Apolipoprotein E Mediates Enhanced Plasma HDL Cholesterol Clearance by Low Dose Streptococcal Serum Opacity Factor via Hepatic LDL Receptors In Vivo

Abstract

Objective

Recombinant streptococcal serum opacity factor (rSOF) mediates the in vitro disassembly of human plasma HDL into lipid-free (LF) apolipoprotein (apo) A-I, a neo HDL that is cholesterol-poor, and a cholesteryl ester-rich microemulsion (CERM) containing apolipoprotein E. Given the occurrence of apolipoprotein E on the CERM, we tested the hypothesis that rSOF injection into mice would reduce total plasma cholesterol clearance via apo E-dependent hepatic LDL receptors (LDLR).

Methods and Results

rSOF (4 μg) injection into wild type C57BL/6J mice forms neo HDL, CERM, and LF apo A-I, as observed in vitro, and reduced plasma total cholesterol (−43%, t_1/2 = 44 ± 18 min) whereas control saline injections had a negligible effect. Similar experiments with apo E^−/− and LDLR^−/− mice respectively reduced plasma total cholesterol ~0 and 20%. rSOF is potent; injection of 0.18 μg rSOF produces 50% of maximum reduction of plasma cholesterol 3 h post injection, corresponding to a ~0.5 mg human dose. Most cholesterol is cleared hepatically (>99%), with rSOF treatment increasing clearance by 65%.

Conclusion

rSOF injection into mice forms a CERM that is cleared via hepatic LDLR that recognize apo E. This reaction could provide an alternative mechanism for reverse cholesterol transport.

Serum opacity factor, (SOF), a virulence determinant produced by the group A streptococcus, Streptococcus pyogenes, is expressed by approximately half of the clinical isolates of S. pyogenes,¹ an important human pathogen that causes pharyngitis, tonsilitis, impetigo, necrotizing fasciitis, and toxic shock syndrome. SOF is a ~110 kDa protein that is found in both culture supernatants and attached to the surface of streptococci via its cell wall-anchoring motif, LPASG. The N-terminal two-thirds of SOF is composed of alternating variable and conserved regions and contains the fibulin-1 binding domain and a domain that causes plasma to opacify.^2–3 The carboxyl terminus of SOF is highly conserved and contains a repeating peptide domain that binds to fibronectin⁴ and fibrinogen⁵ and mediates streptococcal adhesion to host cells⁶.

SOF opacifies serum by disrupting HDL, its exclusive target,³ and forms a large cholesteryl ester- rich microemulsion (CERM; r ~100 - 200 nm), lipid-free (LF) apolipoprotein (apo) A-I and small neo HDL that are cholesterol-poor and phospholipid-rich.⁷ The CERM contains apo E and its heterodimer with apo A-II as its sole proteins⁸ and the neutral lipids of ~400,000 HDL particles.⁷ SOF is potent and catalytic; within 30 min at 37 °C, 10 nM rSOF quantitatively converts 4 μM HDL to CERM, neo HDL, and LF apo A-I.⁹ Kinetic studies^{7, 9} led to a model of rSOF as a heterodivalent fusogen that uses a high affinity docking site to displace labile apo A-I and bind to exposed CE on HDL; the rSOF-HDL complex recruits additional HDL with its binding-delipidation site and through multiple fusion steps forms a CERM. On the basis of the high CE and apo E content of the CERM particle, we hypothesized that SOF would increase reverse cholesterol transport (RCT) by targeting the CERM to multiple apo E-dependent hepatic receptors.^7–8 Treating plasma with rSOF makes it a better acceptor of macrophage cholesterol efflux and a better lecithin:cholesterol acyltransferase substrate (LCAT),¹⁰ and targets the CERM to cellular uptake via low density lipoprotein receptor (LDLR), the heparin sulfate proteoglycan receptor (Syndecan-1), the LDL-receptor related protein-1 (LRP-1), and the scavenger receptor class B type 1 (SR-BI).⁸ However, the effects of SOF on the final RCT step, hepatic uptake, remained to be tested in vivo. Here we show that intravenous injection of rSOF into mice forms neo HDL, CERM and LF apo A-I in vivo and enhances hepatic clearance of CERM-cholesterol in an apo E- and LDLR-dependent way.

Methods

Materials

A polyhistidine-tagged, truncated recombinant form of sof2 (rSOF) encoding amino acids 38–843 was cloned and expressed in Escherichia coli and purified by metal affinity chromatography as described previously.⁴ All mice (Wild type (WT) C57BL/6J, apo E^−/−, and LDLR^−/−) mice were obtained from the Jackson Laboratory. Mice were maintained on normal chow and used at 11 – 12 weeks of age.

Analytical Methods

Unless otherwise indicated, Tris buffered saline (TBS, 10 mM Tris, 100 mM NaCL, 1 mM EDTA) was used throughout. Total mouse lipoproteins were isolated by flotation at d = 1.21 g/mL and separated into subclasses by size exclusion chromatography over Superose HR6 (GE HealthCare) as previously described.⁷ Typically, 200 – 500 μL of sample was injected and the effluent monitiored by absorbance at 280 nm and by immunoblot and lipid analysis of collected 1 mL fractions. Due to its large size, the CERM elutes close to the column void volume (~15 mL). Plasma cholesterol, phospholipid and triglyceride were determined using kits (Wako Chemicals USA, Richmond VA). All animal procedures were carried out in accordance with Institutional Animal Care and Use Committee (IACUC) guidelines.

Apo A-I Analysis

Selected chromatographic fractions were analyzed by SDS-PAGE on 4-20% Tris-glycine pre-cast gels (Invitrogen, Carlsbad CA). Bands were visualized by probing with rabbit anti-mouse apo A-I (Academy Bio-Medical, Houston TX) and anti-mouse apo E (Abcam, Cambridge, MA). Bands were revealed using SuperSignal® West Femto Chemiluminescence Substrate (Thermo-Fisher Scientific, Rockford, IL).

Mouse Plasma Cholesterol Decay Kinetics

Three types of kinetics studies were conducted in mice. In the first experiment, various doses of rSOF or saline were injected via tail-vein, and at various times following injection the mice were anesthetized, their chest cavities opened and the blood collected by heart puncture, and transferred to wet ice. The plasma was separated from the cells by low speed centrifugation at 4 °C and the total plasma cholesterol determined. Each time point represents triplicate mice and each cholesterol measurement was performed in triplicate. In some experiments plasma was also assayed for phospholipid and triglyceride.

Mice were also injected with various rSOF doses (n = 3 mice/dose), their plasma collected after 3 h, chilled on wet ice, and split into two aliquots. One was analyzed for total cholesterol whereas total plasma lipoproteins (TLP) were isolated from the other aliquot by flotation at d = 1.21 g/mL and analyzed by size exclusion chromatography.

In the second experiment, [³H]cholesterol in ethanol (50 μL) was added to pooled mouse plasma (~1.5 mL) and incubated overnight at 37 °C after which analysis of the plasma showed ~98% of the [³H]cholesterol was converted to its ester. The plasma was split into two equal volumes, treated with rSOF (4 μg/mL) or an equal volume of saline, and incubated for 18 h at 37 °C. Mice were injected intravenously via tail-vein with saline or rSOF-treated plasma, sacrificed and bled by heart puncture at various times. Plasma was isolated by low speed centrifugation, β-counted, and the percent injected radioactivity calculated.

In the third experiment, [³H]cholesterol was incubated over night at 37 °C as described above, and split into one aliquot that was treated with rSOF (2 μg/mL) and another that was treated with an equal volume of saline. Size exclusion chromatography showed that during after incubation of whole mouse plasma with [³H]cholesterol overnight at 37 °C, [³H]cholesterol transferred from small to large HDL with the remainder appearing in the LDL range (Supplementary Figure 1 A, B); this interconversion was also reported in the rat ({Ponsin, 1993 #563}. After subsequent overnight incubation with rSOF all [³H]cholesterol in HDL disappeared with concurrent appearance of [³H]cholesterol in CERM without any change in the [³H]cholesterol in the LDL range (Supplementary Figure 1 C).

The samples (0.35 μCi/50 μL) were injected into two groups of mice (n = 3) and after 1 h the mice were anesthetized and perfused with saline via the aorta, and the liver, spleen, testes, lungs, and kidneys were collected, weighed, lyophilized, and extracted with hexane/isopropanol(3/2), which was transferred to scintillation vials and the solvent removed by evaporation. Scintillation fluid was added to the vials, which were β-counted, and the total organ uptake calculated (cpm). Between 35 and 40% of the radioactivity was recovered in the organs.

Statistical Analysis

All values are expressed as mean ± SD. Data were compared using ANOVA followed by Student’s t-test for pair-wise comparisons against rSOF-treated and saline-treated groups. Values of p < 0.05 were considered statistically significant. Statistical analysis and calculation of regression curves was performed using SigmaPlot 11.2 (Systat Software, Inc.). Rate constants were calculated from regression fits of the data points vs time.

Results

Effects of rSOF on Plasma Cholesterol

rSOF (4 μg) or an equal volume of saline was injected into WT C57BL/6J, apoE^−/−, and LDLR^−/− mice, which were sacrificed at various times post injection (Figure 1). Data were fitted to a first order polynomial except for the cholesterol decay curves for WT and LDLR^−/− mice injected with rSOF which were fitted to a 3-parameter exponential decay. Whereas WT mice receiving saline had similar plasma cholesterol levels irrespective of the time post-injection (Figure 1A), plasma cholesterol decay half time in WT mice receiving rSOF was t_1/2 = 44 ± 18 min (r² >0.94). Plasma total cholesterol was significantly reduced by 30 min and remained low, with values at 20 h reduced by 43% or 57 mg/dL. We also measured phospholipids and triglycerides. In mice receiving rSOF, plasma phospholipid concentrations were not significantly changed between t = 0 to 3 hr, but were reduced 36% (p <0.01) at 20 hr. In some but not all experiments, rSOF transiently increased plasma triglyceride concentration, which returned to baseline by 20 hr (Supplementary Table 1). Thus, rSOF injection rapidly reduces plasma total cholesterol without a global decrease in other plasma lipids.

Figure 1.

Plasma cholesterol kinetics following injection of saline or rSOF into (A) WT, (B) apo E^−/− or (C) LDLR^−/− mice. A. Decrease in plasma cholesterol + rSOF: t_1/2 = 44 + 18 min⁻¹, r² > 0.94; B, saline vs. rSOF ns; C, Decrease in plasma cholesterol + rSOF: t_1/2 = 132 + 24 min⁻¹, r² > 0.99.

To probe the role of apo E and the LDLR in rSOF reduction of plasma cholesterol, we also tested apo E^−/− and LDLR^−/− mice. Apo E−/− had elevated plasma total cholesterol at baseline, and plasma cholesterol was not affected by rSOF injection (Figure 1B.) Response of LDLR−/− mice was intermediate between WT and apo E^−/− mice. rSOF injection effected a 20-25% decrease in plasma cholesterol at 1 to 3 hr, with a return to control levels by 20 hrs (Figure 1C).

Size exclusion chromatographic analysis suggests that in the WT mice, the loss of cholesterol was due to the conversion of HDL-C to CERM which is then hepatically removed (Figures 2 A, 6). At t = 0 (no rSOF) the VLDL peak at an elution volume (EV) of 15.5 mL is nil compared to the peak for plasma protein to which all data were normalized. At 30 min a prominent peak is observed at EV = 14.9, the EV of CERM in vitro.⁷ At 60 and 180 min post injection, the magnitude of the CERM peak had increased and shifted to an EV corresponding to smaller particles. At 1200 min, the peak is greatly reduced and has a peak EV = 15.2, between those of VLDL and the largest CERM. Thus, rSOF injection into mice induces formation of CERM, which is cleared from the plasma with a concomitant 43% reduction in plasma cholesterol. Although based on light scattering and not absorption, the reduction in the peak at EV ~ 15 mL suggests that most of the cholesterol remaining in plasma at 1200 min is not CERM-associated. This is confirmed by cholesterol assays of the plasma profiles from the same WT mice as Figure 1 A. At t = 30, 60, and 180 min respectively, 74, 88, and 93% of the formerly HDL-cholesterol was associated with the CERM (Figure 2 E – H). These data also clearly show a small fraction of cholesterol eluting as a peak with the same EV as that of neo HDL (EV = 32 – 34 mL). Finally, at 1200 min the data show that the CERM peak has cleared and the reappearance of a cholesterol-positive peak eluting as HDL. Assays of the same fractions for phospholipid and triglyceride shown in Supplementary Figure 2 showed that the total cholesterol/phospholipid weight ratio in HDL at t = 0.60 and decreased to 0.12 and 0.08 in neo HDL at 60 and 180 min respectively and recovered to 0.32 in HDL at 1200 min. These total cholesterol/phospholipid weight ratios for HDL and neo HDL in vivo are comparable to what we observed in vitro with human HDL, 0.57 for HDL and 0.10 for neo HDL respectively {Gillard, 2007 #3}.

Figure 2.

A – C: Size exclusion chromatography of the CERM/VLDL peak of plasma isolated from mice (n = 3) at various times post rSOF injection. (A) WT C57BL/6J, (B) ApoE^−/−, and (C) LDLR^−/− plasma of rSOF-treated mice was analyzed to follow the CERM size profile as a function of time. The samples were from the experiment of Figure 1 and are normalized to the maximum absorption peak for plasma proteins. D – H: Chromatographic profile of plasma total cholesterol following rSOF (4 μg) injection into WT mice at various times as shown. Left and right vertical lines denote elution volumes for HDL and neo HDL respectively. Plasma (35 μL each) from three mice was pooled and diluted to 140 μL and injected into the size exclusion column using a 100 μL loop. One mL fractions were assayed for total cholesterol. Filled curves in E – G correspond to the profiles for neo HDL.

Figure 6.

Organ uptake of [³H]CE from plasma containing HDL-[³H]CE without (■) and with (■) pre incubation with rSOF. Uptake time = 1 h post injection. The bar heights for spleen, testes, lung, and kidney have been multiplied by 10 for better comparison. *p = 0.034.

The chromatographic profiles of plasma from apo E^−/− and LDLR^−/− mice following injection of saline were distinct from those of WT mice, both showing profound CERM accumulation. As expected for both of these models of hyperlipidemia, the magnitude of the absorbance corresponding to VLDL (EV =15.5) at t = 0 min was higher than that of WT mice, particularly for the apoE^−/− mice (Figure 2 A – C). Following rSOF injection into apo E^−/− mice, the absorbance due to CERM formation increased with time. Unlike the profiles of WT mice, which showed a shift in the EV toward the void volume at 30 min, the CERM that formed in the plasma of apo E^−/− mice were much smaller, approximately the size of VLDL seen at t = 0. The CERM peak accumulated with time and by 1200 min larger particles were observed (Figure 2 B).

LDLR^−/− chromatographic profiles were similar to those observed for the plasma of WT mice, with peaks corresponding to the CERM in LDLR^−/− mice shifted into the void volume between 60 and 180 min, indicating formation of large CERM (Figure 2 C). At 180 min the peak was bimodal with the shoulder seen at 60 min appearing as a prominent peak. By 1200 min, the CERM peak was partially cleared, only twice that of the VLDL seen at t = 0.

Effects of rSOF on the Turnover of Plasma [³H]CE

We compared the turnover of plasma [³H]CE with and without rSOF treatment. A 3-parameter exponential fit of our data (Figure 3 A) gave a rate constant, k, of 0.055 ± 0.02 min⁻¹ (r² >0.94, t_1/2 = 18.2 ± 0.6 min⁻¹) for the disappearance of [³H]CE without rSOF treatment and rates that were 1.4 times greater (k = 0.132 ± 0.02 (r² >0.98, t_1/2 = 7.7 ± 0.1 min) after incubation with rSOF. The respective reductions in plasma radioactivity at 90 min were −58 ± 5% vs. −91 ± 3 without and with rSOF incubation. Thus, in mice, rSOF profoundly increased the rate and magnitude of clearance of total plasma cholesterol (Figure 1 A) and plasma [³H]CE (Figure 3 A).

Figure 3.

(A) Clearance of [³H]cholesterol-labeled plasma. Plasma decay curves for [³H]cholesterol-labeled plasma (100,000 cpm) with (●) and without (o) rSOF treatment. (B) Dose response of total plasma cholesterol 3 h after injection of various rSOF doses into mice (n = 3 mice/dose).

rSOF Dosing in Mice

The plasma total cholesterol showed a dose-dependent decrease with increasing dose and a half maximum decrease at 0.18 μg (Figure 3 B). To analyze the effect of rSOF dose on product formation in vivo, the TLP from the same mice were analyzed by size exclusion chromatography, which separates VLDL/CERM, HDL, and LF apo A-I (Figure 4 A I, II, and V respectively). Under similar conditions, neo HDL formed in vitro elutes at 32 mL (Figure 4 A, III, Arrow).⁷ At 1 ng/mouse the chromatographic profile was only slightly altered relative to that of control mice receiving saline (Figure 4 A). However, as the dose was increased to 100 ng, the Peaks I and II respectively were shifted toward earlier and later elution volumes. These respectively correspond to those of a larger CERM and a putative neo HDL that shifts to a later elution volume (smaller particle size) with increasing dose (Figure 4 B and C). As the dose increased to 10,000 ng, Peak I became nearly symmetrical and eluted as an authentic CERM sample,⁷ and the peak for neo HDL decreased to a very low but still detectable level, shown on an expanded scale (Figure 4D - G). These data along with the cholesterol chromatographic profiles (Figures 1D-H) show that rSOF generates two of its in vitro products, CERM and neo HDL, in vivo.

Figure 4.

Chromatographic profiles of mouse TLP 3 h post injection of various doses of rSOF as shown (A-G). Grey filled curve is the profile of control untreated plasma. Unfilled curves are profiles for plasma treated with the indicated amounts of rSOF. I - V respectively are the elution positions of VLDL, HDL, neo HDL (in vitro), neo HDL (this study) and LF apo A-I. In D – G, the scale for the Y-axes for the grey curves is 1% of that of the black curves.

rSOF Forms LF Apo A-I In Vivo

No peak for LF apo A-I was observed in Figure 4 because flotation to obtain TLP sediments unlipidated proteins. To establish formation of LF apo A-I in vivo, we collected selected chromatographic fractions of whole plasma from mice at various times following rSOF injection (4 μg/mouse) and probed for LF apo A-I by immunoblot analysis. The respective elution volumes of HDL and neo HDL and LF apo A-I formed in vitro are 31.3, 32.4 and 33.9 mL (± 0.025 mL).^{7, 11} Immunoblots of chromatographic fractions of control plasma (no rSOF) show the expected association of apo A-I with HDL (elution volumes 30 and 31 mL) with respectively little and none seen at elution volumes of 34 and 35 mL, which correspond to LF apo A-I in vitro⁷ (Figure 5 B). In contrast, 1 h post rSOF injection, apo A-I was shifted to later eluting fractions centered on neo HDL (EV = 32 and 33 mL) with a smaller but much greater amount relative to control in fractions 34 and 35 where LF apo A-I elutes. At 3 h, the distribution was similar although the intensities of the bands for LF apo A-I were lower. By 20 h, the distribution of apo A-I returned to a profile similar to that of plasma from untreated mice. Immunoblot analysis also showed that the CERM contains apo E but no apo A-I and that neo HDL contains apo A-I and a trace of apo E (Figure 5 C). Thus, rSOF forms CERM, neo HDL and LF apo A-I in vivo in a reaction that is qualitatively similar to the in vitro reaction.⁷ Similarly, apo E appeared in neo HDL at early time points post rSOF but was associated with HDL at 1200 min (Supplementary Figure 3).

Figure 5.

Immunoblot analysis of chromatographic fractions 1 h post rSOF injection into mice. (A) Chromatographic absorbance (280 nm) profile of in vitro rSOF reaction mixture at completion. (B) Immunoblot of mouse plasma for apo A-I at various times post rSOF injection. (C) Immunoblot analyses for apos A-I and E in the CERM and neo HDL fractions.

Organ Distribution of Plasma-[³H]CE

Mice were injected with [³H]CE-labeled plasma aliquots with or without prior treatment with rSOF. As shown in Figure 6, the liver was the major site for clearance of HDL-[³H]CE after injection. Similarly, the liver was also the major site of [³H]CE removal of rSOF-treated plasma containing CERM-[³H]CE with the uptake of radiolabel at 1 hr being ~65% higher than that for control. The total radioactivity associated with the spleen, kidneys, and testes was much less than that found in liver and differences between [³H]CE uptake from control and + rSOF-treated plasma for these organs were small and not significant. Thus, the primary in vivo effect of the rSOF reaction is to potentiate HDL-[³H]CE uptake by the liver.

Discussion

We previously proposed that each in vitro rSOF reaction product—neo HDL, LF apo A-I and CERM—had the potential to enhance RCT.⁷ Both LF apo A-I and neo HDL are ligands for macrophage cholesterol efflux via the ATP-binding cassette transporter A1 (ABCA1).^{10, 12–13} and CERM-CE uptake by hepatoma cells is greater than that of HDL-CE and occurs via multiple apo E-dependent receptors.⁸ However, for the SOF reaction to have its greatest therapeutic potential, it must form the same products in vivo. Our data clearly show that the rSOF reaction in mice produces all of the products formed in vitro.⁷ However, the relative amounts and/or sizes of each product were different from what was observed in vitro. At early time points, the LF apo A-I accumulation in vitro⁷ and in vivo are comparable but at later time points the plasma apo A-I concentration is greatly diminished or absent (Figure 5 B). As with nascent HDL and rHDL, which have sizes and compositions similar to those of neo HDL, this could occur through fusion mediated by multiple lipoprotein-modifying activities, including LCAT, which mediates apo A-I fusion with HDL.^14–15 Alternatively, apo A-I could form nascent HDL via interaction with the ABCA1 transporter on cells at multiple tissue sites, or clear renally. The disappearance of neo HDL could follow similar pathways. Like preβ HDL, which is cleared renally,¹⁶ neo HDL is small⁹ and has preβ mobility (data not shown). The same lipoprotein-modifying activities that mediate fusion of nascent HDL^{15, 17} and rHDL¹⁸ could mediate fusion of neo HDL with itself or with coexisting HDL giving fusion products that reenter the rSOF reaction pathway to produce additional CERM that are subsequently cleared. Another possibility is that neo HDL is directly cleared hepatically.

Injection of rSOF into WT mice produced a robust reduction in plasma total cholesterol (Figure 1 A) that was associated with the formation of CERM that disappeared from plasma over a similar time frame (Figure 2 A, Figure 2 D - H). The large CERM formed at 30 min was replaced by a smaller CERM at 60 and 180 min suggesting a preference for the clearance of large CERM. Some CERM particles are very large (200 – 400 nm at t = 30 min),⁷ so there was some concern that they might be too large to enter the hepatic perisinusoidal space, which contains lipoprotein receptors. However, the kinetic (Figure 1A and 1C) and tissue data (Figure 6) indicate clearance via interaction with lipoprotein receptors, including LDLR. Given the quantitative rSOF-mediated transfer of cholesterol from HDL, its exclusive target, to CERM in vitro and the formation of CERM in vivo,⁷ we attribute the in vivo decrease in plasma cholesterol to the successive conversion of HDL to CERM and its hepatic removal. The concurrent reduction of cholesterol (Figure 1 A), disappearance of the CERM (Figure 2 A, 2D–2H) and accumulation of CERM-[³H]CE in liver (Figure 6), supports our hypothesis that the reduction in plasma cholesterol is via hepatic CERM uptake. Remarkably, in WT mice at 1200 min the cholesterol levels remained low although the CERM had essentially cleared and HDL reappeared, although at levels still below baseline (Figure 2D-H). In LDLR^−/− mice, other apo E receptors⁸ are still present and CERM is cleared although to a lesser extent than in WT mice, and plasma cholesterol has returned to baseline by 1200 min (Figure 1C and 2C). In apo E^−/− mice, rSOF injection results in CERM formation, but no clearance due to absence of apo E, and plasma cholesterol levels are not altered (Figure 1B and 2B).

Studies with [³H]cholesterol complemented these results (Figure 3 A). Injection of [³H]cholesterol after conversion to its [³H]CE gave a plasma clearance time (t_1/2 = 18 ± 5 min) that was reduced by pretreatment with rSOF (t_1/2 = 7.6 ± 1.1 min); this halftime is shorter than that observed for plasma cholesterol decay after rSOF injection (Figure 1A) likely because the [³H]CERM is formed before injection whereas in Figure 1A rSOF must first convert HDL to CERM. As with the clearance of plasma cholesterol, the reduction in [³H]CE was more profound and faster for rSOF-treated plasma than for untreated plasma. Changes in the rates of disappearance of cholesterol though rapid are not unprecedented. In mice, the fractional catabolic rate of HDL-CE ether, 5.7 ± 0.60 day⁻¹ (t_1/2 ~ 4 hr), is increased to 21.7 ± 6.4 day⁻¹ (t_1/2 ~ 45 min) by the over expression of phospholipid transfer protein.¹⁹

Our in vitro studies⁸ showed that in vitro CERM uptake by hepatoma cells occurs via multiple apo E-dependent lipoprotein receptors²⁰—LDLR, the low density lipoprotein-related protein 1 (LRP1) and Syndecan-1, the primary heparan sulfate proteoglycan that clears triglyceride-rich lipoproteins.²¹ Thus, we reasoned that the cholesterol clearance kinetics in apo E^−/− and LDLR^−/− mice would provide mechanistic support to our earlier hypothesis⁷ that the occurrence of apo E on a CERM would mediate increased plasma CE clearance as CERM in vivo via LDLR. In apo E^−/− mice, Figure 2 B clearly shows rSOF-mediated CERM formation, although the maximum amount formed is less than that seen in WT (Figure 2 A vs B) in spite of the disappearance of CERM in the WT mice. Also, the size of the CERM formed in apo E^−/− mice up to 180 min is smaller than those formed in WT mice. The smaller amount and initial size of CERM formed in the apo E^−/− vs. WT and LDL^−/− mice is likely due to the lower plasma concentration of HDL, the CERM precursor, in apo E^−/− mice of HDL,^22–23 and the correlation of decreasing the HDL/rSOF ratio reducing CERM size.⁷ However, at 1200 min a larger and more prominent CERM peak is observed in apo E^−/− mice which may be due to CERM fusion. Despite CERM formation in the apo E^−/− mice, according to Figure 1 B, plasma cholesterol remains unchanged up to 1200 min. Given that rSOF forms CERM in apo E^−/− mice but it is not cleared we conclude that apo E is required for CERM-CE clearance.

A similar kinetic analysis for rSOF-treated LDLR^−/− mice revealed a greatly modulated reduction in plasma cholesterol clearance (−20% vs. −43% for WT). These data were corroborated by the chromatographic profiles showing a slower reduction in CERM absorbance (Figure 2 C) with the CERM peak absorbance at 1200 min post rSOF injection being more profound than that of WT. Given the CERM formation in both WT and LDL^−/− mice but a slower rate of cholesterol clearance in the latter case, we conclude that the LDLR is an important but not exclusive ligand for CERM uptake. Although our in vitro hepatocyte studies showed that CERM-CE uptake is mediated by multiple apo E-dependent receptors as well as SR-BI,⁸ our in vivo data suggests that most hepatic CERM-CE uptake occurs via the LDLR and the final near total clearance at t = 1200 min occurs more slowly via other receptors.

The liver was the major site for clearance of both control and rSOF-treated HDL-[³H]CE, with liver uptake being ~65% higher with rSOF treatment (Figure 6) accounting for >99% plasma cholesterol removal. Uptake by other organs was small and not significantly affected by rSOF treatment. Thus, the primary in vivo effect of the rSOF reaction is to potentiate the transfer of HDL-[³H]CE to the liver. Dosing tests revealed rSOF in vivo potency; an ED_1/2 = 0.18 μg/mouse (Figure 3 B) corresponds to a concentration of 2 pM and size exclusion chromatography revealed rSOF activity at a dose of 1 ng/mouse = ~ 10 fM (Figure 4 A). These respective doses correspond to ~0.5 mg and 3 μg for a 75 kg human. Given that the LDLR mediates a major component of hepatic CERM removal, it may be possible to boost rSOF potency with HMG-CoA reductase inhibitors, which increase the number of hepatic LDLR.

Could the rSOF Reaction be Used Therapeutically?

Our tests add support our initial hypothesis that the rSOF reaction can be used to enhance RCT. As a virulence factor it is reasonable to question the therapeutic value of rSOF. Although S. pyogenes causes several diseases with varying degrees of morbidity (impetigo, tonsillitis, and pharyngitis), they are usually not fatal. More importantly, there are examples of bacterial virulence determinants and toxins that are safe therapeutically. Anthrax toxins have been used to target and kill cancer cells and the isolated toxins are not always health hazards.²⁴ The botulinum toxin (Botox) is used as a muscle relaxant and to treat facial wrinkles.²⁵ Streptokinase, a streptococcal virulence factor, is used as a thrombolytic agent.²⁶ Thus, while bacterial virulence factors may contribute to the pathogenesis of infections, harm from the purified virulence factors is not obligatory. Although rSOF itself may not reach human therapy, the reaction, which is a physical and not enzymatic process, might be catalyzed by rSOF analogs or even small molecules based on the mechanism that we described previously.⁷

Although increasing HDL-C is thought to be cardioprotective, the relationship is not axiomatic and there is a growing body of evidence that efficient RCT, which may or may not be associated with high HDL-C, is more important to cardioprotection than is high HDL. Macrophage cholesterol efflux to plasma better correlates with the levels of preβ1 HDL, presumably a preferred cholesterol acceptor, rather than total HDL.²⁷ In murine models of atherosclerosis, hepatic over expression of SR-BI decreases plasma HDL-C,^28–30 increases HDL-CE clearance,^29–31 biliary cholesterol and its transport into bile,^{28, 30, 32} and reduces atherosclerosis.^33–35 Conversely, ablated or attenuated hepatic SR-BI expression elevates plasma HDL-C and reduces selective HDL-CE clearance³⁶ and is atherogenic.^36–37 Finally, a study in humans showed that even after adjustment for HDL-C and apo A-I levels, the magnitude of macrophage cholesterol efflux capacity is inversely associated with atherosclerosis.³⁸ Thus, future therapies might better focus on enhancing RCT irrespective of effects on plasma HDL-C concentrations.³⁹ The rSOF-meditated reduction of plasma cholesterol via increased hepatic CE uptake suggests that new therapies based on this reaction may be feasible. This will ultimately be determined by tests of the effects of rSOF on RCT and atheroregression in mouse models of dyslipidemia and atherosclerosis.