Both diet and Helicobacter pylori infection contribute to atherosclerosis in pre- and postmenopausal cynomolgus monkeys

A number of viruses and bacterial species have been implicated as contributors to atherosclerosis, potentially providing novel pathways for prevention. Epidemiological studies examining the association between Helicobacter pylori and cardiovascular disease have yielded variable results and no studies have been conducted in nonhuman primates. In this investigation, we examined the relationship between H. pylori infection and atherosclerosis development in socially housed, pre- and postmenopausal cynomolgus macaques consuming human-like diets. Ninety-four premenopausal cynomolgus monkeys (Macaca fascicularis) were fed for 36 months an atherogenic diet deriving its protein from either casein lactalbumin(CL) or high isoflavone soy (SOY). Animals were then ovariectomized and fed either the same or the alternate diet for an additional 36 months. Iliac artery biopsies were obtained at the time of ovariectomy and iliac and coronary artery sections were examined at the end of the study. Evidence of H. pylori infection was found in 64% of the monkeys and 46% of animals had live H. pylori within coronary atheromas as determined by mRNA-specific in situ hybridization. There was a significant linear relationship between the densities of gastric and atheroma organisms. Helicobactor pylori infection correlated with increased intimal plaque area and thickness at both the premenopausal and postmenopausal time points and regardless of diet (p< 0.01), although animals consuming the SOY diet throughout had the least amount of atherosclerosis. Additionally, plasma lipid profiles, intimal collagen accumulation, ICAM-1, and plaque macrophage densities were adversely affected by H. pylori infection among animals consuming the CL diet, while the SOY diet had the opposite effect. Plaque measurements were more highly associated with the densities of cagA-positive H. pylori within coronary atheromas than with the densities of gastric organisms, whereas plasma lipid changes were associated with H. pylori infection, but not cagA status. This study provides strong evidence that live H. pylori infects atheromas, exacerbates atherosclerotic plaque development, and alters plasma lipid profiles independently of diet or hormonal status. Finally, socially subordinate animals relative to their dominant counterparts had a greater prevalence of H. pylori, suggesting a stress effect. The results indicate that early H. pylori eradication could prevent or delay development of cardiovascular disease.

in women or suitable female animals are essential for clarifying CVD risk factors in young and older women.
Cynomolgus monkeys (Macaca fasciularis) have been used extensively to model the development of atherosclerosis in relation to diet, hormonal conditions, psychosocial factors, and pharmaceutical manipulations [33]. Atherosclerosis in this species closely mimics the pattern observed in human beings, including the relative inhibition of disease progression in reproductively mature females compared to males, the extensive involvement of the coronary and carotid arteries, and the exacerbating effect of traditional risk factors (e.g., hypertension and a diet high in cholesterol and saturated fat) as well as psychosocial stress [34][35][36]. Finally, responses to a wide variety of pharmaceutical interventions also resemble those observed in people [37].
The purpose of the present study was to determine in this well-established monkey model whether infectious agents-including H. pylori and a number of viruses-influence atherosclerosis and whether other risk factors, such as diet, reproductive hormones, and psychosocial stress modify any infection-associated effects. Hence, we measured atherosclerosis and plasma lipid composition in pre-and postmenopausal cynomologus macaques following consumption of diets that derived their protein from either animal (casein lactalbumin, CL) or high isoflavone soy isolate (SOY). Most notably we found that H. pylori, particularly cagA-positive H. pylori, positively correlates with atherosclerosis lesion size and dyslipidemia irrespective of diet treatment, whereas viral infections did not correlate with indices of atherosclerosis extent or severity. Psychosocial stress, as indicated by subordinate social status, was associated with an elevated prevalence of H. pylori.

Animals and study design
The research described in this manuscript (including the acquisition of monkeys and all protocols for their use) was approved by the Wake Forest School of Medicine Institutional Animal Care and Use Committee (Protocols A97-157, A01-195, and A04-116, encompassing approximately eight years of in vivo investigation preceded by a one year run in and a one year close down). Wake Forest is an AALAC accredited institution and all animal care procedures followed the NIH guide and the Principles for the Ethical Treatment of Non-human Primates of the American Society of Primatologists. In addition to the foregoing, the research described here was conducted in accordance with the standard operating procedures of the Wake Forest Environmental Enrichment Program, which involves the provision of cage devices and structures that include at a minimum hanging devices, mirrors, and perches in each pen, and regular surveillance under a behavioral management program designed to identify and mitigate atypical behavior. Finally, all animals were housed continuously in stable social groups-a situation designed to maximize the opportunity to engage in normal, species typical behavior. Macaques were fed twice per day with sufficient food that some remained uneaten.
Ninety-four (94) monkeys were randomized into the treatment conditions for this study. Eighteen (18) died over the following seven years of in-vivo investigation and an additional 2 monkeys became diabetic and had to be removed from the study. This meant that seventyfour monkeys survived until experimental necropsy, which for all cases involved use of sodium pentobarbital to induce euthanasia.
Of the 18 in-study deaths, 2 occurred in year two, 5 in year three, 2 in year four, 6 in year five, 1 in year six, and 2 in year seven. A complete anatomic and histopathologic necropsy was performed on each dead animal and a specific cause of death could be determined in 15 of the 18 cases. Causes of death were varied and included gastric dilation (2 cases), bladder carcinoma, acute cardiac arrest, congestive heart failure, oligodendroglioma, hypoglycemia, hyperglycemia, arterial disease, right atrial thrombosis, and intraoperative hypoxia. There were no deaths due to trauma among the animals in this study. This number of deaths and the development of diabetes is not unusual in a study that persists for almost eight years in-vivo, involves consumption of a human-like diet, and includes a major surgery (oophorectomy) at the midpoint (years four and five).
All surgery and euthanasia was performed under aseptic conditions using sodium pentobarbital anesthesia to minimize suffering. Specifically, this investigation was categorized as USDA class D (animals upon which experiments, teaching, research, surgery, or tests are conducting involving accompanying pain or distress to the animals and for which appropriate anesthetic, analgesic, or tranquilizing drugs are provided). Survival surgery was initiated with ketamine HCl (10-15 mg/kg, IM) followed by butorphanol (0.25 mg/kg, IM). After surgery, all animals were returned to single cages and monitored at 30 minute intervals until they were sitting unaided. Butorphanol was administered for pain relief. Hunched posture, lack of appetite, or different from that typically observed after anesthesia for nonsurgical procedures was interpreted as signs of discomfort or pain. Nonsurgical procedures (including blood and fluid sampling, bone density scans) were conducted using ketamine HCl restraint. Animals were observed from the time of anesthesia until they were fully capable of sitting unaided. Procedures were generally conducted in the A.M. Further monitoring (after the animal was sitting unaided) was done on an hourly basis. Veterinary care was sought if any of the aforementioned signs of distress were observed following surgical or non-surgical procedures.
A crossover study was conducted as previously described (Fig 1) [38]. Briefly, adult female cynomolgus macaques (Macaca fascicularis) living in permanent social groups (n = 5 or 6 per group x 16 groups; N = 94) were randomized to receive either an atherogenic diet containing casein and lactalbumin (CL) as the protein source or an otherwise identical diet that substituted high-isoflavone soy protein isolate (SOY) for the animal protein. Both diets were formulated to be equivalent in caloric content of protein, fat, carbohydrates and cholesterol (0.28 mg cholesterol/Cal) and to mimic a typical diet consumed by women in the United States [39]. After 36 months, the animals were ovariectomized to induce menopause and a segment of the left common iliac (LCI) artery was removed to measure premenopausal atherosclerosis. Half of the social groups were then randomly selected to receive the alternate diet, while the remaining social groups continued with the original diet. The four resulting diet groups were designated CL-CL (casein/lactalbumin only), CL-SOY, (casein/lactalbumin followed by high isoflavone soy), SOY-CL (soy followed by casein/lactalbumin), or SOY-SOY (soy isolate only). The postmenopausal phase continued for an additional 36 months. Using procedures developed in numerous prior studies [40,41], animals within each social group were categorized on the basis of competitive interactions as being socially dominant (n = 2 or 3/group) or subordinate (n = 3/group); thus approximately half of the animals were considered dominant and half subordinate.

Plasma lipid measurement
Total plasma cholesterol (TPC), high-density lipoprotein-associated cholesterol (HDLc), and plasma triglycerides (TGs) were determined on numerous occasions as well as at the end of the study by the Wake Forest Primate Center Clinical Chemistry Laboratory [42,43], which is fully standardized with the Centers for Disease Control and Prevention-National Institutes of Health Lipid Standardization Program specifications for accuracy and precision. Non-HDLc, which approximates the sum of low-density lipoprotein (LDL) cholesterol and very-low-density lipoprotein cholesterol, was calculated by subtracting HDLc from TPC.

Necropsy procedures
At the conclusion of the study, surviving monkeys (n = 74) were euthanized using sodium pentobarbital (100mg/kg, IV; i.e. consistent with the recommendations of the panel on euthanasia of the American Veterinary Medical Association) and perfused with lactated Ringer's solution. A complete necropsy was performed and all major organs, including the stomach, were collected and fixed in 10% neutral buffered formalin (NBF). The right common iliac artery (RCI) was removed for comparison with the left common iliac artery that was removed at the time of ovariectomy. Hearts were removed and coronary arteries were perfused-fixed with 10% NBF for 1h at 100 mm Hg pressure. The left anterior descending (LAD), left circumflex (LCX), and right coronary arteries (RCA) were then collected. Each artery was sectioned into five blocks (3 mm each in length), cut perpendicular to the long axis of the artery, processed routinely, and embedded in paraffin to generate formalin-fixed paraffin-embedded (FFPE) blocks. Whole sections of the left circumflex artery were also sent to the Uniformed Services University of the Health Sciences for detection of H. pylori. Due to postmortem degradation, usable gastric tissue could not be obtained from three animals. Coronary artery tissue from these animals was included in the analysis only if H. pylori was detected at that site, since lack of H. pylori in the artery does not guarantee that the animal was H. pylori-negative in the stomach.

Histopathology and morphometry
For each animal, the first 5mm FFPE block of the LCX (n = 74) was serially sectioned at 5um thickness and sections were mounted on slides. The first five sections were stained with Verhoeff -van Gieson stain to highlight the internal elastic lamina, and plaque size was determined using computer assisted morphometric analysis (ImageProPlus 5.1); the cross-sectional area of plaque (defined as the area between the endothelial lining and internal elastic lamina) in each of the five sections was measured and plaque area for each animal was calculated as the mean of the five sections examined. Section 6 was stained with Hematoxylin-Eosin (H&E), examined by light microscopy for the presence of mineralization and/or necrosis within the tunica intima, and semi-quantitatively scored as follows: 0 = none, 1 = minimal, 2 = mild, 3 = moderate, 4 = marked, 5 = severe. Section seven was stained with Masson's Trichrome to highlight areas of collagen deposition (fibrosis). The proportion of intima occupied by fibrosis was determined by overlay of a 25 x 25 pixel grid with evaluation of each crosshatch as positive or negative for collagen by a board-certified pathologist (JAC) blinded to treatment. The remaining LCX serial sections were utilized for immunohistochemistry as detailed below. Additional 5 μm sections of stomach tissue and LCX artery were generated, mounted on slides and stained with H&E and Genta-Robason (H&E, silver stain and Alcian blue), which is a conventional method for detecting the presence of H. pylori. Slides were examined by light microscopy for the presence/absence of argyrophilic curved or spiral-shaped bacteria. The remaining slides were used for in situ hybridization as detailed below.

Histochemistry and immunohistochemistry
For immunohistochemistry, the slides were incubated with mouse anti-human HAM56 (Dako Cytomation, Carpinteria, CA), mouse anti-human CD3 (BDBiosciences, San Jose, CA), mouse anti-human CD4 (Abcam, Cambridge, MA), and mouse anti-human CD8 (Abcam, Cambridge, MA) for 1.5 h at 25˚C. Primary antibodies were localized with appropriate biotinylated secondary antibodies, streptavidin-alkaline phosphatase (Biogenex, San Ramon, CA) and Vector Red (Vector Labs, Burlingame, CA) substrate. Sections were counterstained with Mayer's Hematoxylin and examined by the study pathologist (JAC) using light microscopy. Thymus and lymph node were used as tissue controls; as an assay control, the primary antibodies were substituted with non-immune sera. Cell densities for macrophages, CD3+ cells, CD4+ cells, and CD8+ cells were determined by overlay of a 25 x 25 pixel grid with evaluation of each crosshatch for positive or negative staining by the pathologist (JAC), who was blinded to treatment. Cell densities are expressed as the percentage of intima occupied by positive staining.

RNA isolation and quantification
RNA was isolated from intima-media sections as described previously [44-46] using a Tri-Reagent protocol. Total RNA concentration and quality was assessed using an Agilent 2100 Bioanalyzer and aliquots were reverse transcribed using a high-capacity cDNA archive kit (Applied Biosystems). Quantitative real-time reverse transcriptase-polymerase chain reaction (PCR) was performed on an ABI Prism 7000 system using TaqMan FAM-MGB primer-probe assays (Applied Biosystems

In situ hybridization
H. pylori was detected in coronary arteries (i.e., the LCX) and stomach tissue following euthanasia. All procedures were performed on paraffin sections according to previously described methodologies [Liu, 18648543;[50, 51]]. H. pylori-16S rRNA (antisense 5'-TAC CTC TCC CAC ACT CTA GAA TAG TAG TTT CAA ATGC-3' and sense 5' CTA TGA CGG GTA TCC GGC-3') and H pylori cagA (antisense 5'CTG CAA AAG ATT GTT TGG CAG A-3' and sense 5'-GAT AAC AGG CAA GCT TTT GAG G-3') probes were used. The 5' ends of the oligonucleotides were labeled with biotin or digoxygenin (see below). All probes were synthesized in the Synthesis and Sequencing Facility, Biomedical Instrumentation Center (USUHS). Negative controls were routinely run as described previously [52]. These included using sense instead of antisense probes, hybridization buffer without probe, probe without biotin or digoxigenin label, pre-treatments with DNase I, RNase A or both, and incubation with an irrelevant probe, scorpion Buthus martensi Karsch neurotoxin (5'-GGC CAC GCG TCG ACT AGT AC-3')[53].
In situ hybridization was performed as previously described [52,54,55]. Briefly, sections were deparaffinized in xylene, rehydrated, treated with proteinase K and washed with standard saline citrate solution (SSC). For pre-hybridization, sections were incubated with hybridization mixture (Denhardt's solution, NaCl, formamide, dextran sulfate EDTA, bovine serum albumin) and hybridization solution (SSC and extracted salmon testis DNA). For hybridization, sections were incubated with the denatured primer overnight at 37˚C. For post-hybridization, unbound probe was removed by successive washes in decreasing concentrations of SSC. For visible light studies, biotin-labeled probes were detected using streptavidin-alkaline phosphatase and nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl-phosphate (NBT/BCIP). Sections were counterstained with Nuclear Fast red. Serial sections were used to detect 16S rRNA and cagA mRNA. For fluorescence dual detection studies, 16S rRNA probes were labeled with biotin and detected using avidin-Texas red (red color), while cagA probes were labeled with digoxigenin-3-0-methylcarbonyl-ε-aminocaproic acid-N-hydroxy-succinimide ester (DIG-NHS ester, Roche Diagnostics, Indianapolis, IN) and detected using anti-digoxigenin-FITC (green color).
An Eclipse Nikon microscope was used to examine the section at X100, X400 and X1000 magnification for bright light and for fluorescence. Bacterial density was quantified as described previously [52]. Briefly, the number of positive-stained bacteria were quantified at 400X magnification in three randomly-selected fields according to a modification of the pointcounting stereological method and using an intraocular reticle (27 mm diameter, covering 3,578 μm 2 , i.e. 17,892 μm 3 for 5 μm-thick sections; No. KR-409 (Klarmann Rulings, Inc. Litchfield, NH), located in a Nikon Eclipse E 800. Tissues were analyzed randomly and blindly regardless of category. All data were expressed as mean ± standard error of the mean (SEM) number of bacteria per 10 6 μm 3 (representing an imaginary cube with sides of 100 μm or 0.1 mm).

Statistical analysis
Descriptive statistics and graphical plots, such as means, standard deviations and standard errors, and comparative boxplots, were calculated and obtained for H. pylori densities, cagA densities, H. pylori infection status, and social status. Linear models and analysis of variance (ANOVA) were used to determine the influences of H. pylori infection, diet, and social status on severity of premenopausal and postmenopausal atherosclerosis, arterial inflammatory markers, and lipid profiles. Pairwise group comparisons were performed using Welch twosample t-test and/or nonparametric rank tests (Kruskal-Wallis H test), the latter employed when assumptions for the validity of the t-test were not satisfied. Linear models and correlation analyses were used to compare the effect of H. pylori density in the artery and stomach on plaque area and thickness. Fisher's exact test was used where indicated. The (effective) sample sizes in some of the calculations were smaller than the overall sample sizes due to missing values.

Evidence of H. pylori infection in gastric and artery tissue
The study design is outlined in Fig 1. Curved or spiral organisms consistent with H. pylori were first detected in gastric mucosa using the Genta-Robason staining procedure. Subsequently live H. pylori were specifically detected in gastric mucosa and LCX coronary artery sections by in situ hybridization using RNA anti-sense probes specific for H. pylori 16S rRNA or cagA mRNA. These probes recognize RNA and the method has been previously validated [52]. The normal infection pattern for H. pylori is for the majority of organisms to reside in the mucus just above the epithelium, attached to epithelial cells, or within gastric crypts. H. pylori is occasionally intracellular or found within gastric capillaries [56]. We found similar results in our study. Within the stomach, the majority of H. pylori were located in the superficial gastric mucosa (Fig 2), but a few organisms were located in gastric blood vessels. In situ hybridization demonstrates co-expression of 16S rRNA and cagA in the same bacteria (Fig 3), further confirming that the organisms are H. pylori and demonstrating that some macaques were infected with cagA-positive strains, some were infected with cagA-negative strains, and some were infected with a mixture of strains. Infection with multiple H. pylori strains is not unusual. Of potentially greater interest, H. pylori was detected within coronary atherosclerotic plaques as isolated organisms, or small colonies, randomly distributed throughout the fibrofatty admixture of foam cells and extracellular matrix. Fig 4 shows examples of uninfected, moderately infected, and highly infected LCX arteries as measured using colorimetric ISH with a 16S probe. Tunica intima and media were often involved, but no colonization was observed in the tunica adventitia, suggesting that H. pylori originates from the vessel lumen and does not easily cross the external elastic membrane. Bacteria often appeared in clusters, suggesting limited replication (Fig 4D). Using serial sections, some H. pylori were found to express cagA. This latter observation was confirmed by dual fluorescence in which 16S rRNA was stained with Texas Red and cagA mRNA was stained with FITC (green) (Fig 5). Occasional individual bacteria were noted in collateral vessels (Fig 6), but there was no evidence of bacterial growth in non-atherosclerotic vessels. Animals were defined as H. pylori positive if H. pylori was visualized in either gastric or coronary artery sections following euthanasia. Iliac arteries were not assessed for H. pylori infection. Of the animals with evaluable data sets, 64% were H. pyloripositive. Thus, there were adequate numbers of infected and uninfected animals for statistical determination of the effects of H. pylori infection on atherosclerosis, arterial inflammation, and plasma lipids. All but seven of the 45 H. pylori-positive animals were infected with at least some cagA-positive strains, with the percentages of cagA-positive bacteria varying between 11% and 100% in a single artery section. Four animals were infected with only cagA-positive strains, while the remainder had mixed infections with both cagA-positive and cagA-negative strains, as seen in Fig 5. Of the H. pylori-positive animals, 64% had evidence of H. pylori infection in both the stomach and arteries and 29% had H. pylori only in the stomach. As expected, the density of H. pylori was highest in the gastric mucosa, but impressive numbers of H. pylori were found in the LCX arteries (up to 43 per 10 6 μm 3 ). When the densities of H. pylori in the coronary arteries were plotted against gastric mucosal densities, a substantial linear correlation (coefficient of determination of R 2 = 0.6747, p< 0.0001) was found (Fig 7). A correlation matrix of H. pylori 16S and cagA densities in the stomach and LCX coronary artery revealed correlations of r �0.73 for all pairings, further indicating that the densities of cagA-positive or-negative H. pylori in the artery are significantly associated (p< 0.0001) with densities in the stomach. In other words, live H. pylori were frequently present in coronary arteries of H. pylori-infected monkeys and the density of bacteria found in plaques was strongly related to colonization density in the stomach.

H. pylori infection influences severity of premenopausal and postmenopausal atherosclerosis
The influence of H. pylori infection (as determined in postmenopausal gastric/coronary tissue) was evident at both premenopausal and postmenopausal time points. Premenopausal atherosclerosis was assessed in the LCI at the time of ovariectomy and postmenopausal atherosclerosis was assessed in both the RCI and coronary arteries (LCX, LAD, RCA). Both the intimal area (IA) and maximum intimal thickness (MXIT) were strongly associated with H. pylori infection (p<0.005) even when diet was included in the statistical model (Table 1). Few H. pylori-negative animals had plaques of appreciable size at the premenopausal time point. Only 16% of H. pylori-negative animals had intimal areas greater than 0.5 mm 2 , whereas 55% of infected animals had premenopausal plaque areas above that value (p< 0.001 by Fisher's exact test). There was no significant association between H. pylori infection status and premenopausal estrogen receptor expression. Other premenopausal measurements shown in Table 1 will be discussed below.
As shown in Fig 1, following ovariectomy and for an additional 36 months, animals were either fed the same diet they had consumed premenopausally or switched to the other diet. MXIT and IA measurements were taken on the right common iliac artery and three coronary arteries (LCX, LAD, and RCA). Postmenopausal results are summarized in Table 2. Fig 8B and  8D show measurements of iliac artery plaques at the postmenopausal time point as compared with premenopausal values. Progression of iliac plaque area, as determined by subtracting premenopausal MXIT and IA from postmenopausal values in each animal, trended higher in H. pylori-infected animals, but progression of plaque thickness was unaffected by infection status.
The micrographs in Fig 9A-9C represent examples of extreme LCX artery pathology, which was seen in animals infected with cagA-positive H. pylori. Large necrotic regions and bulging plaques in the LCX were found in six of the cagA-positive animals and none of the cagA-negative or uninfected animals (Fisher's exact test p< 0.05). Plaques representative of median thickness values for each cagA-positive, cagA-negative, and uninfected macaques are shown in Fig 9D-9F. Note that these values represent the maximum thickness of the LCX slice in which H. pylori density was determined, not the maximum thickness of the entire artery section. Plaque measurements for the three coronary arteries were averaged and the differences between H. pylori-positive and-negative animals irrespective of diet group are shown in Fig   Fig 3. Identification of cagA-  Almost all H. pylori-positive animals that received a casein/lactalbumin diet at any point had moderate to severe atherosclerosis. Analyses of the effects of diet on atherosclerosis in these animals have been previously published [38,44], and therefore are only briefly mentioned in this report.
The outlier seen in Fig 10 represents a cagA-positive animal receiving the casein-lactalbumin diet for the duration of the study. There is no a priori basis for invalidating the measurements from this animal. Moreover, the results of analyses were similar whether the outlier was included or not, and all statistical analyses discussed herein are inclusive of all animals.

Effects of H. pylori on plaque size are due to local rather than systemic effects
We next sought to determine whether the effects of H. pylori on plaque measurements were due to local effects of bacteria within the plaques or due to systemic effects stemming from gastric infection. To this end, we performed multiple regression analysis and correlation analysis. As shown in Table 3, the density of cagA-positive H. pylori in the LCX coronary artery showed a stronger correlation to plaque area and thickness than did the density of cagA-positive bacteria in the stomach. On the other hand, the overall densities of H. pylori, as measured by 16S in the stomach or artery, did not correlate with plaque measurements (Table 3). Taken together,   these data suggest that the presence of cagA is critical for artery pathology and that bacteria are acting locally rather than systemically.

H. pylori infection influences arterial inflammatory markers
Additional measurements of inflammation and plaque physiology were performed on sections of the LCI artery at the premenopausal time point. Intimal area containing collagen was significantly increased in H. pylori-infected animals ( Fig 11A and Table 1), while mineralization was modestly increased (Fig 11B and Table 1). Macrophage populations were increased by more than two fold in H. pylori-infected monkeys as measured by both immunohistochemistry (HAM56; Table 1) and CD68 expression ( Fig 11C and Table 1). ICAM1, IL-6 and expression of the T cell markers CD3 and CD4 were significantly increased and expression of the cytokine MCP-1 tended to be higher in infected animals (Fig 11 and Table 1). Our findings indicate an increased inflammatory response due to H. pylori infection. Overall, the effects of   H. pylori infection were the opposite of the anti-inflammatory effects seen with exposure to the SOY treatment, which decreased macrophage and T cell infiltration, ICAM-1, MCP-1, and IL-6 in these animals [44].

H. pylori alters plasma lipid profiles
H. pylori has been linked to dyslipidemia in humans, as indicated by epidemiological studies and following measurement of plasma lipids before and after H. pylori eradication [57,58]. Similarly, we found that H. pylori infection in monkeys correlated with decreased HDLc, increased non-HDLc (low density and very low density lipoprotein-associated cholesterol), and increased TPC/HDLc (Castelli's ratio) (all p<0.05; Table 2). In contrast with plaque measurements, the density of H. pylori in the gastric mucosa (16S and CagA) correlated better with plasma lipids than did H. pylori density in the artery. Also, the presence of cagA did not appear to be an important contributor to dyslipidemia, as indicated by the stronger correlations between plasma lipids and stomach H. pylori 16S than plasma lipids and artery or stomach cagA. This suggests that either local effects of H. pylori in the stomach or systemic effects of infection are responsible for plasma lipid alterations. Body weight, total cholesterol, and triglycerides were not associated with H. pylori infection.

Known viral infections do not influence atherosclerosis in these animals
Human epidemiological studies have noted a relationship between atherosclerosis and viral infections [59]. To explore this possibility, monkeys were tested for Herpes B virus, simian immunodeficiency virus (SIV), measles, Rhesus cytomegalovirus (RhCMV), simian T cell leukemia virus (STLV), types 1 and 2, and simian type D retrovirus (SRV/D). None of the viral infections correlated with the atherosclerosis biomarkers tested in this study.

Subordinate social status is associated with increased H. pylori burden
Premenopausal female cynomolgus macaques of low social status are typically at greater risk of atherosclerosis than their dominant counterparts [41]. Could H. pylori infection influence this risk? We found that subordinate status was significantly associated with increased risk for H. pylori infection and increased infection density (p< 0.05 using either Fisher's exact test or Pearson's Chi-squared test with Yates' continuity correction). Only 48% of the dominant monkeys were infected with H. pylori as compared with 78% of the subordinate animals; however, rank did not influence either premenopausal or postmenopausal plaque measurements when H. pylori infection status was included in the predictive model. Therefore, H. pylori infection may be a more important contributor to atherosclerosis than social status or, alternatively, status may exert its effects in part through modified resistance to this bacterial infection.

Discussion
This work is the first study to examine the effects of H. pylori on development of atherosclerosis in non-human primates and the first to demonstrate that H. pylori within atherosclerotic plaques are alive. Natural infection with H. pylori has been documented in both rhesus (M. mulatta) and cynomolgus macaques [60][61][62], and rhesus macaques have long been used to study H. pylori gastric pathogenesis, evolution, and carcinogenesis [60,63,64]. However, the potential for natural infection by H. pylori to influence extragastric pathology in non-human primates had not been explored previously. Detection of H. pylori DNA by PCR in human atherosclerotic plaques was first reported by Farsak et al. in 2000 and has since been confirmed by a number of other investigators [26][27][28][29]; nonetheless, PCR of DNA does not prove the presence of live H. pylori within plaques. In situ hybridization has been routinely used to detect mRNA in paraffin-embedded tissue and is a sensitive and specific method for detecting H. pylori in gastric biopsies [65,66]. Our detection of 16S and cagA mRNA, combined with corresponding structures matching the size and morphology of H. pylori, confirm that live H. pylori inhabit atherosclerotic plaques. The presence of bacterial clusters further suggests that limited bacterial growth occurs within these same plaques. Seeding of heart valves by hematogenously spread oral flora is a well-documented risk and H. pylori within gastric and oral blood vessels has been previously documented [56,67], so perhaps it is not surprising that H. pylori entering the bloodstream can reach the heart and adhere. H. pylori requires host cholesterol for optimal membrane integrity [68,69] and the cholesterol-rich atherosclerotic plaques may encourage survival and growth of H. pylori. The distinction between living or dead H. pylori within plaques is critical because live H. pylori can inject bacterial components including the CagA toxin, peptidoglycan, DNA, and metabolites of lipopolysaccharide into the host cell cytoplasm via the type IV secretion apparatus [70,71]; dead H. pylori cannot influence host cell signaling in this manner. Therefore, in terms of atherosclerosis H. pylori is likely to directly activate inflammatory pathways within the atheroma in addition to acting indirectly via distal activity at the site of gastric colonization.
H. pylori significantly worsened several plaque parameters, including common iliac artery intimal thickness and area in the same animals evaluated pre-and postmenopausally. Intimal collagen density and mineralization were also affected. Although the majority of macaques were infected with a mixture of cagA-positive and cagA-negative H. pylori strains, we were able to examine the contributions of cagA by determining whether LCX artery measurements were more greatly influenced by total bacterial density (as determined by 16S) or by cagA-positive bacterial density. That analysis revealed that cagA-positive strains were more atherogenic than cagA-negative strains. Moreover, H. pylori densities found within the arteries had a greater impact than bacterial density in the stomach. Though not conclusive, these data suggest that plaque development is due to local effects of cagA-positive H. pylori. Notably, these findings are consistent with epidemiological studies conducted in humans. While early epidemiological studies investigating the role of H. pylori in cardiovascular diseases yielded mixed results [17][18][19][20], recent studies that differentiate between cagA-positive vs. cagA-negative strains have yielded more consistent positive associations [24,[72][73][74]. Interestingly, Franceschi et al reported cross-reactivity between CagA-specific antibodies and artery tissue [75]. This immunoreactivity was assumed to be due to molecular mimicry, yet the authors did not determine whether antibodies specific for other H. pylori proteins reacted with the tissue or prove that the "cross-reactive" antigen was of human origin. In light of the evidence shown here, it is likely that CagA itself was being detected.
Another intriguing finding of this study was that the densities of H. pylori in the gastric mucosa were highly correlated with the densities in the coronary artery plaques. Gastric colonization density is known to vary among individuals and is believed to be related to both the H. pylori strain and the type of immune response generated by the host [76][77][78]. We could not determine whether colonization density varied over time, but the high correlation between postmenopausal bacterial density and both pre-and postmenopausal plaque measurements may suggest that colonization density is stable. Additional studies will be needed to determine how H. pylori densities in both locations change over time and whether gastric biopsies could be used to estimate the risk of severe atherosclerosis.
In addition to affecting plaque measurements, H. pylori infection was associated with increases in some arterial inflammatory markers, providing a possible mechanism for H. pylori-mediated atherosclerosis exacerbation. VCAM-1, ICAM-1, and MCP-1 all contribute to recruitment of macrophages to plaques [79,80] and all tended to be increased in H. pyloriinfected macaques. IL-6, which induces ICAM-1 [81], was also elevated by H. pylori. Our findings are consistent with in vitro induction of P-selectin, E-selectin, VCAM-1, and ICAM-1 and enhanced transendothelial migration of activated CD4 + and CD8 + T cells in human umbilical vein endothelial cell cultures following H. pylori exposure [82,83]. We similarly found elevated plaque CD4 + T cells, which are pro-atherogenic [84], in H. pylori-positive animals. Others have reported that H. pylori induces MCP-1, IL-6, and ICAM-1 in various cell types [85][86][87][88] and endothelial ICAM1 expression strongly correlates with the presence of H. pylori DNA in human atherosclerotic plaques [89]. Furthermore, Figura et al. report that acute coronary artery disease patients infected with cagA-positive H. pylori have higher circulating IL-6 levels than uninfected patients [90]. Our results confirm and expand upon previous results, since we have established that live H. pylori within atherosclerotic lesions can directly interact with macrophages and/or endothelial cells to induce cytokine and adhesion molecule production.
Clinical studies have also reported decreased HDL and increased LDL in association with H. pylori, and some studies have found reversal of dyslipidemia following H. pylori eradication [57,58,91], although not all studies have found such a link [92]. One Japanese study found H. pylori-associated dyslipidemia in male, but not female subjects [93]; we clearly saw dyslipidemia in female, postmenopausal macaques. The varied findings in human studies likely result from the considerable differences among humans in terms of diet, lifestyle, genetics, and many other factors, many of which were controlled for in our monkey study. Some studies have linked dyslipidemia with atrophic gastritis in H. pylori-positive patients [94,95]. This is consistent with our observation that plasma lipid changes were more closely related to gastric H. pylori densities than arterial densities.
Atherosclerosis development is delayed in women compared with men and estrogens likely mediate this effect based on studies showing that premenopausal oophorectomy or premature ovarian failure are associated with increased coronary heart disease risk, an effect that can be mitigated by treatment with exogenous estrogen in younger, but not older women [96]. Soy phytoestrogens improve cardiovascular disease biomarkers in some human studies involving men and post-menopausal women [97]. This may be in part due to the effects of soy on ER-α and ER-β, which are expressed in arteries [48]. H. pylori increased atheroma measurements both pre-and post-menopausally and generally counteracted the beneficial effects of SOY on inflammatory markers in cynomolgus macaques (reported by Walker et al [48]); however, unlike the SOY diet, H. pylori did not significantly alter expression of ER-α or ER-β. Therefore, the increased risk associated with H. pylori infection seems unaffected by estrogen.
The differing H. pylori burdens among dominant and subordinate animals have not been previously reported. Female subordinate macaques are under considerably more stress than dominant females and typically develop more atherosclerosis than their dominant counterparts when consuming an atherogenic diet [41], an effect that co-occurs with substantial ovarian disruption [96]. In humans, lower socioeconomic status positively correlates with H. pylori prevalence, presumably due to poor hygiene and overcrowding [98]. Although human socioeconomic status is not homologous to monkey dominance relationships [40], subordinate female macaques are directly stressed as a result of being physically harassed by their dominant counterparts during competitive interactions over food, space, and preferred grooming partners [99]. Moreover, in the small group context used in the study reported here, dominant animals demonstrably outcompete subordinates for fresh food and water and desirable perching locations, potentially exposing subordinates to more fecal contamination as they consume food and water that drop or spill to the ground. Such contamination, in turn, is one proposed transmission route for H. pylori [100]. Of note, several large epidemiological studies report an increased risk of cardiovascular disease in humans with low socioeconomic status. Although, again, human socioeconomic status differs conceptually from monkey dominance status [40], the increased H. pylori burden in humans with low socioeconomic status could contribute to higher atherosclerosis risk in this group.
The primary weakness of this study is that we are not certain when the animals became infected or whether bacterial prevalence changed over time. Humans most often become colonized during childhood, but transmission can occur later in life, particularly between spouses [101][102][103]. In rhesus macaques, H. pylori transmission occurs within 90 days via social contact, rather than fomites, but some animals remain uninfected, possibly due to host genetic factors [104,105]. H. pylori infection persists for at least a decade in Rhesus macaques [63] indicating that infection is as persistent in non-human primates as it is in humans. If the animals in this study were not previously infected, then transmission likely occurred soon after assignment to social groups. The substantial correlations between postmenopausal H. pylori status and premenopausal plaque parameters further indicate that animals were infected before the study midpoint.
For the first time, this study conclusively demonstrated that live H. pylori can reside within atherosclerotic plaques and that H. pylori is likely to have local effects on plaque size. Furthermore, H. pylori influenced macrophage infiltration, collagen deposition, ICAM1 expression, and plasma lipids. The effects of H. pylori on plaque parameters appeared to be primarily due to local effects of cagA-positive H. pylori, whereas changes in plasma lipids related to gastric infection by either cagA-positive or-negative strains. H. pylori status was relevant during both the premenopausal and postmenopausal phases as demonstrated by correlations between H. pylori infection and iliac artery plaque parameters measured at both ovariectomy and necropsy. Given the many similarities in atherogenesis between humans and non-human primates, future inclusion of cagA-positive H. pylori status in human epidemiological and clinical atherosclerosis studies is warranted.
Supporting information S1