Lipid accumulation, lipid oxidation, and low plasma levels of acquired antibodies against oxidized lipids associate with degeneration and rupture of the intracranial aneurysm wall

Tissue samples were collected intraoperatively from the fundi of 54 aneurysms after microsurgical clipping of the aneurysm neck. Samples were snap frozen in liquid nitrogen (n = 44), or fixed in 4% paraformaldehyde (PFA) for 6 hours (n = 10) and embedded in paraffin. As controls, two snap frozen samples of non-aneurysmatic MCA wall were obtained from ELANA bypass surgeries and two formalin fixed MCA bifurcations from autopsies. Blood samples were drawn with venipuncture on the 4th or 5th postoperative day from patients that underwent microsurgical clipping of unruptured or ruptured sIAs (n = 125). EDTA plasma was isolated by centrifugation. Medical records of the patients were reviewed for demographic data, medical history (smoking, hypertension, prior SAH, presence of other sIAs, family history of SAH, and history of cardiovascular or other major disease), and aneurysm size (Table 1 and Table 2). The study was approved by the Instutional Review Board and Ethical committee for the Departments of Neurology, Opthalmology, Otorhinolaryngology, and Neurosurgery of the Helsinki University Central Hospital.

Due to small size and limited availability of the tissue samples, all stainings were not performed from all samples, but instead subseries of samples were studied with different stainings.

For general morphology, paraffin and cryosections were stained with hematoxylin or hematoxylin-eosin stainings. To visualize lipids, cryosections were stained with Oil-Red-O followed by hematoxylin background staining for nuclei.

For the antibodies and dilutions used in immunohistochemistry, please see Table 3. Snap frozen tissue samples were cryosectioned at 4 um and fixed with −20°C acetone or 4% PFA, followed by endogenous peroxidase block with 3% hydrogen peroxide in methanol. Alternatively the peroxidase block was performed with 3% H202 in PBS between incubations with the secondary antibody and the avidin-biotion complex. The sections underwent serum block with 3% normal horse serum in PBS for monoclonal antibodies, or with 5% goat or rabbit serum for polyclonal antibodies. Following serum block the sections were incubated with the primary antibody either for 1 h at room temperature or overnight at +4°C, followed by PBS washes and incubation with a biotinylated secondary antibody (Vector, Burlingame, CA, USA). After subsequent PBS washes, the sections were incubated with either horseradish peroxidase or alkaline phosphatase conjugated avidin-biotin complex for 30 min – 60 min in room temperature. Peroxidase activity was visualized with DAB (Sigma-Aldrich, St. Louis, MO, USA and Zymed Laboratories Inc, San Fransisco, CA, USA) and alkaline phosphatase acivity with Vector Blue (Vector). Sections incubated with peroxidase conjugated primary antibodies underwent DAB incubation directly after primary antibody incubation and PBS washes. Mayer’s hematoxylin or Nuclear fast red were used as background stains for peroxidase or phosphate conjugated stainings. Stainings with the primary antibody omitted or replaced with an irrelevant antibody served as controls.

For double immunostaining, the sections underwent immunostaining with one antibody as described above, followed by a second serum block for 1 h or overnight, and immunostaining with a second primary antibody as described above, or with a fluorochrome conjugated secondary antibody (1:200, Alexa Fluor 488 green conjugated goat anti-mouse IgG, Molecular Probes Inc., Eugene, OR, USA) that was used to detect anti-alfa smooth muscle actin primary antibody (Table 3). Sections stained with fluorochrome conjugated secondary antibody were mounted with Vectashield containing DAPI (Vector).

Immunostainings were also performed on paraffin embedded tissue samples that were fixed for 6 h in room temperature with 4% PFA. Immunohistochemistry was performed as described for the frozen sections, but before the staining protocol the paraffin sections were deparaffinized in xylene and alcohol and underwent antigen retrieval in heated 0.01 M citrate buffer (pH 6).

The Oil-Red-O stained and copper oxidized LDL immunostained sections were classified according to the wall structure, as described previously (A: endothelialized wall with linearly organized SMC, B: thick myointimal hyperplasia-like wall with disorganized SMC, C: hypocellular wall with fresh or organizing thrombosis on the luminal surface, D: an extremely thin thrombosis-lined hypocellular wall. In this context, hypocellular refers to the low number of smooth muscle cells or myofibroblasts in the wall. The walls classified as hypocellular may have had significant inflammatory cell infiltration) [7].

Oil-Red-O stainings and immunostainings for copper oxidized LDL were imaged witht Zeiss Axiovision light microscope (Carl Zeiss GmbH, Germany). Consecutive serial microphotographs were taken with 10x magnification covering total surface area of the stained sections excluding luminal thrombus. From these microphotographs, the number of hematoxylin stained nuclei was counted with ImageJ (NIH software) from areas with and without lipid accumulation (Oil-Red-O staining) separately (n = 148 from 38 aneurysms), as well as from areas with and without oxidized LDL (immunostaining with YE antibody against copper oxidized LDL) separately (n = 101 from 15 aneurysms). Because the lipid accumulation varied a lot from a sample to another, and because areas with or without lipid accumulation were not of similar size within a section or sample, the number of counted nuclei was divided with the surface area of the region of interest from which the number of nuclei was counted from. This operation resulted to a variable describing the density of nuclei in wall areas with or without lipid accumulation. In addition, positive areas in Oil-Red-O and copper oxidized LDL stainings were scored for staining pattern according to the localization of the positive signal (intracellular or in the extracellular matrix) and according to signal intensity (few or many positive cells; weak, moderate, strong staining of the matrix, please see X-axis on Figure 1G for the scoring of Oil-Red-O stainings and X-axis on Figure 2H for the scoring of copper oxidized LDL immunostainings). Since multiple measurements were obtained from each sIA, Friedman test for non-parametric testing of related samples between multiple groups was used for statistical comparison of nuclear densities and staining patterns in Oil-Red-O or copper oxidized LDL immunostainings.

In addition to the multiple measurement from different sections of each individual sample described above, the ratio of positively stained surface area and total surface area were measured using Image J (NIH software) from reconstruction images made with Photoshop (Adobe Inc.) from consecutive microphotographs taken from the Oil-Red-O stainings and from immunostainings against copper oxidized LDL (YE antibody). Furthermore, for statistical comparison with clinical parameters, an average density of nuclei was calculated for the Oil-Red-O positive or negative and copper oxidized LDL positive and negative areas of each sIA sample from the multiple measurements performed from multiple sections of each sample.

Immunostainings for adipophilin, ApoB100, and oxidized lipid epitopes were scored as positive or negative and according to the localization of immunopositivity as follows: i) in the matrix, ii) in mural cells, or iii) in the thrombus. Microphotographs were taken with an Olympus AX70 (Olympus Optical, Japan) light and epifluorescent microscope using microscope mounted digital camera and analySIS (SoftImagingSystem GmbH, Germany) software. Overlay images and figure panels were created with Image J (NIH software) and Adobe Photoshop 8.0 software (Adobe Systems Inc., San Jose, CA, USA).

Cholesterol and triglycerides levels were measured from plasma samples using standardized enzymatic methods at Kuopio University Hospital Central Laboratory.

Titers of IgG against a modified ApoB-100 peptide (p244 sequence corresponding to the amino acids 3144–3163 of ApoB-100) were measured with a commerically available ELISA kit (Ark Therapeutics Group Inc., London, UK). Plasma samples were incubated in plates coated with the modified peptide. Following subsequent washes, incubation with peroxidase conjugated anti- human IgG reagent and subsequent new round of washes, the plates were incubated with peroxidase substrate (H202) and TMB chromogen for 30 min. Color development was stopped with 0.5 mol/L H2SO4 , and absorbances were measured at 450 nm with a Multiskan microplate reader (Thermo LabSystems, Waltham, MA, USA).

Titers of IgG against copper oxidized and native LDL were first measured as described previously by Närvänen et al. [22]. In brief, plasma samples were incubated in 96 well plates coated with either native and copper-oxidized LDL as described previously [22]. After subsequent washes, bound IgG was detected with a peroxidase conjugated anti-human IgG reagent and TMB chromogen as described [22], using the Multiskan microplate reader. Circulating IgG antibodies against “native” and copper oxidized LDL were measured in three groups: first in 24 patients with SAH and 11 patients with unruptured sIAs, then in other 21 patients with SAH and 11 patients with unruptured sIAs, and finally in a series of 65 patients with SAH and 32 patients with unruptured sIAs, which included the patients in both prior groups. All measurements in both types of ELISA were performed as duplicates in each plates. To control for variance between the plates, 34% of the samples were tested in two or three plates.

Plasma IgG and IgM antibody levels to copper oxidized LDL and malondialdehyde modified LDL, were also determined with chemiluminescent immunoassay as previously described [23]. Copper oxidation and malondialdehyde (MDA) modification of LDL were prepared as described [24]. Antigens were immobilized on white 96-well microtiter plates at 5 μg/ml in PBS containing 0.27 mM EDTA overnight at +4°C. Plasma samples were diluted 1:500 and incubated for 1 h at room temperature. The amount of plasma antibodies bound was measured with alkaline phosphatase labeled anti-human-IgG and anti-human-IgM antibodies (Sigma-Aldrich). LumiPhos 530 (Lumigen Inc., Southfield, MI, USA) was used as substrate and the chemiluminescence was measured with Wallac Victor3 multilabel counter (Perkin Elmer, Waltham, MA, USA). The results were expressed as relative light units measured in 100 ms (RLU/100 ms).

Proportions, medians, and range were calculated for categorical and continuous variables and compared with 2-sided Fisher’s exact test, Friedman test, Mann–Whitney U-test, or Spearman’s correlation test as approriate. Statistics were calculated using SPSS 17.0 statistical software (Apache Software Foundation). Alpha-level was 0.05. Graphs were drawn with OriginPro 8.6 software (OriginLab Corporation, Northhampton, MA, USA).

Lipid accumulation in Oil-Red-O stainings was detected in all studied snap frozen sIA walls (n = 38, 13 unruptured). Lipid accumulation in sIA walls was observed despite normal plasma cholesterol and triglyseride levels (median chol 4.0 mmol/l, range 2.1-4.7 and trigly 1.1 mmol/l, range 0.6-3.2, n = 10 patients included in the histological series, please see Figure 1B for plasma cholesterol levels in patients studied for circulating antibodies reactive against oxidized lipids). In samples obtained from two non-aneurysmal middle cerebral artery walls, in one accumulation of lipid was observed in the subendothelium but not in the media (Figure 1C), and in the other throughout the wall.

In sIA walls that were abundant in mural cells, lipids were found in the matrix in a rim between the outer third and the inner two thirds of the wall (Figure 1D), but mostly intracellularly (foam cells) (Figure 1E). In sIA walls that had less mural cells and more degenerated matrix, the accumulated lipid was found mostly in the matrix and throughout the wall (Figure 1F). Although lipid accumulation was found in both unruptured and ruptured sIAs, the pattern of lipid accumulation (scored as described on the X-axis of Figure 1G) was associated with wall degeneration (scored from A-D as described, P < 0.001, Fisher’s exact in both unruptured and ruptured sIAs respectively) and loss of mural smooth muscle cells (number of nuclei per standardized surface area, P < 0.001, Friedman test, Figure 1G). The more lipid in the matrix, the less that region had cells.

Expression of adipofilin, a protein induced by OxLDL ingestion on smooth muscle cells and macrophages [25], was observed in the sIA wall (6 ruptured and 5 unruptured sIAs studied). Adipofilin positive cells were found in areas of cellularized, intimal hyperplasia-like wall (Figure 3A), whereas in decellularized and degenerated walls adipofilin staining localized to the debris in the extracellular matrix (Figure 3B).

Immunostaining for ApoB100 (the core protein of LDL) was observed in sIA walls (10/10, 5 unruptured Figure 2A-B). In many sIA walls, ApoB100 had an intracellular staining pattern concordant with ingestion of LDL particles by mural cells (Figure 2A). ApoB100 was, however, found also in the extracellular matrix in some samples (Figure 2B).

The sIA walls underwent immunostainings with a panel of previously characterized antibodies generated against oxidized LDL epitopes (Table 3). Immunopositivity for epitopes of oxidized LDL were found in all stained samples (23/23, 5 unruptured, data not shown). Antibodies against copper oxidized LDL colocalized with immunopositivity for minimally modified LDL (Ox4E6 antibody) that stained a larger area of the wall. Immunopositivity for both malondialdehyde and hydroxynonenal (neoepitopes formed by oxidation of LDL) was also found in sIA walls, and colocalized in double stainings with natural mouse IgM antibodies reactive to oxidized LDL (EO6). Double stainings with alfa smooth muscle actin and hydroxynonenal showed ingestion of oxidized lipids by mural smooth muscle cells (Figure 2C-D).

Similarly to Oil-Red-O stainings, staining pattern (scored as described on the X-axis of Figure 2H) for the copper oxidized LDL (n = 15, 2 unruptured) was associated with loss mural smooth muscle cells (number of nuclei per standardized surface area, P < 0.001, Friedman test, Figure 2H). In walls abundant in cells, oxidized LDL was found more intracellularly, although some matrix staining was also observed (Figure 2E-F). In the degenerated walls with loss of mural cells, staining for OxLDL was mostly located to the matrix (Figure 2G). The extent of copper oxidized LDL immunostaining (measured as positive surface area / total surface area of the histological section) correlated (rho = 0.71, P = 0.003, Spearman rank) with the extent of lipid accumulation in Oil-Red-O stainings (measured as positive surface area / total surface area of the histological section ). In the two autopsy samples from non-aneurysmal middle cerebral artery bifurcation, immunopositivity for copper oxidized LDL was found in pads of intimal hyperplasia near the bifurcation (data not shown).

Plasma IgM antibodies against copper oxidized LDL or malondialdehyde modified LDL, did not statistically significantly differ between patients with unruptured sIAs or SAH, although there was a trend for lower levels in patients with a history of SAH (Figure 4C-D).

Plasma IgG antibodies against a modified peptide of the ApoB100 protein were found in 40/92 (43%) of the sIA patients, and had an increasing trend in SAH patients (33% vs. 50%) but failed to reach statistical significance (n = 92). Inspired by the trend, IgG antibodies against oxidized or native LDL were measured as described by Närvänen et al. [22] and by Karvonen et al. [23].

Contrary to the antibodies against modified ApoB100, levels of OxLDL or native LDL reactive IgGs measured as described by Närvänen et al. [22] were significantly higher in patients with unruptured sIAs when compared to SAH patients (n = 97 of which 32 unruptured, median for OxLDL reactive antibodies 0.59 vs 0.47, P = 0.021 Mann–Whitney U, for unruptured sIAs and SAH patients respectively and median for native LDL reactive antibodies 0.43 vs 0.35 for unruptured sIAs vs. SAH patients, P = 0.008 Mann–Whitney U). Their ratio did not significantly differ. To confirm the findings, plasma IgG levels were also measured by chemiluminescent immunoassay against copper oxidized LDL and malondialdehyde modified LDL. Levels of IgG antibodies against these epitopes were significantly higher in patients with unruptured sIAs when compared to SAH patients (n = 125, P = 0.001 and <0.001, Mann–Whitney U, Figure 4A-B).

Patient demographic data and aneurysm characteristics are shown in Table 2.

The association of IgG antibodies against OxLDL or malondialdehyde modified LDL with SAH was clear in females (median RLU for antibodies against OxLDL: 34 276 in unruptured sIA patients vs 25 127 in SAH patients, P < 0.006, Mann–Whitney U and median RLU for malondialdehyde modified LDL: 48 100 in unruptured sIA patients vs 34 490 in SAH patients, P < 0.001, Mann–Whitney U) but only a statistical trend in males (P = 0.067, 0.112) respectively, Mann–Whitney U). In female patients, age correlated inversily with malondialdehyde modified LDL reactive IgG antibodies (rho = −0.234, P = 0.027, Spearman rank).

The association of IgG antibodies against OxLDL or malondialdehyde modified LDL with SAH was significant in patients with single sIAs (median RLU for antibodies against OxLDL: 34 727 in unruptured sIA patients vs 22 373 in SAH patients, P = 0.001, Mann–Whitney U and median RLU for malondialdehyde modified LDL: 51 204 in unruptured sIA patients vs 31 821 in SAH patients, P < 0.001, Mann–Whitney U), but in patients with multiple sIAs the association of IgG against OxLDL was only a trend (P = 0.220) and only the IgG against malondialdehyde modified LDL was significantly different in patients with prior SAH vs. unruptured sIAs (median RLUs: 39 410 vs. 51 417, P = 0.030, Mann–Whitney U).

Smoking was slightly more prevalent in patients with unruptured sIAs (50% vs 42% in SAH patients). However, the smoking status of 32% of the SAH patients was not obtained, making it possible that smoking was much more prevalent than reported among SAH patients (Table 2). In SAH patients, confirmed current ex-smokers (n = 35) had higher levels of IgG against malondialdehyde modified LDL than confirmed non smokers (n = 18) (P = 0.017, Mann–Whitney U). In patients with unruptured sIA smoking did not, however, associate with IgG titers.

Hypertension was not clearly associated with SAH in this sample of patients (40% vs 36%, Table 2). Neither was hypertension associated with the titers of oxidized lipid reactive IgG or IgM titers in patients with SAH or unruptured sIAs, in female or male sIA patients, nor in patients with single or multiple sIAs.

Diagnosed cardiovascular diseases were found only in a few cases. Total cholesterol and triglyseride levels were measured from the plasma of 35 patients with SAH and 12 patients with unruptured sIAs, included in the serology studies. No difference was found in the plasma levels of total cholesterol or triglyserides between the groups (Median plasma cholesterol: 4.6 mmol/l, range 2.8-6.2 mmol/l for unruptured sIA patients and 4.6 mmol/l, range 2.1-6.1 mmol/l for patients with SAH history; median plasma triglycerides: 1.6 mmol/l, range 0.6-5.0 mmol/l for unruptured sIA patients and 1.0 mmol/l, range 0.5-5.7 mmol/l for patients with SAH history, Figure 1B). Furthermore, neither cholesterol nor triglyseride levels were unusually high in either group. No correlation was found between either between the plasma cholesterol or triglyseride levels and IgG or IgM antibodies against oxidized LDL epitopes. Smoking and hypertension did not either associate with plasma lipid profile in this patient population.

Visible atherosclerotic changes were found during operation in the parent arteries in 37% of patients with unruptured sIAs included in this study, and in 37% of patients with SAH and included in this study. In histological studies, however, accumulation of lipids was found in all studied sIA walls. Accumulation of lipids was found also in the 4 “normal” MCAs that were studied. Taken together, this suggests that accumulation of lipids in the cerebral vasculature is a somewhat common phenomenon, probably related to aging. What differs between unruptured and ruptured sIA walls is the pattern of lipid accumulation, which associates with loss of cells and degeneration of the wall.

The mechanisms how LDL and other lipids accumulate in the sIA wall remain unknown. Exposure of the endothelium to increased hemodynamic stress, e.g. in hypertension, predisposes to disturbances in the lipid metabolism of the arterial wall and subsequent intimal accumulation of lipids that become chemically modified [16]. Aneurysm walls are exposed to non-physiologic hemodynamic shear stress [26], suggesting that altered lipid metabolism of the sIA wall endothelia or complete or partial loss of endothelia and its barrier function, might be a mechanism that leads to accumulation of lipids to the sIA wall.

Our results show that the lipids that have accumulated to the aneurysm wall become ingested by mural smooth muscle cells and infiltrating macrophages, both of which eventually may turn into lipid ladden foam cells. In these foam cells, the ingested lipids are known to be enzymatically modified, by e.g. 15-lipoxygenase (15-LOX) [27], which we found expressed in the sIA wall cells (Figure 3E-F). It seems therefore likely, that the lipids in the sIA wall become oxidized and modified in the cells that have ingested them, similarly to what happens in atherosclerotic lesions [27]. It is, however, also possible that some of the lipids are oxidized extracellularly by oxidants released from luminal thrombus, infiltrating inflammatory cells, or mural cells that undergo necrotic cell death [5, 28].

The oxidized epitopes that we found in the sIA wall matrix and in smooth muscle cells, are potent inducers of cell death in vascular smooth muscle cells [18]. We showed an association between accumulation of lipids or oxidized LDL and the loss of mural cells and degeneration of the sIA wall. This suggests that accumulation of lipids or oxidized lipids might be an inducer of cell death in the sIA wall.

Oxidized lipids induce chronic inflammation in the atherosclerotic intima [10‐20]. As in atherosclerosis, infiltration of macrophages and T-cells is increased in ruptured sIA walls [5‐7]. Although we cannot establish a causality link between accumulation of oxidized lipids in the sIA wall and sIA wall inflammation, our observations and prior knowledge of the effects of oxidized lipids in the arterial wall suggest that accumulation of oxidized lipids in the aneurysm wall may trigger or maintain the cellular and humoral inflammation observed in sIA walls. Moreover, the oxidized lipids that are ingested by phagocytosis in antigen presenting cells (macrophages) in the sIA wall, may trigger a systemic humoral inflammatory response against oxidized lipid epitopes. Inflammatory cells were observed in the sIA wall areas with lipid accumulation (Figure 3C-D), although most foam cells observed were CD45 negative (not leukocytes).

Accumulation of oxidized lipids in the vasculature is known to trigger an immune response that leads to the formation of acquired antibodies against these oxidized lipids [12, 13, 15‐17, 20]. The plasma titers of these acquired oxidized lipid reactive antibodies correlate with the risk of clinical manifestations of atherosclerosis, although depending on the method used in the measurements, the correlation can be positive or negative [20, 22‐24, 29, 30]. Although the role that ox-LDL reactive antibodies have in atherosclerosis and lipid associated vascular diseases is not completely known, it has been shown in animal models that induction of acquired antibodies against oxidized lipids with immunization can reduce and even protect from the formation of atherosclerosis [15, 20]. It seems that at least some of the antibodies against oxidized lipids are protective and may facilitate the clearance of oxidized lipids from the vascular wall [15, 20].

In the patients we studied, lower levels ox-LDL reactive IgG antibodies were associated with a history of sIA rupture and SAH in patients with sIAs. Our results could suggest that patients with unruptured sIAs have acquired a stronger humoral immunity against oxidized lipids than patients in whom sIAs have ruptured and caused SAH. Since it is possible that consumption of IgG antibodies after SAH might have influenced the results, although no difference was found in IgM class antibodies reactive to oxidized lipids, our observation and interpretation should be confirmed with prospective serology studies on patients with unruptured sIAs.

Despite contradictory results, screening for anti-OxLDL antibodies has been suggested to be a useful diagnostic tool to detect patients at an increased risk of acute cardiovascular events [20, 22‐24, 29, 30]. The value of anti-OxLDL antibodies to detect patients at risk of hemorrhagic stroke has been investigated in one study [30], which found a negative result [30]. However, in that study two etiologically different forms of hemorrhagic stroke were grouped together (spontaneous intracerebral hemorrhage and aneurysmal hemorrhage) [30], which very likely influenced the outcome.

In order to use the titers of circulating antibodies reactive against oxidized lipids as a meaningful tool in the estimation of SAH risk in patients with unruptured sIAs, the results of this study should be replicated in separate, large prospectively collected patient cohorts and the possible associations with SAH risk factors investigated. Standardization of the methods will be of utmost importance in order to get consistent results in different cohorts.