Direct Assessment of Lipoprotein Outflow From In Vivo–Labeled Arterial Tissue as Determined in an In Vitro Perfusion System
Abstract—The rate of cholesterol deposition during the atherosclerotic process is determined by the balance between the inflow and outflow of plasma lipoproteins in the arterial wall. Whereas the rate of inflow may be measured directly, the rate of outflow has most often been calculated indirectly from lipoprotein uptake by using the 2-compartment model. One objection against such calculations is that lipoprotein binding is not being considered. In the present study 2 different protocols were used to obtain a direct measure of the outflow of lipoproteins from atherosclerotic rabbit aortas. Thus, 3 rabbits with experimental atherosclerosis were given 125I-LDL intravenously and 3 were given [14C]cholesterol perorally. Twenty-four hours later the aortas were removed and the outflow of label was monitored during in vitro perfusion. Despite the different protocols, our results were consistent and indicated that fractional loss relative to whole tissue was ≈0.01 pool/h, which is 1 order of magnitude lower than current estimates based on the 2-compartment model (0.04 to 0.4 pool/h). Furthermore, whereas as much as 2/3 to 3/4 of the tracer that had entered the arterial wall was effectively trapped, the remainder equilibrated at a faster rate (0.06 pool/h). In conclusion, it seems that tissue binding constitutes a prominent and possibly underrated mechanism of lipoprotein deposition, at least in the atherosclerotic rabbit aorta. Furthermore, this means that current estimates of lipoprotein exchange parameters based on the 2-compartment model (eg, fractional loss) may rest on invalid assumptions and should be regarded with caution.
- Received October 7, 1998.
- Accepted May 29, 1998.
Cholesterol deposition is 1 of the prominent features of the developing atherosclerotic lesion, and it seems that the major part of the cholesterol originates from LDL in the blood. The dynamics of the uptake of lipoproteins by the arterial wall has been subjected to a number of studies in animals and humans, both in vivo and in vitro, and it is generally agreed that the uptake is characterized by a bulk-phase transport, whereby the rate of accumulation is inversely related to the size of the lipoprotein. In most cases the rate of entry has been estimated from measurements of the accumulation of labeled lipoproteins in the arterial tissue. However, it is obvious that the net deposition also depends on the outflow of lipoproteins. Estimates of the rate of exit have been made under the assumption that plasma and arterial tissue may be regarded as a 2-compartment system in dynamic equilibrium. Most likely, however, this approach is too simplified, because it is well known that lipoproteins may bind to the matrix of the arterial wall and thus become withdrawn from the freely equilibrating pool to a variable extent.
To address this issue more directly, we used an in vitro perfusion system using the rabbit aorta with or without experimental atherosclerosis. In this system we have studied the outflow of lipoproteins from tissue that had accumulated radioactive lipoproteins in vivo.
Male New Zealand White rabbits were used throughout the study. At 2 to 3 months of age, experimental atherosclerosis was induced through a combination of a cholesterol-enriched diet (1% cholesterol, wt/wt, in standard rabbit chow fed ad libitum) and mechanical injury by use of an embolectomy catheter. The method to achieve the mechanical injury was originally described by Baumgartner in 1963,1 and the process of endothelial injury and healing in the rabbit aorta take place during several weeks and months.2 Careful control of the technique makes it possible to obtain a mild injury.3 At the end of the induction period (13±2.4 weeks) a combination of fatty foam cell–rich lesions and fibrous plaque–like lesions were obtained to a varying extent: 0.52±0.18 of the surface area of the perfused aortas was occupied by lesions, corresponding to 0.27±0.12 on a weight per weight basis.
At the time the rabbits were killed (pentobarbital sodium [≈50 mg/kg], plus bleeding to death), serum cholesterol and triglyceride levels in these animals were 50.3±19.5 and 3.57±2.58 mmol/L, respectively. In rabbits in which125I-LDL was given in vivo, cholesterol feeding was interrupted 3 weeks before termination to achieve a high specific activity in plasma. Their cholesterol and triglyceride levels were lower at the time of injection: 23.4±22.6 and 2.86±3.72 mmol/L, respectively.
Perfusion of Rabbit Aorta In Vitro
The dissection procedure and the in vitro perfusion system that was used have been developed to ensure optimal tissue integrity.4 In brief, under general anesthesia (premedication: ketamine 7.5 mg/kg and xylazine 3 mg/kg im; anesthesia: ketamine (50 mg/mL)/xylazine (20 mg/mL), 1:1, 0.35 mL/kg iv initially, plus 0.17 mL/kg per 15 minutes thereafter) the descending thoracic aorta was removed and dissected, and the obtained intima-media preparation was fitted into Teflon cylinders of appropriate size. The cylinders were inserted into 1 of the legs of a U-shaped glass tube, which served as the perfusion chamber. The flow of the perfusion medium was adjusted to a steady rate of 1.5 to 2.0 mL/h. In this system there is no convective flow across the arterial wall.
In perfusions where 125I-LDL had been given in vivo, the dissection procedure was simplified. Thus, meticulous removal of adventitial fat and blood vessels was performed, but the dissection was not carried down into the media. In this way the interval between the death of the animal and the start of the perfusion could be reduced to <1 hour.
Throughout the study, Eagle’s minimum essential medium with Earle’s salts and 10 mmol/L NaHCO3 was used, supplemented with 1% nonessential amino acids, 100 μg/mL streptomycin, 100 IU/mL penicillin, 60 mg/mL BSA (Serva Feinbiochimica), and 10% heat-inactivated rabbit lipoprotein–deficient serum (prepared by ultracentrifugation at d>1.21 g/mL). All animal procedures were in accordance with institutional guidelines.
Isolation of LDL
LDL (1.019 to 1.063 g/mL) was isolated by sequential centrifugation from fresh, fasting human plasma in the presence of 10 mmol/L EDTA. The size of LDL particles was determined by gradient polyacrylamide gel electrophoresis,5 giving a peak diameter of 27.4 to 27.8 nm. Ferritin, thyroglobulin, and a pooled LDL standard were used to calibrate the gel.
Labeling of LDL With 125I
LDL was iodinated with 125I using the ICl technique as described by McFarlane6 and modified by Shepherd et al.7 Specific activities of the lipoproteins were 340 to 770 counts per minute (cpm)/ng apoB protein. ApoB protein was determined by a quantitative immunoassay.8 Trichloroacetic acid (TCA) precipitability of the preparations was between 0.97 and 0.99. The LDL preparation was diluted with nonradioactive lipoprotein from the same donor to obtain the desired concentration. The final preparation was given to rabbits intravenously (5.7 to 12.5 mL; 3.2 to 7.0 mg apoB per mL; 7.5 to 12.6 cpm/ng apoB) in the left marginal ear vein.
In Vivo Labeling With [14C]Cholesterol
In selected rabbits the plasma lipoproteins were labeled with [14C]cholesterol in vivo. Thus, 0.13 to 0.25 mCi (2.3 to 4.6 μmol) [4-14C]cholesterol in 0.2 mL ethanol was added to 0.1 g sodium taurocholate in 10 mL saline and given via a gastric tube to sedated (ketamine/xylazine) rabbits 24 to 37 hours before the rabbits were killed. By this method, the major part of the label in plasma is said to be recovered as cholesterol ester.9 10 In our study this proportion was 0.84 (0.80 to 0.89), mean and (range). Much of the label (0.78; 0.66 to 0.86; mean and range) was recovered in the fraction floating at d=1.019 g/mL.
Three rabbits were given [14C]cholesterol via gastric tube and plasma samples were collected at intervals until the aorta was removed after 25, 36, or 37 hours. In 3 other rabbits125I-LDL was injected into the left marginal ear vein, and plasma samples were collected after 15 minutes and at intervals until the aorta was removed after 24 hours.
After dissection, the most proximal 4- to 5-mm portion of the aorta was set aside for analysis, whereas the major part of the aorta was perfused for 24 hours with medium. The whole volume of the perfusate was collected and divided into hourly fractions. At the end of the experiment the tissue was dissected and analyzed.
Overt atherosclerotic lesions were cut out and dissected free from the underlying media under a dissecting microscope. Plaques, tissue underlying plaques, and noninvolved areas were analyzed separately. When the adventitia was included in the perfusions, this layer was also dissected free and analyzed separately. Wet weight (wet wt) and surface area were measured.
Analysis of samples in experiments with125I -LDL is illustrated in Figure 1⇓. Total radioactivity was measured on the nonprocessed tissue. After homogenization in 200 μL saline, samples were centrifuged, and TCA-precipitable, nonlipid-soluble radioactivity was determined on the supernatant (mobile pool) and the pellet (bound pool) separately. The overall recovery of the analysis was 0.81±0.04.
In experiments with [14C]cholesterol-labeled lipoproteins, lipids were extracted in chloroform/methanol, 2:1 vol/vol/, and washed.11 Total cholesterol mass (Boehringer Mannheim CHOD-PAP12 ) and total radioactivity were determined on the eluates. Lipids were separated by thin-layer chromatography,13 and the proportion of radioactivity in cholesterol and cholesterol ester was determined from videorecorded images of autoradiograms by using KS400 (Carl Zeiss) image analysis software.
In experiments with 125I-LDL, apoB mass and total and TCA-precipitable radioactivity were determined in plasma samples collected at intervals during the 24 hours preceding the perfusion. Furthermore, the proportion of the radioactivity recovered in the d>1.019 g/mL fraction was assessed after ultra centrifugation.
In experiments with [14C]cholesterol, total cholesterol mass and total and lipid-soluble radioactivity were determined on the plasma samples obtained during the 24 to 37 hours before perfusion. Total and lipid-soluble radioactivity at d<1.019 and d>1.019 g/mL was determined after ultracentrifugation. Lipids were separated by thin-layer chromatography, and the ratio between the amount of radioactivity in cholesterol and that in cholesterol ester was determined through image analysis of videorecorded autoradiograms by using KS400 software (Carl Zeiss).
Total radioactivity was measured in all collected fractions of the perfusate. In experiments where the aorta had been labeled with 125I-LDL, lipoprotein-associated radioactivity was also assessed after separation by exclusion chromatography on a PD10 column (Pharmacia).
Measurement of Radioactivity
125I radioactivity was measured on a 1282 Compugamma gamma counter (LKB) and 14C radioactivity on a 1214 Rackbeta liquid scintillation counter (LKB) with Readygel (Beckman Instruments Inc) as the scintillation cocktail. Samples were counted to achieve a counting error better than 3% (total counts >4000), which in some cases meant counting periods of up to 1 hour.
The content of radioactivity in the tissue at a given time point during the perfusion could be extrapolated by adding the sum of the outflowing activity after this time point to the activity remaining in the tissue at the end of the experiment. The steadily falling level of tissue activity that was extrapolated in this way was fitted to y=A(1+Be−Ct), where A+B denotes the activity before perfusion, A the remaining tissue activity at infinity, and C the fractional loss. Consequently, A/(A+B) denotes the proportion of the tracer that was not available for equilibration that is described by this approximation.
Because the impression was that tissue activity was best described as a 2-phase process, successive curve fits were made to approximate the time course of the second phase in the best possible manner. In this procedure all data points were initially included. Then data points were removed 1 at a time from early time points and onward, the best fit being finally accepted when the SEs of the included parameters were minimal. This fit included the time period from 3 to 5 hours and onward of the obtained data. Marquardt’s algorithm as applied in SigmaPlot (Jandel Scientific) was used for the calculations. If not otherwise specified, mean and SD were used to describe the data.
The mean total radioactivity in plasma 25 to 37 hours after an oral dose of 0.13 to 0.25 mCi [14C]cholesterol was 0.23 to 1.0 cpm/nL with a specific activity of 3.4 to 15 cpm/nmol cholesterol. In Figure 2⇓ the radioactivity is expressed as the proportion of the given dose per milliliter of plasma. If one assumes a distribution volume of 120 mL,10 this means that ≈0.30 of the administered radioactivity was recovered in plasma at 24 hours. After 11 to 37 hours an average 0.86±0.03 of plasma radioactivity was recovered as cholesterol ester; 0.77±0.07 of the radioactivity was associated with lipoproteins floating at d=1.019 g/mL.
After the administration of a bolus of125I-LDL intravenously (5.7 to 12.5 mL; 3.2 to 7.0 mg apoB per mL; 7.5 to 12.6 cpm/ng apoB), the TCA-precipitable radioactivity in plasma rose instantly and then decreased during the following 24 hours, with a mean of 0.97 to 1.28 cpm/nL. In Figure 2⇑ the radioactivity is expressed as a proportion of the given dose per milliliter of plasma. The average apoB concentration in plasma was 0.20 to 0.33 mg/mL and the specific activity 2.82 to 4.49 cpm/ng apoB, which was 0.40±0.13 of the specific activity in the given preparation. The TCA-precipitable portion was >0.97 in the injected preparation and 0.93±0.09 in plasma; 0.94±0.05 of the plasma activity was recovered at d>1.019 g/mL.
[14C]Cholesterol Loading and Washout
Twenty-five, 37, and 36 hours before the animals were killed, an oral dose of [14C]cholesterol was given to 3 rabbits, resulting in a steady increase in plasma radioactivity, with a time-averaged mean of 0.63±0.39 cpm/nL (Figure 2⇑). The aorta was taken out and perfusion ensued. Normalized to the average total radioactivity in plasma, the uptake in the whole aortic tissue of the sample before perfusion was 67.9±21.7 nL/mg wet wt, whereas the activity was 41.7±21.3 nL/mg wet wt in the incubated tissue at the end of the perfusion after 24 hours. In dissected plaques the uptake was 198±29 and 92.3±27.5 nL/mg wet wt before and after perfusion, respectively. The activity in noninvolved tissue was very low, and reliable values were obtained in only 2 aortas before perfusion, ie, 12.6 and 12.6 nL/mg wet wt and in 1 aorta after perfusion, 9.6 nL/mg wet wt. Paired analysis indicated that 0.47±0.14 of the radioactivity remained in the plaques after the washout phase. In noninvolved tissue the proportion was ≈3/4. In whole tissue the activities before and after perfusion are not directly comparable, because the proportion of plaque to noninvolved area is not necessarily the same in the sample obtained before perfusion and in the incubated aorta. However, when this difference was taken into account, the uptake in the whole perfused aorta before washout could be extrapolated to 64.6±28.3 nL/mg wet wt, indicating that the proportion of the activity that remained after perfusion was 0.66±0.05 (paired analysis).
The content of radioactivity in the tissue at a given time point during the perfusion could be extrapolated by adding the sum of the outflowing activity after this time point to the activity remaining in the tissue at the end of the experiment (Figure 3⇓). With this procedure the normalized uptake of label before perfusion could be estimated at 55.2±28.3 nL/mg wet wt, which is 0.82±0.11 (paired ratio) of the content estimated from sample analysis. Nonlinear regression indicated that there was a proportion of the label, 0.65±0.08, that was not equilibrating with the perfusion medium at the time scale that was used. The fractional loss from the proportion that seemed to equilibrate was 0.051±0.016 pool/h. Based on the total content of label in the tissue, however, the estimate was lower, ie, ≈0.011±0.003 pool/h, averaged over the 24 hours of perfusion. Almost 90% (0.86±0.03) of the radioactivity in plasma was recovered as cholesterol ester. In the tissue, the proportion was somewhat lower both before perfusion (0.71±0.07) and after (0.74±0.02).
The average specific activity of total cholesterol in the tissue (0.56±0.39 cpm/nmol) was 0.059±0.026 of that in plasma (9.4±5.8 cpm/nmol). The specific activity in the tissue did not change during the perfusion, and the specific activity was similar in plaque and in nonplaque areas.
125I-LDL Loading and Washout
Three rabbits with experimental atherosclerosis were given ≈100 μCi 125I-LDL intravenously 24 hours before they were killed. After an initial peak the TCA-precipitable radioactivity in plasma fell gradually (fractional catabolic rate ≈0.06) with a time-averaged mean of 1.11±0.17 cpm/nL at a specific activity of 3.86±0.74 cpm/ng apoB during the 24 hours (Figure 2⇑).
In samples collected during the initial dissection, the normalized uptake of TCA-precipitable radioactivity before perfusion was 35.3±11.5 nL/mg wet wt in whole tissue. After the 24 hours of cold perfusion, the remaining uptake was 22.3±8.2 nL/mg wet wt, constituting 0.64±0.10 of the activity at the beginning of the washout phase (paired analysis). In dissected plaques the uptake was 84.7±12.4 and 70.7±37.2 nL/mg wet wt and in noninvolved tissue, 10.3±4.1 and 5.33±1.78 nL/mg wet wt at the respective time points. These figures indicated that 0.64±0.10, 0.80±0.32, and 0.53±0.07 of the radioactivity remained after the washout phase in whole tissue, plaques, and noninvolved tissue, respectively. During these experiments the difference in the degree of atherosclerosis was minimal between the preperfusion sample and the incubated aorta and required no compensatory corrections during the calculations.
The outflow of radioactivity from the incubated tissue was used to extrapolate the tissue activity during perfusion as described before (Figure 3⇑). With this procedure the normalized uptake of label before perfusion could be estimated at 31.9±11.6 nL/mg wet wt, which is 0.91±0.13 of the uptake measured on the initial samples. The data indicated a 2-phase pattern of outflow, with a more rapid release during the first few hours and a slower decline during the major part of the washout period (Figure 3⇑). Nonlinear regression during the latter phase indicated that the tissue concentration of LDL did not approach zero with time, suggesting that a proportion of the label, 0.76±0.06, belonged to a nonequilibrating pool. The fractional loss of the label from the equilibrating pool was 0.067±0.020 pool/h. Calculations based on total tissue activity would give a considerably lower fractional loss, ie, ≈0.010±0.006 pool/h, averaged over the 24 hours of perfusion. Exclusion chromatography of the outflowing perfusate indicated that 0.62±0.18 of the activity was associated with lipoprotein particles and this proportion did not seem to change with time during the perfusion. Under the assumption that the outflow of degradation products is much faster than that of lipoprotein particles, this means that the fractional loss of the lipoprotein particles per se might be even lower, possibly ≈0.04 pool/h, from the equilibrating pool and 0.006 pool/h relative to the total tissue activity.
The samples were homogenized in saline to differentiate easily equilibrating LDL (the mobile pool, ie, the TCA-precipitable radioactivity in the supernatant) from the slowly equilibrating LDL (the bound pool, ie, the TCA-precipitable activity in the pellet).
Before perfusion 0.11±0.03 of the LDL belonged to the mobile pool. After perfusion the proportion was somewhat less, 0.078±0.030, or 0.73±0.11 of the initial value. Although the LDL uptake in plaques was 1 range of order higher than that in the adjacent media, the proportion of mobile LDL was similar.
In most cases calculations of the outflow of lipoproteins from the arterial wall have been based on lipoprotein uptake and the use of a simple 2-compartment model.3 14 15 16 In such a model, lipoprotein binding is not considered, and this introduces a systematic error that is negligible in short-term experiments but may become considerable after 24 hours, an end point that is commonly used.3 14 15 16 Another source of error is, of course, the crucial assumption that the arterial wall may be regarded as 1 homogeneous, readily equilibrating compartment. This error was effectively circumvented by Tozer and Carew in a recent study,17 in which they used a stochastic approach insensitive to the characteristics of the dynamic exchange process. However, other approximations were introduced, and determinations of outflow still rested on indirect observations.
The main idea with the present study was to develop a method for direct measurement of the outflow of lipoproteins from the arterial wall. To this end, aortas of rabbits with experimental atherosclerosis were loaded with radioactive lipoproteins in vivo, and the outflow of radioactivity was studied in a well-defined in vitro perfusion system.
Two different techniques were used to load the aortic tissue:125I-LDL injected intravenously (I) and 14C-labeled lipoprotein after an oral dose of [14C]cholesterol (C). In I, the injected LDL was continuously degraded and the arterial lumen was exposed to gradually falling levels of lipoprotein (Figure 2⇑). Much (0.93±0.09) of the radioactivity in plasma was TCA-precipitable, and 0.94±0.05 was recovered at d>1.019 g/mL, indicating that a major part of the radioactive label belonged to the LDL fraction. In C, the administered [14C]cholesterol appeared in plasma at steadily rising levels of radioactivity during the loading phase (Figure 2⇑). A major part of the radioactivity appeared as cholesterol ester (0.86±0.03). As much as 0.77±0.07 of the label most likely belonged to the β-VLDL fraction floating at d=1.019 g/mL. Thus, whereas in I the aorta was exposed to human LDL particles labeled in the protein moiety, the major part of the label in C resided in the cholesterol ester portion of homologous lipoprotein particles, most likely β-VLDL.
The normalized uptake in whole tissue at the beginning of the washout phase was 35.3±11.5 and 55.2±28.3 nL/mg wet wt in I and C respectively, which is similar to published data for atherosclerotic monkeys14 18 and rabbits with16 or without15 19 atherosclerosis. In I the uptake reflected the entry of LDL particles into the arterial wall, whereas in C the uptake represented a combination of lipoprotein uptake and physical exchange of unesterified cholesterol. If cholesterol ester is used as a label of lipoprotein particles,11 the normalized uptake would be similar (57.2±16.5 nL/mg wet wt), indicating that a major part of the tissue radioactivity at the beginning of the washout phase represented lipoproteins that had been taken up. The uptake in C was considerably higher than in I, which may partly reflect the longer loading period in C (25, 37, and 36 hours) than in I (24 hours). Another explanation might be that whereas the final degradation product in I (free iodine) readily left the tissue, this was not the case in C. In fact, 0.37 of the outflow in I was of low molecular weight, indicating that the lipoprotein uptake in I could have been as high as 56 nL/mg wet wt [35.3/(1−0.37)] if the degradation products had equilibrated at the same rate as the LDL particles.
Thus, 2 quite different aortic preparations were used to study the washout of radioactivity (Figure 4⇓): In C the major part of the label probably resided in the cholesterol ester portion in the core of homologous β-VLDL particles, whereas the remainder most likely was distributed between intracellular cholesterol ester from degraded lipoprotein particles and free cholesterol, reflecting inflow through physical exchange and hydrolysis of cholesterol ester. In I the radioactivity resided in the protein moiety of human LDL particles and in their degradation products.
During the 24-hour washout phase the rate of outflow seemed somewhat faster during the first few hours than during the rest of the perfusion (Figure 3⇑). During the latter phase (from 3 to 5 hours and onward) outflow data fitted well to a negative exponential function, y=A(1+Be−Ct).
In neither I nor C was the outflow of radioactivity constant, but it decreased slowly with time at a rate that indicated that only a proportion of the radioactivity that had been taken up during the loading phase was freely equilibrating with the incubation medium (Figure 3⇑). It seemed that as much as 0.76±0.06 and 0.65±0.08 of the activity in I and C, respectively, belonged to a tissue pool that did not equilibrate at the time scale used. This implies that a major portion of the lipoprotein that had entered the arterial wall at the beginning of the washout phase had become effectively trapped. Our interpretation of the data is illustrated in Figure 5⇓, in which this bound pool is represented by the dotted line and the experimental data by the heavy line. The thin line depicts the washout curve that would have been expected if all the tracer had equilibrated freely. The size of the bound pool is compatible with early data by Srinavasan et al,20 who estimated that 0.47 of LDL was tissue bound after 4 hours. Similarly, approximately half of LDL was tissue bound in early, short-term experiments at our laboratory.21 Also in a recent study, 0.50 to 0.60 of LDL was defined as tightly bound after 23 hours.15 As an alternative measure, the bound LDL pool was also estimated as the saline-nonextractable radioactivity of the tissue homogenates. By this method the bound pool seemed larger (0.90±0.03 and 0.92±0.03 before and after the washout phase, respectively), which might reflect that some of this nonextractable activity was released during perfusion. Furthermore, only 1 extraction in saline was performed, whereas others have used protracted20 or repeated21 extractions, even including EDTA.15 In I the proportion of the outflowing radioactivity that was associated with LDL particles was 0.62±0.18 as assessed by exclusion chromatography. The remaining 0.37 represented low-molecular-weight degradation products, indicating a degradation rate similar to values that we have calculated earlier by other methods.16 This proportion did not seem to change with time, suggesting a fairly constant degradation rate of the mobile LDL pool during the 24-hour washout phase.
The fractional loss in I and C calculated from the outflow data (0.067±0.020 and 0.051±0.017 pool/h, respectively) was similar to the rate deduced from uptake data only in animals (0.08 to 0.09,14 0.04 to 0.10,3 and 0.11 to 0.42 pool/h15 ) and in humans (0.11 to 0.13 pool/h22 ). However, our data cannot be compared directly with these figures, because our results refer to the fractional loss from only a part of the tissue lipoprotein pool, ie, the mobile pool (cf Figure 5⇑), which only constituted 0.24±0.06 and 0.35±0.08 of the total tissue content in I and C, respectively. For comparison, our data were recalculated relative to the total tissue content, as was the case in the quoted studies. This recalculation gave an apparent fractional loss that was considerably lower, ie, 0.010±0.006 and 0.011±0.003 pool/h, respectively, for I and C. Roughly the same low value of the fractional loss was also obtained when calculations were made from our data on tissue content only before and after perfusion, ie, 0.017 pool/h. These rates are 1 order of magnitude lower than the quoted determinations based on the accumulating radioactivity in the 2-compartment model, in which lipoprotein binding was disregarded. Ghosh et al14 deliberately ignored lipoprotein retention, stating that it amounted to <1/20 the inflow, but our data and those of others15 20 do not vindicate this statement. However, it is disturbing that an increased accumulation due to lipoprotein binding would lead to a flattening of the uptake-versus-time curve, which would give too low rather than too high an estimate of the fractional loss. One way to reconcile these observations is to introduce a “very fast equilibrating pool” or to adopt the concept of plasma contamination23 (Figure 5⇑). This would lead to high uptake at early (3 hours) time points, thus restoring the shape of the curve, whereby the 2 types of errors might actually tend to cancel each other. Our observation that the rate of outflow of label from the tissue was higher during the very first hours of the washout phase (Figure 3⇑) supports this view.
The findings in our study rest on the pattern of outflow of tracer from the incubated tissue, ie, the observation that the outflow seemed to level off with time earlier than would have been expected in a simple 2-compartment model. The calculated parameters (bound pool and fractional loss) were remarkably similar between the 2 incubation setups despite the fact that the outflowing tracees were different and not distinctly defined. One obvious explanation is, of course, that the collected radioactivity in effect represents outflow of lipoprotein particles to a high extent. In I we showed that at least 0.62 of the outflowing activity could be attributed to LDL particles. In C the physical state of the outflowing activity was not studied for technical reasons, but it is likely that a large proportion of the activity represented outflow of lipoprotein particles, because cholesterol ester from degraded lipoproteins is highly insoluble in aqueous media and no vehicle was available to facilitate the outgoing transport of free cholesterol via physical exchange.
The dynamics of the exchange process during the loading phase (Figure 5⇑) is not essential for our calculations, but the size of the bound pool suggests that the rate of deposition must lie within the same range as the rate of entry, pointing to a considerable capacity of the arterial wall to bind lipoproteins. Although it cannot be ruled out that the size of the bound pool, as defined in our in vitro system, partly reflected the incubation conditions, it should be pointed out that binding itself took place in vivo. At the same time it may seem hard to imagine a rate of deposition within the same range as the inflow in the long run, since the bound pool also will eventually reach saturation. Then again, crude calculations do in fact indicate that several weeks would be required to load the cholesterol pool of the atherosclerotic rabbit aorta even if all inflowing lipoprotein cholesterol was deposited. (Assume a cholesterol clearance rate of 2 to 10 nL/mg wet wt per hour11 14 16 at a plasma cholesterol level of 25 mmol/L and a cholesterol concentration in the tissue of 100 nmol/mg wet wt [data from the present study].)
One inherent limitation in our approach was that it did not allow the differentiation between lesion and nonlesion areas but gave a composite picture of the outflow. However, the decrease in tracer concentration during the washout phase did not differ substantially between total tissue, plaque, and nonlesion areas, ie, 0.41±0.16, 0.32±0.28, and 0.47±0.07. Furthermore, although the concentration of tracer in lesions was higher than in other areas, the proportion of the total tracer content of the aorta that resided in lesions was not higher than 0.66±0.19. This means that the total outflow pattern was not dominated by either, so it seems that the major conclusions that we have drawn were valid for both.
Lipoprotein deposition in the arterial wall is determined by lipoprotein inflow and lipoprotein outflow. Whereas measurement of the rate of inflow is fairly straightforward, the assessment of lipoprotein outflow is more controversial and most commonly rests on indirect conclusions from determinations of lipoprotein uptake. In the present study we used 2 different protocols to obtain a direct measure of the fractional loss of lipoproteins from atherosclerotic rabbit aorta. Our results indicated that fractional loss related to whole tissue was ≈0.01 pool/h, which is 1 order of magnitude lower than current estimates based on the 2-compartment model (0.04 to 0.4 pool/h). Furthermore, whereas as much as 2/3 to 3/4 of the tracer that had entered the arterial wall was effectively trapped, the remainder equilibrated at a faster rate (0.06 pool/h).
It appears likely that binding to resident proteoglycans constitutes the major mechanism of the observed lipoprotein entrapment.21 24 25 Such entrapment has been suggested to facilitate lipoprotein modification, thereby contributing to the atherogenic potential of the deposited lipoproteins.26 27
In conclusion it seems that tissue binding constitutes a prominent and possibly underrated mechanism of lipoprotein deposition, at least in atherosclerotic rabbit aorta. Furthermore, this implies that current estimates of lipoprotein exchange parameters based on the 2-compartment model (eg, fractional loss) may rest on invalid assumptions and should be regarded with caution.
This study was supported by the Swedish Heart Lung Foundation and the Swedish Medical Research Council (project No. 4531) (to Göran Bondjers).
Baumgartner HR. Eine neue Methode zur Erzeugung von Thromben durch gezielte Überdehnung der Gefässwand. Z Ges Exp Med. 1963;137:227–232.
Stemerman MB, Spaet TH, Pitlick F, Cintron J, Lejnieks I, Tiell ML. Intimal healing: the pattern of re-endothelialization and intimal thickening. Am J Pathol. 1977;125:87.
Schwenke DC, Zilversmit DB. The arterial barrier to lipoprotein influx in the hypercholesterolemic rabbit, 1: studies during the first two days after mild aortic injury. Atherosclerosis. 1989;77:99–103.
Björnheden T, Bylock A, Hansson GK, Bondjers G. A system for long-term perfusion of rabbit aorta in vitro. Arteriosclerosis. 1983;3:366–382.
Gambert P, Bouzerand-Gambert C, Athias A, Farnier M, Lallemant C. Human low density lipoprotein subfractions separated by gradient gel electrophoresis: composition, distribution and alterations induced by cholesteryl ester transfer protein. J Lipid Res. 1990;31:1199–1210.
Laurell B. Electroimmunoassay. Scand J Clin Lab Invest.. 1972;29:21–37.
Stender S, Zilversmit DB. Transfer of plasma lipoprotein components and of plasma proteins into aortas of cholesterol-fed rabbits: molecular size as a determinant of plasma lipoprotein influx. Arteriosclerosis. 1981;1:38–49.
Folch J, Lees M, Sloane Stanley GH. A simple method for the isolation and purification of total lipids from animal tissues. J Biol Chem. 1957;205:920–922.
Nielsen LB, Stender S, Kjeldsen K, Nordestgaard BG. Specific accumulation of lipoprotein(a) in balloon-injured rabbit aorta in vivo. Circ Res. 1996;78:615–626.
Tozer EC, Carew TE. Residence time of low density lipoprotein in the normal and atherosclerotic rabbit aorta. Circ Res. 1997;80:208–218.
Lin DS, Conner WE, Wissler RW, Vesselinovitch D, Hughes R. A comparison of the turnover and metabolism of cholesterol in normal and atherosclerotic monkey aortas. J Lipid Res. 1980;21:192–201.
Haarbo J, Nielsen LB, Stender S, Christiansen C. Aortic permeability to LDL during estrogen therapy. Arterioscler Thromb. 1994;14:243–247.
Olsson G, Wiklund O, Bondjers G. Effects of injury on apoB kinetics and concentration in rabbit aorta. Arterioscler Thromb Vasc Biol. 1995;15:930–936.
Shaikh M, Wootton R, Nordestgaard BG, Baskerville P, Lumley JS, La Ville AE, Quiney J, Lewis B. Quantitative studies of transfer on vivo of low density, Sf 12–60, and Sf 60–400 lipoproteins between plasma and arterial intima in humans. Arterioscler Thromb. 1991;11:569–577.
Nielsen LB, Nordestgaard BG, Stender S, Kjeldsen K. Aortic permeability to LDL as a predictor of aortic cholesterol accumulation on cholesterol-fed rabbits. Arterioscler Thromb. 1992;12:1402–1409.
Hurt-Camejo E, Olsson U, Wiklund O, Bondjers G, Camejo G. Cellular consequences of the association of apoB lipoproteins with proteoglycans: potential contribution to atherogenesis. Arterioscler Thromb Vasc Biol. 1997;17:1011–1017.
Williams KJ, Tabas I. The response-to-retention hypothesis of early atherogenesis. Arterioscler Thromb Vasc Biol. 1995;15:551–561.