry and molecular modeling to map binding sites of LPL and ApoA5 on ANGPTL3/8. Remarkably, we found that LPL and ApoA5 both bound a unique ANGPTL3/8 epitope consisting of N-terminal regions of ANGPTL3 and ANGPTL8 that are unmasked upon formation of the ANGPTL3/8 complex. We further used ANGPTL3/8 as an immunogen to develop an antibody targeting this same epitope. After refocusing on antibodies that bound ANGPTL3/8, as opposed to ANGPTL3 or ANGPTL8 alone, we utilized bio-layer interferometry to select an antibody exhibiting high-affinity binding to the desired epitope. We revealed an ANGPTL3/8 leucine zipper-like motif within the anti-ANGPTL3/8 epitope, the LPL-inhibitory region, and the ApoA5-interacting region, suggesting the mechanism by which ApoA5 lowers TG is via competition with LPL for the same ANGPTL3/8-binding site. Supporting this hypothesis, we demonstrate that the anti-ANGPTL3/8 antibody potently blocked ANGPTL3/8-mediated LPL inhibition in vitro and dramatically lowered TG levels in vivo. Together, these data show that an anti-ANGPTL3/8 antibody targeting the same leucine zipper-containing epitope recognized by LPL and ApoA5 markedly decreases TG by suppressing ANGPTL3/8-mediated LPL inhibition.

Keywords: angiopoietin-like protein (ANGPTL); apolipoprotein (Apo); epitopes; hydrogen-deuterium exchange mass spectrometry (HDXMS); leucine zipper; lipoprotein lipase (LPL); molecular modeling; nano-imaging; transmission electron microscopy (TEM); triglycerides (TG).

Fig. 1

Discovery of an anti-ANGPTL3/8 therapeutic antibody that blocks the binding of ANGPTL3/8 to LPL. A: Anti-human ANGPTL3/8 antibodies isolated following mouse immunizations were profiled for binding to ANGPTL3/8 complex versus ANGPTL3 or ANGPTL8. A large number of individual clones bound specifically to the ANGPTL3/8 complex, with 153 out of the 172 clones tested exhibiting selectivity for ANGPTL3/8 compared to ANGPTL3 or ANGPTL8 alone. B: Biotinylated anti-ANGPTL3/8 antibody immobilized on streptavidin biosensors was incubated with ANGPTL3, ANGPTL8, or ANGPTL3/8 and transferred to buffer-only wells to monitor dissociation. The results obtained confirmed the specific binding of the antibody to ANGPTL3/8 and not ANGPTL3 or ANGPTL8. The results are representative of 3 independent experiments. C: Biolayer interferometry was used to examine the binding of ANGPTL3/8 to GPIHBP1-LPL in the absence or presence of the anti-ANGPTL3/8 antibody. Avidin-tagged LPL complexed with GPIHBP1 was first immobilized on streptavidin biosensors and incubated with ANGPTL3/8 that had been preincubated with no antibody (trace a), an irrelevant isotype-matched control antibody (trace b), or the anti-ANGPTL3/8 antibody (trace c). The biosensors were transferred to buffer-only wells to monitor dissociation. The anti-ANGPTL3/8 antibody almost completely blocked the binding of ANGPTL3/8 to GPIHBP1-LPL. The results are representative of 3 independent experiments.

Fig. 2

ANGPTL3/8 complex formation creates a neo-epitope recognized by LPL, ApoA5, and the anti-ANGPTL3/8 antibody. Deuterium uptake was measured after 10-s, 2-min, 10-min, and 1-h incubations in deuterated buffer. Boxed areas in Fig. 2A, B show increased HDX occurring in peptide regions comprising the unique epitope formed in the ANGPTL3/8 complex. Boxed areas in Fig. 2C–H show peptide regions with decreased HDX that are protected when the ANGPTL3/8 complex binds LPL, ApoA5, or the anti-ANGPTL3/8 antibody. A: Deuterium uptake into ANGPTL3 complexed with ANGPTL8 relative to deuterium uptake into ANGPTL3 alone. B: Deuterium uptake into ANGPTL8 complexed with ANGPTL3 relative to deuterium uptake into ANGPTL8 alone. C: Deuterium uptake into ANGPTL3 in ANGPTL3/8 complex bound by GPIHBP1-LPL relative to deuterium uptake into ANGPTL3 in unbound ANGPTL3/8 complex. D: Deuterium uptake into ANGPTL8 in ANGPTL3/8 complex bound by GPIHBP1-LPL relative to deuterium uptake into ANGPTL8 in unbound ANGPTL3/8 complex. E: Deuterium uptake into ANGPTL3 in ANGPTL3/8 complex bound by ApoA5 relative to deuterium uptake into ANGPTL3 in unbound ANGPTL3/8 complex. F: Deuterium uptake into ANGPTL8 in ANGPTL3/8 complex bound by ApoA5 relative to deuterium uptake into ANGPTL8 in unbound ANGPTL3/8 complex. G: Deuterium uptake into ANGPTL3 in ANGPTL3/8 complex bound by anti-ANGPTL3/8 antibody relative to deuterium uptake into ANGPTL3 in unbound ANGPTL3/8 complex. H: Deuterium uptake into ANGPTL8 in ANGPTL3/8 complex bound by anti-ANGPTL3/8 antibody relative to deuterium uptake into ANGPTL8 in unbound ANGPTL3/8 complex. ApoA5, apolipoprotein A5.

Fig. 3

ANGPTL3 CCDs and the ANGPTL8 CCD that become unmasked in the ANGPTL3/8 complex bind LPL, ApoA5, and the anti-ANGPTL3/8 antibody. Blue regions show low HDX, while orange and red regions show high HDX. A: AlphaFold prediction of the ANGPTL3 CCD structure. B: AlphaFold model of the ANGPTL3 CCD in the ANGPTL3/8 complex with mapped HDX relative to ANGPTL3 alone, indicating the presence of a novel leucine zipper-like motif consisting of L40, L46, L50, L51, L53, and L57. C: AlphaFold model of the ANGPTL3 CCD in the ANGPTL3/8 complex bound to GPIHBP1-LPL with mapped HDX relative to unbound ANGPTL3/8 complex. D: AlphaFold model of the ANGPTL3 CCD in the ANGPTL3/8 complex bound to ApoA5 with mapped HDX relative to unbound ANGPTL3/8 complex. E: AlphaFold model of the ANGPTL3 CCD in the ANGPTL3/8 complex bound to the anti-ANGPTL3/8 antibody with mapped HDX relative to unbound ANGPTL3/8 complex. F: AlphaFold prediction of ANGPTL8 structure. G: AlphaFold model of ANGPTL8 in the ANGPTL3/8 complex with mapped HDX relative to ANGPTL8 alone, indicating the presence of a novel leucine zipper-like motif consisting of L35, L38, L45, L49, L60, and L67. H: AlphaFold model of ANGPTL8 in the ANGPTL3/8 complex bound to GPIHBP1-LPL with mapped HDX relative to unbound ANGPTL3/8 complex. I: AlphaFold model of ANGPTL8 in the ANGPTL3/8 complex bound to ApoA5 with mapped HDX relative to unbound ANGPTL3/8 complex. J: AlphaFold model of ANGPTL8 in the ANGPTL3/8 complex bound to the anti-ANGPTL3/8 antibody with mapped HDX relative to unbound ANGPTL3/8 complex. CCD, coiled coil domain.

Fig. 4

Binding of anti-ANGPTL3/8 antibody Fab to ANGPTL3/8 complex. A: Multiple 2D class averages for the anti-ANGPTL3/8 antibody Fab binding ANGPTL3/8 complex obtained from negative stain TEM experiments show a well-defined approximate 8 nm density with a bi-lobed structure typical for an IgG Fab (cyan). There is an elongated density about 6–15 nm long and approximately 3 nm wide, that is not well-defined (orange), likely consisting of three ANGPTL3 CCDs associated with the ANGPTL8 CCD. Located on the opposite end of the narrow, elongated density, there is a 7–11 nm less well-defined density (yellow), likely consisting of the three ANGPTL3 FLDs. B: Detailed view of a representative thumbnail class average image obtained from the third round of alignments obtained by negative stain TEM. All eight images from this round of alignments are shown in supplemental Fig. S11. The image shown here is that of the first of the eight images (top left image from supplemental Fig. S11). The image is consistent with an ANGPTL3:ANGPTL8 ratio of 3:1 in the ANGPTL3/8 complex. C: A molecular model shows the anti-ANGPTL3/8 antibody Fab binding the ANGPTL3/8 complex. D: Head-on view of the ANGPTL3/8 complex showing interaction of the ANGPTL3 and ANGPTL8 CCDs via leucine zipper motifs. The FLDs of the three ANGPTL3 molecules in the ANGPTL3/8 complex are shown in the background. TEM, transmission electron microscopy; FLDS, fibrinogen-like domains.

Fig. 5

Modeling of ANGPTL3 and the ANGPTL3/8 complex. Leucine and isoleucine residues are highlighted. A: A model for the full-length ANGPTL3 structure made by joining three ANGPTL3 CCDs via amide bonds to the crystal structure of the FLD trimer. B: A model for the ANGPTL3/8 complex showing a tetrameric assembly containing the three CCDs of ANGPTL3 and the single ANGPTL8 CCD, revealing how the three ANGPTL3 CCDs and ANGPTL8 associate together, consistent with the rules of coiled coil multimeric states described by LOGICOIL. C: Head-on view of the ANGPTL3 trimer, showing leucine and isoleucine interactions amongst the three ANGPTL3 CCDs. D: Head-on view of the ANGPTL3/8 tetrameric complex, showing leucine and isoleucine interactions between the ANGPTL8 CCD and the three ANGPTL3 CCDs. CCDS, coiled coil domains.

Fig. 6

Models of the ANGPTL3/8 complex binding the anti-ANGPTL3/8 antibody Fab and LPL. Blue regions show the areas of decreased HDX. A: Model showing binding of the ANGPTL3/8 complex to the anti-ANGPTL3/8 antibody Fab with HDX relative to unbound ANGPTL3/8 complex. The location of W105 in the Fab heavy chain is highlighted in red. B: Model showing binding of the ANGPTL3/8 complex to GPIHBP1-LPL with HDX relative to unbound ANGPTL3/8 complex. C: Head-on view of the ANGPTL3/8 tetrameric complex bound to the antibody Fab, with the location of W105 in the Fab highlighted in red. D: Head-on view of the ANGPTL3/8 tetrameric complex bound to LPL, suggesting that LPL and the anti-ANGPTL3/8 antibody bind ANGPTL3/8 in a similar manner.

Fig. 7

The anti-ANGPTL3/8 antibody blocks the ability of ANGPTL3/8 to inhibit LPL. A: The ability of the anti-ANGPTL3/8 antibody or evinacumab to block human ANGPTL3/8-mediated inhibition of human LPL activity was assessed using human LPL-stable expression cells with fluorescent lipase substrate. Results are shown as the mean ± SD (n = 6 from 3 independent experiments). B: The ability of the anti-ANGPTL3/8 antibody or evinacumab to block mouse ANGPTL3/8-mediated inhibition of mouse LPL activity was assessed using mouse LPL-stable expression cells with fluorescent lipase substrate. Results are shown as the mean ± SD (n = 6 from 3 independent experiments).

Fig. 8

The anti-ANGPTL3/8 antibody does not block ANGPTL3-mediated LPL inhibition. A: The ability of the anti-ANGPTL3/8 antibody or evinacumab to block human ANGPTL3-mediated inhibition of human LPL activity was assessed using human LPL-stable expression cells with fluorescent lipase substrate. Results are shown as the mean ± SD (n = 6 from 2 independent experiments). B: The ability of the anti-ANGPTL3/8 antibody or evinacumab to block mouse ANGPTL3-mediated inhibition of mouse LPL activity was assessed using mouse LPL-stable expression cells with fluorescent lipase substrate. Results are shown as the mean ± SD (n = 6 from 2 independent experiments).

Fig. 9

The anti-ANGPTL3/8 antibody blocks the ability of ANGPTL3/8 to inhibit EL much less potently than it blocks the ability of ANGPTL3/8 to inhibit LPL. A: The ability of the anti-ANGPTL3/8 antibody or evinacumab to block ANGPTL3/8-mediated inhibition of human EL activity was assessed using EL-stable expression cells with fluorescent PLA1 substrate. Results are shown as the mean ± SD (n = 4 from 3 independent experiments). B: The ability of the anti-ANGPTL3/8 antibody or evinacumab to block ANGPTL3-mediated inhibition of human EL activity was assessed using EL-stable expression cells with fluorescent PLA1 substrate. Results are shown as the mean ± SD (n = 4 from 3 independent experiments). EL, endothelial lipase.

Fig. 10

The anti-ANGPTL3/8 antibody potently lowers TG in vivo. Hypertriglyceridemic CETP/ApoA1 transgenic mice were administered a single dose of the anti-ANGPTL3/8 antibody (at 1, 3, 10, or 30 mg/kg) or an irrelevant isotype-matched control antibody (at 30 mg/kg). Blood was collected predose and 1 day, 7 days, 15 days, 21 days, and 28 days following administration of the single dose of the antibody. Serum TG levels were measured using a Roche Cobas assay. Results are shown as the mean ± SEM for 7 mice for each antibody group for each time point. Asterisks denote p-values of < 0.05. ApoA1, apolipoprotein A1.