Influence of a hyperlipidic diet on the composition of the non-membrane lipid pool of red blood cells of male and female rats

Introduction

Most blood-borne lipids are carried by plasma lipoproteins which play a critical role in the transfer and distribution of energy between organs and tissues. Red blood cell (RBC) integrity is essential, and largely depends on their external membranes; thus, there is abundant literature on RBC lipids, and their modulation by diet (Gibson et al., 1984; Rotimi et al., 2012; Volek et al., 2004) or disease (Ferreri et al., 2005; Labagambe et al., 2008; Lemaitre et al., 2002).

Analysis of RBC lipid classes showed the presence of different types of phospholipids, cholesterol (and its esters), but also significant amounts of triacylglycerols (TAG) (Rotimi et al., 2012; Volek et al., 2004). However, TAG are not components of cell membranes; mammal RBCs could not oxidize lipids, and thus are devoid of reserves. Consequently, the presence of TAG in RBC, widely demonstrated (Volek et al., 2004), remains unexplained, since mammalian RBCs do not contain intracellular membranous structures. The outstanding presence of TAG adds to the ample variability on the proportions of lipid classes (and fatty acids) found by different studies (Carlson, Carver & House, 1986; Rotimi et al., 2012) in RBC. These data, taken together, hint at the existence of lipid depots in RBCs that are different from (true) membrane lipids. As far as we know, however, in RBC no specific lipid stores have been described which could explain the presence of TAG and other lipid classes in excess of those forming part of cell membrane structure.

RBCs are in constant contact with plasma lipids (and lipophilic hormones); they have a direct relation with apolipoproteins (Cooper, Durocher & Leslie, 1977), bind lipoproteins (Hui, Noel & Harmony, 1981), and have been found to carry free cholesterol to the liver (Hung et al., 2012). RBCs carry/harbour a number of steroid hormones (Romero et al., 2012), as well as extra glucose and amino acids loosely adsorbed on their surface (Proenza, Palou & Roca, 1994). This disposition has been suggested to speed up the interchange of substrates with the epithelial cells lining the capillary walls (Elwyn et al., 1972).

RBC membranes are periodically in close contact with those of endothelial cells due to constant blood cycling. This also causes cyclic exposure of the membranes to contact with different tissues and blood changes (arterial/venous) in pH and pO₂. These changes compound the continuous variations in other plasma components (fatty acids, lipoproteins, cholesterol, glucose, etc.) and exposure to free radicals (superoxide, peroxynitrite, nitric oxide) in tissues (Kagota et al., 2009). Consequently, the structure and fatty acid composition of RBC membrane lipids change with their age, in spite of an active lipid turnover, affecting especially its external layer (Dise, Goodman & Rasmussen, 1980; Quarfordt & Hilderman, 1970), which composition is markedly different from the more stable inner layer (Zwaal & Schroit, 1997). In any case, constant exposure results in a relative loss of function, with increased rigidity, a condition aggravated in metabolic syndrome (Van Blitterswijk, Van der Meer & Hilkmann, 1987).

Studies on the incorporation of labelled fatty acids showed a small but significant incorporation of them in RBCs in just 24 h (Leyton, Drury & Crawford, 1987); this implies the existence of a rapid system for plasma fatty acid interchange with RBC lipids. This could not be fully explained by phospholipid or cholesterol turnover (Quarfordt & Hilderman, 1970; Reed, Murphy & Roberts, 1968), which are slower and related to median RBC half-life: 45–50 days in rats (Burwell, Brickley & Finch, 1953). In addition, the substitution of internal membrane phospholipids is extremely slow (Zwaal & Schroit, 1997). We hypothesized that the rapid lipid interchange repeatedly observed in RBC may be related to a more accessible pool, in equilibrium with plasma, rather than to the necessarily stable cell membrane lipid bilayer.

A report suggest the presence of a fuzzy (probably lipid) cover on the RBC of hyperlipidemic blood when examined under the microscope (Miller, Hirani & Bain, 2013), and the overall lipid content of packed blood cells is higher than what can be expected from its membrane lipids alone. We hypothesized that the ability of RBC to transport lipophilic compounds and the presence of non-membrane lipids may be a consequence of the more or less loose presence of lipoprotein-derived lipids attached/bound to the surface of the RBC. This pool may participate in the turnover/repletion of RBC membrane lipid fatty acids. In order to test the hypothesis we analysed RBC lipids using a known mild lipid extraction procedure (i.e., assumedly not affecting their membrane lipids), comparing the fatty acid patterns and quantitative analysis of the lipids extracted in function of sex and exposure to a hyperlipidic diet.

Materials & Methods

Results

Plasma lipids

There were no significant differences in plasma triacylglycerols and packed cell volume between the four groups of rats studied (Table 2). However, total cholesterol was modified by sex and diet (higher for females and cafeteria diet). Non-esterified fatty acids were unaffected by diet, but females showed higher plasma levels than males. There were no clear relationships between the plasma lipid parameters. Table 2 also shows the proportion of lipids recovered from packed RBC. There were no differences attributable to sex and diet, the data was remarkably uniform.

The patterns of distribution of individual fatty acids in plasma total-lipids were also similar between the sex/diet groups (Fig. 1). There were only a few statistical differences between individual fatty acids between them. In plasma, exposure to a cafeteria diet resulted in significant differences in palmitoleic and oleic acids. Sex affected the levels of stearic, α-linolenic, eicosadienoic, gondoic and heneicosanoic acids. The similitude of overall distribution pattern was repeated when fatty acids were grouped in their main classes: saturated (SFA), monounsaturated (MUFA), polyunsaturated (PUFA, ω-3 and ω-6) and trans (Table 3); SFA and PUFA (mostly ω-6) were predominant, there were only trace amounts of ω-3 PUFA and trans fatty acids were practically absent. Only a few female rats treated with the cafeteria diet showed measurable levels of trans fatty acids. However, the grouped sums of fatty acids showed a significant effect of diet for MUFA and ω-6 PUFA, and effects for sex in SAT fatty acids. Total plasma fatty acids (i.e., the sum of all individual fatty acids analysed, expressed as mM) were 4.1 ± 0.4 (male control), 9.8 ± 3.3 (male cafeteria), 5.3 ± 0.6 (female control), and 9.2 ± 3.3 (female cafeteria); the differences were significant (P = 0.029) for diet but not for sex.

Table 3:

Sum of SFA, MUFA, PUFA and trans-FA levels in plasma and blood lipids of male and female rats subjected to a control or a cafeteria diet for 30 days.

All values are the mean ± sem of 6 different animals per group. Statistical analysis was done using a 2-way ANOVA.

	Units	Male rats		Female rats		P values
		Control	Cafeteria	Control	Cafeteria	Sex	Ciet
Plasma
SFA	mM	1.53 ± 0.17	3.25 ± 1.02	2.16 ± 0.25	3.67 ± 1.18	0.0300	NS
MUFA	mM	0.76 ± 0.12	2.42 ± 0.91	0.96 ± 0.10	2.05 ± 0.88	NS	0.0235
ω-3 PUFA	mM	0.027 ± 0.005	0.038 ± 0.014	0.027 ± 0.005	0.032 ± 0.014	NS	NS
ω-6 PUFA	mM	1.75 ± 0.24	4.12 ± 1.35	2.11 ± 0.31	3.43 ± 1.32	NS	0.0435
trans-FA	mM	0.007 ± 0.004	0.006 ± 0.004	0.005 ± 0.001	0.014 ± 0.008	NS	NS
RBC
SFA	µmol/g	12.43 ± 0.77	11.52 ± 3.22	13.54 ± 1.25	9.52 ± 0.61	NS	NS
MUFA	µmol/g	6.78 ± 0.47	6.60 ± 2.49	6,62 ± 0.79	6.04 ± 0.83	NS	NS
ω-3 PUFA	µmol/g	0.00	0.00	0.00	0.00
ω − 6 PUFA	µmol/g	2.58 ± 0.08	2.59 ± 1.25	2.92 ± 0.28	2.50 ± 1.08	NS	NS
trans-FA	µmol/g	0.12 ± 0.02	0.11 ± 0.03	0.08 ± 0.03	0.12 ± 0.01	NS	NS

DOI: 10.7717/peerj.1083/table-3

Notes:

FA

fatty acids

Red blood cell extractable lipids

Figure 2 shows the individual fatty acid levels in packed cell extracts of the four groups of rats. The patterns were highly similar, with only small differences induced by diet (palmitic and heneicosanoic acids), and none due to the effect of sex. In addition, there was a lesser variety of fatty acids than in plasma, with lower levels of PUFAs, and a small but clear presence of trans fatty acids (essentially elaidic acid). When considering the classes of fatty acids (Table 3), all groups showed a similar pattern, with low ω-6 PUFA, nil presence of ω-3 PUFA and a token presence of trans fatty acids.

Double bond distribution

The differences between plasma and RBC fatty acids widened when the number of double bonds was computed. The lower proportion of PUFA in RBC compared with plasma, and the nil influence on this parameter of SFA, resulted in a proportion of double bonds in the range of 0.5 per fatty acid molecule in RBC versus almost 1.5 in plasma lipids (Fig. 3). The proportion of double bonds in PUFA with respect to the total sum of double bonds showed the same pattern; in plasma, insaturation was mostly due to PUFA, but in RBC, the share of MUFA was much higher. Finally, the proportion of trans double bonds with respect to the sum of total double bonds was small but significantly higher in the RBC lipids of all experimental groups than in plasma, where trans fatty acids showed higher variability and lower values.

Discussion

A critical aspect of the validity of the data presented here showing the existence of a secondary lipid pool in RBC, different from that constituted by membrane lipids, is the comparison of the methodology used and the fatty acid profiles. Those found were different from those expected from membrane composition as described by other authors (Cooper, Durocher & Leslie, 1977; Ferreri et al., 2005; Lemaitre et al., 2002).

The usual procedure for the estimation of blood cell-membrane fatty acids, followed almost universally is: (A) Separation and washing of RBC in order to remove all traces of plasma lipids. (B) Breakup of the cells and separation of a membrane fraction clean of non-membrane proteins, especially haemoglobin, but also structural fibres such as spectrin. (C) Solvent extraction (twice in most cases) with a suitable solvent and extraction conditions and time (Rose & Oklander, 1965), the remaining lipid (bound to proteins) being usually discarded. (D) Overall methylation, usually in strong acid to hydrolyse complex phospholipids. (E) Analysis through GC or CG-MS. (F) Presentation of the results as percentages of total fatty acids measured, since step (B), and in part step (C), could hardly be quantitative, and the effectiveness of step (D) is often problematic.

Since we assumed that the most probable place for the RBC non-membrane-bilayer lipid “stores” (i.e., the site containing the TAG found in RBC lipid extracts) was the cell surface, we decided not to wash the cells and extrude the plasma by centrifugation (but taking care to limit cell breakage). We estimated the mass of trapped plasma, which allowed discounting its contribution to the total (mildly) extracted lipids of packed cells. Evidently, our intention was to keep the membranes as inaccessible as possible to our extraction method, and thus we used a procedure suitable for the extraction of all lipid classes (Folch, Lees & Sloane-Stanley, 1957) which has been found inadequate/ ineffective for RBC membrane lipid extraction (Eder, Reichlmayr-Lais & Kirchgeßner, 1993; Rose & Oklander, 1965).

The extracted lipids were methylated with a method sufficiently powerful to release methyl-fatty acids from most phospholipids (Christie, 1993), but which, in turn, was not sufficiently strong for all classes of complex membrane phospholipids (Eder, 1995); in addition, we used internal standards to check the effectiveness of methylation.

The results obtained suggest that the RBC lipid extract was not representative of membrane lipids. First because of the practical absence of critical PUFA membrane components (such as docosahexaenoic), even when compared with plasma. Second because of the impoverishment in double bonds, compared with plasma (which we assume was its origin), essentially at the expense of PUFA. There was, also, a relative abundance of SFA. Both factors reflected a situation extremely different from that found in membranes (Carlson, Carver & House, 1986; Lemaitre et al., 2002). In any case, these results do not preclude the possibility that an unknown proportion of RBC membrane lipids would be extracted with the procedure used. The trichloromethane: methanol method indeed can dissolve most lipids from complex matrices, including membranes, as in brain (Folch et al., 1951), but we did not break massively all cells, only compressed and snap froze them, preserving in part their structure. In addition, we discounted the lipids of trapped plasma. Nevertheless, since the mass of lipid recovered from cells was similar to the calculated weight of membrane lipids, a higher share of PUFA should be detected if membrane lipids were extracted in a significant proportion. Contrary to that, the results obtained strongly suggest that membrane lipids were not extracted in a proportion high enough to allow us to detect their typical PUFA presence/pattern.

Most studies on RBC lipids first purified cell membranes from washed cells’ ghosts (Eder, Reichlmayr-Lais & Kirchgeßner, 1993; Rose & Oklander, 1965; Vayá et al., 1993). Using these methods, PUFA levels were higher, in contrast with studies analyzing “erythrocyte lipids” (Labagambe et al., 2008; Rotimi et al., 2012; Volek et al., 2004). The differences between these approaches support our assumption that the RBC lipid we obtained represented largely non-membrane RBC lipids.

The quantitative importance of the extracted lipids was highly dependent on diet, but not on sex, with a proportion of 0.26 to 0.89% w/wet RBC weight, i.e., 1–2% w/dry RBC weight. However, the generalized lack of correlation of dietary fatty acid intake and the levels found in plasma lipids showed that these relationships are not straightforward and may be modulated by overall energy (largely lipid/ glucose) metabolism and substrate plasma turnover.

The easy availability for extraction of these RBC lipids, together with their abundance in MUFA and SFA agree with a lipoprotein origin of the deposits and their placement on the RBC surface. The findings of fuzzy borders in direct microscopic examination of blood cells from hyperlipidic plasma (Cooper et al., 1975; Miller, Hirani & Bain, 2013) may support this assumption. In addition, as far as we know, no internal RBC membrane or lipid depots have been described for mammals. However, the lack of significant effects of diet (hyperlipidic in our case) on the mass of recovered RBC lipids suggests that the lipid “cover” of RBC should be rather thin (it would be similar in thickness to the membrane lipid bilayer if it were spread uniformly) and not directly dependent on plasma lipid content. The lack of relationship seems to preclude the occasional weak bonding of lipoproteins as such, a possibility based on the presence of anchor proteins for apolipoproteins on the RBC surface (Hui, Noel & Harmony, 1981). The thin layer (or discontinuous blobs) of lipid is bound, probably, rather permanently to the actual membrane, since its proportion of double bonds (compared with plasma), essentially from PUFA, and the small presence of transfatty acids suggests a sustained exposure of these lipids to oxidizing or nitrating agents (at least to a higher degree than their plasma counterparts). In any case, these differences show that their turnover is slower than that of plasma lipids.

Nitric oxide is known to favour the conversion of cis to trans double bonds (Proell et al., 2002), lowering the fluidity of membranes and affecting their function. Oxidative attacks by superoxide and other free radicals tend to break down unsaturated fatty acids, mainly PUFA (Mattson & Grundy, 1985; Trotschansky & Rubbo, 2008), which levels tend to decrease in structures continuously exposed to oxidizing environments. This is in part corrected by turnover of membranes in cells, and by interchange with lipoproteins in RBC (Cooper et al., 1975; Dise, Goodman & Rasmussen, 1980; Quarfordt & Hilderman, 1970; Reed, Murphy & Roberts, 1968). However, the marked lack of PUFA (and absence of ω-3) indicate that: (A) The postulated outer layer of lipid in RBC should be rather permanent, at least enough to show the effects of oxidation and nitration on its fatty acids. (B) This lipid is repeatedly exposed (for all its functional life) to highly oxidative microenvironments in capillary beds. (C) The lipid occupies a limited and defined space on the cells, which is not directly affected by the availability (or turnover) of lipids in plasma.

In any case, there must be a certain degree of interchange of lipids between the RBC outer lipid layer and plasma lipids since its comparison with lipoprotein fatty acid patterns shows a considerable degree of similitude if PUFA are excluded. Labelled fatty acids are rapidly incorporated into RBC (Leyton, Drury & Crawford, 1987), and interchange or reposition of PUFA in the outer layer of RBC membranes has been previously described (Dise, Goodman & Rasmussen, 1980; Reed, Murphy & Roberts, 1968). Furthermore, diets high in PUFA decrease the stiffness of RBC membranes in metabolic syndrome (Katan et al., 1997; Pöschl et al., 1999). Probably there is a direct relationship between these phenomena, and this can be a function, so far not defined, of the external lipid layer of RBC. We postulate that it may act as an intermediate step for repairs (or protection) of the RBC membrane, since in mammals most maintenance systems must be external to the RBC, because they lack nuclei, ribosomes and most of the cell turnover machinery.

We expected, at least in the rats with overweight, that as a consequence of the cyclic exposure and close contact of RBC with endothelia there would be a marked increase in trans fatty acids (Alemany, 2012), a consequence of the higher production of nitric oxide and other oxidative and nitrating agents (Ghasemi, Zahediasl & Azizi, 2012). The levels of trans fatty acids we actually found were small, but could not be justified by the residual levels found in the diet (<0.1 nmol/day). Endogenous production of trans fatty acids is linked to the production of nitric oxide by RBC themselves or by the neighbouring endothelial lining (Zambonin et al., 2006). In any case, PUFA are easily affected by oxidative and nitrative processes (Trotschansky & Rubbo, 2008), inducing membrane damages (Van Blitterswijk, Van der Meer & Hilkmann, 1987). Perhaps, the presence of the outer RBC lipid layer may help the transfer (or interchange) of trans or damaged fatty acids (in exchange for “fresh” PUFA) to plasma lipid for disposal elsewhere, helping to extend the functionality of RBCs for a longer time.

In the end, the presence of this lipid pool results in a net loss of PUFA, probably a consequence of oxidative processes affecting first the lipid layer over the RBC, protecting the underlying membrane. The external lipid layer could, then, constitute a first line of defence against deleterious oxidative processes that shorten the lifespan and functionality of RBCs. In metabolic syndrome, RBC half-life is reduced, probably because of increased fragility and loss of flexibility (Kung, Tseng & Wang, 2009). Increased dietary supply of PUFA tends to reduce the extent of this damage (Pöschl et al., 1999), probably via lipid interchange with lipoproteins or cells (blood, endothelial, foamy, etc.) and the external lipid pool of RBCs.

Conclusions

We postulate the presence of a small lipid pool on the RBC surface. This layer may help minimize the effects of oxidative damage on RBC membranes, which affects the functionality and lifespan of RBCs, as shown by its marked deficit of PUFA. We speculate that the loss of PUFA is probably compensated through interchange with lipoproteins. Only residual trans fatty acids remained. The external lipid pool may, also help explain the role of RBCs in the transport of lipophilic compounds. The possible importance of the external lipid RBC layer should be analysed under the light of available information on the role of blood cells under conditions of hyperlipidemia and inflammation. So far, these questions have not been studied in depth, in spite of their potential importance in the transport of lipids and regulatory agents between organs and in the cell-to-cell interactions (transfer, signalling) within the tight space of tissue capillary lumen and its cell lining.

Supplemental Information