Increased Expression of Genes for Platelet-Derived Growth Factor in Circulating Mononuclear Cells of Hypercholesterolemic Patients

Methods

Quantitative RT-PCR

Quantitation of PDGF mRNA by RT-PCR was essentially according to Wang et al.¹⁵ Reverse transcription was performed with 1 μg total cellular RNA, 5×10³ to 2×10⁶ copies AW109 cRNA as internal standard (Perkin-Elmer), 50 mmol/L Tris-HCl at pH 8.3, 75 mmol/L KCl, 3 mmol/L MgCl₂, 10 mmol/L DTT, 10 U RNasin (Promega), 5 μmol/L random hexanucleotide primers (Pharmacia), 0.125 mmol/L dNTPs, and 200 U Superscript Moloney murine leukemia virus reverse transcriptase (GIBCO-BRL), in a total volume of 20 μL at 37°C for 2 to 5 hours. Serial 1:3 dilutions of the cDNA products were amplified with 0.5 μmol/L PDGF-A or -B specific primers,¹⁵ 5 μC [α-³²P]dCTP, 0.025 mmol/L dNTPs, 10 mmol/L Tris-HCl at pH 9.0, 50 mmol/L KCl, 1.5 mmol/L MgCl₂, 0.1% Triton X-100, and 2.5 U Taq DNA polymerase (Promega), in a volume of 100 μL for 25 cycles (0.5 minute at 94°C, 2 minutes at 55°C, and 3 minutes at 72°C). PCR products were separated by electrophoresis on 3% Nusieve 3:1 agarose gels (FMC). Products of PDGF-A mRNA template (225 bp) and AW109 cRNA (301 bp) or PDGF-B mRNA (217 bp) and AW109 cRNA (300 bp) were cut from the gel, and ³²P was quantitated by Cerenkov counting. Linear double log plots of counts per minute against amounts of template were drawn¹⁵ and numbers of copies of PDGF-A or -B mRNA in total cellular RNA determined by equating with the number of molecules of AW109 cRNA that generated the same amount of radioactive PCR product (Fig 1⇓). Double log plots for AW109 cRNA and each mRNA gave parallel straight-line plots, indicating that the amplification efficiency was the same for both templates. Corrections were made for differing percent guanine+cytosine contents of each amplified template. Small variations in amounts of RNA template were corrected by independently measuring the poly(A) content of each sample by dot blot hybridization against oligo(dT)₁₈,¹⁶ and results were expressed as copies mRNA per nanogram total cellular RNA, assuming 0.02 ng poly(A)/ng total cellular RNA.¹⁶ This quantitative RT-PCR assay was found to be reproducible. Repeat amplifications of one cDNA sample gave mRNA concentrations differing by ±3.5%. Reassay of mononuclear PDGF-B mRNA from one individual on three occasions over 12 months gave values differing by ±9.5%.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Quantitative RT-PCR analysis of PDGF-A and -B mRNA levels in blood mononuclear cells of normal and hypercholesterolemic individuals. Variable template concentrations of internal standard AW109 RNA and mononuclear cell total RNA are plotted against radioactivity of their PCR products. A, PDGF-A mRNA and B, PDGF-B mRNA analyses for participant Normal 5. C, PDGF-A mRNA and D, PDGF-B mRNA analyses for patient FH6. In each case, values extrapolated from the graphs are corrected for differing percent guanine+cytosine content of templates and variations in polyA content of cellular RNA before presenting data in Table 2⇓ (see “Methods”).

Results

Concentrations of mRNA for PDGF-A and PDGF-B in PBMN were measured using the quantitative RT-PCR assay developed by Wang et al.¹⁵ This procedure depends on the inclusion in the RT-PCR reaction of known amounts of an internal RNA standard (pAW109 RNA) that can be amplified concurrently with the specific mRNA, using the same PCR primers but generating a product of a different size. The internal standard allows correction for variable amplification efficiency. In this system, PDGF-specific primers amplify regions of the cDNA from exon 3 to the exon 4/5 junction (PDGF-A) or between exon 3/4 and exon 4/5 (PDGF-B) and thus will detect all forms of mRNA transcript containing exon 4 of each PDGF gene, including the alternatively spliced forms of PDGF-A mRNA.^{17 18}

PDGF mRNA levels in normal and hypercholesterolemic individuals were compared. Hypercholesterolemic patients were classified as either FH, on the basis of clinical data and family history, or as HL, showing hypercholesterolemia and variable hypertriglyceridemia. Clinical details of patients analyzed are given in Table 1⇓, together with information on normal subjects, and corresponding blood lipid measurements are given in Table 2⇓. In addition to hypercholesterolemia, other positive risk factors for coronary heart disease (eg, hypertension, smoking) were associated with some individuals. Presence or absence of overt vascular and other diseases and medication is also indicated in Table 1⇓. Typical results of analyses of PDGF-A and -B mRNAs in one normal and one hypercholesterolemic individual are presented in Fig 1⇑. Low concentrations of PDGF-A and PDGF-B mRNA were present in PBMN of eight normocholesterolemic individuals (Table 2⇓). In contrast, hyperlipidemic patients showed an approximate 15-fold and FH patients an approximate 20-fold increase in amounts of both mRNA species (Table 2⇓). The FDB patient carrying a mutation in the region of apoB-100, which interacts with the LDL receptor,⁹ had PDGF mRNA levels similar to other members of the FH group. Preliminary data suggest that >90% of PDGF-A mRNA in these patients is the short form, lacking exon 6.^{17 18} Since it is known that PDGF gene expression is rapidly induced when monocytes adhere to plastic surfaces,¹³ no attempt was made to isolate monocytes from PBMN preparations for direct determination of their PDGF mRNA concentrations. However, when monocytes from both normocholesterolemic and hypercholesterolemic individuals were allowed to adhere to plastic, the PDGF-A and PDGF-B mRNA content of the nonadherent cells (lymphocytes) was undetectable, whereas mRNA concentrations for LDL receptor¹⁵ and HMG CoA reductase¹⁵ were similar to those of total PBMN (data not shown). This confirms that the majority of both PDGF-A and PDGF-B mRNA detected in PBMN of normocholesterolemic and hypercholesterolemic individuals are derived from monocytes. PDGF-A and -B mRNA species are not detectable in granulocytes.⁵

Although concentrations of PDGF-A and PDGF-B mRNA tended to be higher in the FH group than the hyperlipidemics (Table 2⇑), there was in fact a better correlation with plasma cholesterol levels than with this patient classification per se. Indeed, clear positive correlations were seen for both mRNA species with plasma total cholesterol (PDGF-A mRNA: r=.890, P<.001; and PDGF-B mRNA: r=.846, P<.001; Fig 2⇓) and with LDL cholesterol levels (PDGF-A mRNA: r=.888, P<.001; and PDGF-B mRNA: r=.876, P<.001; data not shown) when normocholesterolemic and hypercholesterolemic individuals were analyzed together. When hypercholesterolemic individuals were considered alone, significant positive correlations of mRNA concentrations with plasma total cholesterol (PDGF-A mRNA: r=.879, P<.001; and PDGF-B mRNA: r=.665, P<.05; data not shown) and with LDL cholesterol (PDGF-A mRNA: r=.874, P<.001; and PDGF-B mRNA: r=.728, P<.05; data not shown) were again apparent. Corresponding positive correlations for normocholesterolemic individuals failed to reach significance, although PDGF-A mRNA did show a weak positive correlation with the ratio of LDL cholesterol to HDL cholesterol (r=.746, P<.05). Stronger positive correlations were apparent between this ratio and PDGF-A and -B mRNA levels for hypercholesterolemic individuals alone (r=.772, P<.01 and r=.746, P<.01, respectively). All of the above correlations were also significant at the 1% or 5% level when FH patients were analyzed separately. Interestingly, PDGF-A and -B mRNA levels also demonstrated weak but significant negative correlations with HDL cholesterol concentrations, when all subjects were considered together (PDGF-A mRNA: r=−.472, P<.05; and PDGF-B mRNA: r=−.547, P<.05; data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Correlation between PDGF mRNA levels in blood mononuclear cells and plasma total cholesterol concentrations. Lines were drawn using a least-squares algorithm. A, PDGF-A mRNA: r=.890, P<.001. B, PDGF-B mRNA: r=.846, P<.001. □ represents normocholesterolemic individuals; ⋄, HL individuals; and ○, FH individuals.

Although the data in Fig 2⇑ demonstrate significant correlations of mononuclear cell PDGF mRNA levels with plasma total cholesterol, closer examination of Fig 2A⇑, in particular, suggests the presence of a threshold effect. Thus, whereas among normal individuals, there is a modest rise in PDGF-A mRNA as total cholesterol increases from 3.8 to 5.2 mmol/L, the lowest hypercholesterolemic patient (cholesterol, 6.0 mmol/L) has at least a 10-fold higher PDGF-A mRNA level. This implies that the greatly increased PDGF mRNA concentrations in mononuclear cells of hypercholesterolemic patients might at least in part be associated with an additional confounding factor other than cholesterol. This factor would be present in patients but not normal subjects and might explain the apparent discontinuity in the relationship between cholesterol and PDGF mRNA. We have therefore investigated further potential correlations of PDGF mRNA levels with other risk factors for coronary heart disease (hypertension, smoking), or with age, gender, overt vascular or other disease, plasma triglyceride concentration, and medication. Among hypercholesterolemic subjects, a negative correlation was apparent, which just reached significance at the 5% level, between age and both PDGF mRNA concentrations. However, this was heavily influenced by the very high mRNA levels in the young, severely hypercholesterolemic patient FH7, and if data from this individual were excluded, no significant correlation was observed. No significant correlation with age was apparent for PDGF mRNA levels in normal subjects. Similarly, PDGF mRNA levels showed no correlation with gender in normal or hypercholesterolemic subjects, although the small number of female subjects in this study precludes definitive conclusions. Plasma triglyceride concentration showed weak positive correlations with PDGF-A and PDGF-B mRNA levels in normal individuals (r=.773, P<.05; and r=.776, P<.05, respectively), but no correlation was evident among hypercholesterolemic subjects. None of the other factors mentioned above showed significant correlation with PDGF mRNA concentrations whether subjects were considered as one group or separately as normocholesterolemic and hypercholesterolemic groups. In particular, the two patients with evidence of cardiovascular disease (HL5 and FH7) and one who developed symptoms subsequent to sampling (FH6) exhibited PDGF mRNA concentrations similar to other, asymptomatic hypercholesterolemic patients. Although definitive conclusions on the association of smoking, hypertension, or vascular disease itself with PDGF mRNA levels would require a study of larger subject groups for each category, the present data suggest that plasma cholesterol is the only individual variable showing significant correlation with PDGF expression in the subjects examined here.

In both normocholesterolemic and hypercholesterolemic subjects, there was a strong positive correlation between the levels of PDGF-A and PDGF-B mRNA species present in blood mononuclear cells (Fig 3⇓). This was apparent whether all subjects were considered together (r=.901, P<.001) or normocholesterolemic and hypercholesterolemic subjects were analyzed separately (r=.796, P<.01; and r=.693, P<.01, respectively). This finding indicates that the two genes are coordinately regulated in mononuclear cells in both normal and hypercholesterolemic individuals.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Correlation between PDGF-A and PDGF-B mRNA concentrations in blood mononuclear cells. Lines were drawn using a least-squares algorithm. A, Results for normocholesterolemic and all hypercholesterolemic individuals together (r=.901, P<.001). B, Results for normocholesterolemic individuals (r=.796, P<.01). C, Results for hypercholesterolemic individuals (r=.693, P<.01). □ represents normocholesterolemic individuals; ⋄, HL individuals; and ○, FH individuals.

All three isoforms of PDGF can upregulate LDL receptor mRNA, and PDGF-BB can upregulate HMG CoA reductase mRNA in fibroblasts and smooth muscle cells.¹⁹ It is, therefore, interesting to note that in the mononuclear cells analyzed in the present experiments, there were strong negative correlations between mRNA levels for PDGF-A and -B on the one hand, and mRNA levels for LDL receptor and HMG CoA reductase (also assayed by the Wang procedure¹⁵ ) on the other, when data for all individuals were considered together (Table 3⇓). These reciprocal relationships are consistent with our findings that mononuclear cell mRNA levels for both HMG CoA reductase and LDL receptor negatively correlate with plasma total and LDL cholesterol (see Reference 20 and M.A.B. et al, 1995, unpublished observations), whereas PDGF mRNA levels positively correlate with these parameters (see above and Fig 2⇑).

Discussion

We have demonstrated 15- to 20-fold increases, relative to normal individuals, in the expression of PDGF-A and PDGF-B genes (reflected in mRNA levels) in circulating monocytes of HL and FH patients. These measurements of PDGF mRNA concentration were performed after isolation of mononuclear cells by density gradient centrifugation of blood, but with no further incubation. Although these are not strictly direct in vivo measurements, the very clear differences in PDGF mRNA observed between hypercholesterolemic and normocholesterolemic individuals argue strongly that the mononuclear cell isolation procedure has not distorted the data and that they reflect real differences in vivo.

Northern blotting techniques could not detect PDGF-A or PDGF-B mRNA in freshly isolated, unstimulated monocytes,^{4 5} but PDGF-B mRNA was detectable after adherence of monocytes to a plastic surface.¹³ Bacterial lipopolysaccharide, interferon-γ, and phorbol esters have been variously shown to stimulate PDGF-A and/or PDGF-B mRNAs in monocyte-derived macrophages and in the monocytic cell line U937.^{1 15 21 22 23} More recently, sensitive RT-PCR assays detected low levels of both PDGF-A and PDGF-B mRNA in circulating monocytes, but not granulocytes, of normal individuals.⁵

The increases in PDGF mRNA levels observed in the present work could be due to an increase in transcription, a decrease in mRNA degradation, or a combination of both. Many mitogens, cytokines, and other factors, including transforming growth factor-β, tumor necrosis factor-α, interleukin-1, and shear stress, can stimulate PDGF gene expression,^{1 22 24 25} whereas others (eg, ω3 fatty acids) decrease expression.^{1 5 22 26} Although some of these factors modulate PDGF mRNA stability,²⁶ most agents directly affect PDGF gene transcription.^{1 22 24 25} Whereas the two PDGF genes are independently regulated, some agents have similar effects on both genes.^{1 22} We have clear evidence in the present work of coordinate regulation of PDGF-A and PDGF-B genes in mononuclear cells of normal and hypercholesterolemic patients.

Although we do not know the primary stimulus for PDGF expression in these patients, the simplest explanation would be a direct positive upregulation by cholesterol or a metabolite thereof. The data presented here are consistent with such a direct effect, particularly through LDL cholesterol, eg, positive correlations of PDGF-A and -B mRNA with total and LDL cholesterol (which are significant whether all subjects, hypercholesterolemic, or FH patients alone are considered) and negative correlations with LDL receptor and HMG CoA reductase mRNAs and with HDL cholesterol. Indeed, there is some preliminary evidence that cholesterol oxides (25-hydroxycholesterol, 7-ketocholesterol) can increase PDGF mRNA levels in cultured rabbit aortic smooth muscle cells.²⁷ However, it is equally possible that some other factor is independently responsible for both the hypercholesterolemia and the raised PDGF mRNAs or that hypercholesterolemia initiates alterations in cytokine signaling between endothelium, monocytes, lymphocytes, or other cells, which cause induction of PDGF expression in monocytes. The potential for involvement of cytokines is illustrated by the recent demonstration that LDL cholesterol levels in children correlate with the number of cells in different T lymphocyte subsets.²⁸

Furthermore, the correlations of cholesterol concentration with PDGF-A mRNA show a sudden sharp increase in mRNA level between normal and hypercholesterolemic individuals, indicative of a threshold effect (Fig 2A⇑). In the case of PDGF-B mRNA, however, the transition is more gradual. In view of the evidence for the involvement of PDGF in atherogenesis,³ one might presume that the increased PDGF mRNA levels in hypercholesterolemic patients reflect the degree of vascular damage in each individual and when added to the cholesterol correlation might produce the apparent threshold effect. However, many factors can influence the atherogenic process, positively or negatively, including LDL and HDL cholesterol, age, duration of hypercholesterolemia, medication (especially lipid-lowering drugs such as statins), smoking, hypertension, and others.³ Only two patients in the current study had established vascular disease, although one asymptomatic patient developed symptoms a year after sampling. Nevertheless, it is quite possible that active or early forms of atherogenesis were present in other individuals yet were not extensive enough to cause overt symptoms. Thus, the increased PDGF expression might reflect vascular pathology and any other factors contributing to it. However, in the present work, the only individually identifiable factor that showed significant correlations with PDGF mRNA levels in hypercholesterolemic subjects was cholesterol itself.

Previous data on PDGF protein concentrations in plasma or serum from hypercholesterolemic patients are limited. One study measured PDGF in plasma indirectly and found a twofold increase in hypercholesterolemic patients with proven coronary atherosclerosis.²⁹ In a similar study, plasma PDGF was found to correlate with the number and severity of distinct stenoses in atherosclerotic patients.³⁰ A further study found a twofold increase in PDGF in serum of hypercholesterolemic patients, although there was no significant correlation between serum PDGF and total cholesterol in individuals.³¹ Whereas the majority of PDGF in serum is believed to reflect release from platelets,^{1 2 30 31} the much lower concentrations in plasma may be derived from platelets, monocytes/macrophages, or endothelial cells. There have been reports of increases in plasma PDGF levels in chronic arterial obstructive disease³² and in the coronary circulation, but not the aortic root, in patients with unstable angina.³³ However, these observations are complicated by uncertainty as to the source of the PDGF and by suggestions that plasma levels are affected by changes in consumption of PDGF by damaged or repairing tissues.^{32 33}

Little is known about the regulation of PDGF mRNA translation, postranslational processing, or secretion of the mature PDGF protein.¹ However, the parallel effects of dietary fish oil supplementation on concentrations of monocyte PDGF mRNAs and on plasma PDGF protein levels suggest that monocyte mRNA levels may well reflect PDGF protein production.^{5 34} The half-life of PDGF in the circulation is no more than 2 minutes,^{1 2} indicating rapid uptake into target tissues or binding to other plasma proteins, such as α₂-macroglobulin, or to proteoglycans on the endothelium.³⁵ Our preliminary analysis of the different spliced variants of PDGF-A mRNA suggests that monocytes of hypercholesterolemic patients would predominantly produce the short form of PDGF-A protein, lacking the exon-6 encoded retention motif, which mediates binding to proteoglycans.³⁵ However, since mature PDGF-B contains a constitutive retention motif, it seems likely that PDGF-AB and -BB will, at least, bind to proteoglycans of the extracellular matrix. In this context, estimates of the rate of PDGF production from mRNA concentrations in cells may well be more informative than monitoring steady state levels of the protein in the circulation. Furthermore, whether or not the elevated PDGF mRNA levels that we have observed are reflected in increased PDGF protein in the circulation, they serve as a striking marker for monocyte activation in these patients.

Increased expression of both PDGF-A and PDGF-B genes is detected in human atherosclerotic carotid artery.^{3 4 15} Immunochemical and in situ hybridization analyses indicate that smooth muscle cells are the main site of PDGF-A mRNA, and macrophages, which contain little or no lipid, are the major site of PDGF-B mRNA in atherosclerotic lesions.^{3 4} Lipid-laden foam cells appear to contain little PDGF-B, consistent with the observation that oxidized LDL inhibits PDGF-B gene expression in monocyte-derived macrophages.^{4 21} The consensus view is therefore emerging that during development of atherosclerotic lesions, differentiating macrophages release PDGF-B, which acts as a chemoattractant to smooth muscle cells migrating into the intima.³ Cytokines, such as interleukin-1, tumor necrosis factor-α, and transforming growth factor-β, released from activated macrophages induce PDGF-B gene expression in endothelial cells and PDGF-A gene expression in smooth muscle cells. Both PDGF-AA and PDGF-BB could then stimulate smooth muscle cell proliferation in the intima.³ Furthermore, in both humans and animal models, PDGF is required for smooth muscle cell accumulation during restenosis after arterial balloon angioplasty.^{36 37} Thus, PDGF is seen to have a pivotal role in the fibroproliferative process of atherosclerosis.³

It has been claimed, on the basis of experimentally induced atherosclerotic lesions in animals, that continuous recirculation of lipid-laden macrophages between lesions and bloodstream occurs.^{3 4} If this is so, our results might reflect recirculation of activated monocytes/macrophages prior to lipid loading, since this process is thought to downregulate PDGF expression. However, current evidence suggests that macrophages in atherosclerotic lesions contain predominantly PDGF-B mRNA, whereas our mononuclear cells contain similar concentrations of both PDGF-A and -B mRNAs. Whatever the cause of the increased PDGF expression in circulating mononuclear cells of hypercholesterolemic individuals, it seems likely that it represents some form of monocyte activation that may possibly be associated with early stages in the generation of lesions in the vessel wall/vasculature. It will clearly be important to determine whether elevated PDGF expression in circulating monocytes is associated with other positive risk factors for atherogenesis (eg, hypertension, smoking), especially in the absence of hypercholesterolemia, and to establish whether there is a clear relationship between monocyte PDGF expression and stages of atherosclerosis in larger groups of patients with differing severity of vascular disease. In the present studies, plasma cholesterol was the only individual variable correlating significantly with monocyte PDGF mRNA. It seems likely that elevated monocyte PDGF expression is a significant component of the mechanism that leads to increased atherosclerotic risk in hypercholesterolemia.

Selected Abbreviations and Acronyms