Measurement of apolipoprotein E and amyloid β clearance rates in the mouse brain using bolus stable isotope labeling

Several methods have been described to study protein turnover in animals using stable isotope labeling. Traditionally, stable isotope incorporation was measured by gas chromatography mass spectrometry (GC-MS) quantitation of labeled amino acids obtained following protein derivatization [9]. However, this technique is limited to measuring large quantities of proteins, and consequently has only been used to analyze total tissue protein turnover or the kinetics of highly abundant proteins [31‐33]. Recently the more sensitive analytical technique of LC-MS has been applied to quantify the turnover of specific proteins following administration of a stable amino acid in the diet of animals [14, 15]. Though these techniques have provided useful information on the turnover of abundant proteins in various organs, the requirement of labeled amino acid delivery via the diet has been a technical issue. Creating a diet enriched with an isotopic label is costly, and inability to control feeding patterns in animals such as mice and rats requires long term exposure (several hours to days) to the labeled diet to achieve reliable and consistent isotope levels in tissues. This is especially problematic in studying proteins with rapid turnover rates, as the difficulty in accurately measuring label incorporation over short time periods (minutes to hours) limits the sensitivity of kinetic analysis. To create a more practical and efficient method of labeling proteins, we have tested whether pulse labeling of mice could be used to measure protein turnover rates in the mouse brain.

The outline of our experimental design for the labeling of mice and tissue processing is shown in Figure 1A. Mice were intraperitoneally (IP) injected with a bolus of ¹³ C₆-leucine, a non-radioactive stable isotope form of the amino acid leucine. We chose ¹³ C₆-labeled leucine because it is one of the essential amino acids that rapidly crosses the blood brain barrier via facilitative neutral amino acid transport [34]. Intraperitoneal administration of the label was chosen because it is straightforward and quick, and it allows for high bioavailability upon absorption into the bloodstream. Following the injection, we observed a rapid increase in the amount of ¹³ C₆-leucine in the plasma of the mice over the first hour, as measured by the ratio of labeled to unlabeled free leucine quantified by GC-MS (Figure 1B). After predetermined time points, the mice were euthanized and the brains were quickly removed and frozen. Upon collection of all of the brain samples for each time course, the tissue was then lysed in a 1% Triton X-100 lysis buffer and the protein of interest (apoE and Aβ for this study) was immunoprecipitated with protein-specific antibodies covalently coupled to protein G sepharose beads. Only the cortex of the brain was used in this study; however this technique could easily be applied to measure turnover rates in other regions of the brain. The isolated proteins were eluted off the beads using formic acid, and the concentrated samples were digested with trypsin to generate protein-specific peptides for each protein. These peptide mixtures were then subjected to LC-MS analysis for identification and characterization as described below.

We used a targeted LC-MS approach to accurately and precisely quantify the amount of labeled apoE and Aβ in the brain. Multiple reaction monitoring (MRM) assays were developed for each protein by first selecting a peptide that shows high MS signal intensity, contains only one leucine residue, and is specific for each protein (LQAEIFQAR for apoE, LVFFAEDVGSNK for Aβ). Synthetic peptides were then directly injected into the MS to select and optimize the MRM product transitions for each parent ion (Figure 2A, Table 1, details of the MRM optimization protocol are found in the METHODS section). These parent/precursor ion groupings were then used for the relative quantitation of the labeled and unlabeled peptides from the brain sample. The area under the curve (AUC) of the MRM ion count during the course of the parent ion elution was calculated for both the labeled and unlabeled peptide peaks (Figure 2B). The AUC for the labeled peak was then divided by the AUC for the unlabeled peak to calculate the tracer-to-tracee ratio (TTR) for each sample.

In order to accurately compare the TTR values between individual brain samples and across cohorts of animals, we developed standard curves for both apoE and Aβ using the stable isotope labeling of amino acids in cell culture (SILAC) method [35]. The standard curve for Aβ was generated as previously described [17]. Since astrocytes are the main cell type in the brain that produce apoE [20], we used primary astrocyte cultures to produce a labeled apoE standard curve. To label newly synthesized apoE, the astrocytes were cultured in leucine-free media supplemented with different ratios of ¹³ C₆-leucine to unlabeled leucine. Under these conditions, all apoE that was synthesized and secreted into the cell media are labeled with the percentage of ¹³ C₆-leucine provided to the cells. After a 48 hr incubation, the cell media was collected and apoE was immunoprecipitated. Following trypsin digestion, the apoE peptides were subjected to LC-MS as described above. The measured amount of percent labeled apoE gave values that were very close to the expected values (Figure 2C). The linear fit had a slope of 0.976 and an R² value of 0.9954. The apoE and Aβ media standards were used in all subsequent experiments to calibrate the quantitation of the mouse brain samples.

In order to calculate the fractional clearance rates (FCRs) of apoE and Aβ from the brain for each cohort of mice, mice were injected with ¹³ C₆-leucine and the brains were removed at predetermined time points following the label administration. To determine the optimal time course for analyzing apoE and Aβ clearance, a preliminary experiment was performed with 1 to 2 mice at each time point. For the actual experiments, several mice (n = 5 to 6) were labeled for each time point and the TTR values were averaged. The averaged TTR values were then plotted over time for the whole cohort of mice to obtain the kinetic time course of label disappearance (Figure 2D). The negative of the slope of the monoexponential curve was then calculated in order to determine the fractional clearance rate for each protein from the brain. Through the use of cell-derived labeled protein standards, this technique yielded highly reproducible results. For instance, the labeling of three different cohorts of wild-type (Wt) control mice resulted in similar FCR values for apoE (Table 2; 0.093, 0.10, and 0.09 pools/hr, mean = 0.094 pools/h ± 0.003). These cohorts did not have the same amount of apoE protein as measured by ELISA [Pool size (PS) in Table 2, likely because of the different genetic backgrounds of the mice. Studies have shown that different mouse strains can have up to a four-fold difference in plasma apoE levels [36, 37]. Since we observed similar FCR values from the Wt mice across the different genetic strains, our results suggest that the different brain apoE levels are not caused by variations in apoE catabolism.

To verify that our labeling technique could measure differences in the clearance rates of proteins, we analyzed the effect of overexpressing LDLR on the apoE clearance rate from the brain. LDLR is a receptor that binds to apolipoprotein B and apoE in the periphery to facilitate the uptake of cholesterol-laden lipoproteins by cells [38]. Previously we have shown that LDLR transgenic mice that overexpress LDLR in the brain have significantly decreased levels of brain apoE [27]. We therefore hypothesized that the apoE clearance rate in the brain would be increased in mice that have elevated LDLR levels. Wt and LDLR Tg mice were labeled with ¹³ C₆-leucine and the apoE TTR values were measured after pre-determined time points. Plots of the TTR values (presented as labeled/unlabeled apoE) over time along with the monoexponential slopes of these curves are shown in Figures 3A and 3B, respectively. Table 2 shows the pool sizes (PS), fractional clearance rates (FCRs), and production rates (PR) for apoE from each genotype. The FCR of apoE was 2.7-fold faster in the LDLR Tg mice in comparison to the Wt mice, while the apoE pool size was 2.8-fold higher for the Wt mice in comparison to the LDLR Tg mice. These values were used to estimate the PR values for both genotypes (see METHODS section for explanation of PR calculation), and no statistical difference in the PR was observed between Wt and LDLR Tg mice. These results convincingly demonstrate that apoE clearance is enhanced in the brains of LDLR Tg mice, providing the likely explanation for the decreased total apoE protein levels. We therefore concluded that the pulsed ¹³ C₆-leucine injection labeling technique is effective for measuring the clearance of proteins from the brain, and could be used to detect differences in FCRs between genetically modified mouse models.

ABCA1 is a transmembrane protein that plays an important role in the efflux of cholesterol and phospholipids to lipid-poor apolipoproteins [39]. In the brain, the level of ABCA1 has been shown to modulate the extent of apoE lipidation and apoE levels. Surprisingly, both deletion and overexpression of ABCA1 in the mouse brain led to a decrease in apoE levels [28‐30]. However, the apoE containing lipoprotein particles isolated from the cerebral spinal fluid (CSF) of ABCA1^−/− mice were poorly lipidated, while those from the CSF of mice overexpressing ABCA1 had higher levels of lipidation compared to Wt animals [28, 30]. Because lowering apoE levels decreased amyloid deposition in the mouse brain[25], it was hypothesized that altering ABCA1 levels would also alter amyloid deposition in APP transgenic mice. Despite decreased apoE levels with both ABCA1 deletion and overexpression, only ABCA1 overexpression caused a significant decrease in amyloid load in the mouse brain of APP transgenic animals [30]. The amyloid load in APP transgenic mice deficient in ABCA1 did not change, or even increased, when ABCA1^−/− mice were crossed to various APP transgenic models [40‐42]. Because of the opposing effects of ABCA1 deletion and overexpression on Aβ accumulation, ABCA1 likely alters Aβ levels through a mechanism distinct from modulating apoE levels. One proposed mechanism is ABCA1 levels could alter Aβ clearance from the brain [30, 43]; however this hypothesis has never been tested in vivo. Therefore, we used our stable isotope labeling technique to study the effect of ABCA1 levels on both apoE and Aβ clearance rates in the mouse brain.

To generate APP transgenic mice that either overexpressed or were deficient in ABCA1 levels, we crossed PDAPP mice with ABCA1 Tg and ABCA1^−/− mice. These animals were then injected with ¹³ C₆-leucine and the FCRs of both apoE and Aβ were measured as described above for the LDLR Tg animals. To limit complications due to Aβ extraction from tissue with amyloid plaques, all experiments were performed on young animals (3.5 months old) prior to the onset of detectable plaque deposition. Plots of the labeled/unlabeled protein values over time along with the monoexponential slopes of these curves are shown in Figures 4 and 5 for apoE and Aβ, respectively. The PS, FCR, and PR values for apoE and Aβ are given in Table 2 and Table 3, respectively. The apoE FCR was 1.5-fold faster in ABCA1 Tg mice and 1.9-fold faster in ABCA1^−/− mice compared to Wt mice; however the difference was only significant for the ABCA1^−/− mice. The apoE PS decreased by 20% in ABCA1 Tg mice and by 51% in ABCA1^−/− mice compared to Wt mice. No differences were observed in the PR of apoE. For Aβ, no differences were observed in the FCR, PS, or PR values (Table 3).

These results demonstrate that both ABCA1 overexpression and deletion increase the fractional clearance rate of apoE from the brain, but have no effect on the Aβ fractional clearance rate. In terms of apoE, the results from the ABCA1^−/− mice parallel nicely with plasma kinetic studies performed in humans with loss-of-function mutations in ABCA1. These studies demonstrated that the catabolism of both high-density lipoprotein (HDL) and low-density lipoprotein apolipoprotein B-100 (LDL B-100) were increased in the plasma of these individuals [44, 45], suggesting decreased stability of poorly lipidated lipoprotein particles. The lipidation of apoE also significantly alters its propensity to bind to LDLR, with increased lipid levels leading to enhanced binding [46]. As a result, the faster apoE clearance rate in the ABCA1 Tg mice may be due to increased LDLR-mediated clearance of the more highly lipidated apoE-containing lipoprotein particles. Therefore, though both ABCA1 overexpression and deletion led to enhanced apoE clearance, the mechanism underlying the difference in each case may be different.

In regards to Aβ, our results suggest that at a young age prior to the onset of plaque deposition, ABCA1 levels do not alter Aβ clearance from the mouse brain. A previous study measuring the disappearance of radiolabeled Aβ injected into the brains of ABCA1^−/− mice also found no effect on Aβ clearance across the blood–brain barrier [47]. Though we cannot rule out the possibility that changes in Aβ clearance develop as the mice age, our current data suggests that the effect of ABCA1 on Aβ deposition in the brain may not occur due to altered metabolism of Aβ. Rather, it is possible that the effect of ABCA1 relates to altered modulation of differentially lipidated forms of apoE on Aβ aggregation or fibrillogenesis.

¹³ C₆-leucine was obtained from Cambridge Isotope Laboratories (Andover, MA, USA). HJ5.2 (Aβ) and HJ6.3 (ApoE) antibodies were made in-house. Protein G Sepharose 4 Fast Flow beads were obtained from GE Healthcare (Piscataway, NJ, USA). Formic acid (Optima LC-MS) was obtained from Fisher Scientific and triethylammonium bicarbonate was obtained from Sigma-Aldrich (St. Louis, MO, USA). Trypsin Gold (mass spec grade) was purchased from Promega (Madison, WI, USA).

The production and characterization of the LDLR transgenic and ABCA1 transgenic mice have been previously described [27, 30]. ABCA1^+/− mice on a DBA background were obtained from the Jackson Laboratory (Bar Harbor, ME, USA). PDAPP mice on a C57/BL/6J background were a generous gift from Eli Lilly (Indianapolis, IN, USA). LDLR Tg^+/− mice were bred to Wt mice to generate mice that were LDLR Tg^+/− and LDLR Tg^−/−. ABCA1 Tg mice were backcrossed to C57/BL/6J mice for 8 generations, and then crossed to DBA mice. PDAPP mice were also crossed to DBA mice and ABCA1^+/− mice were crossed to C57/BL/6J mice to create strains that were on a 50% C57/BL/6J/50%DBA background. The ABCA1 Tg^+/− and PDAPP^+/− mice were then bred to each other to generate ABCA1 Tg^+/−/PDAPP^+/− and ABCA1 Tg^−/−/PDAPP^+/− mice that were used for the experiments. ABCA1^+/− were crossed to PDAPP^+/− mice to generate mice that were PDAPP^+/−/ABCA1^+/−. These mice were then bred to ABCA1^+/− mice to generate mice that were PDAPP^+/−/ABCA1^+/+, PDAPP^+/−/ABCA1^+/−, and PDAPP^+/−/ABCA1^−/−. The PDAPP^+/−/ABCA1^+/+ and PDAPP^+/−/ABCA1^−/− mice were used for all experiments. Mice were maintained under constant light/dark conditions and had free access to food and water. All experimental protocols were approved by the Animal Studies Committee at Washington University in St. Louis.

Prior to injection, the ¹³ C₆-leucine was dissolved in medical-grade normal saline to a concentration of 7.5 mg/mL. The mice were weighed and then intraperitoneally injected with the ¹³ C₆-leucine (200 mg/kg of body weight). After predetermined time points, the animals were anesthetized and the blood was collected by cardiac puncture. The mice were then perfused with PBS-heparin and regional brain dissection was performed. All brain samples were subsequently frozen on dry ice.

Primary astrocytes were cultured from postnatal day (P1) C57/BL/6J mouse pups as described previously [27]. Cells were cultured in serum-containing growth media (DME/F12, 15% fetal bovine serum, 10 ng/mL epidermal growth factor, 100 units/mL penicillin/streptomycin, and 1 mM sodium pyruvate) until they reached 70% confluency. The cell media was then changed to serum free media that did not contain any leucine (DME/F12 without leucine prepared by the Washington University Tissue Culture Support Center, N2 growth supplement, 100 units/mL penicillin/streptomycin, and 1 mM sodium pyruvate) and cultured for 12 h. ¹³ C₆-leucine was then diluted into unlabeled leucine to make labeled/unlabeled percentages that were either 0, 1.25, 2.5, 5, 10, or 20%. These different percent-labeled leucine solutions were then added to separate flasks of primary astrocytes, and the cells were cultured for an additional 48 h. The media was then collected from the cells, spun down at 1500 rpm to clear cellular debris, and stored at −80°C.

Antibody beads were prepared by covalently binding either HJ6.3 (apoE) or HJ5.2 (Aβ) to Protein G Sepharose 4 Fast Flow beads. The beads initially were washed 3 times with ice-cold PBS and then resuspended in ice-cold PBS to make a 50% slurry of beads. 300 μL of the washed 50% beads were then mixed with antibody (0.4 μg/μL of 50% bead mixture), 10 μL of 1% Triton X-100 and ice-cold PBS to make a final volume of 1000 μL. This mixture was then tumble incubated overnight at 4°C. The beads were then washed 3 times with 1% Triton X-100 lysis buffer (Triton X-100, 150 mM NaCl, 50 mM Tris–HCl) and 2 times with 0.2 M triethanolamine (pH = 8.2). Freshly prepared dimethyl pimelimidate in 0.2 M triethanolamine (pH = 8.2) was then added to the beads, followed by a 30 min incubation with tumbling at room temperature to allow for crosslinking. The beads were then washed once with 50 mM Tris (pH = 7.5) to stop the crosslinking reaction, and twice with 0.1% Triton X-100 in PBS. The washing solution was removed by vacuum aspiration, and the beads were resuspended in PBS to make a 50% bead slurry.

Brain cortex samples were weighed and 1% Triton X-100 lysis buffer (Triton X-100, 150 mM NaCl, 50 mM Tris–HCl, 1 X Roche Complete Protease Tablet) was added at a concentration of 150 mg brain tissue/μL of lysis buffer. The samples were then sonicated (2 rounds of 20 1-sec pulses) and centrifuged at 14,000 rpm for 30 min. The supernatant was collected and used for subsequent immunoprecipitation steps. Brain lysates and cell media were pre-cleared with beads not conjugated to antibody by tumble incubating the samples with 50 μL of the 50% bead slurry for 4 h at 4°C. The pre-cleared lysate and media samples were then tumble incubated with antibody-conjugated beads overnight at 4°C. The beads were then washed 3 times with PBS and 3 times with 25 mM triethylammonium bicarbonate (TEABC). Following the last TEABC wash, the TEABC was removed via vacuum aspiration with a pipette tip. Formic acid was then added to the beads to elute the bound proteins, and the mixture was vortexed for 20 min. The beads were then centrifuged at 14,000 rpm for 5 min and the supernatant was collected from the beads. The formic acid supernatant was transferred to a new microcentrifuge tube and evaporated in a Savant SpeedVac for 60 min (37°C). The dried proteins were then resuspended in 20% acetonitrile/80% 25 mM TEABC and vortexed for 30 min. The samples were then digested with 500 ng of mass spectrometry-grade trypsin (Promega) and incubated at 37°C for 16 h. The digested samples were dried again by vacuum evaporation, resuspended in 10% acetonitrile and 0.1% formic acid in water, and transferred to mass spec vials.

LC-MS/MS measurements were performed on a Waters Xevo TQ-S triple quadrupole mass spectrometer (Waters Inc., Milford, MA) coupled to a Waters nano-ACQUITY ultra performance liquid chromatography (UPLC) system, equipped with a Waters nano-ESI ionization source. To identify multiple reaction monitoring (MRM) transitions, the synthetic apoE peptide LQAEIFQAR and synthetic Aβ peptide LVFFAEDVGSNK were purchased from AnaSpec, Inc. (Fremont, CA), and directly infused into the LC-MS for automatic tuning of optimized MRM transitions produced by the peptide. For both the apoE and Aβ peptide, optimal conditions were identified as a capillary voltage of 3.3 kV, source temperature of 80°C, cone voltage of 52 V, purge gas flow rate set at 100 L/hr, and cone gas at 50 L/hr. Obtained MRM transitions (Table 1) were then validated by the analysis of apoE and Aβ cell culture media standards. For the actual experiments, all digested peptide samples were kept at 4°C and 1 μL aliquots were injected onto a Waters BEH130 nanoAcquity UPLC column (C18 particle, 1.7 μm, 100 μm × 100 mm). The peptide mixtures were separated on a reverse-phase nanoUPLC operated at a flow rate of 500 nL/min with a gradient mixture of solvents A (0.1% formic acid in water) and B (0.1% formic acid in acetonitrile). For apoE, the column was initially kept at 99% solvent A for 1.5 min, followed by a separation gradient of 1% to 97% solvent B from 1.5 to 18 min. The column was then kept at 97% solvent B for another 5 min followed by 1% solvent B to re-equilibrate for 10 min to prepare for the next injection. For Aβ, the column was initially kept at 90% solvent A for 7.0 min, followed by a separation gradient of 10% to 45% solvent B from 7 to 12 min. The column was then kept at 45% to 95% solvent B from 12 to 14 min, and at 95% solvent B for another 3 min followed by 10% solvent B to re-equilibrate for 15 min to prepare for the next injection. All raw data were acquired and quantified using Waters MassLynx 4.1 software suite. The labeled/unlabeled ratio was obtained by dividing the area under the curve (AUC) of the MRM total ion for the labeled peptide by the AUC for the unlabeled peptide, and converted to tracer-to-tracee ratios (TTRs) by reference to the standard curve.

The free leucine tracer-to-tracee ratio was measured from the mouse plasma using GC/MS. Plasma proteins were precipitated with ice-cold acetone, and lipids were extracted using hexane solvent. The resulting aqueous fraction was then dried with a vacuum (Savant Instruments, Farmingdale, NY) and converted to t-butyldimethylsilyl derivatives. The free leucine TTR was then measured by monitoring ions with m/z ratios of 200 (unlabeled) and 203 (labeled) [52].

The mice were in steady-state conditions, since the amount of apoE and Aβ did not significantly change over the time period of the kinetic analysis. This was determined by measuring the protein level (via ELISA as described below) for the cohorts of mice at each time point following the stable isotope injection, and comparing across groups. At metabolic steady state, the fraction of the pool that is synthesized per unit time equals the fraction of the pool catabolized per unit time (FCR), which can be calculated as the negative of the slope of the natural log of TTR plotted over time [53]. Production rates (PRs) were determined as: PR (protein amount/mg/hr) = [FCR (pools/hr) × protein concentration (protein amount/mL) × lysate volume (mL)]/brain weight (mg). The half-lives (t_½) were calculated using the equation t_½ = ln 2/FCR. Protein concentrations of apoE and Aβ in the lysates were determined by protein-specific sandwich ELISAs using in-house antibodies. For apoE, HJ6.2 was used as the coating antibody and biotinylated HJ6.3 as the detection antibody. Pooled C57/BL/6 J mouse plasma was used as a standard. For Aβ, HJ2 (anti-Aβ35-40) and biotinylated HJ5.1 (anti-Aβ13-28) were used as the coating and detection antibody, respectively.