A link between diabetes and atherosclerosis: Glucose regulates expression of CD36 at the level of translation

/Home/Complications of Diabetes/Diabetes and Atherosclerosis/Diabetic Atherosclerosis Research

A link between diabetes and atherosclerosis: Glucose regulates expression of CD36 at the level of translation
Erik Griffin, Alessandro Re, Nance Hamel, Chenzong Fu, Harry Bush, Timothy McCaffrey & Adam S. Asch
Nature Medicine

Both the risk and the rate of development of atherosclerosis are increased in diabetics, but the mechanisms involved are unknown. Here we report a glucose-mediated increase in CD36 mRNA translation efficiency that results in increased expression of the macrophage scavenger receptor CD36. Expression of CD36 was increased in endarterectomy lesions from patients with a history of hyperglycemia. Macrophages that were differentiated from human peripheral blood monocytes in the presence of high glucose concentrations showed increased expression of cell-surface CD36 secondary to an increase in translational efficiency of CD36 mRNA. We obtained similar data from primary cells isolated from human vascular lesions, and we found that glucose sensitivity is a function of ribosomal reinitiation following translation of an upstream open reading frame (uORF). Increased translation of macrophage CD36 transcript under high glucose conditions provides a mechanism for accelerated atherosclerosis in diabetics.

Atherosclerosis is one of the major vascular complications of diabetes and has enormous impact on the health of those affected1, 2, 3. Atherosclerosis is thought to develop as a result of lipid uptake by vascular-wall macrophages leading to the development of foam cells and the elaboration of cytokines and chemokines resulting in smooth muscle−cell proliferation4, 5. In macrophages, CD36 is a scavenger receptor that mediates uptake of oxidized low-density lipoprotein (OxLDL) and subsequent foam-cell development6, 7, 8, 9, 10, 11, 12, 13.

Increased CD36 expression in vascular lesions
CD36 expression is regulated at the transcriptional level by various cellular stimuli including soluble mediators, differentiation and adhesion events. Recent data indicate that in response to glucose, increases in peroxisome-proliferator&#8722;activated receptor- (PPAR-) might lead to an increase in macrophage expression of CD36 and contribute to accelerated atherosclerosis14. Consistent with this, increased CD36 protein expression has been found in the myocardium of hyperglycemic mice15. In order to explore the relative expression of CD36 in diabetes, we examined vascular tissue sections from patients undergoing surgical revascularization for atherosclerotic disease. We examined vascular lesions from 24 patients and scored them for CD36 expression by immunohistochemistry. Patients whose fasting glucose levels in the perioperative period were greater than 140 mg/dl were scored as hyperglycemic. CD36 expression was significantly more prevalent in vascular lesions from patients with a history of hyperglycemia (8/12) than in lesions from normoglycemics (3/12) (P < 0.001, 2; Fig. 1). Quantitative analysis confirmed these findings and revealed positive staining for CD36 in an average of 2578 pixels per high-power field in the hyperglycemic group, versus 414 pixels per high-power field in the normoglycemics (P = 0.01, Student's t-test). The macrophage-specific antigen Ham56 was detected by staining in all of the lesions examined (data not shown). These data demonstrate a correlation between increased CD36 expression and hyperglycemia in atherosclerotic vascular lesions and raised the possibility that increased expression was mediated transcriptionally—perhaps by PPAR-. These data do not, however, establish a causal relationship between glucose and CD36 expression, given that many other factors such as insulin level might have a role in expression of CD36 in the vessel wall.


Figure 1. CD36 expression by tissue macrophages in vascular lesions.

Vascular lesions from 4 of 24 vascular lesions stained for CD36 expression using avidin-biotin-peroxidase and DAB. a and b, Hyperglycemic. c and d, Normoglycemic. Magnification, 200.

Full Figure and legend (347K)


Expression of CD36 correlates with glucose level
To examine the relationship between CD36 expression and glucose, we measured the surface expression of CD36 by flow cytometry in macrophages derived from peripheral blood mononuclear cells that were incubated for five days in medium containing 100, 200 or 600 mg/dl glucose. CD36 expression was significantly increased over the range of glucose concentrations examined (P < 0.001, Kolmogorov&#8722;Smirnov statistical analysis), with mean channel fluorescence increasing from 16 to 89, corresponding to a five-fold increase in surface expression of CD36 (Fig. 2a). Another monocyte/macrophage antigen, CD14, remained unchanged under these conditions (mean channel fluorescence ranged from 41 at 100 mg/dl to 35 at 600 mg/dl).


Figure 2. Immunologic and functional expression of CD36.

a, Flow cytometric analysis of monocyte-macrophages. Monocyte-macrophages incubated in medium containing 100 (blue), 200 (red) or 600 (yellow) mg/dl glucose for 5 d were examined for CD36 and for CD14 expression. CD36 expression was significantly (P < 0.001 by Kolmogorov&#8722;Smirnov statistical analysis) increased over the range of glucose concentrations examined, with mean channel fluorescence increasing from 16 to 89 corresponding to a 5-fold increase in surface expression of CD36. Another monocyte-macrophage antigen, CD14, remained unchanged under these conditions (mean channel fluorescence ranged from 41 to 35). Data are from 1 of 4 similar experiments. b, Monocyte-derived macrophage Ox-LDL uptake. Monocyte-macrophages incubated at 100, 200 or 600 mg/dl glucose conditions for 5 d were examined for their ability to bind and ingest 125I-labeled Ox-LDL. CD36-dependent uptake () is defined as fucoidin-resistant Ox-LDL uptake. Non-CD36 dependent uptake () is defined as fucoidin-inhibitable (fucoidin is an inhibitor of the Type I scavenger receptor). Data are the means of triplicate points c.p.m. s.e.m. c, Radiolabled Ox-LDL uptake over a range of glucose concentrations. Lesion-derived phagocytic cells were incubated with 125I-labeled Ox-LDL and ingested radiolabeled lipid particles measured after 12 h. Ox-LDL uptake increased 4-fold in response to glucose () but not to mannitol (). Data are from 1 of 3 similar experiments. d, Ox-LDL uptake by atherosclerotic lesion-derived phagocytic cells. Lesion-derived cells were incubated over a range of indicated glucose concentrations in media containing 40 g/ml Ox-LDL for 12 h and stained with Oil Red O to visualize ingested lipid.



Full Figure and legend (327K)


Because CD36 is responsible for much of the Ox-LDL uptake of macrophages, we investigated whether the increase in surface expression correlated with an increase in Ox-LDL uptake in the cells. Although there was no significant change in non-CD36-mediated Ox-LDL uptake, the increased CD36 expression correlated with an almost 10-fold increase in CD36-mediated Ox-LDL uptake (Fig. 2b). We obtained similar results with primary vascular cells derived from atherosclerotic lesions (ref. 16 and F. Pearce et al., manuscript submitted) that express several monocyte-macrophage characteristics including phagocytosis, CD36 and CD14 expression. Analysis revealed increased expression of CD36 correlating with Ox-LDL uptake over a range of glucose concentrations. No effect was seen when mannitol was used as an osmotic control (Fig. 2c and d).

Increased CD36 translational efficiency with hyperglycemia
Although increased CD36 expression has been reported in hyperglycemia in the myocardium of mice15, it is not known whether this response or the observed increase in CD36 expression was mediated transcriptionally (by PPAR-) or post-transcriptionally. We performed RNAse protection assays on RNA from monocyte-derived macrophages that were incubated with various glucose concentrations. As expression did not correlate with changes in the steady-state level of CD36 mRNA, the increase in CD36 expression in response to high glucose level must occur post-transcriptionally. The steady-state level of CD36 mRNA relative to actin mRNA was 0.517 at 100 mg/dl and 0.395 at 600 mg/dl as determined by densitometric analysis (Fig. 3a).


Figure 3. RNA analysis.

a, Steady-state levels of CD36 mRNA from monocyte-macrophages cultured under low ('L'; 100 mg/dl) or high ('H'; 600 mg/dl) glucose was examined using RNase protection with antisense probes specific for CD36 and -Actin. b, Monocyte-derived macrophage sucrose-gradient polysome analysis. Fractions derived from a sucrose gradient centrifugation of polysomes derived from monocyte-macrophages incubated in low (100 mg/dl) or high (600 mg/dl) glucose were analyzed by RT-PCR to amplify a 311-bp fragment of CD36 cDNA. Migration in the gradient of CD36 mRNA derived from high glucose monocyte-macrophages is shifted to the heaviest fractions corresponding to an increased number of ribosomes/transcript and increased translational efficiency. Actin mRNA did not change its position in the gradient in response to glucose concentration. The positions of 18 and 28S rRNA within the gradient were determined by semi-quantitative RT-PCR. c, Lesion-derived phagocytic cell sucrose-gradient polysome analysis. The translational efficiencies of CD36, glyceraldehyde phosphate dehydrogenase (GAP), -actin, and CD14 in lesion-derived cells incubated at 50 or 200 mg/dl glucose were examined. The optical density (OD) at 254 nm is given for each of the 2 conditions, and the positions of 18 and 28S rRNA were determined by semi-quantitative RT-PCR. CD36 transcript, but not GAP, actin or CD14 transcripts is seen predominantly in the fractions that appear to correspond to the 80S ribosome (monosome) peak. Data are from 1 of 5 similar experiments.

This finding indicated that translational regulation might be responsible for the increase in CD36 expression. To examine whether the translational efficiency of CD36 mRNA is sensitive to glucose concentration, we fractionated on sucrose gradients polysome preparations derived from monocyte-derived macrophages exposed to 100 or 600 mg/dl glucose, and we performed reverse transcriptase (RT)-PCR on gradient fractions using CD36 and -actin specific primers. We used primers specific for 18S and 28S mRNA to determine the relative positions of the corresponding ribosomal subunits in the gradient. We detected CD36 transcripts from macrophages grown in 600 mg/dl glucose in the heaviest fractions (more ribosomes), whereas we found CD36 transcripts from macrophages grown in 100 mg/dl in lighter fractions (fewer ribosomes) (Fig. 3b). This finding indicates that increased expression of CD36 in hyperglycemic conditions is mediated by increased translational efficiency. The distribution of -actin transcript in the polysome fractions was unaffected by glucose concentration. We obtained similar results with the human vascular-derived phagocytic cells (Fig. 3c). Our studies show that translation of CD36 but not control transcripts is sensitive to glucose concentration and leads to greater ribosomal loading of the transcript at higher glucose levels.

The 5'-UTR of the CD36 transcript is glucose-sensitive
Increased mRNA levels have been reported to mediate the effects of glucose on the expression of key proteins in matrix deposition such as TGF-1 (ref. 17), and translational regulation in response to glucose levels has also been described in yeast18. Glucose is thought to regulate the translation of several transcripts important in glucostasis including proinsulin19,gluco-kinase20, fatty-acid synthase21 and others22; however, the evidence for translational control of these transcripts is somewhat indirect. The metabolic state of a cell can affect gene expression at the translational level in part by modulating the availability and state of phosphorylation of multiple initiation factors involved in the complex process of translational initiation23, 24. Translational efficiency can also be regulated by features of the 5' untranslated region (UTR) such as length, secondary structural features or the presence of upstream uORFs (ref. 25). The interaction between these elements and the availability of translational cofactors can provide precise control over the translation of individual transcripts. The CD36 transcript has a 5'-UTR length of more than 200 bp, and contains 3 uORFs and several stable predicted secondary structures. All these features are consistent with tight regulation of CD36 translation (Fig. 4).


Figure 4. Secondary structure of CD36 5'-UTR.

The 5'-UTR of CD36 was analyzed by M-fold43, 44 to determine possible stable structures due to base pairing. The structures obtained contained several regions of stable base-pairing resulting in a complex folded pattern with numerous hairpin loops. Arrows indicate the positions of start codons for each of the uORFs and for CD36.



Full Figure and legend (25K)


Glucose-mediated control of CD36 translation may depend upon this relatively complex 5'-UTR, and translation of an ORF downstream of the CD36 5'-UTR might be induced by elevated glucose levels. To test this, we fused a luciferase reporter downstream of the CD36 5'-UTR. We examined translation of CD36 5'-UTR&#8722;bearing and control leader-reporter transcripts in a monocytoid cell line over a range of clinically relevant glucose concentrations using a control leader-reporter construct. Translation of the CD36 5'-UTR&#8722;bearing transcript relative to the control transcript increased with higher glucose concentration in the medium (Fig. 5), indicating the importance of the CD36 5'-UTR for the glucose-mediated regulation of CD36 expression.


Figure 5. Role of CD36 5'-UTR in glucose-mediated increase in translation.

Reporter expression in response to glucose concentration was measured in the THP-1 monocyte cell line 48 h following transfection with a luciferase reporter construct bearing the 5' untranslated region of CD36 (5'-UTR-luc) or a reporter control (luc). Data are from 1 of 3 similar experiments. Translational efficiency is calculated here as disinhibition of reporter expression: 1/(1&#8722;5'-UTRluc/luc). Data are from 1 of 4 experiments with similar results s.e.m.



Full Figure and legend (10K)


Translation reinitiation is critical for the glucose sensitivity
The presence of three uORFs in the CD36 5'-UTR indicates their possible involvement in the regulation of CD36 translation. To test this, we mutated all of the upstream AUG codons to create M123 or the first or third upstream AUGs to create M1 and M3 constructs and fused the new leaders to a reporter. If any of the uORFs were critical to glucose sensitivity of CD36 translation, the M123 construct would be expected to show decreased sensitivity to glucose and thus more efficient translation at low glucose concentrations than the wild-type leader-reporter. We found that expression of the M123 construct increased five-fold over the wild-type construct at low glucose concentrations and that the mutated construct was no longer responsive to glucose. Loss of the first AUG (M1) resulted in a loss of glucose responsiveness, but loss of the third AUG (M3) remained glucose responsive (Fig. 6). These findings again demonstrate an inhibition of translation exerted by the wild-type leader at low glucose concentrations, and disinhibition of translation as glucose levels increased. Moreover, the uORFs appear to be the features within the 5'-UTR responsible for glucose-mediated regulation of CD36 translation, and uORF 1 appears to be necessary for glucose sensitivity. To ensure that the changes over the range of glucose concentrations were not due to changes in message stability, we examined mRNA levels for each of the constructs at each glucose level by semi-quatitative RT-PCR (Fig. 6).


Figure 6. Mechanism of glucose-mediated increase in translation.

a, SEAP expression was measured following transfection of reporter constructs with the wild-type leader, M3, M1, M123, termination mutant 1 (Ter1), termination mutant 2 (Ter2) or termination mutant 3 (Ter3). b, The wild-type leader and M3 show glucose responsiveness over a range of glucose concentrations. M1, M123, Ter1, Ter2 and Ter3 constructs show a loss of glucose responsiveness. Colors of lines correspond to construct labels in a. Data are the normalized means s.d. of 5 experiments. c, Steady-state mRNA at each of the glucose levels was assessed for each of the constructs by semi-quantitative RT-PCR.



Full Figure and legend (63K)


Such regulation of translation by uORFs is thought to occur principally by one of two different mechanisms—either modulation of leaky ribosomal scanning or ribosomal reinitiation. To distinguish between these possibilities, we used a method designed to define the regulation of the yeast GCN4 system23, 26. We examined the translation of reporter constructs in which each of the termination codons of the uORFs in the 5'-UTR were mutated so that the new uORFs overlapped with the start codon of the reporter construct. In such constructs, reporter translation is due only to leaky scanning beyond the upstream AUG. If leaky scanning past any of the upstream ORFs was sensitive to glucose, then reporter expression would remain glucose-sensitive. Alternatively, if reinitiation following translation of the wild-type uORF was the glucose-sensitive feature of the transcript, then the glucose sensitivity of the termination mutants would be abolished. We examined expression of the reporter secretory alkaline phosphotase (SEAP) following transfection over a range of glucose concentrations from 100&#8722;600 mg/dl (Fig. 6). The wild-type UTR showed a more than three-fold increase in translation over this range of glucose. In contrast, the termination mutants showed complete loss of glucose responsiveness, supporting reinitiation following translation of the first ORF as a critical step in the glucose responsiveness of the CD36 transcript. These findings show that an increase in the efficiency of reinitiation following translation of upstream ORFs is the primary mechanism of glucose responsiveness of this transcript. The 85% inhibition of translation at 100 mg/dl seen with termination mutant 1 relative to M123 shows that translation of this uORF is highly efficient. The fact that the start AUG of uORF 2 overlaps with the stop codon of uORF 1 (see Fig. 5) makes it unlikely that uORF 2 is translated except by those ribosomes that scan past the AUG of uORF 1. The data indicate that fewer than 15% of the ribosomes could recognize the start of uORF 2 when in the context of the wild-type transcript. Thus, we suggest that reinitiation following translation of uORF 1 is likely to be primarily responsible for the glucose sensitivity of the transcript.

Discussion
Several potential links between diabetes and atherosclerosis have been identified and many clinical observations point to the correlation between risk of vascular complications in diabetes and poor glycemic control27. Abnormalities of apoprotein and lipoprotein particle distribution, hyperinsulinemia, alterations in growth-factor expression and cytokine cascades are some of the potential mechanisms1. Some evidence also points to the role of advanced glycation end products (AGEs) and their receptors (RAGE) as possible mediators for diabetic vascular damage, and CD36 has been identified as a receptor for advanced glycation end products3, 28, 29. In addition to these proposed mechanisms, substantial evidence supports the central role of Ox-LDL in atherogenesis9, 10, 11, 12. Consistent with this idea, CD36-deficient mice bred on an apolipoprotein-E&#8722;deficient backround show delayed atherosclerosis compared with the background13. Several factors are thought to promote lipoprotein oxidation in diabetes: enhancement of auto-oxidative glycosylation (or glycoxidation) in the presence of high glucose concentration; increased production of free radicals and lipid peroxidation in the setting of hyperglycemia; and the decrease in antioxidant pathways described in diabetics6, 30.

Although its pathophysiologic role in the development of atherosclerotic foam cells is clear, the true physiology of CD36 is much less so. Some data indicate that CD36 might participate in fatty-acid metabolism by functioning as one of several receptors for fatty-acid transport, particularly in adipocytes31, 32. The role of PPARs in regulating adipogenesis as well as cardiac and placental development has recently been shown33. This class of nuclear hormone receptors appears also to have a substantial role in regulating insulin responsiveness and fatty-acid metabolism34. PPAR- has also been shown to regulate CD36 transcription in macrophages7, 8, 14. Further evidence indicates that although PPAR- is sufficient to induce an adipogenic program in cultured fibroblasts, the establishment of an insulin-sensitive glucose transport might require expression of CCAAT/enhancer binding protein- (C/EBP-)35, a transcription factor the expression of which is known to be regulated at a translational level.

We have shown that CD36 expression in human monocyte-derived macrophages and in macrophage-like cells is tightly regulated by a translational control mechanism dependent on ribosomal reinitiation. Why is expression of CD36 so closely regulated? What is the physiology underlying the tight translational regulation that in higher eukaryotes is more closely associated with transcripts encoding cytokines, oncogenes or their receptors, as well as a small but growing list of transcriptional activators? Perhaps the evolution of this regulation has more to do with the physiology of starvation than with response to increased glucose levels. Expression of CD36, insulin and PPAR are all glucose-sensitive, such that expression of all three would be curtailed at low glucose levels—and at least two of these are regulated at both the transcriptional and translational level. Our ability to survive between meals or when meals are scarce depends on our ability to use free fatty acids. In response to low levels of glucose, insulin levels decrease leading to increased lipolysis and release of free fatty acids. Expression in adipocytes of a transporter capable of removing free fatty acid (FFA) from the circulation would be counterproductive. Whatever the evolutionary pressure, the regulation of CD36 is aimed at preventing the inappropriate expression of this molecule. From a therapeutic standpoint, these findings might be good news for those taking PPAR agonists for diabetes. A concern has been raised that PPAR agonists might worsen atherosclerosis by increasing transcriptional of CD36 (ref. 14). One possible interpretation of these findings is that whereas CD36 mRNA might increase in response to PPAR agonists, CD36 translation will be limited by good glycemic control.

The mechanism by which CD36 translation is achieved appears to have some similarities to the mechanism first described in the yeast GCN4 transcript. The GCN4 transcript has a long 5' leader containing four upstream ORFs that, along with the degree of phosphorylation of initiation factor 2 (eIF2), control the translation of the major ORF (ref. 23). There are, however, significant differences from GCN4 as shown by our data. In contrast to the GCN4 system, where induction of translation relies on impaired reinitiation allowing ribosomes to skip downstream uORFs and end up at the GCN4 ORF, reinitiation is increased following translation of uORF 1/2 in CD36. The 5'-UTR regions of several transcripts regulate translation during embryonic development in a spatiotemporal manner. Members of the antennapedia and retinoic-acid receptor families, insulin like growth factor II, platelet-derived growth factor 2, transforming growth factor- , fibroblast growth factor 2 and vascular endothelial growth factor as well as c-myc, c-mos and C/EBP- and CEBP- are all transcripts whose translation is closely regulated36, 37, 38, 39. Recently a 5'-UTR database has been subjected to classification and regression-tree analysis, categorizing 5'-UTRs according to common structural features such as predicted stable folds, UTR length greater than 100 bp, upstream AUG codons, upstream ORFs or terminal oligopyrimidine tracts39. The CD36 5'-UTR shares several features that are typical of transcripts that are closely regulated.

Our data provide a direct mechanistic link between hyperglycemia and the pathophysiology of accelerated atherosclerosis in diabetes. To our knowledge, this is the first example of human pathophysiology due to altered translation of a normal human mRNA transcript in response to glucose. So far, few disease processes have been clearly tied to alterations in physiologic translational control mechanisms. Several viruses adopt the translational machinery of the cell, shutting down cap-dependent translation of the host mRNA and use internal ribosomal entry to facilitate preferential translation of viral transcripts. Inherited mutations of the 5'-noncoding region of ferritin lead to loss of function in the iron-responsive element and are associated with hyperferritinemia and congenital cataracts40. The cataracts might be due to the abnormal circulating levels of ferritin. We suggest that translational regulation of transcripts other than CD36 in response to glucose might have a role in the pathophysiology of diabetes and atherosclerosis.

Methods
Tissue specimens and lesion-derived cells.
Vascular specimens were obtained during revascularization procedures at The New York Presbyterian Hospital/Weill Medical College at endarterectomy for carotid artery disease or coronary artery bypass. Lesion-derived cells (LDC) were cultured as described16.

Immunocytochemistry.
Paraffin sections of human carotid artery lesions were immunostained with FA6, an antibody against CD36 (1:100 dilution) by standard procedures for avidin-biotin enhanced immunoperoxidase detection (Vector, Burlingame, California). Positive immunostaining was detected with DAB and counterstained with hematoxylin. Sections from 11 of the patients were stained with FA6, and adjacent sections were stained with HAM56 (DAKO, Carpinteria, California) as a macrophage marker.

Cell preparation.
Human peripheral blood mononuclear cells (MNCs) were isolated in Ficoll&#8722;Paque, washed in PBS and resuspended in RPMI1640, containing penicillin/streptomycin 1%, and 10% heat-inactivated human AB serum. Plastic non-adherent cells were removed and cells incubated in media containing 100, 200 or 600 mg per 100 ml for 5 d at 37 °C in a humidified 5% CO2 environment. Cells were collected by gentle scraping. Viability was assessed by Trypan blue exclusion (> 90%) and monocyte purity determined by flow cytometry (> 80%).

Ox-LDL uptake.
Culture-derived human monocyte-macrophages were exposed to 125I-labeled Ox-LDL for 2 h at 37 °C (40 mg/ml) and CD36-dependent uptake measured by inhibiting Type I scavenger receptor-mediated uptake with fucoidan (50 mg/ml). Cells were washed to remove unbound ligand, solubilized (1 N NaOH for 30 min) and the radioactivity measured. Lesion-derived cells were incubated at 50 mg/dl overnight and over a range of glucose concentrations in media containing Ox-LDL 40 g/ml for 12 h. Oil Red O was used to visualize ingested lipid. Alternatively, lesion-derived phagocytic cells were incubated with 125I-labeled Ox-LDL and ingested radiolabeled lipid particles quantified after 12 h.

Fluorescence flow cytometric analysis
Cell aliquots (2 105) were exposed to a murine monoclonal antibody against CD 36 IgG, FA6 (5 g/ml) followed by PE-conjugated F(ab')2 preparation of goat anti-murine antibody (Tago, Burlingame, California). Free Fab sites were blocked with murine monoclonal antibody 19.1 and incubation performed with FITC-conjugated murine monoclonal antibody against CD14 IgG, UCHM-1 (Sigma). Cells (1 104) were analyzed by flow cytometry using appropriate gating.

RNAse protection assay.
Assays were performed using RPA II ribonuclease protection assay kit (Ambion, Austin, Texas). Total cellular RNA was obtained from cells (TRI REAGENT, Sigma) and hybridized for 16 h at 45 °C with antisense 32P-labeled RNA probes. CD36 RNA probe was generated from a 401-bp fragment corresponding to nucleotides 288&#8722;688 (PCR 2.1 vector, Invitrogen, San Diego, California). Following linearization, T7 polymerase was used to transcribe antisense probe. After hybridization samples were digested with a combination of RNAse A and T1 and protected fragments analyzed on a 5% native polyacrylamide electrophoresis gel. A human -actin probe protecting 173 bp was transcribed from pTRI-b-actin-human antisense control template (Ambion) digested with HaeIII and used as a control for these studies. Autoradiograms of dried gels were assessed by densitometry (NIH Image).

Preparation of polysomal RNA and RT-PCR analysis.
Polysomal RNA was prepared as described41 with slight modifications. For studies on monocyte-derived macrophages approximately 1 10 5 cells and for studies on lesion-derived cells, 1 106 cells were washed in PBS and resuspended in 0.75 ml of low-salt buffer (LSB, 20 mM Tris-HCl, ph 7.4, containing 10 mM NaCl and 3 mM MgCl2) prior to homogenization in LSB containing 1.2% (v/v) Triton N-101 and 0.2 M sucrose. After centrifugation, lysate was applied to a 0.5&#8722;1.5 M linear sucrose gradient (10 ml in LSB), prepared by GM-20 gradient maker (CBS Scientific, Del Mar, California) in a 14 39 mm polyallomer centrifuge tube (Beckman Instruments, Palo Alto, California) and centrifuged in a SW 40 Ti rotor at 36,000g for 110 min at 4 °C. Gradients were analyzed using a Beckman spectrophotometer and fractionated. A 311-bp fragment of CD36 cDNA was amplified from RNA derived from fractions using primers corresponding to CD36 (GAGGACTGCAGTGTAGGACTT and CAGTTCCGGTCACAGCCCAT) and a GeneAmp RNA PCR Kit (Perkin Elmer, Foster City, California) according to the manufacturer' s protocol. Other primer pairs used: -actin (230-bp product), 5'-GTGGTGGTGAAGCTGTAGCC-3' and 5'-GAGACCTTCAACACCCC-3'; CD14 (100-bp product), 5'-TCACAAGTGTGAAGCCTG-3' and 5'-CTCCATGGTCGATAAG-3'; glyceraldehyde phosphate dehydrogenase (GAP) (350-bp product), 5'-CGGTGTCAACGGATTTGG-3' and 5'-GGAGATGATGACCCTTTTGG-3'; 18S rRNA (150-bp product), 5'-GTAACCCGTTGAACCCCATT-3' and 5'-CCATCCAATCGGTAGTAGCG-3'; 28S rRNA (100-bp product), 5'-TTGAAAATCCGGGGGAGAG-3' and 5'-ACATTGTTCCAACATGCCAG-3'. Ribosomal RNA products were amplified for 28 cycles.

Reporter studies
Reporter expression in response to glucose concentration was measured in THP-1 cells following transfection with a luciferase reporter construct bearing the 5'-UTR of CD36 (5'-UTR-luc) or a reporter control (pGL2). Transfection was performed using DEAE-dextran. After 48 h incubation in media containing glucose in concentrations ranging from 50&#8722;600 mg/dl, cells were collected and cell lysates assayed for luciferase expression (20/20 luminometer, Turner Designs, Sunnyvale, California). Disinhibition of reporter expression over a range of glucose concentrations is calculated as 1/(1 &#8722; 5'-UTRluc/luc) and is the mean of triplicate values s.e.m.

5' leader constructs.
SEAP reporter constructs were designed in which the AUG codons of uORFs 1 and 3 were mutated or the sequence of the transcripts altered such that open reading frames originating from each of the three upstream AUG codons extended beyond the start site for the reporter. Mutagenesis was performed as previously described42 and constructs cloned in the vector. Orientation and sequence of clones were confirmed by sequence analysis. Transfection in C32 cells was performed using calcium phosphate with a luciferase plasmid (pGL2) cotransfected as a control for transfection efficiency. Luciferase activity and SEAP expression were assayed according to manufacturers protocol (Clonetech, Palo Alto, CA). To ensure that measured reporter expression was a function of translational efficiency, steady state mRNA levels were assessed by semi-quantitative RT-PCR (25 cycles) using primers specific for the SEAP reporter (5'-GGAGTGGTAACCACCACACGAG-3' and 5'-CAGTGACGAGGCTCAGCGTGTC-3').

REFERENCES
Semenkovich, C.F. & Heinecke, J.W. Perspectives in diabetes: the mystery of diabetes and atherosclerosis. Diabetes 46, 327–334 (1997).
Schmidt, A. et al. RAGE: A novel cellular receptor for advanced glycation end products. Diabetes 45, 77–80 (1996).
Vlassara, H. Recent progress in advanced glycation end products and diabetic complications. Diabetes 46, 19–24 (1997).
Berliner, J.A. Atherosclerosis: Basic mechanisms. Circulation 91, 2488–2496 (1995).
Boring, L., Gosling, J., Cleary, M. & Charo, I.F. Decreased lesion formation in CCR2-/- mice reveals a role for chemokines in the initiation of atherosclerosis. Nature 394, 894–897 (1998).
Witzium, J.L. & Steinberg, D. Role of oxidized low density lipoprotein in atherogenesis. J. Clin. Invest. 88, 1785–1792 (1991).
Nagy, L., Tontonoz, P., Alvarez, J., Chen, H. & Evans, R. Oxidized LDL regulates macrophage gene expression through ligand activation of PPAR. Cell 93, 229–240 (1998).
Tontonoz, P., Nagy, L., Alvarez, J., Thomazy, V. & Evans, R. PPAR promotes monocyte/macrophage differentiation and uptake of oxidized LDL. Cell 93, 241–252 (1998).
Endemann, G. et al. CD36 is a receptor for oxidized low density lipoprotein. J. Biol. Chem. 268, 11811–11816 (1993).
Huh, H., Pearce, S.F., Yesner, L.M., Schindler, J.L. & Silverstein, R.L. Regulated expression of CD36 during monocyte to macrophage differentiation: potential role of CD36 in foam cell formation. Blood 87, 2020–2028 (1996).
Nozaki, S. Reduced uptake of oxidized low density lipoproteins in monocyte-derived macrophages from CD36-deficient subjects. J. Clin. Invest. 96, 1859–1865 (1995).
Podrez, E. et al. Macrophage scavenger receptor CD36 is the major recptor for LDL modified by monocyte-generated reactive nitrogen species. J. Clin. Invest. 105, 1095–1108 (2000).
Febbraio, M. et al. Targeted disruption of the class B scavenger receptor, CD36, protects against atherosclerotic lesion development in mice. J. Clin. Invest. 105, 1049–1056 (2000).
Spiegelman, A. PPAR in monocytes: less pain, any gain? Cell 93, 153–155 (1998).
Greenwalt, D.E., Scheck, S.H. & Rhinehart-Jones, T. Heart CD36 expression is increased in murine models of diabetes and in mice fed a high fat diet. J. Clin. Invest. 96, 1382–1388 (1995).
McCaffrey, T. et al. Decreased Type II/Type I TGF-1 receptor ratio in cells derived from human atherosclerotic lesions: Conversion from an antiproliferative to profibrotic response to TGF-1. J. Clin. Invest. 96, 2667–2675 (1995).
Kolm-Litty, V., Sauer, U., Nerlich, A., Lehmann, R. & Scheicher, E.D. High glucose-induced transforming growth factor b-1 production is mediated by the hexosamine pathway in porcine glomerular mesangial cells. J. Clin. Invest. 101, 160–169 (1998).
Ashe, M., DeLong, S. & Sachs, A. Glucose depletion rapidly inhibits translation initiation in yeast. Mol. Biol. Cell 11, 833–848 (2000).
Skelly, R.H., Schuppin, G.T., Ishihara, H., Oka, Y. & Rhodes, C.J. Glucose-regulated translational control of proinsulin biosynthesis with that of the proinsulin endopeptidases PC2 and PC3 in the insulin-producing MIN6 cell line. Diabetes 45, 37–43 (1996).
Iynedjian, P.B. et al. Differential expression and regulation of the glucokinase gene in liver and islets of Langerhans. Proc. Natl. Acad. Sci. USA 86, 7838–7842 (1989).
Semenkovich, C.F., Coleman, T. & Fiedorek, F.T.J. Human fatty acid synthase mRNA: tissue distribution, genetic mapping, and kinetics of decay after glucose deprivation. J. Lipid Res. 36, 1507–1520 (1995).
Guest, P.C., Bailyes, E.M., Rutherford, N.G. & Hutton, J.C. Insulin secretory granule biogenesis. Co-ordinate regulation of the biosynthesis of the majority of constituent proteins. Biochem J. 274, 73–78 (1991).
Hinnebusch, A.G. in Translational Control (eds. Hershey, J.W.B., Mathews, M.B. & Sonenberg, N.) 199–244 (Cold Spring Harbor Laboratory Press, Cold Spring Harbor, 1996).
Gilligan, M. et al. Glucose stimulates the activity of the guanine nucleotide-exchange factor eIF-2B in isolated rat islets of Langerhans. J. Biol. Chem. 271, 2121–2125 (1996).
Morris, D. & Geballe, A. Upstream open reading frames as regulators of mRNA translation. Mol. Cell. Biol. 20, 8635–8642 (2000).
Abastado, J., Miller, P., Jackson, B. & Hinnebusch, A. Suppression of ribosomal reinitiation at upstream open reading frames in amino acid-starved cells forms the basis for GCN4 translational control. Mol. Cell. Biol. 11, 486–496 (1991).
Laakso, M. & Muusisto, J. Epidemiological evidence for the association of hyperglycemia and atherosclerotic vascular disease in non-insulin-dependent diabetes mellitus. Ann. Med. 28, 415–418 (1996).
Schmidt, A., Yan, S., Wautier, J. & Stern, D. Activation of receptor for advanced glycation end products: a mechanism for chronic vascular dysfunction in diabetic vasculopathy and atherosclerosis. Circ. Res. 84, 489–497 (1999).
Ohgami, N. et al. CD36, a member of the class B scavenger receptor family, as a receptor for advanced glycation endproducts. J. Biol. Chem. 276, 3195–3202 (2001).
Hunt, J. V., Bottoms, M. A., Clare, K., Skamarauskas, I. T. & Mitchinson, M. Glucose oxidation and low-density lipoprotein-induced macrophage ceroid accumulation: possible implications for diabetic atherosclerosis. Biochem. J. 300, 243–249 (1994).
Aitman, T. et al. Identification of CD36 (FAT) as an insulin-resistance gene causing defective fatty acid and glucose metabolism in hypertensive rats. Nature Genet 21, 76–83 (1999).
Febbraio, M. et al. A null mutation in murine CD36 reveals an important role in fatty acid and lipoprotein metabolism. J. Biol. Chem. 274, 19055–19062 (1999).
Barak, Y. et al. PPAR is required for placental, cardiac and adipose tissue development. Mol. Cell 4, 585–595 (1999).
Barroso, I. et al. Dominant negative mutations in human PPAR associated with sever insulin resistance, diabetes mellitus and hypertension. Nature 402, 880–883 (1999).
El-Jack, A., Hamm, J., Pilch, P. & Farmer, S. Reconstitution of insulin-sensitive glucose transport in fibroblasts requires expression of both PPAR and C/EBPalpha. J. Biol. Chem. 274, 7946–7951 (1999).
Le, S.-Y. & Maizel, J.V.J. A common RNA structural motif involved in the internal initiation of translation of cellular mRNAs. Nucleic Acids Res. 25, 362–369 (1997).
Koromilas, A. E., Lasaris-Karatzas, A. & Sonenberg, N. mRNAs containing extensive secondary structure in their 5' non-coding region translate efficiently in cells overexpressing initiation factor eIF-4E. EMBO J. 4153–4158 (1992).
Kozak, M. Regulation of translation in eukaryotic systems. Annu. Rev. Cell Biol. 8, 197–225 (1996).
Davuluri, R., Suzuki, Y., Sugano, S. & Zhang, M.Q. CART Classification of HNuman 5' UTR Sequences. Genome Research 10, 1807–1816 (2000).
Beaumont, C. et al. Mutation in the iron responsive element of the L ferritin mRNA in a family with dominant hyperferritinaemia and cataract. Nature Genet. 11, 444–446 (1995).
Mach, M., White, M.W., Neubauer, M., Degen, J.L. & Morris, D.R. Isolation of a cDNA clone encoding S-adenosylmethionine decarboxylase: expression of the gene in mitogen-activated lymphocytes. J. Biol. Chem. 261, 11697–11703 (1986).
Ho, S.N., Hunt, H.D., Horton, R.M., Pullen, J.K. & Pease, L.R. Site-directed mutagenesis by overlap extension using the polymerase chain reaction. Gene 77, 51–59 (1989).
Zuker, M., Mathews, D. & Turner, D. in RNA Biochemistry and Biotechnology (ed. Clark, J.B.) 11–43 (Kluwer Academic, Dordrecht, the Netherlands, 1999).
Mathews, D., Sabina, J., Zuker, M. & Turner, D. Expanded sequence dependence of thermodynamic parameters improves prediction of RNA secondary structure. J. Mol. Biol. 288, 911–940 (1999).

Acknowledgments
We thank J. Han for the gift of labeled Ox-LDL; F. Pearce for advice on the Ox-LDL studies; A. Cerutti for help with some of the flow cytometry; and M.J.R. Echevarria, D. Falcone and R. Nachman for their critical review of the manuscript. This research was supported by grants from the NIH (DK48698) and American Heart Association (to A.S.A.) and from Universita degli Studi di Parma, Parma, Italy (to A.R.).