This study represents the first thorough investigation of the origin of circulating CD36 in human plasma. The results confirm our initial findings, which concluded that the proposed soluble form of CD36 (sCD36)5
is not truly soluble but entirely associated with plasma MPs.12
In the present study, we have gone on to further demonstrate that these CD36+MPs are readily detectable in both T2DM and non-T2DM individuals, occurring at levels corresponding to approximately half that of platelet MPs. Importantly, the cellular source for the majority of CD36+MPs could be accounted for using markers specific to platelet, endothelial and erythrocyte antigens. However, the relative amounts of each source differed between diabetic individuals and controls, with CD36+MPs being mainly derived from erythrocytes in T2DM as compared with endothelial cells in healthy controls.
There have been many reports of increased levels of circulating MPs in various pathological states associated with chronic inflammation and hypercoagulation, including obesity,25
the metabolic syndrome,26
Indeed, MPs are well known to exhibit both pro-coagulant and pro-inflammatory properties.17
Thus, the association of high MP counts with increased leucocyte count, a very basic but widely recognised indication of inflammatory conditions, is not unexpected. The most commonly examined MPs are those derived from platelets and endothelial cells and often incorporate various activation markers;28
while others target monocyte-derived MPs;29
and some focus on expression of molecules that may be specific to a particular function, such as Tissue Factor for its role in atherogenesis.30
Regardless of the MP species, levels are nearly always found to be increased in T2DM compared with controls. Our results are consistent with this trend, showing significantly higher levels of MPs expressing PS, CD41, CD235a, CD14 and CD36 in obese T2DM compared with obese controls. Thus, the absolute increase of CD36+MP levels could have simply been due to an overall increase in MP production associated with inflammation. However, analysis of specific subsets of CD36+MPs revealed that these were not universally increased but rather particularly derived from erythrocytes.
Very few investigations of circulating MPs have considered erythrocyte-derived MPs in any disease context. Most of the interest has occurred with respect to blood transfusion and the mechanisms of erythrocyte aging. Throughout their short existence, erythrocytes gradually lose 20% of their haemoglobin and corresponding surface area through the shedding of vesicles31
in a vital mechanism that is postulated to rid the erythrocyte of damaged membrane in order to prolong its lifespan.32
As a consequence, the favourable surface area-to-volume ratio decreases, rendering old erythrocytes less deformable. In support of increased erythrocyte MP release in diabetes, a recent report has shown that erythrocyte deformability is impaired in patients with coronary artery disease coupled with diabetes, and this was inversely correlated to both glucose concentration and HbA1c
Furthermore, analysis of MPs in stored erythrocyte transfusion bags has revealed that these vesicles are selectively enriched with the major modified haemoglobin species, including HbA1c
Thus, it would logically follow that high plasma glucose levels in poorly controlled diabetes that lead to increased percentage of HbA1c
would result in increased shedding of MPs from erythrocytes. It was surprising then that we found no correlation between individual HbA1c
and circulating erythrocyte-derived MP levels in the our diabetic cohort. This unexpected finding clearly warrants further investigation, but nevertheless, there is some evidence to associate the level of diabetes control and the increased erythrocyte MPs in diabetes.
Outside of the current work we know of only two small studies that have previously considered erythrocyte-derived MPs in T2DM, each giving rise to different conclusions. Agouni et al.26
found that erythrocyte-derived MPs were significantly increased in a cohort of 43 patients with metabolic syndrome, a third of whom also had diabetes. By contrast, Diamant et al.30
showed no difference between CD235a+MPs in 18 controls and 16 patients with normal BMI and uncomplicated diabetes of short duration. The differences may lie within the cohorts themselves as diabetic subjects analysed by Agouni et al.26
had relatively high HbA1c
levels compared with those employed by Diamant et al.30
(6.3±0.13 (ref. 26), vs 5.8±1.2 (ref. 30), respectively, compared with the normal range of 3.5–5.5%). In the present study, the HbA1c
levels were even higher (7.7±1.1% in obese diabetic individuals), thereby providing a rationale to reconcile erythrocyte MPs levels with the extent of diabetes control.
Levels of circulating endothelial cell MPs are believed to reflect cellular damage and are increasingly being used as markers of endothelial cell dysfunction in various conditions, such as cardiovascular disease.35
However, we found no difference in overall endothelial cell-derived MPs between T2DM and controls, consistent with independent reports by Sabatier et al.
and Tsimerman et al.37
This is in contrast to a report of increased total endothelial (CD144+) MPs in the metabolic syndrome,38
and a case control study by Tramontano et al.28
who found CD105+ MP levels to be the only independent predictor of diabetes. These conflicting reports may simply reflect the diabetic cohort being analysed, as endothelial MP profiles have been shown to vary specifically with accompanying diabetic vascular complications,39
which were not considered in the present study. Alternatively, medications consumed by the cohort to minimise vascular complications could also affect endothelial MP levels, as suggested by the findings that the latter are reduced by pioglitazone therapy in patients with the metabolic syndrome40
but increased in response to atorvastatin (statins were prescribed to over 60% of our T2DM cohort as indicated in Supplementary Table S1
) in patients with peripheral arterial occlusive disease.41
Independent of these considerations, we found that levels of endothelial MPs measured by detection of CD105 were unusually high; even the control cohorts had levels that surpassed that of platelet MPs. Because CD105 is not specific for endothelial cells (it can also be expressed on activated macrophages, fibroblasts and smooth muscle cells), we have omitted any CD105+MPs that co-expressed the leucocyte-specific marker CD45; however, this was negligible. Others have observed endothelial MP levels of similar enormity in controls using antibodies to comparable non-specific markers such as CD31 (found on platelets, monocytes, granulocytes, and B-cells), even when combined with CD42 to exclude platelets.42
Alternatively, particular species of endothelial MPs have been shown to increase after prolonged storage at −80
The controls for the present study were collected before the cases and thus may have endothelial MP levels that are artefactually increased due to longer storage times. Although all of the samples were processed in identical fashion, there is no consensus on the effects of long-term storage for MP enumeration, and this may also explain why the control endothelial MP levels appeared to be equal to that of the T2DM patients.
This is the first study to quantify CD36 protein levels in human plasma. These were found to vary widely from undetectable (<78
) to as high as 22.9
. These values are relatively high compared with other plasma proteins such as sCD40L and sICAM (soluble intercellular adhesion molecule), which are in the order of 3 and 300
, respectively. The only other published study to use a quantitative CD36 ELISA was performed by Chmielewski et al.
but their assay differed from ours in that they used serum rather than plasma samples. Although serum has been reported to contain 10 times more platelet MPs than plasma,45
it is not clear how much CD36 is contributed by these vesicles that have been produced by artificial platelet activation in vitro
. In addition, other CD36-expressing MPs present before activation may become entangled in the clot that forms. Chmielewski et al.
found the median CD36 protein concentration in serum to be a low 25.3
in non-diabetic patients, suggesting that the latter may be true. The recently released commercial ELISA kits for quantification of CD36 in human plasma have a reported sensitivity of 1.0
(Adipo Bioscience, Santa Clara, CA, USA) and should facilitate further studies in this area.
The high variability in plasma CD36 protein concentrations detected by ELISA may reflect actual individual differences in cellular expression of CD36. Protein levels on the surface of monocytes and platelets are largely determined by the inheritance of common CD36 gene polymorphisms,46, 47
but this expression is also subject to further exogenous modulation. Of particular relevance to the present cohort, statins have been shown to decrease CD36 expression on monocytes and have been associated with lower levels of sCD36 in diabetic individuals.44, 48
Conversely, CD36 expression can be increased by both fat in the diet49
If the number of CD36 molecules on MPs is reflective of that of the parent cell, then any analysis of absolute circulating CD36 protein levels will be subject to these various confounders. By contrast, quantification of the number of MPs expressing CD36 will be less compromised by individual protein expression levels. For these reasons, it is perhaps not surprising that plasma CD36+MP levels were a much better marker of T2DM than CD36 protein concentration per se