Marked increase in rat red blood cell membrane protein glycosylation by one-month treatment with a cafeteria diet

Introduction

Glucose, in addition to being the main intercellular energy staple, is a reducing aldose. Thus, it may easily react with a number of chemical groups in proteins and other biological compounds. The direct condensation with protein free amino groups (Maillard reactions) (John & Lamb, 1993) is fairly common, to the degree that a significant proportion of circulating plasma proteins are glycosylated (Gragnoli et al., 1982), as well as proteins in the red blood cell (RBC) membrane (Miller, Gravallese & Bunn, 1980) and the hemoglobin they contain (Bunn, Gabbay & Gallop, 1978). The proportion of hemoglobin glycosylated in the terminal valine of chain B (Hb_1AC) is currently used as an index of overall exposure to free plasma glucose over time (Siu & Yuen, 2014). Glycosylation products may undergo Amadori reorganizations, producing a number of complex compounds known as advanced glycosylation products (AGP) (Henning et al., 2011), which play a significant role in the control of substrate utilization (Wu et al., 2011), cell function (Guo et al., 2012) and inflammation (Poulsen et al., 2014).

The chemical reactivity of glucose is often overlooked because of its overwhelming function in energy supply and rapid turnover, but direct non-enzymatic glycosylation remains a common mechanism of alteration of protein function and interference in signaling pathways (Asahi et al., 2000; Itkonen & Mills, 2013). It is commonly accepted that higher sustained circulating levels of glucose, as in diabetes, result in increased proportions of glycosylated proteins in plasma, RBCs and endothelial cells, Hb_1AC being a case in point (Carson et al., 2010). In fact, equations based on the correlation between mean estimated plasma glucose concentration and Hb_1AC proportion are currently in use (Borg & Kuenen, 2009; Nathan et al., 2008).

The self-selected cafeteria diets (Sclafani & Springer, 1976) are essentially hyperlipidic (Prats et al., 1989), and its consumption by rats causes hyperphagia, insulin resistance and obesity (Correa Pinto & Monteiro Seraphim, 2012; Prats et al., 1989). Exposure for one month of young adult rats to a cafeteria diet induces a number of metabolic changes that are in the limit of normalcy and correspond to the initial stages of the metabolic syndrome (Romero et al., 2010). The effects are more marked in male than in female rats (Romero et al., 2012), probably because of the anti-inflammatory effects of estrogen (Thomas et al., 2003); but, in any case, the obesity is already patent. Short-term treatment with cafeteria diets induce a mild hyperglycemia and hyperinsulinemia (Romero et al., 2010), but not frank diabetes, which is more developed after prolonged exposure (Castell-Auví et al., 2012; Correa Pinto & Monteiro Seraphim, 2012).

In the present study, we have analyzed whether the glycosylation degree of total plasma or RBC proteins, as well as those of RBC membranes, are a direct correlate of their prolonged contact with plasma glucose in an early stage of development of hyperglycemia. We wanted, also, to check whether sex exerts any influence on the glycosylation response to comparable glucose concentrations.

Materials and Methods

Results

Table 1 presents the changes in body weight experienced by the rats during one month of exposure to a cafeteria diet. The initial weight difference between female and male rats widened with time, since control males increased about 20% of their weight, compared with 15% of females; cafeteria diet increased body weight 35% in males and 34% in females. Males ate more energy than females: 36% (control diet) or 19% (cafeteria diet). Males’ food (expressed as energy content) intake was 2.5-fold higher in cafeteria than in control diet; the value for females was 2.8×.

Standard plasma glucose (measured on day 29) showed both an effect of sex (female levels being lower) and diet (cafeteria diet data being higher).

Table 2 shows the data obtained from the analysis of blood extracted under isoflurane anesthesia. In this case, all plasma glucose data were higher than those obtained on day 29 under basal conditions, and there were no statistical differences between the groups attributable to sex or diet. Total blood glucose values were lower than those of plasma, and showed neither differences by sex or diet. However, the estimated data for cell glucose showed a clear effect of diet (Fig. 1). In both groups of cafeteria rats, the levels were minimal, not statistically different from zero, while those of rats under the control diet were lower than in plasma but clearly positive, the differences being not significant for “sex” but significant for “diet”. In control rats, when water content of plasma (about 92%) and packed cells (about 70%) was included in the calculations, the molal concentrations of cell glucose were in the range of 1/3rd of those of plasma; female rats presented similar values. Cafeteria diet-fed rats showed values in the range of only 4–7%.

The proportions of glycosylated protein, both in RBC and in plasma, are presented in Fig. 2. No significant differences were observed between the groups for “sex” and “diet”. However, cell protein was more heavily glycosylated than plasma proteins. In the case of cells, since most of the protein (>95%) is hemoglobin, it can be assumed that most glycosyl residues were bound to this protein; since its molecular weight (tetramer) is about 64,000, the molar ratio of HMF to hemoglobin was about 320, i.e., about 80 glycosyl residues per hemoglobin subunit. This value is about six-fold higher than the number of sites representing 7% Hb_1AC, which is limited to the terminal chains of hemoglobin. In the case of plasma, since albumin makes about 55% of plasma proteins and its molecular weight is close to 66,500, we obtain about 90 glycosyl residues per molecule. Evidently, this is only an imprecise approximation but shows that under the particular conditions of this experiment, protein glycosylation was significant and about 3.5 times more intensive in cells (on a molar ratio) than in plasma proteins as a whole.

Figure 3 depicts the rate of glycosylation observed in purified membranes of blood cells. Since purification of membranes is not even close to quantitative, we could not determine in which proportion RBC membranes were glycosylated. In fact, we were not able to ascertain the degree of the purity of samples. Thus, membrane proteins could be contaminated by hemoglobin (in spite of the appearance of total elimination at the expense of dwindling recovery of membranes), spectrin or other molecules. Thus, we decided to also relate the degree of glycosylation to phospholipid, an exclusive membrane component in RBC. The molar ratio of released HMF to phospholipid phosphate (Fig. 3) showed an image quite different from that of Fig. 2. There were no statistical differences between groups attributable to “sex”. This was clear for control diet, but the post-hoc test showed a significant (P < 0.05) sex-related difference in cafeteria-fed rats. The effect of “diet” was significant, with several-fold higher values in cafeteria- than in control-fed rats. Presentation of the data of HMF per mg of membrane preparation protein shown in Fig. 3, yields almost the same pattern, but statistical significance was lower because the individual variation of data was higher.

Discussion

In the development of this apparently simple study, we tried to maintain a close control of methodology, since the problems of glycosylation of blood components have generated a sizeable number of studies, but their integrated analysis is scarce, in a way that only limited comparisons have been studied. We intended to present homologous data for plasma and RBC proteins, including also samples of RBC membranes, and using a model in which the metabolic syndrome pathologies, especially insulin resistance and hyperglycemia, were not fully set in.

The problem of anesthesia as hyperglycemic agent (Arola et al., 1981; Zuurbier et al., 2008) has not been solved; we opted by using this avenue to obtain sufficient blood to carry out all the compartmentation and membrane experiments in the same samples. Consequently we had to obtain separate plasma samples to compare the basal results with previous studies (Palou et al., 1980). The changes elicited by cafeteria diet agree with previously published studies (Ferrer-Lorente et al., 2005). We assumed that the brief isoflurane anesthesia-induced hyperglycemia (Zuurbier et al., 2008) (less than 5 min from start to exsanguination) changes plasma glucose levels, but its effects on RBC glucose (if any) would be at least partly buffered. In any case, it is highly improbable that these changes would affect differentially the rats depending on their diet. The uniformity of the data obtained seem to support this weak point of our experimental setup. We have not been able to circumvent the problem within the ethical standards of our Laboratory.

The lower blood vs. plasma glucose levels, more marked in cafeteria diet-fed rats, attest directly to a lower cell compartment glucose content (there were minimal differences in packed cell volume). The accuracy of the calculations used to quantify the cell glucose content notwithstanding, do not change the fact that cafeteria rats had higher glucose content in the blood plasma fraction compared with that of cells; precisely the glucose in direct contact with hemoglobin.

Metabolic syndrome, diabetes and in general, high exposure to inflammation and hyperglycemia increase the glycosylation of plasma proteins (Matsuura, Hughes & Khamashta, 2008; Roohk & Zaidi, 2008). In fact, glycosylated albumin has been proposed as an indicator of maintained hyperglycemia (i.e., exposure of plasma proteins to higher aldose levels for long periods) (Abe & Matsumoto, 2008). However, the most used indicator of long-time maintained hyperglycemia is the measurement of glycosylated hemoglobin (Siu & Yuen, 2014), which initially was applied to whole RBC hemoglobin (Carson et al., 2010), but soon was focused on the terminal amino groups of hemoglobin (Hb_1AC) alone, giving rise to a much more sensitive (and extended) assay methodology (Little et al., 2008; Weykamp et al., 2008). The study of glycosylated hemoglobin (Hb_1AC) has become one of the standard elements for the evaluation of diabetes (metabolic syndrome) and, in general, sustained hyperglycemia (Ong et al., 2010). The critical point, however, is that all hemoglobin is contained within the RBC, and is not in direct contact with plasma glucose. This obvious circumstance would make a priori glycosylated albumin a more acceptable indicator of hyperglycemia. The conundrum of a marker of glycosylation not in direct contact with the parameter it measures has not been sufficiently explained so far. Nevertheless, its widespread use and clinical reliability are powerful reasons in favor of its continued use despite the largely unexplained nature of its origins.

In mammals, the direct permeability of the RBC membrane to glucose is low, if any (Britton, 1964; Rich et al., 1967), however, interchange of plasma and RBC glucose is active thanks to a facilitated-diffusion transport system (Levine, Oxender & Stein, 1965). The transport has been attributed, mainly to GLUT1 (Graybill et al., 2006), which function may be regulated by insulin, glucocorticoids and other factors (Kahn & Flyer, 1990). However, no differences in glucose transport through erythrocyte membranes were found between diabetic and euglycemic children (Mortensen & Brahm, 1985). There is, also, a high variability in the permeability of RBC membranes to glucose, due to species differences, individual factors and transporter modulation/saturation (Khera et al., 2008).

Compartmentation of blood glucose between plasma and cells may be an important regulatory factor by itself (Palou et al., 1980), since glucose carried by blood cells is rapidly interchanged with tissues (Jacquez, 1984). This is in overt contradiction with the slow velocity of glucose interchange of RBC when measured in vitro (Sen & Widdas, 1962). In addition, given the glycolytic nature of mammalian RBC, it can be expected that a sizeable part of the glucose entering the cell is rapidly glycolyzed to lactate, a process that is the only significant source of ATP for the cell. This inefficient mechanism converts blood in a sizeable source of lactate, which implies that a variable part of glucose will be converted to hexoses-P on entering the cell, and thus (at least in part, when isomerized to ketose-P) lose its aldose-related glycosylating capacity.

The high proportions of Hb_1AC found under conditions of assumed sustained hyperglycemia (Giuffrida et al., 2010) contrast with the physical existence of barriers between hyperglycemic plasma and hemoglobin. We could not explain why Hb_1AC is so highly correlated with hyperglycemia, since total hemoglobin glycosylation does not reflect only hyperglycemia (Adams et al., 2009; Chan et al., 2014), which is in agreement with our data. In addition, diabetogenic conditions, such as those presented here and those found in the literature (Koga et al., 2007; Miyashita et al., 2007; Wakabayashi, 2012) do not show the expected changes in hemoglobin glycosylation. Our data on the lack of significant changes elicited by diet on plasma protein glycosylation do not agree with the common occurrence of increased glycosylated proteins in plasma of humans and rodents alike under already settled metabolic syndrome or its associated pathologies (Gornik & Lauc, 2008). The probable differences lie in the fact that in our model of initial stages of metabolic syndrome, the pathologic markers have not been yet developed fully, as we have previously found (Ferrer-Lorente et al., 2005). It must be also taken into account that metabolic syndrome-induced modifications on plasma proteins (Marliss et al., 2006; Welle et al., 1992) and RBC (Cohen, Franco & Joiner, 2004; Manodori & Kuypers, 2002) increase their cell turnover rates which compounds the problem and makes more difficult the comparisons unless the data maintain their homology.

The elevated degree of glycosylation found in RBC membranes, however, shows that even the small differences in basal glycemia found in our model are enough to already induce several-fold changes in the glycosylative activity of plasma glucose. Probably, other factors so far not identified, may help explain the increased glycosylation observed even at early stages of the development of metabolic syndrome. The relationship with high-energy (lipid) diet is clear, but the common assumption that these changes are a correlate of hyperglycemia remain unproven, and largely based only on indirect evidence.

In our experiment, the degree of glycosylation of hemoglobin was high, even under conditions in which practically no free glucose was found within the RBC. We only measured glycosyl residues, not those recombined by Amadori rearrangements, i.e., those able to liberate hydroxymethylfurfural. However, the ratio of up to 90 glycosyl residues per subunit of hemoglobin is close to the level of saturation of glycosylable sites. This was necessarily achieved in at most one month, a time shorter than the mean rat RBC half-life, 45–50 days (Burwell, Brickley & Finch, 1953), a value considerably decreased in rats with metabolic syndrome (Kung, Tseng & Wang, 2009). However, compared with albumin which median life span is close to 2 days (Reed et al., 1988), the differences can be better explained, since exposure of hemoglobin was 15-fold higher (30 days out of 45–50) than that of plasma proteins (2 days, assuming a behavior comparable to that of albumin). The shorter exposure was predictably more intense (as that of RBC membrane protein) because plasma proteins were in constant contact with plasma glucose.

The levels of glycosylation of plasma proteins and hemoglobin observed do not reflect the (limited) changes in plasma glucose, however, RBC membranes do. The results we obtained are puzzling; they agree with the known fact that exposure to hyperglycemia results in increased protein glycosylation, as shown by membrane proteins’ differences, but not observed in plasma proteins; this may be due to their shorter half-lives and limited span of glucose level change.

On the other side, the low free glucose levels observed in cafeteria diet-fed rat RBC, agree with a slower rate of uptake (Prats et al., 1987) compared with plasma, but cannot directly explain how the overall glycosylation of hemoglobin was unaffected by one month of consumption of a hyper-energetic obesogenic diet.

Conclusions

We conclude that blood glucose compartmentation, as previously indicated, may play a role, in the regulation of plasma/blood versus tissue glucose transport/transfer, more important than usually assumed, but also, that glycosylation of blood proteins widely affects non-diabetic young experimental animals, both under standard or hyper-energetic diet conditions. This extensive glycosylation does not seem to be markedly affected by sex; and could be hardly justified from an assumedly sustained plasma hyperglycemia. More detailed—and comprehensive—analyses should be carried out to study the dynamics of glucose in vivo in the RBC to understand how so extensive protein glycosylation as that found here could take place, including an special emphasis on the hormonal regulation of RBC glucose transporters.

We have also found that RBC membrane glycosylation is a sensitive indicator of developing metabolic syndrome-related hyperglycemia, more sensitive than the general measurement of plasma or RBC protein glycosylation.

Supplemental Information