|Home | About | Journals | Submit | Contact Us | Français|
We previously demonstrated that vascular injury-induced neointima formation is exaggerated in human C-reactive protein transgenic (CRPtg) compared to non-transgenic (NTG) mice. We now test the hypothesis that complement is required for this effect.
CRPtg and NTG with a normal complement system versus their counterparts lacking expression of complement protein C3 (CRPtg/C3-/- and NTG/C3-/-) underwent carotid artery ligation. 28 days later, the injured vessels in CRPtg had thicker neointimas and more immunoreactive C3 in the surrounding adventitia compared to NTG. In CRPtg/C3-/- there was no increase in neointimal thickness compared to NTG or NTG/C3-/-. Decreasing human CRP blood levels (via administration of a selective antisense oligonucleotide) eliminated the depletion of serum C3 associated with vascular injury and reduced immunoreactive C3 in the resultant lesions. In injured vessels C3 co-localized with F4/80 (macrophage marker) and in-vitro, human CRP elicited increased expression of C3 by bone-marrow derived macrophages.
Human CRP exaggeration of neointima formation in injured mouse carotid arteries associates with decreased circulating C3 and increased tissue-localized C3. C3 elimination or pharmacological reduction of human CRP prevents CRP-driven exacerbation of the injury response. In the CRPtg model system, mouse C3 is essential for the effect of human CRP.
We recently demonstrated that C-reactive protein (CRP) exacerbates the neointimal response to acute vascular injury in a human CRP transgenic mouse (CRPtg) model.1, 2 Thus when ovariectomized (OVX) mice were subjected to ligation of the common carotid artery, CRPtg had twofold greater neointima formation than non-transgenic (NTG) controls, as well as extensive deposition of human CRP in the injured vessels. Human CRP can bind to the immunoglobulin G Fc receptors FcγRI (CD64), FcγRII (CD32) and FcγRIII (CD16) in both mice3 and humans4, and we showed that in the absence of FcγRI (but not in the absence of FcγRIIb or FcγRIII) the deleterious effect of CRP on the neointimal response to injury is greatly attenuated. We provided additional evidence that in this model of vascular injury the action of CRP is likely propagated by resident F4/80+ macrophages.1 In those experiments we used CRPtg because in that model human CRP expression closely resembles the pattern seen in humans, and because the protein is expressed endogenously during the course of experimentally induced disease.5 The CRPtg model has been used by us and others to demonstrate that CRP promotes vascular thrombosis6, induces endothelial dysfunction7, and accelerates atherosclerosis8, though not all reports by us and others show that human CRP is detrimental to the vasculature of CRPtg.9, 10
Human CRP can activate human11 and mouse complement12, 13, and CRP and complement are known to co-localize in atherosclerotic lesions14, suggesting that in the context of vascular injury some of the actions of CRP might be mediated by local activation of complement.15 In this study, we demonstrate that in CRPtg mice the enhancing effect of CRP on the vascular injury response requires the presence of a fully functional complement system.
CRPtg mice are fully described elsewhere.5 Importantly, human CRP is present in the blood of CRPtg at concentrations relevant to human health and disease (i.e., levels at or below 3μg/ml under steady-state conditions and levels up to 500μg/ml during an acute phase response). In contrast, endogenous mouse CRP is expressed at only ~2μg/ml and does not act as an acute phase reactant.16-18 Mice carrying a null mutation for complement component 3 (C3-/-)19, 20 have been described elsewhere; C3-/- produce no C3 due to targeted disruption of the murine C3 gene promoter and thus they lack the ability to generate C3 and C5 convertases. Double mutant mice (CRPtg/C3-/-) were generated by breeding CRPtg with NTG/C3-/-.21 We showed previously21 that serum human CRP levels in CRPtg/C3-/- were not different from those in CRPtg and that the human transgene was fully responsive to inflammatory stimuli. All mouse strains used in this study were backcrossed at least 10 generations onto the C57BL/6 background, and C3 deficient versus sufficient and CRPtg versus NTG progeny were obtained in the expected Mendelian ratios. Each of the genotypes appeared phenotypically normal, and no remarkable differences in lifespan have been noted. When needed, some C57BL/6 wildtype mice were obtained from The Jackson Laboratory (Bar Harbor, Me).
All mice were fed a standard mouse pellet diet (Ralston Purina) and maintained at constant humidity (60 ± 5%) and temperature (24 ± 1°C) with a 12 hour light cycle (6 AM to 6 PM). All protocols were approved by the Institutional Animal Care and Use Committee at the University of Alabama at Birmingham and were consistent with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH publication 96-01, revised 1996).
Female mice were anesthetized with ketamine (80 mg/kg IP; Abbott Laboratories) and xylazine (5 mg/kg IP; Rompun, Bayer Corp) and subjected to OVX. Five days later, mice were subjected to right common carotid artery ligation1, 2, 22 as detailed in the Appendix. Twenty-eight days after ligation injury, mice were anesthetized and euthanized and the carotid arteries were fixed, sectioned, stained and subjected to computer assisted morphometric analysis to assess the extent of the injury response (see Supplement Material).1, 2 For immunofluorescence microscopy (see Supplement Material) representative arteries were harvested either 24 hrs or 14 days after vascular injury and processed with DAPI stain to detect nuclei and with antibodies to detect human CRP, mouse complement C3 and C5, and the mouse macrophage marker F4/80.
To specifically lower the level of human CRP in the blood of CRPtg, we used a human CRP specific antisense oligonucleotide (ASO) (ISIS Pharmaceuticals Inc.) that targets human CRP mRNA.23 CRPtg mice (23.63 ± 2.96 μg human CRP and 756 ± 158 μg mouse C3 per ml serum at baseline) were randomly separated into two groups and one group was injected with the CRP-specific ASO (200μl; 50mg/kg) twice weekly (i.p.) for 2 weeks prior to carotid artery ligation and for 2 weeks after ligation. Control animals in the second group were injected with a non-specific ASO comprised of a scrambled oligonucleotide sequence that is not complementary to any known mRNA sequence. The CRP-specific ASO has been demonstrated to lower human CRP in CRPtg mice selectively with no effect on mouse CRP (ref. 23 and data not shown).
Bone marrow was harvested from the long bones of CRPtg and NTG mice. The cells were grown in culture for one week in the presence of RPMI 1640 (Invitrogen) with 10% fetal bovine serum and 30% L-cell conditioned medium as a source of M-CSF. After 7 days, the cells were >90% positive for the F4/80 macrophage marker as judged by flow cytometry. These bone marrow macrophages were made quiescent by culturing them in the absence of fetal bovine serum for 24 hours. Quiescent bone marrow macrophages were then treated with RPMI media or human CRP (50μg/ml) for 4 hours. The human CRP that we used for these experiments was from United States Biological (product C7907-26A) and is purified to >95% by the manufacturer. We verified this level of purity on overloaded SDS-PAGE (single band at 21 kDa). The CRP was never exposed to sodium azide, and by Limulus amebocyte lysate assay (Chromo-LAL, Associates of Cape Cod Incorporated) contained less than 0.01 endotoxin units/ml.
ELISA was used to measure serum mouse C3 and human CRP as previously described (see Supplement Material).1 Real-time quantitative RT-PCR was used to assess local expression of C3 and C5 mRNA by the injured arteries and by cultured bone marrow-derived macrophages under resting conditions and following CRP treatment, as described in detail in the Supplement Material.
Results of morphometric analyses, RT-PCR, and ELISA are expressed as the mean ± SEM without transformation. All statistical analyses were performed using SPSS® version 11.5 for Windows® (SPSS® Inc., Chicago, IL). Comparisons among experimental groups were performed with one-way ANOVA followed by pair-wise multiple comparisons using the least significant difference test. Differences were considered significant when the associated p value was <0.05.
The architecture of carotid arteries was examined 28 days after ligation injury. In animals with a normal complement system (i.e., in C3 sufficient animals) the neointima was thicker in mice that expressed the human CRP transgene (CRPtg) than in NTG controls (Fig.1A; compare the upper two panels). This is in concert with our earlier reports that endogenous expression of human CRP exacerbates the neointimal response to acute vascular injury in CRPtg.1,2 In contrast, in animals whose complement system was not fully intact (i.e., in C3 deficient animals) there was no readily observable difference in neointimal thickness for CRPtg/C3-/- compared to NTG/C3-/- (Fig. 1A; compare lower panels). Quantitative morphometric analysis confirmed this observation; in C3 sufficient mice neointimal area was more than twofold greater in CRPtg than NTG (p<0.05) (Fig. 1B, left pair of columns) but in the absence of C3 expression no significant difference in neointimal area was seen in CRPtg/C3-/- compared to NTG/C3-/- (Fig. 1B, right pair of columns). Importantly, the neointimal response of CRPtg mice that did not express mouse C3 was more than 50% reduced compared to CRPtg mice that expressed C3 (p<0.05; Fig. 1B) and in the absence of human CRP expression the elimination of mouse C3 had no measurable effect. In contrast to the neointima, the area of the media did not differ among the four genotypes (3.88 ± 0.24, 3.51 ± 0.34, 3.01 ± 0.53, and 3.32 ± 0.81 × 104 μm2 for CRPtg, NTG, CRPtg/C3-/-, and NTG/C3-/-, respectively). These data establish that human CRP exacerbates the neointimal response to vascular injury in CRPtg only if the endogenous complement system is intact.
We performed immunofluorescence staining of cross-sections of carotid arteries from C3 sufficient mice, harvested 24hrs versus 14 days after ligation injury. At 24hrs, immunoreactive C3 and C5 protein were detectable in the periadventitial regions of the injured vessels, but there was no discernable difference in the amount of C3 or C5 in CRPtg versus NTG (data not shown). However on day 14 after injury, staining for immunoreactive C3 was substantially more robust in ligated vessels from CRPtg than NTG (Fig. 2, p<0.05). Most of the immunoreactive C3 was in the periadventia. In both CRPtg and NTG mice, comparatively low levels of C5 were detected in the neointima on day 14 post injury, and there was no significant difference in C5 immunostaining between the genotypes (Fig. 2). In CRPtg the periadventitial immunoreactive C3 co-localized with F4/80+ macrophages (Fig. 3, A to E), and in CRPtg human CRP was evident in the lesions whether mice were C3 sufficient (Fig 3G) or C3 deficient (Fig. 3I). These results indicate that within two weeks after acute vascular injury, periadventitial expression and/or deposition of C3 is enhanced by the presence of endogenously expressed human CRP.
Since in our CRPtg model C3 is required for the human CRP-mediated exacerbation of the vascular injury response following arterial ligation (Fig. 1), and since this effect is associated with an increase in the amount of immunoreactive C3 in the periadventitia surrounding the vascular lesion (Fig. 2), we sought to determine whether the C3 identified in the injured vessel is derived from increased local synthesis of C3, increased local deposition of circulating C3, or both. To assess local expression of complement in the ligated artery we measured C3 and C5 mRNA levels in injured arteries harvested 24hrs after ligation, and compared these to levels in control uninjured arteries. Consistent with the immunostaining results obtained on arteries collected 24hrs after injury (see above), we found no difference in the expression of C3 or C5 mRNA in CRPtg versus NTG at 24hrs or at 14 days after injury. Neither C3 nor C5 mRNA level was increased 24hrs following injury in either genotype, but expression of C3 was increased in both genotypes at 14days after injury (Supplement Material Fig. I).
We had previously shown that CRP-induced exacerbation of the neointimal response to ligation injury depends on FcγRI, and that in vessels harvested 28 days after injury CRP colocalized with F4/80+ cells recruited to the injured vessel.1 In the current study we found that both mouse C3 (Fig. 3E) and human CRP (Fig. 4A, large arrowhead) colocalized with the macrophage marker F4/80 in the artery wall. Therefore, we hypothesized that human CRP might drive complement expression and/or its activation (and deposition) in the ligated carotid artery by stimulating macrophages resident in or recruited to the periadventitial region surrounding the site of injury. To test this hypothesis we utilized a surrogate in vitro assay, wherein we cultured bone marrow-derived macrophages from CRPtg and NTG mice and stimulated them with human CRP. We found that both C3 and C5 mRNA was expressed by the cells, and that expression was approximately equal in both genotypes (Fig. 4, C and D). In concert with our immunostaining findings, C3 mRNA was upregulated in response to stimulation with human CRP in both genotypes (Fig. 4C). Also in concert with our immunostaining, there was little effect of CRP exposure on C5 mRNA expression (Fig. 4D).
The data described above suggest that macrophages resident in the periadventitial region surrounding the injured carotid artery respond to human CRP by expressing C3 (but not C5). Indirect evidence in support of this conclusion was obtained by treating CRPtg with an antisense oligonucleotide (ASO) that specifically targets human CRP mRNA, thereby depleting the blood of the human protein. For the CRPtg treated with a non-specific ASO (Fig. 5A; 19.3 ± 9.4 μg human CRP and 564 ± 68 μg mouse C3 per ml serum at baseline) and subjected to carotid artery ligation, serum levels of human CRP rose and levels of mouse C3 fell, consistent with C3 activation. In contrast, for the CRPtg treated with a human CRP-specific ASO (Fig. 5B; 27.3 ± 3.1 μg human CRP and 825 ± 213 μg mouse C3 per ml serum at baseline) there was selective lowering of human CRP and no reduction in circulating C3 in the blood. Further, there was a dramatic reduction in the intensity of mouse C3 immunoreactivity in the periadventitial region of the injured arteries in the mice treated with the human CRP-specific ASO (Fig. 6). The intensity of C3 staining in injured blood vessels from CRPtg receiving the CRP-specific ASO (Fig. 6) was comparable to that seen in untreated NTG (Fig. 2). The combined results suggest that human CRP both stimulates local C3 expression and activates mouse complement after carotid artery ligation in the CRPtg mouse.
We confirmed here our prior finding that in CRPtg, the neointima that develops in response to vascular ligation is increased ~twofold compared to NTG that do not express human CRP.1, 2 The major new finding of this study is that mouse C3 is required for this action of human CRP. Thus in CRPtg/C3-/- mice that lack expression of complement C3 (the activation of which is the converging point of all three complement pathways and the source of the effector functions of complement), we observed that the neointimal response to injury was significantly less than that of CRPtg mice that have a fully functional complement system. In fact, neointima formation in CRPtg/C3-/- was no more robust than that seen in mice that did not express human CRP but did express mouse C3.
While the liver is the primary source for the synthesis of C3,24 many other specialized cells - including macrophages25, 26 - can synthesize C3. The results generated here indicate that the requirement of mouse C3 for the effect of human CRP in the vascular injury response is related to human CRP mediated induction of expression of C3 by mouse macrophages residing in the periadventitia. Given that CRP also activates complement, this sets the stage for a feed-forward cycle in the injured blood vessel, wherein CRP both induces expression of complement and activates it.
Similar to CRPtg mice, it has been demonstrated that complement components27 and CRP28 are present in the atherosclerotic intima, and that CRP colocalizes with C5b-9 in early atherosclerotic lesions in humans.14 The effect of complement activation on the development of the neointima in experimental animals has been controversial, with some studies showing that interruption of the complement cascade reduces the extent of the lesion,29-31 while others showed no effect,32 or even an enhancement of neointima formation.33, 34 Although there is no consensus on the interpretation of these apparently conflicting findings, these differences could be dependent on the models utilized to induce neointima formation. While studies showed that blocking complement led to a reduction in neointima formation in animals with diet-induced disease,29, 30 other studies that were carried out in animals with genetically-induced disease; i.e. APOE-/- and LDL-/- mice suggested an increase in neointima formation or no effect.32-34 Caution is warranted however, because the vascular remodeling process associated with carotid artery ligation versus atherogenesis is not the same. A report by Shagdarsuren et al., the only study so far to address the effect of complement on neointima formation in the setting of mechanical vascular injury, suggested that blocking the complement cascade at the level of C1q retarded neointima formation.31 Our findings are consistent with this view in that the neointimal response to carotid artery ligation, a model of vascular injury that is similar (not identical) to the one used by Shagdarsuren et al., was significantly reduced in the absence of the C3 gene.
Importantly, complement activation by CRP is distinguished by selective activation of early components with little or no formation of the distal C5-convertase or C5b-9 membrane attack complex.reviewed in 11 CRP also recruits factor H, inhibiting the alternative complement pathway amplification loop and the C5 convertases.35 In accordance with these observations, the deposition of C3, but not C5, in the lesion was increased in our CRPtg compared to NTG mice, and CRP induced macrophages in culture to increase their expression of C3 but not C5.
It is now generally accepted that inflammation plays an integral role in the pathogenesis of vascular injury.36, 37 In parallel with evolution of the inflammatory model of vascular disease, epidemiological data emerged that point to CRP as a strong predictor of vascular events.38 Recently, a large prospective randomized study demonstrated the additive value of serum levels of CRP to that of cholesterol in directing therapy in a population with no other indication for cholesterol lowering therapy.39 The evidence that CRP is present in vascular lesions and that it stimulates and activates inflammatory cells, endothelial cells and vascular smooth muscle cells, has sparked a debate on whether it is directly involved in the pathogenesis of vascular disease or is simply a marker of its ongoing activity.5 Although there are well known genetic determinants of serum CRP levels,40 two large studies have failed to show an association between these genetic determinants and cardiovascular disease outcomes, raising suspicion that elevated serum levels could be an effect rather than a cause of vascular disease.41, 42 These indirect studies are inherently incapable of establishing or refuting a direct cause-and-effect relationship linking CRP to vascular injury. Ultimately, human studies in which CRP is selectively lowered in randomized controlled trials will be needed to answer this question.43 Meanwhile, studies that explore the actions of human CRP in vivo, in a system that expresses human CRP in a fashion that parallels the human condition, can provide very useful information. In this study we showed that by decreasing circulating levels of human CRP in CRPtg with a human CRP-specific ASO drug, both circulating human CRP levels and C3 expression/deposition in injured arteries were reduced. Both outcomes would be predicted to be of clinical benefit in patients.
Our new results need to be interpreted in the context of our previous finding that FcγRI is essential for CRP to mediate vascular remodeling, i.e. in the absence of expression of the FcγRI ligand-binding α-chain or the cell-signaling γ-chain, human CRP does not enhance the neointimal response to injury in CRPtg.1 Based on those earlier findings and the new ones reported here we are now testing the following model: In the context of vascular injury CRP activates resident/recruited periadventitial macrophages, likely by interacting with FcγRI,1, 44 thus stimulating them to increase expression of C3. Locally expressed C3, perhaps in conjunction with circulating (hepatically derived) C3, itself an acute phase protein,45 accumulates and is readily available for activation by CRP. Ultimately this generates anaphylotoxic and chemotactic fragments of C3, which recruit more macrophages to the site of injury, so the process feeds forward. This model accounts for why both FcγRI deletion and C3 deletion eliminate the CRP-mediated exacerbation of the neointimal response to injury in CRPtg: in the absence of FcγRI CRP is unable to stimulate macrophage expression of C3, and in the absence of C3 the feed-forward loop is broken. Although smooth muscle cells are known to be intimately involved in the development of the neointima after carotid artery ligation, we found no evidence that human CRP colocalizes with smooth muscle actin.1 Rather, human CRP colocalized with the macrophage marker F4/80. Efforts are currently underway to validate our CRP→macrophage FcγRI→macrophage C3 model in CRPtg. If these are fruitful, and the model is confirmed in humans, these findings could have important therapeutic implications.
Sources of Funding: This work was supported, in part, by American Heart Association NCRP Scientist Development Grant 0930098N and the Walter B. Frommeyer, Jr., Fellowship in Investigative Medicine (to F.G.H.), National Heart, Lung, and Blood Institute grants HL07457, HL64614, HL75211, HL087980 (to S.O.); HL080017, HL044195 (to Y.F.C.), American Heart Association Greater Southeast Affiliate grant 09BGIA2250367 (to D.X.)., and funding from Isis Pharmaceuticals Inc. (A.J.S.).
Disclosures: Dr Alexander J Szalai has received grant funding from Isis Pharmaceuticals Inc.