Characterization of blood lipoproteins and validation of cholesterol and triacylglycerol assays for free-ranging polar bears ( Ursus maritimus )

Abstract

Blood triacylglycerol (TG) and lipoproteins are important variables for evaluating nutritional status of wildlife, but measurements are often expensive and difficult. Performance of a small, portable blood analyzer intended for human medical diagnostics was evaluated in measuring these variables in plasma and serum from free-ranging polar bears (Ursus maritimus), which are experiencing nutritional stress related to sea ice loss. The analyzer accurately tracked changes in concentration of total cholesterol (C_total), cholesterol associated with high-density lipoprotein (C_HDL), and TG during a validation protocol of diluting samples and spiking them with exogenous cholesterol and glycerol. Values of C_total and TG agreed well with values obtained by other methods (ultracentrifugation followed by colorimetric assays); agreement was variable for values of cholesterol associated with specific lipoproteins. Similar to a study of captive polar bears, ultracentrifugation methods revealed greater TG in very low-density lipoproteins than in low-density lipoprotein, which is unusual and merits additional study.

Whether an animal is fasting or feeding is reflected by blood values of cholesterol associated with high-density lipoproteins (C_HDL), low-density lipoproteins (C_LDL), very low-density lipoproteins (C_VLDL), and total cholesterol (C_total), as well as by values for triacylglycerols (TG_HDL, TG_LDL, TG_VLDL, TG_total). For example, C_total and TG_total often rise during mobilization of endogenous fat while animals are fasting. ⁸ Thus, lipoprotein measurements provide important inferences regarding the nutritional state of free-ranging animals that may or may not be fasting. ¹¹

Several methods are available for lipoprotein quantification, including colorimetric assays in lipid panels and ultracentrifugation techniques. ¹³ However, ultracentrifugation is expensive, labor-intensive, and not easily performed in research field camps; therefore, many researchers instead send samples to laboratories for colorimetric assays. ¹ The objective of the current study was to validate assay applicability and performance for samples from wildlife species, particularly assays that have the potential to be performed in field camps, allowing immediate analysis and decreasing risks of sample deterioration. ⁵

Blood samples from polar bears (Ursus maritimus) were analyzed with a portable blood analyzer ^a designed for human diagnostic procedures. The analyzer is also available for veterinary diagnostics, ^b but the human version ^a includes software for lipoprotein analysis. Both instruments^a,b are small (height 32 cm, width 15 cm, depth 20 cm, mass 5 kg) and can easily be used in field camps provided there is electricity (AC 100–240 volts and 50–60 Hz, DC 15 volts and 5.0 Amperes) and that ambient temperatures remain within 15°C and 32°C. Both instruments use disposable rotors ^c requiring 100 µl of sample. Thus, total equipment required for lipoprotein analysis includes the analyzer, ^a disposable rotors, ^c and a pipette. A printer is built into the analyzer, ^a and results are produced in approximately 12 min. Disposable lipid rotors ^c include assays for C_total, ⁹ C_HDL, ¹⁹ and glycerol as a TG proxy.^4,12 Other variables are calculated: C_VLDL = TG_total/5, and C_LDL = C_total – C_HDL – TG_total/5. ⁷

The 3 assays have been certified by the Centers for Disease Control and Prevention (United States) for blood from fasted human beings and validated for point-of-care use with human patients. ¹⁹ The current study tested the assays on blood from free-ranging polar bears and compared the results to values obtained by ultracentrifugation followed by analysis using commercial kits. ^d Accurate lipoprotein measurements are particularly useful in polar bear research. Research suggests that sea ice loss is leading to extended fasting and potential population declines in these carnivores, ¹⁸ and lipoprotein measurements have been successfully used to understand polar bear nutritional state. ⁸ Additionally, polar bears are unique because they consume a high-fat diet (e.g., 80% seal blubber), ² and a previous study of captive bears reported greater TG_LDL than TG_VLDL. ⁸ This is unusual because in human beings, LDLs are synthesized from TG-depleted VLDLs, implying that TG_LDL is less than or equal to TG_VLDL.

Portable analyzers enable the ideal scenario in which analyses are performed immediately after sample collection, avoiding deterioration risks of storage. ⁵ However, it is also important to validate assays if logistics prevent immediate analysis and samples must be stored. Validations were performed with samples collected during ongoing studies of polar bears in the Southern Beaufort and Chukchi seas, Alaska, USA (details on capture and sample collection described elsewhere ¹⁰ ). Blood was collected from the femoral vein or artery in ethylenediamine tetra-acetic acid (EDTA) and no-additive containers, ^e centrifuged at 2,000 RCF for 10 min, then plasma or serum was siphoned into cryovials and stored at −20°C or −40°C until analysis.

Four validation experiments (A–D) were conducted as follows: A) plasma was diluted, then exogenous C_HDL and TG were added, and percent recovery measured with lipid rotors ^c after both manipulations; B) results were compared from lipid rotors when analyzing serum versus EDTA plasma samples; C) results were compared from lipid rotors and kits ^d for C_total and TG_total; and D) ultracentrifugation was used to separate HDL, LDL, and VLDL, then cholesterol and TG were quantified for each class.

In experiment A, preliminary trials with 2 bears indicated plasma contained C_HDL levels above the detection limit of the blood analyzer ^a (100 mg/dl). Bear 1 required a 20:1 dilution (volume of plasma-to-sterile saline), and bear 2 required a 3:4 dilution, to reduce C_HDL to a measurable concentration. Additional dilutions were then performed (bear 1—5:2, 4:3, 3:4, 2:5; bear 2—1:2, 1:3, 1:4, 1:7). For each dilution, C_HDL and C_total were measured on 3 separate rotors; mean values were calculated, then divided by predicted values (based on the first dilution), and the results were converted to percent recovery (%). To each dilution, 4 μl of exogenous C_HDL (606 mg/dl) ^f was added, and new expected concentrations were calculated. Measurements for C_HDL and C_total were repeated and the percent recovery calculated.

Triacylglycerol was measured on bear 3 without dilution, as the concentration was within detection limits. The sample was then diluted (7:2, 9:7, 1:2, and 1:7) and percent recovery calculated for each dilution. Next, 5 μl of exogenous glycerol used as a TG standard (1,000 mg/dl) ^g was added to each dilution, TG measurements were repeated, and the percent recovery was calculated.

For consistency with previous studies of lipoproteins in bears, ⁶ the plasma was analyzed. However, the anticoagulant EDTA may negatively bias some colorimetric measurements of C_HDL ²⁰ although measurements of C_total seem unaffected. ⁹ In experiment B, lipid rotor ^c assays were performed on both serum and EDTA plasma for 10 bears, and results were compared using linear regression.

In experiment C, C_total and TG_total were measured in plasma from 8 bears, using kits ^d and lipid rotors. ^c Results were compared with linear regression. In experiment D, ultracentrifugation was used to create a gradient based on density, separating HDL, LDL, and VLDL. For 8 plasma samples, 21 fractions of lipoproteins ⁶ were identified: VLDL (1–3), LDL (3–11), and HDL (11–20). Some VLDL remained on the meniscus surface in the separation column, forming fractions 20 and 21. In each fraction, cholesterol and TG were measured using the same kits as in experiment C. The area under the curve was then calculated for each fraction and multiplied by C_total and TG_total to derive cholesterol and TG for each lipoprotein.

In experiment A, lipid rotors ^c accurately measured diluted and spiked samples of plasma. After dilution and spiking, mean recovery rates (± standard deviation) for C_HDL and C_total were 91% (±3%) to 105% (±3%). Recovery rates for TG after dilution and spiking were 100% (±2%) to 106% (±7%).

In experiment B, EDTA plasma samples yielded data similar to serum samples. Of the 10 bears for which both sample types were analyzed, 1 sample of EDTA plasma had extremely low values of C_total (38 mg/dl), C_HDL (22 mg/dl), and TG_total (43 mg/dl), whereas the serum appeared normal (353 mg/dl, 187 mg/dl, and 255 mg/dl, respectively); this sample was excluded from further analyses. For TG_total, EDTA plasma yielded results nearly identical to serum (in linear regression, 95% confidence intervals for slope included 1, and for intercept included 0). In contrast, slopes differed from 1 and intercepts differed from 0 for C_total and C_HDL, indicating these variables were measured differently (Fig. 1).

Figure 1.

Relation between measurements performed on ethylenediamine tetra-acetic acid (EDTA) plasma samples and serum samples from free-ranging polar bears (Ursus maritimus), of (A) total cholesterol (C_total) and (B) high-density lipoprotein (HDL)-cholesterol (C_HDL). Data were generated with lipid rotors (Lipid panel 400-7144; Abaxis Inc., Union City, California).

In experiment C, measurements of C_total and TG_total were similar between lipid rotors ^c and kits ^d (Fig. 2). In experiment D, values of C_HDL, C_LDL, and C_VLDL were inconsistent between the lipid rotor and the ultracentrifugation separation followed by measurements with the same kits used in experiment C (Fig. 3). The following TG content (mean ± SD; mg/dl) was determined: TG_HDL (52 ± 13), TG_LDL (221 ± 78), and TG_VLDL (23 ± 11).

Figure 2.

Relation between measurements from commercial kits and lipid rotors for (A) total cholesterol (C_total) and (B) triacylglycerol (TG_total) in ethylenediamine tetra-acetic acid plasma samples from free-ranging polar bears (Ursus maritimus). Commercial kits (cholesterol E 439-17501 and L-type TG M 461-08992) obtained from Wako Diagnostics, Richmond, Virginia. Lipid panel (400-7144) obtained from Abaxis Inc., Union City, California.

Figure 3.

Relation between measurements from lipoprotein separation by ultracentrifugation and analysis with a commercial kit and lipid rotors, for ethylenediamine tetra-acetic acid plasma samples from free-ranging polar bears (Ursus maritimus). (A) high-density lipoprotein (HDL)-cholesterol was measured in both approaches, whereas (B) low-density lipoprotein (LDL)-cholesterol and (C) very low-density lipoprotein (VLDL)-cholesterol were measured in the ultracentrifugation approach but calculated based on other variables in the lipid rotors approach. Commercial kits (cholesterol E 439-17501 and L-type TG M 461-08992) obtained from Wako Diagnostics, Richmond, Virginia. Lipid panel (400-7144) obtained from Abaxis Inc., Union City, California.

Lipid rotors ^c accurately measured diluted samples and samples with added HDL and glycerol, indicating there were no interfering substances in polar bear plasma. The EDTA plasma samples yielded lower C_total and C_HDL than serum samples. This may be because EDTA causes osmotic shifting of water out of red blood cells, ²⁰ influencing the concentration of C_total and C_HDL in plasma (although TG_total was not reduced). Alternatively, it is possible that EDTA interferes with the chemical reactions within the rotor. Regardless, significant slope parameters and high R² values suggest a consistent and precise relationship, allowing for reliable correction of data derived from plasma samples. One bear (of 10) exhibited extremely low values of all variables in EDTA plasma, but normal values in serum; it is unclear whether this is a risk when analyzing EDTA plasma or a chance occurrence. It is recommended that 1 sample type be used when comparing individuals or sampling events and that the results be compared to the published range of the species.

Measurements of C_total and TG_total from lipid rotors ^c and kits ^d accurately tracked each other. It is not clear why C_HDL measurements differed between lipid rotors and ultracentrifugation. The lipid rotor precipitates non-HDL lipoproteins and quantifies the remaining cholesterol that must be associated with HDL. However, the non-HDL precipitation may be inaccurate because it depends on recognition of apolipoprotein-B, which can vary in structure among species ¹⁴ and may differ between human beings and polar bears. The ultracentrifugation method separates lipoproteins by density and then quantifies cholesterol in each layer. In dogs, HDL are divided into 2 subtypes, 1 of which overlaps with LDL in density. ¹⁵ It is not known if polar bears possess HDL subtypes similar to dogs; if so, this may make it difficult to separate HDL and LDL by density. Neither method should be affected by chylomicrons, which likely precipitate with non-HDL in lipid rotors, and which have a much lower density than HDL and thus, in ultracentrifugation, should not appear in the HDL layer. Causes of the discrepancy between methods are unclear and merit further investigation.

Dissimilarities in C_HDL between lipid rotors ^c and ultracentrifugation must have contributed to inconsistencies in C_LDL because the blood analyzer ^a software uses C_HDL to calculate C_LDL. Intriguingly, C_VLDL calculated from the rotor measurement of TG_total was approximately 6× higher than C_VLDL measured after ultracentrifugation. This pattern, in combination with the unusual distribution of TG among VLDL and LDL (discussed in the following), suggests that lipoproteins may be different in polar bears than in human beings.

Ultracentrifugation methods revealed that, similar to polar bears in captivity, ⁸ free-ranging polar bears consistently have greater TG_LDL than TG_VLDL. In contrast, TG_LDL is less than or equal to TG_VLDL in black bears, ⁶ human beings, ³ rats, ¹⁷ and ground squirrels. ¹⁶ Dogs with greater TG_LDL than TG_VLDL have been reported in studies where, similar to the current study, the VLDL fractions may have contained chylomicrons.^17,21 In dogs, however, TG_LDL was about 2× TG_VLDL whereas in the present study, TG_LDL was approximately 10× TG_VLDL. It is speculated that if HDL and LDL do not separate completely because of an overlap in density, TG_LDL may include TG_HDL, elevating the former measurement. This topic requires further study.

In summary, lipid rotors ^c yielded accurate, precise measurements of C_total and TG_total, but more variable measurements of cholesterol associated with specific lipoproteins, in blood from free-ranging polar bears. These results, in combination with ease of operation and field portability, indicate that the portable blood analyzer ^a evaluated in the current study is a useful tool for research on polar bears, and potentially other wildlife species.