A reporter for tracking the UPR in vivo reveals patterns of temporal and cellular stress during atherosclerotic progression.

Progression of human arteriosclerosis is associated with and promoted by induction of the endoplasmic reticulum (ER) stress pathway known as the unfolded protein response (UPR). Most studies that assess UPR markers in atherosclerosis rely on methodologies that suffer from low signal sensitivity, nonspecific immunohistochemistry, or inability to resolve differences between cellular subsets. To accurately monitor the UPR independently of artifacts generated postmortem, we describe here the first in vivo reporter for ER stress during atherosclerosis. Mice transgenic for the fluorescent XBP-1 ER stress indicator Erai were bred onto the Ldlr−/− background and fed an atherogenic diet. Subsequently, ERAI fluorescence at aortic roots was quantified and colocalized with lesional cell type. We found that the ERAI fluorescent signal increased as a function of time on the atherogenic diet and, in advanced lesions, was found close to necrotic cores. The majority of ERAI fluorescence localized to macrophages, and to a lesser extent, to intimal smooth muscle cells and patches of endothelial cells. These mice provide a valuable tool to monitor activation of the UPR in atherosclerosis and will be useful for future studies investigating relationships between pharmacologic and genetic modulators of UPR and atherosclerosis.

with ice-cold PBS and the upper half removed by cutting the ascending aorta midway between the aortic root and brachiocephalic artery. Roots were rapidly harvested and frozen in OCT. Fifty sections per mouse were cut at a thickness of 6 µm starting at the valves at the base of the aortic root. Total intimal lesion area (between internal elastic lamina to the lumen) per cross-section was quantifi ed by taking the average of six sections spaced 30 µm apart beginning at the base of the aortic root. Six cross-sections per mouse, each representing the average lesion area per mouse, were used for quantifi cation of ERAI and cell-type immunohistochemistry. Five transgenic mice and fi ve control mice were analyzed per time point except for week 16, in which 10 transgenic mice were compared with 10 controls.

Lesion analysis
Slides were fi xed in cold-acetone and blocked in 3% BSA prior to epifl uorescence microscopy or immunohistochemistry. All section images were normalized to equal exposure and contrast settings between control and transgenics. Under these settings, less than 1% of cells in the non Erai control lesions were autofl uorescent. To control for possible differences in fat-induced autofl uorescence in Erai mice, we also measured and found no differences in Oil Red O staining versus control. Staining of endothelial cells was performed with rat anti-mouse CD31 (PECAM-1;GPIIa) from AbD Serotec (MCA2388T). Macrophages were stained with anti-F4/80 from Serotec. Anti-actin was from Sigma. Phospho-eIF2 a (Ser51) (119A11) was from Cell Signaling. For endothelial cells, ‫ف‬ 1,000 endothelial cells were counted per mouse. Percent ERAI+ macrophages was determined by colocalization of F4/80+ nuclei with ERAI signal. Apoptotic cells in atherosclerotic lesions were detected by Tdt-mediated dUTP nick end labeling (TUNEL) using the in situ cell death detection kit, TMR red from Roche. The treated sections were incubated in TdT reaction mixture containing TMR red dUTP for 45 min at 37°C in a humidifi ed chamber. Nuclei were counterstained with Hoechst for 5 min. The slides were viewed and imaged by fl uorescent microscopy on an Olympus IX-70 inverted fl uorescent microscope equipped with a Cool Snap charge-coupled device camera and imaging software (Roper Scientifi c). Fluorescent images were captured and analyzed with image Photoshop analysis software (Adobe Systems, San Jose, CA) and ImageJ. All settings were normalized to equal exposures.

Statistics
Data are displayed as mean ± SEM. Statistically signifi cant differences of normally distributed data sets were determined by ANOVA.

A UPR reporter mouse for in vivo studies of ER stress in atherosclerosis
An in vivo ER stress reporter in plaque would be a useful tool for monitoring the effects of experimental and therapeutic manipulations on UPR activation within the complex cellular milieu of atheromata. To achieve this goal, mice transgenic for a fl uorescent ER stress-activated indicator ( Erai ) were crossed to atherosclerosis-prone Ldlr 2 / 2 mice to generate Erai, Ldlr 2 / 2 mice. Erai mice are healthy, fertile, and without developmental defects as previously described ( 14 ). Briefl y, these mice are transgenic for a fusion protein consisting of venus fl uorescent protein downstream of the XBP1 ER stress-specifi c intron and driven by heterogeneous extracts and immunohistochemistry. These techniques represent an indirect measure that is subject to artifi cial antigen modifi cation, nucleic acid degradation, and the possibility of nonspecifi c antibody interactions ( 11 ). These potential problems can lead to false-positive signals or postfi xation artifacts that include loss of antigenic epitopes and therefore signal underestimation or, conversely, false-positive signals. Gene expression analysis by laser capture micro-dissection is useful for monitoring UPR mRNA expression in situ ( 12 ), but this technique may be compromised by RNA degradation effects and single cell examination is technically challenging. Finally, spatial correlations between defi ned regions of interest, such as plaque necrosis and areas of UPR activation, may provide unique insight into the biology of plaque subregions subject to ER stress. Thus, an endogenous fl uorescent reporter induced during ER stress in vivo , could signifi cantly reduce such complications and further increase resolution to the level of individual cells, which is an important goal in the context of a heterogenous plaque population ( 13 ). In this context, Iwawaki et al. ( 14 ) created a mouse transgenic for a UPR reporter consisting of UPR X-box binding protein 1 (XBP-1) fused to venus fl uorescent protein and driven by the universal chicken b -actin promoter. Murine cells that are transgenic for this marker fl uoresce after activation of the UPR and therefore can be used to monitor physiological and pathological ER stress in vivo.
To monitor endogenous cell-type specifi c UPR in vivo during atherosclerosis, we crossed the aforementioned ERAI reporter mouse ( Erai ) to the atherogenic Ldlr 2 / 2 background. Transgenic Erai, Ldlr 2 / 2 mice developed atherosclerotic lesions in which lesional cell fl uorescence could be localized to differing extents to macrophage, endothelial, and vascular smooth muscle cells during lesion progression. Given the strong association between cardiovascular disease and UPR activation, this model will be a useful tool for future studies that test the temporal and spatial relationship between ER stress induction and atherosclerosis progression under various therapeutic and disease stimuli.

Vascular tissue preparation
At the end of the feeding protocols, mice were fasted, weighed, and anesthetized with isofl urane. Hearts were perfused in situ often found associated with intimal macrophages ( Fig. 2A ). The ERAI signal was also found to a lesser extent in endothelial cells ( Fig. 2B ). Endothelial cells that were positive for ERAI signal were often found in clusters. Previous reports documented that ER stress occurs in endothelial cells near regions of disturbed fl ow ( 19 ). Finally, the fl uorescent signal was also found in a small percentage of intimal VSMCs ( Fig. 2C ). The summary of this quantifi cation by cell type is shown in Fig. 2D .
Activation of the UPR in mice and humans has been shown by immunohistochemistry to increase with age and disease burden ( 2,5 ). To determine if fl uorescent signal in Erai, Ldlr 2 / 2 recapitulated these previous reports, we next assessed the kinetics of ERAI+ cell expression during atherosclerosis progression after feeding mice highcholesterol diet for increasing fi xed durations. As exemplifi ed by the images in Fig. 3A , early lesions contained very low levels of ERAI signal relative to more advanced and complex lesions. ERAI signal in advanced lesions was often found proximal to and within necrotic cores, areas that characteristically stain for macrophage debris ( 20 ). A quantifi ed analysis for intimal macrophage staining for mice on the atherogenic diet for 4, 8, 12, 16, and 20 weeks is shown in Fig. 3B . The data show that lesions from mice on diet for 4 and 8 weeks exhibited signifi cantly less reporter gene activation compared with the more advanced lesions starting at 12 weeks of diet ( Fig. 3B ). Previous reports have implicated endothelial ER stress as a relatively early event during atherogenesis (21)(22)(23). Quantifi cation of endothelial ERAI fl uorescence showed a prominent increase at 8 weeks ( Fig. 3C ), which is consistent with earlier activation in this cell type than in macrophages. Finally, previous work by Lin et al. ( 24 ) indicates that under chronic ER stress, the IRE-1/XBP-1 pathway is attenuated whereas PKR-like ER kinase (PERK) /CHOP signaling is sustained during programmed cell death. Consistent with their observation, we have found the vast majority of apoptotic/TUNEL positive cells (greater than 95%) to be ERAI negative. In contrast, signifi cantly more TUNEL positive cells costained for activated markers of the UPR PERK pathway, specifi cally phospo-eIF2 a ( Fig. 4 ).

DISCUSSION
During atherosclerosis, ER stress and ER chaperones are activated at different levels and in multiple cell types of the vascular wall ( 4,5,15 ). In endothelial cells of Erai, Ldlr 2 / 2 mice, ERAI signal was identifi ed. In endothelial cells, the potentially pro-atherogenic molecule homocysteine causes activation of ER stress-induced growth arrest ( 25 ). Also in endothelial cells, IRE-1-mediated XBP1 splicing leads to endothelial apoptosis and atherosclerosis development in response to disturbed arterial fl ow ( 6 ). ER transmembrane signal transducer IRE1 a has been shown to be activated in endothelial cells at atherosclerosis-prone regions of the aortic arch ( 19 ). This is consistent with our fi ndings of ERAI expression, which is downstream of IRE1 a signaling. It will be informative to determine in future studies the relative levels of reporter gene expression the universal chicken b -actin promoter . Under homeostatic conditions, ERAI message is not spliced and translation terminated at a stop codon upstream of the venus open reading frame. During induction of the UPR, the 26 nucleotide intron is spliced out and a frame shift permits translation of fl uorescent venus. The reporter does not contain a DNA binding domain and does not modulate, induce, nor inhibit ER stress responses ( 14 ). As an initial characterization, Ldlr 2 / 2 and Erai,Ldlr 2 / 2 mice were fed a Western diet for 16 weeks. Total plasma cholesterol and body weights of control and ERAI mice were not significantly different (data not shown). In addition, no differences in lesion size or Phospho-eIF2 a by immunohistochemistry were found between control and transgenic groups. Although the arterial media showed green autofl uorescence in both groups, only the Erai,Ldlr 2 / 2 mice exhibited signifi cant levels of green fl uorescence in the intima when the lesions from the two groups were imaged under identical exposure settings ( Fig. 1 ). These data are consistent with our fi ndings and others that have previously documented expression of UPR markers in advanced plaque ( 5,15 ). Atheromatous plaque is populated by a heterogeneous population of cells, including smooth muscle cells, endothelial cells, and macrophages, which are the most numerous cells in lesions (16)(17)(18). To ascertain the identity of cells in plaque that were positive for the UPR reporter, we performed immunohistochemistry for markers of macrophages, VSMCs, and endothelial cells from 16 week dietfed lesions. In advanced lesion, the ERAI signal was most is activation of ER stress signal transduction pathways ( 31 ). In humans, there is a very strong correlation between expression of ER stress markers and plaque necrosis and rupture in human coronary artery lesions ( 1 ). In experimental animals, ER stress effectors are causally linked to advanced atherosclerotic progression ( 2 ). In advanced lesions, ERAI signal was predominantly localized to macrophages. This is in line with well-documented in vitro and in vivo fi ndings that illustrate activation of macrophage ER stress pathways in response to atherogenic stimuli ( 4 ).
To determine the kinetics of ER stress in Erai , Ldlr 2 / 2 mice, we monitored ERAI signal as a function of time and high-cholesterol diet. Earlier lesions from Erai , Ldlr mice exhibited a lesser amount of reporter gene activation compared with more advanced lesions. Increases in fl uorescent signal occurred between 8 and 12 weeks. Under conditions of prolonged ER stress, a major function of the UPR is to trigger programmed cell death/apoptosis ( 32 ). A major death effector of the UPR is CHOP, and we recently reported that CHOP defi ciency reduces apoptotic cell accumulation in advanced atherosclerotic plaques ( 2 ). In cultured ER-stressed macrophages, CHOP induction precedes markers of programmed cell death ( 4 ). CHOP expression typically is elevated after IRE1-XBP-1 signaling ( 24 ). We therefore predicted that the lesional ERAI signal would not colocalize with TUNEL positive cells. This was in endothelial cells at other aortic sites of atherosclerotic susceptibility, including at sites of arterial bifurcation or curvature, where atherogenesis predictably originates ( 26 ). ERAI signal was also found in VSMCs. During advanced stages of atherosclerosis, VSMCs populate the subendothelium to form a protective fi brous cap. Current evidence indicates that athero-relevant ER stressors can promote smooth muscle cell death ( 27,28 ). In advanced plaque, VSMC death compromises plaque integrity and weakens the fi brous cap ( 29 ). Exposure of human coronary arteryderived VSMCs to 7-ketocholesterol can lead to the induction of CHOP and UPR chaperones ( 1 ). This consequently may affect the ability of plaques to generate cap-protective collagen. Moreover, fi brous caps of thin-cap human atheroma lesions were found to be positive for UPR markers ( 1 ). The role of VSMC ER stress is relatively less well understood and further experimental approaches, including in vivo tests of agents that modulate VSMC ER stress, will be benefi cial.
Macrophages play a critical role during atherogenesis by ingesting lipoprotein-derived cholesterol and accumulating cholesterol ester as lipid-laden foam cells in early lesions. In advanced lesions, a number of cellular and signaling events contribute to vulnerable plaque formation ( 30 ). A process known to be prominent in advanced plaque tent with reports in human infarct-associated plaque, these mice will be useful for future studies that accurately test the in vivo relationship between pharmacologic and genetic modulators of the UPR and atherosclerotic progression.
indeed the case as in advanced lesions, though ERAI-TUNEL double positive nuclei could be found, the majority (greater than 95%) of TUNEL positive intimal nuclei were ERAI negative, consistent with reported prosurvival effects of the IRE1-XBP-1 axis. In contrast, signifi cantly more TUNEL positive cells costained for activated markers of the UPR PERK pathway, specifi cally phospo-eIF2 a .
Erai,Ldlr 2 / 2 mice will be useful for a number of scenarios, including the effects of chemical chaperones during ER stress and atherosclerosis ( 33,34 ). These mice will also be useful in monitoring the in vivo dynamics of ER stress in the context of hypercholesterolemia at sites peripheral to the plaque. An important future direction will be to examine the regulation of specifi c branches of the UPR in vivo during athero-progression. Whether or not activation of the three major branches of the UPR and their downstream effectors differentially control athero-progression is an open question. The generation of mice that track the signaling of these distinct branches of the UPR will be informative. Given that Erai, Ldlr 2 / 2 mice exhibit patterns of the UPR consis-