Modified C‐Reactive Protein Is Expressed by Stroke Neovessels and Is a Potent Activator of Angiogenesis In Vitro

Abstract

Native C‐reactive protein (nCRP) is a pentameric oligo‐protein and an acute phase reactant whose serum expression is increased in patients with inflammatory disease. We have identified by immunohistochemistry, significant expression of a tissue‐binding insoluble modified version or monomeric form of CRP (mCRP) associated with angiogenic microvessels in peri‐infarcted regions of patients studied with acute ischaemic stroke. mCRP, but not nCRP was expressed in the cytoplasm and nucleus of damaged neurons. mCRP co‐localized with CD105, a marker of angiogenesis in regions of revascularisation. In vitro investigations demonstrated that mCRP was preferentially expressed in human brain microvessel endothelial cells following oxygen‐glucose deprivation and mCRP (but not column purified nCRP) associated with the endothelial cell surface, and was angiogenic to vascular endothelial cells, stimulating migration and tube formation in matrigel more strongly than fibroblast growth factor‐2. The mechanism of signal transduction was not through the CD16 receptor. Western blotting showed that mCRP stimulated phosphorylation of the key down‐stream mitogenic signalling protein ERK1/2. Pharmacological inhibition of ERK1/2 phosphorylation blocked the angiogenic effects of mCRP. We propose that mCRP may contribute to the neovascularization process and because of its abundant presence, be important in modulating angiogenesis in both acute stroke and later during neuro‐recovery.

INTRODUCTION

Stroke is a leading cause of death and disability in the Western world and usually occurs as a result of progressing atherothrombosis, cardiac or arterial embolism resulting in the loss of membrane integrity, proteolysis, the ability to synthesize proteins and altered signal transduction. Survival of neurons, particularly in peri‐infarcted regions determines the extent of patient recovery (31). Stroke patients with a higher density of blood vessels appear to have reduced morbidity and survive longer (16). It has been demonstrated that both therapeutic and mechanical revascularization in patients with acute ischemic injury may help to salvage tissue and enhance functional recovery after stroke 3, 32. Currently, only a limited number of stroke patients benefit from acute reperfusion therapies. In this context, therapeutic angiogenesis by collateral revascularization and reperfusion of potentially viable tissue may in part determine patient recovery. Our previous work has shown that angiogenesis correlates with patient survival and that survival of neurons is greatest in tissue undergoing angiogenesis (for a review, see 32). Following middle cerebral artery occlusion (MCAO) in a rat model, new blood vessels initiated through vascular buds formed regular connections with intact microvessels within 1 week of ischemia, the patterns being similar to those seen in normal brain. Apoptosis occurred in damaged endothelial cells (EC) in the “penumbra,” resulting in collapsed lumina and a lack of reflow (17).

Increased concentration of high sensitivity of C‐reactive protein (hsCRP) is related to an increased risk of vascular episodes and has been shown to correlate with brain infarct area, with the severity of ischemic episodes, with greater neuronal damage and with a higher risk of future vascular episodes 19, 34. Increased CRP levels in the plasma arise because of the enhanced synthesis by the liver, as a result of interleukin‐6 (IL‐6) induction. Evidence has shown that CRP, an important biomarker that is able to predict the pathogenesis of atherosclerosis, and is most relevant to this application, may also be a direct participant in the modulation of biological progression of the disease 2, 12, 18, 25. CRP is a pentameric oligoprotein composed of five identical 23 KDa subunits that can be irreversibly dissociated to form free subunits or mCRP. mCRP has a reduced aqueous solubility and a tendency to aggregate into matrix‐like lattices in various tissues, in particular, blood vessel walls (10). Cell membranes and liposomes can dissociate native CRP (nCRP) to form this more highly biologically active derivative 14, 41. A distinct difference in the biological activity of these two isoforms has been shown. mCRP is the most effective activator of the complement cascade when bound to low‐density lipoprotein (LDL) or oxidized LDL; however, fluid phase mCRP can bind to and prevent C1q from subsequent downstream activation of the complement cascade. mCRP can induce EC activation at concentrations significantly below the cardiovascular disease risk cut‐off point of 3.7 (ie, approximately 1 µg/mL) (14). In this paper we have shown for the first time that the insoluble form of CRP (mCRP) is expressed in damaged brain tissue after ischemic stroke. mCRP associated particularly with microvessels in the peri‐infarcted and infarcted zones and was found to be strongly angiogenic, promoting proliferation, migration and tube‐like structure formation of vascular EC in vitro. Our previous studies showed that commercially purified nCRP was also highly angiogenic (38), but data presented here show that on removal of contaminating mCRP, almost all of the angiogenenic effects are lost. We also showed the expression of mCRP in damaged neurons, although the possible role of mCRP in neuronal survival needs to be investigated in further detail.

MATERIALS AND METHODS

Patients and samples

Tissue samples were obtained within 4 h of death from 11 patients who died 1 to 29 days after stroke following MCAO. Immediately after death, the bodies were routinely refrigerated at 4^οC, and tissue were collected within 4 h. Tissue samples were taken from gray and white matter and from infarcted and peri‐infarcted areas. The peri‐infarcted region was defined as the area of tissue adjacent to the ischemic core, which we have previously found to exhibit stroke‐related changes including neuronal apoptosis and angiogenesis and was characterized by tissue edema and discoloration, morphology of the neurons and maintenance of structural integrity. Sections were stained with 2, 3, 5‐triphenyltetrazolium chloride, which stains active mitochondria pink, and therefore negatively stained areas represented the stroke area. The contralateral hemisphere served as a control from which tissue samples were removed at the same time. Tissue was immediately frozen in liquid nitrogen, stored at −70^οC, and a sample was processed for histology and stained with hematoxylin and eosin to determine tissue morphology. The usefulness of post‐mortem samples in studies involving measurement of RNA and protein expression has been identified in previous studies (22). All clinical and biochemical data are available for these patients. Samples were obtained from the Brain Bank at the Institut de Neuropatologia, Servei Anatomia Patològica, IDIBELL, Hospital Universitari de Bellvitge, Barcelona, with full institutional ethical approval.

Immunohistochemistry (IHC) and immunofluorescent staining

Paraffin‐embedded tissue samples from 11 patients were processed and serial 5‐µm sections were cut. The avidin–biotin peroxidase complex (ABC Vectastain kit; Vector Laboratories, Peterborough, UK) method was used, and antibodies to nCRP (N2C10), mCRP (M8C10) were obtained and fully characterized as described previously (27). Anti‐CD105 was from Abcam (Cambridge, UK). All antibodies were used at a dilution of 1:50. Paraffin‐embedded sections were deparaffinized, rehydrated and boiled for 10 minutes in an antigen unmasking solution of concentrated citric acid pH 6.0 as described elsewhere (33). Slides were incubated in 0.5% v/v H₂O₂ in methanol for 30 minutes, with normal serum for 20 minutes and then with a primary antibody (diluted in normal serum) for 30 minutes, followed by a 30‐minute incubation with biotinylated secondary antibody (diluted 1:50) and finally with ABC complex (diluted 1:50) for 30 minutes at room temperature. Staining was completed after incubation with 3,3′‐diaminobenzidine‐tetrachloride (DAB) substrate chromogen solution for 3 to 10 minutes. Slides were counterstained with hematoxylin, dehydrated, cleared and mounted in di‐n‐butyl‐phthalate‐polystyrene‐xylene. For immunofluorescence, cultured cells were fixed in 4% paraformaldehyde for 20 minutes, permeabilized with 0.2% Triton X100 for 10 minutes, blocked with normal serum and stained with the primary antibody as described earlier, followed by incubation with primary antibodies (1:50 for 1 h). Sections were washed and then incubated with the appropriate secondary antibodies (1:50). Fluorescein isothiocyanate‐conjugated sheep anti‐mouse IgG (for mCRP; Stratech Scientific Unit, Suffolk, UK) and tetramethylrhodamine isothiocyanate‐conjugated rabbit anti‐goat (for CD105; Jackson Labs) were used. Negative control slides were performed in parallel, where primary antibody was replaced with washing buffer and processed as above. No staining was seen in these sections (data not included).

Western blotting

Proteins were extracted from the tissue of eight patients where sufficient material was available and the protein concentration of each sample was determined using the BioRad assay. For Western blotting, 10 µg of protein was separated by sodium dodecylsulfate‐polyacrylamide gel electrophoresis (SDS‐PAGE; 13% w/v) and the proteins were electroblotted onto nitrocellulose filters as described previously (22). Filters were blocked in 1% w/v bovine serum albumin (BSA) in Tris‐buffered saline Tween (TBS‐Tween) and stained overnight at 4^οC with antibodies to nCRP, mCRP (1:100), phospho‐extracellular signal‐regulated kinase 1/2 (ERK1/2) and total ERK (1:500; Autogen Bioclear, Wiltshire, UK) and α‐actin (Sigma, St. Louis, MO, USA 1:1000) diluted in 1% BSA. Membranes were washed in TBS‐Tween before staining with the appropriate peroxidase‐conjugated secondary antibody, diluted 1:1000 in 5% w/v milk in TBS‐Tween for l h. Blots were developed with the enhanced chemiluminescence (ECL) detection system (Amersham, Buckinghamshire, UK). The relative intensities of the bands were measured in an LKB densitometer (LKB Instruments, Mt Waverley, Vic, USA). Results are semiquantitative and are given as a numerical (fold) change compared with the control (contralateral tissue), which was given an arbitrary value of 1.0. All experiments were performed twice to ensure no errors in the original loading and analysis of the initial samples and a representative example is shown.

ANGIOGENESIS ASSAYS

Preparation, testing and storage of CRP isoforms and antibodies

Antibodies directed to nCRP (clone N2C10) and mCRP (clone M8C10) were produced and characterized as described previously (27), and their specificity was demonstrated by measurement of binding characteristics (Figure 1A). High purity recombinant human nCRP (Calbiochem, San Diego, CA, USA) was stored in 10 mmol/L Tris, 140 mmol/L NaCl buffer (pH 8.0) containing 2 mmol/L CaCl₂ to prevent spontaneous formation of mCRP from the native pentamer. Before use, the nCRP was passed through a protein sepharose column to ensure removal of any contaminating mCRP and then stored in a sterile glass vial. mCRP obtained from human serum was purified by anion‐exchange chromatography as previously described (13). In order to validate CRP modification, electrophoresis was performed to distinguish mCRP from ntCRP, according to the method recently described by Taylor and van den Berg (37). Purity of the ntCRP and mCRP was also shown by dot blotting where doubling dilutions of CRP preparations were bound to nitrocellulose strips and exposed to anti‐n and mCRP antibodies (1:100). Following washing and appropriate secondary antibody incubation (anti‐mouse horseradish peroxidase conjugated; 1:100), blots were developed by ECL (Figure 1B). Controls where primary antibodies were omitted and where CRP was replaced with BSA on the nitrocellulose membrane showed no staining. In all of the experiments, CRP treated with detoxi‐gel columns containing immobilized polymyxin B was used to ensure the absence of pyrogens (AffinityPak™ detoxi‐Gel™ column; Pierce, Rockford, IL, USA). Removal of lipopolysaccharides (LPS) was confirmed using the limulus assay (sensitivity 0.01 EU/mL). Sodium azide was dialyzed out in CRP preparations in a large volume of Tris‐HCL buffer.

Figure 1.

Antibody characterization (A) shows specificity of the anti‐mCRP monoclonal antibody 8C10 used in the study by enzyme‐linked immunosorbent assay, demonstrating no binding with native C‐reactive protein (nCRP) or non‐specific binding to bovine serum albumin even at dilutions as low as 1:10. B. A dot blot with CRP proteins (5 µg) blotted onto nitrocellulose membrane demonstrating antibody 8C10 binding to a commercial source of nCRP contaminated with mCRP but no binding following phenyl‐sepharose column purification (left image). N2C10 antibody specific for nCRP still bound following purification (right image). Abbreviations: mAb = monoclonal antibody; mCRP = monomeric form of C‐reactive protein; PSC = phenyl sepharose column.

Cell culture and CRP binding studies

Our fully characterized primary vascular EC [bovine aortic EC (BAEC)] were used in previous studies 22, 27. Expression of endothelial cell markers von Willebrand factor, CD31 and the uptake of Dil‐labeled acetylated LDL has been demonstrated (data not shown). Cells were grown in complete medium composed of Dulbecco's modified Eagle's medium (DMEM) supplemented with 15% fetal bovine serum (FBS), 2 mM glutamine, 100 U/mL penicillin and 100 µg/mL streptomycin. Cells were seeded into T75 flasks (Nunc, Roskilde, Denmark) pre‐coated with 0.1% gelatin and incubated in humidified 5% CO₂ air at 37°C. Every 3 days, BAEC underwent a passage at a split ratio of 1:2 or 1:3 by treatment with phosphate‐buffered saline (PBS) without Ca²⁺ and Mg²⁺, then cell detachment by enzymatic digestion with 0.05% trypsin/0.02% ethylenediaminetetraacetic acid (EDTA). At confluence, EC had cobblestone morphology and were used throughout the study between passages 4 and 10. In all experiments, fibroblast growth factor‐2 (FGF‐2; 25 ng/mL) was used as a positive angiogenic control. Binding of mCRP and nCRP to BAEC was examined following exogenous administration (5 µg/mL; 5 minutes to 24 h), and fluorescent visualization (tetramethyl rhodamine isothiocyanate label) following staining with CRP‐specific antibodies detailed earlier.

Chemotaxis assay

BAEC were seeded at a concentration of 7.3 × 10⁴ cells/ml in 100 µL of serum‐poor medium, on Transwell Costar® porous membranes (Milian Laboratory Products, Gahanna, OH, USA; 8‐µm pore filter) in a 24‐well plate. Basal medium was supplemented with 0.1% FBS, and in some wells, purified nCRP or mCRP (1–10 µg/mL) or FGF‐2 (25 ng/mL) was added. For each experimental condition, cells were treated in triplicate. After 24 h incubation, the cells that did not migrate on the upper surface of the membrane were removed with a cotton swab soaked with PBS then wiped with a dried cotton swab. The cells that had migrated were fixed with 4% paraformaldehyde, left to air‐dry, stained with Giemsa Dako (Ely, UK) and five microscopic fields from each membrane counted with an optical microscope. All experiments were performed at least three times.

Cell proliferation assay

Cells were seeded in complete medium at a concentration of 2 × 10⁴ cells/mL per well in 24‐well plates. After 4 h, medium was replaced with serum poor medium (SPM), containing 2.5% FBS in which the cells grew at a significantly reduced rate, with or without n/mCRP (1–10 µg/mL or FGF‐2 25 ng/mL). For each experimental condition, EC were treated in triplicate. After 72 h incubation, cells were washed with PBS without Ca²⁺ and Mg²⁺, detached with trypsin‐EDTA then counted on a Coulter counter (Coulter Electronics, Hialeah, FL, USA). Experiments were performed a minimum of three times.

Tube formation

For Matrigel™‐based (BD Biosciences, San Jose, CA, USA) assays, 40 µL of serum‐starved (0.1% fetal calf serum) containing EC at a density of 1.5−2 × 10⁶ cells/mL was mixed with 40 µL of growth factor‐reduced Matrigel™ matrix, in the presence of purified nCRP or mCRP (1–10 µg/mL or FGF‐2 (25 ng/mL), in 48‐well plates and allowed to polymerize for 1 h at 37°C. After polymerization, 1.5 mL of medium of the same composition as used to polymerize the Matrigel™ was added to the wells and the cells were incubated for 24 h. The Matrigel™ gels were fixed with 4% paraformaldehyde (PFA), and the length of tube‐like structures was measured in a double‐blind fashion in five areas from each of the three wells as described previously (38) Experiments were performed a minimum of three times.

ERK1/2 inhibitor studies and mCRP‐induced angiogenesis

Western blotting demonstrated that addition of mCRP to BAEC in culture resulted in a notable increase in phospho‐ERK1/2 expression peaking after 8 minutes. Here, we compared phospho‐ERK1/2 expression in BAEC treated with either purified nCRP or mCRP (1–10 µg/mL; 8 minutes). In order to determine if signaling through ERK1/2 was necessary to elicit the mitogenic effects, BAEC were pre‐incubated for 1 h with the specific ERK inhibitor, PD98059, used previously with these cells (20 µM; 30) before addition of mCRP (5 µg/mL; 8 minutes). Changes in expression of phospho‐ERK1/2 were examined by Western blotting. The effects of inhibition of ERK1/2 on mCRP‐induced angiogenesis were studied using the methods described earlier.

Measurement of the expression of CD16 on BAEC and its relationship to mCRP signaling

As CD16 has been shown to be a receptor for CRP on various cells of immune origin, we determined if it was expressed on BAEC by incubation with anti‐CD16 antibody (1:50) (3G8; Pharmingen, San Diego, CA, USA) and fluorescein‐isothiocyanate (FITC) labeling as described earlier. This function‐blocking antibody was also pre‐incubated with BAEC (1 h; 2.5 µg/mL) as described in previously published studies (24), before addition of mCRP (5 µg/mL) to determine if membrane binding and angiogenic signaling operated through this receptor pathway. Cell migration and Matrigel tube‐like structure formation were examined using the methods described earlier.

Oxygen–glucose deprivation (OGD) experiments

Human brain microvessel EC (HBMEC; TCS Biologicals, Buckingham, UK) were cultured on cover slips in glucose‐free DMEM (Gibco, Paisley, UK) in a humidified atmosphere consisting of 95% N₂ and 5% CO₂ for 12 h, followed by 24 h reperfusion in a medium containing 4.5 g/L glucose (resulting in approximately 30% of cells undergoing apoptosis after OGD). Mixed fetal neuronal/astrocyte cultures were treated as described previously (23). Cells were stained with anti‐CRP antibodies (1:50) using FITC labeling as described earlier. Propidium iodide (PI; 10 µg/mL) an indicator of DNA damage was exogenously administered to the growing cells 1 h prior to the termination of the experiment (23).

Statistical analysis

All in vitro experiments were performed at least three times unless otherwise stated and the results expressed as the mean ± standard deviation. Statistical significance was tested by Student's t‐test and data were considered significant when P ≤ 0.05.

RESULTS

Commercially obtained nCRP contains contaminating mCRP

Figure 1B demonstrates that the commercially obtained source of nCRP contained measurable quantities of mCRP, and consequently, our previous work showing that nCRP has strong angiogenic properties might have been influenced by the presence of this mCRP. Therefore, in these experiments, mCRP was removed directly by passing through a phenyl sepharose column, and the resulting pure nCRP was used immediately or stored before use in glass containers to prevent spontaneous transformation (Figure 1B).

mCRP is expressed predominantly by microvessels in the infarcted core and peri‐infarcted regions of ischemic stroke tissue

IHC demonstrated that the native soluble form of CRP (nCRP) was expressed at very low levels both in normal looking (contralateral; Figure 2A) and stroked brain tissue (Figure 2B). This is perhaps not surprising as it is reported that following cell membrane binding, it dissociates into the mCRP isoform (14), and hence nCRP is normally found in the circulation. We could identify small amounts of nCRP within the lumen of occasional vessels in both contralateral and stroked areas (Figure 2B). Similarly, normal‐looking contralateral areas of brain tissue had almost null expression of mCRP as determined by IHC (Figure 2C). mCRP, however, was abundantly seen in the walls of and associated with the EC of both small and medium‐sized microvessels from both infarcted core and peri‐infarcted regions of all stroke patients studied (Figure 2D–H). We also demonstrated, using double immunofluorescent staining, that many of the mCRP‐positive vessels co‐localized with CD105, a specific marker of active (angiogenic) endothelium (Figure 2I,J). Furthermore, mCRP was associated with the areas of high concentration of newly formed blood vessels with a similar pattern of migrating EC to the ischemic core areas as described previously (Figure 2E,F; 16). In patients with up to 7‐day infarcts, mCRP was almost exclusively expressed in the microvessels. In these young infarcts there was an important inflammatory response in the ischemic tissue, with polymorphonuclear lecocytes and monocytes. However, these cells did not express mCRP. Interestingly, some parenchymal brain cells like microglia expressed mCRP in the surrounding areas (Figure 2K). In the old infarcts, there was a strong expression of mCRP in the core angiogenic microvessels. mCRP‐positive microvessels outnumbered all other stained cells.

Figure 2.

A. Expression of native C‐reactive protein (nCRP) in contralateral brain tissue: almost no expression of nCRP was seen in blood vessels and neurons from normal looking brain tissue [arrows mark the blood vessels; 3,3′‐diaminobenzidine‐tetrachloride (DAB) brown development; Patient 17]. B. Expression of nCRP in peri‐infarcted brain tissue again showing no notable expression, apart from occasional staining within the lumen of larger vessels (arrows; DAB brown development; Patient 17). C. Expression of monomeric form of CRP (mCRP) in contralateral brain tissue: almost no expression of mCRP was seen in normal looking brain tissue (arrows mark the blood vessels; DAB brown development). mCRP was abundantly expressed in peri‐infarcted (D–F) and infarcted (G–H) brain tissue particularly in microvessels (marked by arrows; DAB brown development; Patient 17). I–J. Double immunofluorescent labelling of mCRP and co‐localization with CD105 in active peri‐infarcted blood vessels (left, co‐immunofluorescence; middle, CD105‐stained fluorescein‐isothiocyanate, green and right image; mCRP‐stained tetramethyl rhodamine isothiocyanate, red). K. Expression of mCRP in parenchymal brain cells from inflammatory regions (arrows; DAB brown development; Patient 18). L. Expression of mCRP in peri‐infarcted neurons (arrows; DAB brown development; patient 18). A, C, H, K and L are ×40 the others are ×100.

mCRP was also expressed in the nuclei of some stroke‐affected neurons

Although not the main focus of this paper, it was noted that many of stroke‐affected neurons expressed mCRP in both the cytoplasm and within the nuclei (Figure 2L). The later pattern was mainly present in the most affected dying neurons of the infarcted core. Penumbral neurons were only weakly stained. In the normal areas of the brain, there was a weak cytoplasmic expression of mCRP in cortical neurons.

Western blotting showed increased expression of mCRP in stroke‐affected regions

Our previous published work and that of others have shown that CRP is a relatively unstable pentraxin with the pentametric form dissociating to monomers in acidic conditions and also in the presence of components of the blotting buffers such as SDS and mercaptoethanol. However, as we demonstrated clearly by IHC that almost all of the CRP found in brain tissue was composed of the mCRP isoform, Western blotting was able to provide a good approximation of the overall relative levels of its expression in our samples, although not as sensitive or accurate as IHC. Our results showed that mCRP expression was notably increased in active regions of 7/8 patients and more specifically in 4/8 patients in the peri‐infarcted zone (1.5‐ to 8.5‐fold) and in 6/8 patients in the infarcted region (2.5‐ to 9.9‐fold) compared with contralateral tissue (Figure 3; representative examples of blotting shown).

Figure 3.

Western blotting: Contalateral tissue lysate was compared with peri‐infarcted and dissected infracted areas for eight of the stroke patients (see Table 1 ). Sodium dodecylsulfate‐polyacrylamide gel electrophoresis was used to separate the proteins and mCRP was analyzed by scanning densitometry following blotting, staining and development with enhanced chemiluminescence. Relative fold differences between contralateral regions (given an arbitrary value of 1.0) and stroke‐affected areas are shown in the bar graphs. Blot labels are: 1, contalateral tissue; 2, peri‐infarcted tissue; 3, infarcted tissue. The blots were repeated twice to ensure no errors occurred in the first assay and a representative example is shown.

Table 1.

Expression of mCRP in brain tissue after stroke determined by IHC. Abbreviations: EC = endothelial cell; IS = ischemic stroke.

Code Western blot performed*	Age	Sex	Survival after stroke (days)	Expression of mCRP
				Neurons	In stroke affected areas EC	Glia	Inflammatory cells
256			2	+++	+++	−	−
18*	63	Female	2	+++ type IV	+++	−	−
17*	84	Male	3	−	+++	+	+
20*	51	Male	3	+++	+++	−	−
13*	84	Male	5	+	+++	−	−
8*	84	Male	6	−	+	−	−
112	87	Female	7	−	+++	−	+
4*	51	Male	9	−	+++	−	−
11*			10	−	++	+	−
136	68	Male	19	++	+++	−	−
12*	75	Male	29	+	+	−	−

mCRP was potently angiogenic to BAEC in vitro and induced phosphorylation of ERK1/2, whereas PD98059, a specific inhibitor of ERK1/2 activation, inhibited the angiogenic effects of mCRP

Exogenous administration of mCRP but not purified nCRP resulted in cell membrane association within 10 minutes of treatment. mCRP remained attached to the cells over the period of the study (24 h) (Figure 4A,B). Addition of mCRP or purified nCRP (1–10 µg/mL) to BAEC in culture produced only a small non‐significant increase in cell proliferation after 72 h compared with control‐untreated cells (data not included). Migration of BAEC, assessed using the Boyden chamber, however, was significantly increased in the presence of mCRP (1–10 µg/mL), with maximal increase at 10 µg/mL after 24 h culture (approximately 500% increase; P < 0.01; Figure 5). Purified nCRP had no significant effect on cell migration. When cells were cultured in Matrigel™ in the presence of mCRP for 24 h, a significant increase in the formation of tube‐like structures was observed compared with control cells (175% at 5.0 µg/mL; P < 0.01; Figure 6C). The increases were similar to that produced by FGF‐2 (25 ng/mL), a known potent angiogenic factor for these cells. Purified nCRP (minus contaminating mCRP) produced no angiogenic effect (Figure 6B). The mechanism of induction of angiogenesis did not appear to operate through mCRP binding to the CD16 receptor as we could not identify its expression on BAEC (Figure 7C), and specific blocking of CD16 sites using anti‐CD16 antibodies did not inhibit the angiogenic effects, that is, Boyden chamber cell migration (Figure 7A) and tube‐like structure formation in matrigel (Figure 7B).

Figure 4.

Exogenous addition of the monomeric form of C‐reactive protein (mCRP) but not the native C‐reactive protein (nCRP) resulted in cellular association with bovine aortic endothelial cells (BAEC). A. Shows positive staining and foci or localized expression of mCRP on the surface of BAEC following exogenous administration [5 µg/mL; 20 minutes; mCRP antibodies 1:50; tetramethyl rhodamine isothiocyanate developed, red; arrows]. The mCRP remained associated with the cells for the 23 h of the study. B. Under the same conditions, almost no binding of nCRP was seen. Photomicrographs were taken at ×100. All experiments were performed twice and a representative example is shown.

Figure 5.

Monomeric form of C‐reactive protein (mCRP) but not purified native C‐reactive protein (nCRP) (1–10 µg/mL shown) significantly increased the cell migration of bovine aortic endothelial cells determined using a modification of Boyden's method. Experiments were repeated three times in triplicate and a representative example is shown.

Figure 6.

Monomeric form of C‐reactive protein (mCRP) but not purified native C‐reactive protein (nCRP) (1–10 µg/mL shown) significantly increased formation of tube‐like structures in Matrigel™ in bovine aortic endothelial cells after 24 h. A. Control cells. B. Cells treated with purified nCRP. C. Those treated with mCRP (5 µg/mL). The tube structures are clearly shown in the boxed exert. Experiments were repeated three times in triplicate and a representative example is shown.

Figure 7.

Examination of CD16 involvement in the angiogenic effects of the monomeric form of C‐reactive protein (mCRP). Blocking the CD16 receptor sites with specific anti‐CD16 antibody (5 µg/mL; 1 h) prior to incubation with mCRP (10 µg/mL) did not inhibit migration or tube‐like structure formation in matrigel (A and B, respectively). Furthermore, we could not identify expression of CD16 on bovine aortic endothelial cells (C: immunohistochemistry, anti‐CD16 at 1:50; fluorescein‐isothiocyanate green development; arrows point to phalloidin‐stained actin cytoskeleton; tetramethyl rhodamine isothiocyanate, red). Immunoglobulin G controls showed negative staining (data not shown). Experiments were repeated twice in triplicate and a representative example is shown. Abbreviation: FGF‐2 = fibroblast growth factor‐2.

We have previously shown that ERK is a key downstream signaling protein responsible for early response gene activation and EC activation in vitro (30). Addition of either FGF‐2 (25 ng/mL) or mCRP but not purified nCRP was able to chronically induce phosphorylation of the mitogenic protein ERK1/2 from 5 minutes and for up to 4 h (Figure 8A–C). Here we show that under the same experimental conditions, inhibition of ERK1/2 phosphorylation using the specific inhibitor PD98059 (20 µM; Figure 8B) also resulted in almost complete reduction in mCRP‐induced BAEC migration (Figure 9) and tube‐formation in matrigel (Figure 10C,E).

Figure 8.

Both fibroblast growth factor‐2 (25 ng/mL; used as a positive control) and monomeric form of C‐reactive protein (mCRP) (µg/mL) induced phosphorylation of the mitogenic proteins extracellular signal‐regulated kinase 1/2 (ERK1/2) and this was perturbed following pre‐incubation of bovine aortic endothelial cells with the specific ERK1/2 inhibitor PD98059 (20 µg/mL: A and B). Addition of purified native C‐reactive protein had no effect on ERK1/2 phosphorylation (C). The bar graphs show fold differences in phospho‐ERK1/2 expression compared with the control (given an arbitrary value of 1.0). Experiments were repeated twice and a representative example is shown. Abbreviation: IgG = immunoglobulin G.

Figure 9.

The monomeric form of C‐reactive protein (mCRP) (5 µg/mL) and fibroblast growth factor‐2 (FGF‐2) (25 ng/mL)‐induced bovine aortic endothelial cell migration measured after 24 h were almost completely blocked following the pre‐incubation of the cells with the extracellular signal‐regulated kinase 1/2 (ERK1/2) inhibitor PD98059 (20 µM). Experiments were repeated three times in triplicate and a representative example is shown.

Figure 10.

The formation of tube‐like structures in Matrigel™ by either the monomeric form of C‐reactive protein (mCRP) (5 µg/mL) or fibroblast growth factor‐2 (FGF‐2) (25 ng/mL) was almost completely blocked following pre‐incubation of bovine aortic endothelial cells (BAEC) with the extracellular signal‐regulated kinase 1/2 inhibitor PD98059. B and C show the tube‐like structure formation in FGF‐2 (25 ng/mL) and mCRP (5 µg/mL)‐treated BAEC, respectively, and almost no tubes formed in E and F (PD98059‐treated cells). A. Control untreated cells. D. Cells treated with PD98059 alone. Experiments were repeated three times in triplicate and a representative example is shown.

In vitro OGD experiments on human fetal neurons (HFN) and HBMEC

HFN exposed to OGD (8h, as determined in pilot studies) demonstrated a weak but clear increase in intracellular mCRP expression (Figure 11A; control and B: after OGD). HBMEC cultured without OGD showed weak regular cytoplasmic expression of nCRP (Figure 11C); however, following OGD (13 h), the expression of nCRP was stronger and exhibited a granular appearance consistent with microvesicular localization (Figure 11D). mCRP expression was not observed in control HBMEC (Figure 11E); however, after OGD, strong granular cytoplasmic expression was seen (Figure 11F,G) indicating de novo synthesis of CRP, followed by rapid conversion to the monomeric form following exposure to hypoxic conditions. Many of the cells expressing nCRP or mCRP were not positive for PI, suggesting a possible protective mechanism against cell damage or apoptosis.

Figure 11.

Oxygen–glucose deprivation (OGD) induced increased expression of the monomeric form of C‐reactive protein in human fetal neurons (HFNs) and human brain microvessel endothelial cells (HBMEC). When cells were exposed to stroke‐mimicking conditions of OGD, an increase in cytoplasmic expression of mCRP was observed in both HFN [A: control; B: plus OGD 8 h; mCRP stained with fluorescein‐isothiocyanate (FITC), green; arrows; yellow arrow shows a propidium iodide (PI)‐positive nucleus]. C,D. HBMEC exposed to OGD for 13 h showing weakly increased expression of granular‐looking native CRP (E: FITC green; arrows) compared with the control untreated cells showing regular but weak cytoplasmic staining (C). E–G. There was no detectable expression of mCRP on control untreated HBMEC (E) but a notable increase in expression following OGD (13 h; F–G; FITC green; arrows), particularly in PI‐negative nuclei. Photomicrographs are taken at ×60. Experiments were performed in duplicate and a representative example is shown.

DISCUSSION

This study is the first to demonstrate increased expression of mCRP in human stroke tissue and its association with angiogenic neovessels and ischemic neurons in damaged regions. We further showed that although nCRP, purified by removal of contaminating mCRP, is not angiogenic in vitro, mCRP is highly angiogenic and therefore may contribute to the revascularization process aiding tissue recovery and reperfusion.

As mentioned earlier, CRP is a pentameric oligoprotein composed of identical 23 KDa subunits that can be irreversibly dissociated to form free subunits or mCRP. Only mCRP is able to aggregate into matrix‐like lattices in various tissues, in particular, blood vessel walls and therefore could potentially induce activation of signaling pathways within cells (9). In contact with nCRP, cell membranes and liposomes have been shown to dissociate this molecule into the highly biologically active derivative mCRP (14). Using carefully purified and synthesized mCRP and nCRP, Boguslawski et al (2) showed that only FITC‐labeled mCRP was able to bind to IgG molecules including pentraxins and vitronectin, again suggesting alternative and increased activity of the modified form. In this study we demonstrated an increase in mCRP expression in stroke‐affected regions of human brain tissue and a strong association of mCRP with angiogenic microvessels (CD105‐positive) in the damaged regions of tissue, suggesting a potential role in brain angiogenic responses folowing ischemic stroke. mCRP was present in angiogenic microvessels at very early stages of ischemia, as well as in old infarcted core areas. This may suggest that mCRP plays an important role in modulating angiogenesis at both the acute phase and later when plasticity and neurogenesis determine tissue survival. Interestingly, in our human post‐stroke brains, mCRP expression was almost uniquely expressed in microvessels and not in the inflammatory tissue. This may suggest that upon ischemic/hypoxic stimulus, CRP is being modified into a proangiogenic molecule, losing its main pro‐inflammatory properties. This is in agreement with the work of Schwedler et al (29), who showed that female ApoE (–/–) mice treated with a high‐fat Western diet suffered cardiovascular symptoms through inducible nitric oxide synthase‐dependent impairment of endothelial vasoreactivity only after supplementation with nCRP, whereas in the same model, nCRP promoted inflammation and plaque growth as mCRP actually reduced the size of the aortic plaques (28).

Vascular EC can directly express and secrete CRP. This has been confirmed by in vitro studies using EC and vascular smooth muscle cells (VSMC) cultures, which expressed CRP in response to various inflammatory stimuli 4, 40. Commercial antibodies used in these studies recognized both nCRP and mCRP, and so the relative expression of each was not determinable. However, this study did demonstrate a concomitant rise in IL‐6 (main inducer of CRP expression) and of the chemotactic protein MCP‐1 (an angiogenic protein induced by CRP), which might reflect a mechanism through which CRP of vascular origin contributes to maintaining and promoting the inflammatory process in stroke‐affected regions. LPS stimulation of U937 macrophages resulted in the formation of mCRP, suggesting that extrahepatic cells can produce this protein de novo (6). Local inflammation could therefore lead to conversion of nCRP to mCRP under conditions of low pH, and in the presence of oxygen radicals and enzymatic activity, providing a rich environment of this angiogenic substance in damaged stroke brain tissue. Khreiss et al (15) demonstrated that only mCRP could significantly increase gene expression of IL‐8 and MCP‐1, both angiogenic molecules, within 4 h through a p38 mitogen‐activated protein (MAP) kinase‐dependent pathway in human coronary artery endothelial cells (HCAEC) at physiological concentrations (1 µg/mL). These results suggest that CRP and subsequent conversion to mCRP may be in part a de novo response within the brain tissue after stroke and could impact upon infarct development and vascularization.

(Molins et al, unpublished observations), our group has demonstrated that following acute myocardial infarction in a porcine model, tissue expression of mCRP but not nCRP was increased in hypoxic and tissue damaged penumbral and infarcted regions. Similarly, here we were only able to show nCRP associated with blood cells within the lumen of medium‐large sized blood vessels, not associated with tissue, which is in keeping with the possibility that circulating CRP is pentametric, whereas that associated with necrotic or damaged tissue is monomeric. It is possible that studies performed previously were unable to discriminate between the pentametric and monomeric isoforms because of the lack of available and specific antibodies. It is now accepted that certain commercial antibodies (Sigma, clone 8) actually bind most strongly to the mCRP isoform rather than the nCRP. In vitro data presented here show that mCRP bound to and internalized in BAEC, whereas purified nCRP did not. It is possible that nCRP, when it becomes static and in contact with damaged tissue, tends to convert to the insoluble monomer, and this is what we are seeing.

In relation to the complement activating capacity of nCRP and mCRP, previous studies have shown co‐localization of CRP and complement in damaged cardiomyocytes and subsequent CRP mediated complement activation in infarcted regions (26). Similarly, all components of the complement cascade have previously been identified in infarcted brain lesions (1) and complement activation is associated with poor stroke outcome (35). In our recently published studies, we showed that mCRP enhanced, whilst nCRP had no effect on platelet deposition on a collagen surface, suggesting a modified role of mCRP in activating and/or maintaining pro‐thrombotic activity (24). As in this study we did not measure directly co‐expression and localization of mCRP with complement components, we can only speculate that it is possible that EC‐associated mCRP could activate this process, in particular, as arterial tissue has the ability to produce complement proteins. However, our IHC data show a strong association of the mCRP primarily in the walls of stroke‐affected active microvessels, and in this context and in regard to the focus of this paper, it would most likely exert a direct effect on EC activation, that is, angiogenesis.

The specific role of CRP in modulating angiogenesis has not been determined. Some earlier in vitro studies attributed functions to CRP that may have been due to contamination of the original commercial preparations by endotoxins and/or azide. For example, sodium azide in commercial CRP preparations was found to be responsible for a reduction in proliferation, migration and tube formation in Matrigel™ as well as to induce pro‐apoptotic effects in human umbilical vein EC (HUVEC) 21, 36. In other more carefully performed studies (and our recently published studies), the presence of LPS and azide was strictly controlled. Studies where the contaminants have been removed or controlled for have shown beyond doubt that native purified CRP does initiate activation of cell signaling cascades in EC, although it is highly likely that following contact with cells, the nCRP becomes modified or partially modified to mCRP. This could be the main instigator of angiogenesis, as shown by Ji et al (14), and confirmed by our data that showed beyond doubt that nCRP purified by removal of mCRP present in commercial sources loses its angiogenic capacity.

Cirillo et al (5) and most recently Devaraj et al (8) showed that CRP stimulated increased cell mitogenesis, ERK1/2 phosphorylation and tissue factor expression in thoracic aortic and coronary EC from New Zealand white rabbits and in a rat sterile air pouch model in vivo, respectively. In the study of Cirillo et al, relatively high concentrations of CRP were required to induce the responses (50–100 µg/mL) that may be explained by the presence of lower concentrations of mCRP in their “native” commercial source and/or its spontaneous conversion on contact with cellular membranes. Interestingly, tissue factor, which is upregulated in angiogenic malignancies as well as other vascular diseases, is itself angiogenic, promoting migration and differentiation of HUVEC and human microvessel EC in vitro (11); however, the expression of tissue factor by vessels after ischemic stroke and its relationship with CRP expression has not been investigated so far. In this study, we showed that purified mCRP at much lower doses (1–5 µg/mL) was as effective as FGF‐2 in promotion of in vitro angiogenesis in vascular EC. The effects on tube formation and cell movement were particularly strong, suggesting a mechanism whereby mCRP binding to cell surface receptors on EC elicits an intracellular mitogenic cascade resulting in cellular activation. The receptor for mCRP does not appear to be CD16 (shown on macrophages to be the main receptor), as we could not identify expression of this protein on our BAEC, nor did blocking the receptor sites with a specific monoclonal antibody abrogate the angiogenic effects of mCRP.

In our recently published studies, we found that commercial purified (LPS free) nCRP significantly promoted BAEC‐induced cell proliferation, although the effects on migration and tube formation in Matrigel™ were stronger. The current study most likely reflects most accurately the true angiogenic capacity of pure mCRP. Here, we we carefully removed LPS and sodium azide from the commercial source and then passed it through a phenyl sepharose column, ensuring complete removal of contaminating mCRP. The angiogenic effects were lost. It is possible that the commercial nCRP preparation contains other components able to stimulate cell proliferation that could also be removed after column purification. A complete microanalysis, perhaps using high performance liquid chromatography/mass spectrometry, would be useful in identifying other contaminating compounds in these preparations. In summary, we believe the effects of mCRP are more related to cell movement and differentiation in EC and are therefore focusing our future work on elucidating the cellular mechanisms through which this occurs.

Both mCRP and nCRP upregulated the receptor for advanced glycation end products in human saphenous vein EC, which are known to stimulate increased VEGF gene expression as well as MCP‐1 activation resulting in angiogenesis (44). CRP was also shown to upregulate nuclear factor‐κB and vascular cell adhesion molecule‐1 expression in HUVEC and human aortic EC (HAEC) (20). The equivalent concentration of sodium azide alone had no effect. CRP also induced matrix metalloproteinase‐1/10 gene and protein expression through p38 and mitogen‐activated protein kinase kinase and p38 and c‐jun N‐terminal kinase signaling pathways, respectively (25). Dasu et al (7) demonstrated that CRP, obtained from pleural fluid, stimulated IL‐6/8/1beta, plasminogen activator inhibitor‐1 and endothelial nitric oxide synthase in control and toll‐like receptor‐4 HAEC knockdowns in contrast to LPS.

In our investigation, using carefully purified mCRP, we observed a strong increase in phospho‐ERK1/2 expression in BAEC. Inhibition following incubation with PD98059 also prevented mCRP‐induced angiogenesis, suggesting that intracellular transduction operating through a key mitogenic MAP kinase pathway was responsible for its action. ERK1/2 is the final signaling intermediate, phosphorylated prior to nuclear activation of early response genes in association with mitogenesis. Our studies suggest that mCRP is responsible for this activation and are in agreement with our previous data that showed that the angiogenic effects of CRP could be blocked following pre‐incubation with a CRP‐specific antibody (38) However, whether these effects occur through a direct CRP‐induced receptor‐signaling cascade or via its activation or interaction with other receptor‐mediated mitogenic tyrosine kinase signaling pathways is not yet understood.

We also performed immunoflourescence using anti‐CD16 antibodies and showed that CD16 was not expressed on the cell surface or intracellularly, and, furthermore, blocking CD16 using a specific blocking antibody (24) did not ameliorate the angiogenic potency (cell migration or tube‐like structure formation) of mCRP, suggesting that mCRP does not elicit these effects through this receptor. A detailed analysis of the receptor‐mediated signalling pathways initiated following cellular binding of mCRP to EC is therefore warranted, as is an analysis of the membrane binding and dissociation/conversion characteristics of both mCRP and nCRP in order to determine their mechanisms of action.

We also observed expression of mCRP in both the cytoplasm and the nucleus of stroke‐affected, damaged neurons. In the areas of infarcted core, there was both nuclear and cytoplasmic staining. In surrounding core penumbra regions or in some scattered normal neurones, staining was limited to the cytoplasm. Nuclear binding of nCRP and association of mCRP with intermediate filaments was previously shown by Vaith et al (39) in rat kidney cells, and it is possible that CRP can exert signal transduction activation via this mechanism. Previous studies have demonstrated generation of CRP by and expression in pyramidal neurons and association of CRP and amyloid P (also a pentraxin) with neurofibrillary tangles of Alzheimer's plaques (43). In their study, they did not determine if the CRP was native or modified (monomeric) as the two isoforms had not been discriminated at this time. Previous publications have demonstrated that treatment of myocytes with nCRP in a hypoxic environment resulted in increased cellular apoptosis associated with cytochrome C translocation and activation of Bax and caspase‐3/9 (42). Our data show that in stroke‐affected regions, neurons express mCRP, and in vitro, HFN and, more strongly, HBMEC were able to synthesize CRP de novo following OGD, suggesting that the source may at least in part be de novo as well as via uptake of ECM‐associated CRP derived either from circulating CRP in hemorrhagic vessels and/or that produced by other cells including macrophages or EC. In summary, our results clarify beyond doubt that the highly angiogenic nature of mCRP as opposed to nCRP explains why patients exhibiting high circulating levels of nCRP during infection and disease do not necessarily have excessive inflammation or angiogenic responses. Association of mCRP with peri‐infarcted angiogenic neovessels in ischemic regions after stroke suggests a possible role in modulation of the revascularization process.

Our unpublished data have shown that mCRP is not readily detected in the serum of normal patients. An important question would be whether mCRP levels in the serum of patients can add useful prognostic information following ischemic and hemorrhagic stroke. In this study, both mCRP and nCRP levels and ratios could be measured, and clinical parameters, severity of stroke and extent of recovery can be studied. The potential influence of mCRP on revascularization and therefore its use as a marker of reperfusion could be examined by correlation between circulating levels and improved blood flow using neurological imaging techniques, such as magnetic resonance imaging and positron emission tomography/single‐photon emission computed tomography, as well as utilization of rodent models of stroke revascularization, to give a preliminary indication of its effects. In relation to the intensity of mCRP staining observed by IHC, we showed that where there was evidence of active remodeling with microvessel activation, the intensity of mCRP was strongest. In old areas of infarct containing dead necrotic material, the staining was much weaker or absent. Because of the small number of available patients/stroke samples used in this study, we were unable to correlate intensity of staining and perform any quantitative analysis of concentration and any relationship between time of initial infarct, survival times and ultimate mCRP tissue expression. Further studies should attempt to address these issues.