Leptin Regulates Neointima Formation After Arterial Injury Through Mechanisms Independent of Blood Pressure and the Leptin Receptor/STAT3 Signaling Pathways Involved in Energy Balance
Background— Leptin is an adipocyte-derived hormone critical for energy homeostasis and implicated in vascular disease processes. The relevant cellular leptin receptor pools and signaling pathways involved in leptin-related vascular phenotypes in vivo are unclear.
Methods and Results— Arterial injury was induced in wild-type (wt), leptin-deficient (lepob/ob), and leptin receptor–deficient (leprdb/db) mice. Compared with wt mice, lepob/ob and leprdb/db mice were protected from the development of neointima. Bone marrow transplantation experiments between wt and leprdb/db mice indicated that the vascular protection in leprdb/db mice was not attributable to lack of leptin receptor expression on bone marrow–derived elements. To investigate the role of the lepr-mediated signal transducer and activator of transcription 3 (STAT3) signaling pathway in the response to vascular injury, leprs/s mice homozygous for a leptin receptor defective in STAT3 signaling underwent femoral arterial injury. Despite similar obesity and blood pressure levels, the neointimal area in leprs/s mice was significantly increased compared with leprdb/db mice.
Conclusions— The molecular mechanism by which the leptin receptor mediates neointima formation and vascular smooth muscle cell proliferation is largely independent of the STAT3-dependent signaling pathways involved in energy balance.
Leptin is an adipocyte-derived hormone initially described for its profound effects on energy balance,1 which are largely mediated via leptin receptors expressed on cells in the hypothalamus.2 Leptin receptors have also been shown to be expressed on other cell types that may affect vascular disease processes including white blood cells,3 platelets,4 endothelial cells,5 and vascular smooth muscle cells (VSMCs).6 These receptor pools are likely to be biologically relevant as leptin has been demonstrated to promote angiogenesis,7 thrombosis,8–10 atherosclerosis,9 and neointimal hyperplasia11,12 in animal models. However, the relevant leptin receptor pools responsible for these vascular effects are unclear.
The interaction of leptin at the leptin receptor results in the phosphorylation of critical tyrosine residues13 and subsequent activation of additional intracellular signaling. Several pathways have been implicated in post-leptin receptor signaling events toward appetite and metabolism including PI3 kinase,14 AMP kinase,15 and signal transducer and activator of transcription 3 (STAT3).16 Tyr1138 of the leptin receptor mediates activation of STAT3. Substitution of Tyr1138 with serine blocks STAT3 activation in cultured cells and in mice in which the leptin receptor is replaced by this mutant (leprs/s).16 The obesity of these leprs/s mice closely resembles that of leptin receptor–deficient (leprdb/db) mice.16 However, the role of this pathway in mediating leptin receptor responses on vascular phenotypes in vivo is unknown.
This report examines the effects of leptin and leptin deficiency on neointimal hyperplasia after vascular injury. We demonstrate that the in vivo effect of leptin is not mediated by bone marrow–derived elements, is at least partially independent of leptin receptor-mediated STAT3 signaling, and is associated with the proliferation of vascular smooth muscle cells.
Wild-type (wt), leptin-deficient (lepob/ob), and leptin receptor–deficient (leprdb/db) mice were purchased from the Jackson Laboratory (Bar Harbor, Me). Targeted leptin receptor mutant mice with failure to activate STAT3 signaling pathways (leprs/s) were generated as previously described.16 Leprs/s mice had been backcrossed to the C57BL/6J strain ≥7 generations. Male animals, between 8 and 11 weeks of age, were fed a standard laboratory rodent diet (#5001, TestDiet, Richmond, Ind) in specific pathogen-free facilities. All procedures complied with the Principles of Laboratory and Animal Care established by the National Society for Medical Research and were approved by the University of Michigan Committee on Use and Care of Animals.
Femoral Artery Injury
Mouse femoral artery injury was performed using a model of wire-induced vascular injury as described previously.17 This model causes endothelial damage at the site of injury followed by reproducible intimal hyperplasia. Mice were anesthetized with intraperitoneal sodium pentobarbital (120 mg/kg; Butler Co), except for mice that received a BMT, which were anesthetized with inhaled isofluorane (Baxter Healthare Corporation) using a Veterinary Anesthesia and Monitoring device (SurgiVet Inc). Mice were then secured in the supine position and placed under a dissecting microscope (Nikon SMZ-2T, Mager Scientific Inc). After a midline incision was made on the ventral left thigh region, the femoral nerve was isolated and pulled aside. The distal and proximal ends of the femoral artery were held with surgical suture for temporary control of blood flow. A small arterial branch superior to the muscular branch of the femoral artery was cauterized (DELC Cauterizer, Aaron Medical Industries Inc). One drop of 1% lidocaine was used to prevent arterial spasm. An arteriotomy of the femoral artery muscular branch was then performed. A straight guide wire (0.38 mm diameter, No. C-SF-15-15, Cook) was introduced 1 cm into the femoral artery through this muscular branch and remained in the artery for 1 minute. After the wire was removed, the muscular branch was ligated with 7.0 silk suture. Blood flow through the femoral artery was restored by releasing the surgical thread. Surgical clips were used to close the incision site.
Recombinant murine leptin (R&D Systems) was delivered to mice via intraperitoneal (IP) injection (5.0 μg/g body weight). Adenovirus expressing murine leptin on the RSV promoter (ad-leptin; gift of Dr Savio Woo, Mount Sinai, New York)18 was delivered to mice via tail vein injection (1.25×1011 pfu/mouse). A control empty adenoviral vector was delivered to lepob/ob via the same conditions.
Bone Marrow Transplantation
At 10 weeks of age, wt and leprdb/db mice were irradiated and bone marrow transplantation (BMT) was performed as previously described.8 We have previously shown that this protocol leads to complete engraftment.8 Femoral artery injury was performed six weeks following transplantation.
Histological and Morphometric Analyses
At the end of the protocol, mice were injected with bromodeoxyuridine (BrdU; Zymed Laboratories Inc) (IP, 30 μg/g body weight) 12 and 2 hours before sacrifice. The mice were perfused with saline and fixed using formalin with a 25 gauge needle, inserted into the left ventricle, at a rate of 1 mL/min. The injured femoral artery was excised, embedded in paraffin blocks, sectioned (5um), and stained for elastin using the Accustain Elastic Stain kit (Sigma–Aldrich). For quantitation, 4 cross sections taken every 2 mm were selected from each artery. The images were analyzed using Image-Pro Plus software (Media Cybernetics). Intimal (I) and medial (M) areas were measured and the ratio of I/M was calculated. Immunohistochemistry was performed as described previously.19 Smooth muscle cells were detected by smooth muscle α-actin staining using an α-actin monoclonal antibody (1:100, Cedarlane Laboratories Limited). Cellular proliferation was analyzed with BrdU Staining Kit (Zymed Laboratories Inc) and the number of BrdU-positive cells were counted manually and expressed as BrdU-positive nuclei within the external elastic lamina per cross section.
Body Weight, Metabolic Parameters, and Blood Pressure
Mouse body weight was recorded weekly after arterial injury. At the time of euthanasia, blood was obtained by ventricular puncture for plasma separation. Plasma concentration of cholesterol, triglycerides, glucose (Sigma-Aldrich), and insulin (Crystal Chemical Inc) were determined at the time of sacrifice. Plasma leptin levels were determined weekly after arterial injury using a murine leptin ELISA kit (Crystal Chemical Inc). Arterial blood pressure was determined in 8- to 10-week-old wt, leprdb/db, and leprs/s mice using a radiotelemetry system.20 Mice of each genotype were instrumented with radiotelemetric monitoring devices, which allowed continuous nonrestrained monitoring of blood pressure. A transmitter catheter (Data Sciences) was inserted into the left carotid artery when the mouse was under isoflurane anesthesia, and the blood pressure transmitter body (Data Science, PA-C20) was routed to the abdominal cavity. After recovery, the mouse was housed in an individual cage and provided with regular mouse chow and tap water ad libitum. Mean arterial pressure was recorded at 5-minute intervals for at least 2 weeks duration. After data collection, the 5 minute sampling periods were averaged over 5 consecutive days (once diurnal rhythm was established). The 5 minute averages were then furthered condensed into 24 separate 1 hour values for each hour during the entire light-dark cycle for the purpose of graphical presentation and statistical analysis.
VSMC were isolated from 7 to 14 week-old wt, leprdb/db, and leprs/s mice (n=3 each group) as previously described.21 Cells were grown in Dulbecco modified Eagle medium (DMEM, Gibco Inc) containing 10% fetal bovine serum (FBS, Gibco Inc) and 1% penicillin/streptomycin, and passaged as described previously.22 For experiments, cells between passages 4 and 9 were made quiescent by incubation with serum-free DMEM for 30 hours before use. The VSMC proliferation assay6 and quantification of cell number by crystal violet staining and optical density reading23 were performed as previously described. Briefly, cells were seeded in 96-well plate in DMEM containing 0.5% FBS at a density of 4×103/well and allowed to attach overnight. Medium was then replaced with DMEM containing 2% FBS and Insulin-Transferrin-Selenium premix (Gibco Inc) in the presence of either leptin (100 ng/mL; R&D Systems) or vehicle. After 3 days of incubation, cells were fixed with 10% glutaraldehyde, stained with 0.1% crystal violet solution, and dissolved with 10% acetic acid. The wells were read at 590 nm in a microplate reader (VersaMax, Molecular Devices Corp). All assays were performed in triplicate wells and 2 to 3 replications of the experiment were performed.
Values are expressed as mean±SEM. The statistical significance of differences between groups was determined by 1-way ANOVA followed by Dunn post-hoc analysis when more than 2 experimental groups were included. The student 2-tailed t test was performed when only 2 groups were being compared. For BMT and blood pressure measurements, a 2-way analysis of variance was performed followed by the Tukey HSD test to examine significant main effects. P<0.05 was considered significant.
Effect of Leptin and Leptin Receptor on Neointimal Formation
To determine the influence of leptin and the leptin receptor on neointimal formation, femoral injury was induced in wt, lepob/ob, and leprdb/db mice. Four weeks after injury, the average femoral artery neointimal area and ratio of intima/media in lepob/ob and leprdb/db mice was significantly reduced compared with wt mice (Figure 1 and Table 1). There were no differences in the medial area between wt, lepob/ob, and leprdb/db. The neointimal area was comprised predominantly of VSMCs (Figure 1). The VSMC cellular proliferation index was significantly reduced in lepob/ob and leprdb/db compared with wt mice (Table 1).
Effect of Exogenous Leptin on Neointimal Formation
To further examine whether the reduced neointimal formation in lepob/ob mice was attributable to the lack of leptin, recombinant murine leptin was given to lepob/ob mice beginning 1 day before injury on a daily basis until sacrifice. Leptin levels 12 hours after injection were 45.7±5.8 ng/mL compared with basal levels of 1.0±0.1 ng/mL in wt mice. The average neointima of lepob/ob mice receiving recombinant leptin was significantly higher than lepob/ob mice receiving no treatment (Figure 1). These results confirm that leptin plays a causal role in neointimal growth.
To test another leptin treatment strategy, lepob/ob and leprdb/db mice were infected with ad-leptin or control adenovirus vector. Because leptin may affect thrombosis after vascular injury,8,10 which could in turn affect neointimal formation, ad-leptin was injected 1 day after injury to reduce potential effects on acute thrombosis. Leptin levels of lepob/ob mice treated with ad-leptin reached a peak of 230.3±58.9 ng/mL at one week after ad-leptin injection and gradually declined to 84.0±33.2 ng/mL at 2 weeks, 8.8±1.6 ng/mL at 3 weeks and 2.1±0.6 ng/mL at 4 weeks. The neointima of lepob/ob mice receiving ad-leptin was significantly higher compared with that of lepob/ob mice receiving no treatment (Figure 1). The neointima of lepob/ob mice receiving a control adenovirus was not different from lepob/ob mice receiving no treatment (data not shown). This data supports the effect of leptin on neointimal formation and also demonstrates that leptin is not required at the time of injury to promote neointimal formation. The neointima of leprdb/db mice receiving ad-leptin was also examined to confirm the effects of the leptin receptor on neointimal formation and to serve as a control for nonspecific inflammatory effects of the adenovirus. The neointima of leprdb/db mice treated with ad-leptin was not different from the leprdb/db mice receiving no treatment (Figure 1).
Effect of Leptin Treatment on Body Weight and Metabolic Parameters
Leptin treatment improved many metabolic parameters in lepob/ob mice (Table 2) providing evidence of an improvement in the concentration of many classic risk factors. Despite these salient changes, there was a robust leptin-mediated enhancement of neointimal growth.
Effect of Bone Marrow Transplantation on Neointima Formation
To examine whether the leptin receptor–deficient phenotype was attributable to the lack of leptin receptor on circulating blood elements, BMT was performed from leprdb/db to wt, wt to wt, wt to leprdb/db, and leprdb/db to leprdb/db mice. Control mice not receiving bone marrow after irradiation died 7 days later, demonstrating effective bone marrow ablation after the irradiation. Six weeks after BMT, femoral arterial injury was performed to induce intimal hyperplasia. The neointima of wt mice receiving bone marrow from leprdb/db mice was significantly greater than that of leprdb/db mice receiving bone marrow from leprdb/db mice (Figure 2). Similarly, the neointima of leprdb/db mice receiving wt bone marrow was significantly less than the neointima observed in wt mice receiving bone marrow from wt mice (Figure 2).
Effect of LepR-Mediated STAT3 Signaling on Neointima Formation
To determine the effect of leptin receptor-STAT3 signaling on neointimal formation, leprs/s mice (n=9), underwent femoral arterial injury. Despite development of an obese phenotype similar to that observed in leprdb/db mice (Table 2),16 leprs/s mice developed greater intimal hyperplasia than leprdb/db mice (P<0.02; Figure 3a through 3c).
Effect of LepR-Mediated STAT3 Signaling on Arterial Blood Pressure
To examine whether the neointimal formation of the leprs/s was related to arterial blood pressure we measured the mean arterial pressure of wt, leprdb/db, and leprs/s mice. Consistent with previous studies using radiotelemetry, we detected a reduced arterial pressure in leprdb/db compared with wild-type mice.24 Similarly, the leprs/s were observed to have reduced blood pressure compared with wild-type (Figure 4). No difference in blood pressure was observed between leprs/s and leprdb/db mice suggesting that the leptin-mediated effects on blood pressure are mediated by the Lepr-STAT3 pathway.
Effect of Leptin on SMC Proliferation
To further elucidate relevant signaling pathways, VSMCs were isolated from wt, leprs/s, and leprdb/db mice. Similar to previous studies,6 leptin induced a significant increase in cellular proliferation in wt cells compared with vehicle (leptin=0.40±0.006 versus vehicle=0.37±0.005; P<0.03). As previously shown,6 this leptin effect was completely blocked with the phosphoinositide 3 kinase inhibitor, wortmannin (leptin + wort=0.37±0.006; P<0.02 compared with leptin + vehicle). VSMCs from leprs/s mice also responded to leptin with an increase in proliferation compared with vehicle control (leptin=0.67±0.01 versus vehicle=0.49±0.02, P<0.00001), and this effect was blocked with wortmannin (lep+wort=0.53±0.001, P<0.00003). VSMCs from leprdb/db mice did not respond to leptin treatment.
Leptin is a 16-kDa peptide hormone with structural features similar to those of the long-chain helical cytokine family.1 The expression of receptors for leptin are ubiquitous,25 indicating that leptin may affect multiple physiological processes. To determine the importance of leptin and the leptin receptor on neointimal hyperplasia, we performed femoral wire injury on wt, lepob/ob, and leprdb/db mice. Consistent with previous studies, lepob/ob and leprdb/db mice were protected from neointimal formation.11,12 Similar to the findings of Schafer et al,11 who used an iron chloride model of carotid injury, we have shown that leptin treatment (recombinant murine leptin or ad-leptin) increases neointima in the leptin deficient mice but not in mice lacking the leptin receptor, confirming a causal role for leptin and the leptin receptor in neointimal formation. In addition, cell proliferation analyses revealed that lepob/ob and leprdb/db mice had significantly less cellular proliferation in the femoral arterial wall after injury compared with wt mice. This suggests that leptin may augment VSMC proliferation after arterial injury. Previous studies have demonstrated leptin receptor expression by VSMCs after arterial injury, supporting a direct effect of leptin on vascular smooth muscle cells.11,12
In addition to a potential role of the VSMCs, we sought to directly examine the importance of bone marrow–derived leptin receptor in leptin-mediated effects on neointimal hyperplasia. For these studies, we performed femoral wire injury after BMT from leprdb/db to wt and from wt to leprdb/db mice. Interestingly, the recipient phenotype was unchanged in both of these experiments compared with control groups receiving marrow from genetically identical donors. This suggests that the in vivo effect of the platelet and leukocyte leptin receptor is negligible toward the response to vascular injury.
Similar to other cytokine receptors, the leptin receptor signals by activating a Jak family tyrosine kinase (Jak2) that autophosphorylates numerous tyrosine residues and phosphorylates tyrosine residues (Tyr985 and Tyr1138) on the leptin receptor during leptin stimulation. Leptin can therefore signal through activation of several different receptor-mediated pathways.13 The relevance of these pathways toward energy balance has been established by the analysis of mice in which the leptin receptor gene has been substituted by a mutant leptin receptor (leprS1138) containing a Tyr1138-Ser substitution that is defective in STAT3 binding.16 These mice develop obesity similar to that observed in leprdb/db mice, although other phenotypes associated with leptin receptor deficiency are less affected indicating the relevance of other signaling pathways such as the IRS-protein-PI3 kinase pathway.13 The pathways responsible for the vascular effects of leptin are unknown, although both STAT3 and PI3K lepr signaling pathways have been implicated in VSMC proliferation in vitro.26,27 To dissect these pathways in vivo, leprs/s mice with defective STAT3 signaling were studied in the femoral arterial injury model. These mice developed an obese phenotype similar to that observed in the leprdb/db group, as previously described,16 but they exhibited significantly greater intimal hyperplasia after arterial injury, suggesting that leptin receptor–mediated STAT3 signaling is not responsible for the leptin effect on neointimal formation after injury.
The robust effects of exogenous leptin in these experiments could be partially mediated by effects on arterial blood pressure. Leptin has been shown to influence sympathetic activation with elevations in blood pressure.24,28,29 Furthermore, this effect on sympathetic activity appears to be centrally mediated and may not be subject to the same degree of leptin resistance as observed with leptin-receptor related feeding behavior.30 To examine whether the in vivo effects observed in the leprs/s mice were mediated by elevations in blood pressure, we performed radiotelemetry arterial blood pressure recordings. Consistent with previous publications we observed reduced blood pressure in leprdb/db mice.24 In addition, the blood pressure of the leprs/s mice was also lower than wt and was not different than leprdb/db mice. Therefore, although high doses of exogenous leptin probably elevated blood pressure at some time points in our exogenous leptin experiments, we have demonstrated that elevations in blood pressure are not mediating the increased neointimal response of the leprs/s mice as compared with the leprdb/db mice.
To further examine the potential underlying mechanism of the increased neointimal response of the leprs/s compared with the leprdb/db, we used a cell culture system to examine VSMC proliferation. Leptin has previously been shown to stimulate proliferation of VSMCs in culture6 and our in vivo studies demonstrated reduced intimal VSMC proliferation in leptin deficient and leptin receptor deficient mice. Therefore, we examined whether the leprdb/db and leprs/s mice responded differently to recombinant leptin in vitro. VSMCs isolated and cultured from leprdb/db did not respond to leptin treatment whereas leptin significantly stimulated the proliferation of VSMCs isolated from leprs/s mice, consistent with the in vivo phenotype. Furthermore, this leptin effect on leprs/s VSMC’s was completely blocked after incubation of the VSMCs with a PI3 kinase inhibitor, implicating this pathway in leptin receptor-mediated VSMC proliferation.
Limitations of this study include the supraphysiologic plasma concentrations of leptin obtained in the leptin replacement experiments. These experiments serve as proof of principle that leptin affects neointimal hyperplasia. The increased neointimal hyperplasia observed in the leprs/s mice compared with the leprdb/db mice, where no exogenous leptin was added, demonstrates the potential of physiological elevations of leptin, as observed in obese states, to affect this vascular end point. An additional limitation of this study is lack of definitive proof of the relevant leptin receptor cellular pool. Although we have demonstrated that leptin effects on neointimal hyperplasia are at least partially independent of Lepr-STAT3 signaling, not dependent on blood pressure elevation, and independent of bone marrow–derived elements, it remains possible that other Lepr-mediated central nervous system pathways could influence vascular disease.
In conclusion, these data suggest that leptin enhances neointimal formation through an interaction with the leptin receptor resulting in smooth muscle cell proliferation. These effects are mediated by a leptin receptor signaling pathway that is distinct from the major pathway involved in energy balance. These findings suggest that it may be possible to design therapeutic agents capable of promoting the beneficial metabolic activities of leptin while avoiding the potential hazard of vascular complications.
Sources of Funding
This work was supported by grants HL57346 and AHA EIG (to D.T.E.), American Diabetes Association Junior Faculty Award (to P.F.B.), and DK56731 and DK57768 (to M.G.M.).
Y.S. and P.F.B. contributed equally to this study.
Original received September 14, 2006; final version accepted October 31, 2006.
Chua SC, Jr., Chung WK, Wu-Peng XS, Zhang Y, Liu SM, Tartaglia L, Leibel RL. Phenotypes of mouse diabetes and rat fatty due to mutations in the OB (leptin) receptor. Science. 1996; 271: 994–996.
Zarkesh-Esfahani H, Pockley G, Metcalfe RA, Bidlingmaier M, Wu Z, Ajami A, Weetman AP, Strasburger CJ, Ross RJ. High-dose leptin activates human leukocytes via receptor expression on monocytes. J Immunol. 2001; 167: 4593–4599.
Nakata M, Yada T, Soejima N, Maruyama I. Leptin promotes aggregation of human platelets via the long form of its receptor. Diabetes. 1999; 48: 426–429.
Bouloumie A, Drexler HC, Lafontan M, Busse R. Leptin, the product of Ob gene, promotes angiogenesis. Circ Res. 1998; 83: 1059–1066.
Bodary PF, Gu S, Shen Y, Hasty AH, Buckler JM, Eitzman DT. Recombinant leptin promotes atherosclerosis and thrombosis in apolipoprotein E-deficient mice. Arterioscler Thromb Vasc Biol. 2005; 25: 1634.
Schafer K, Halle M, Goeschen C, Dellas C, Pynn M, Loskutoff DJ, Konstantinides S. Leptin promotes vascular remodeling and neointimal growth in mice. Arterioscler Thromb Vasc Biol. 2004; 24: 112–117.
Stephenson K, Tunstead J, Tsai A, Gordon R, Henderson S, Dansky HM. Neointimal formation after endovascular arterial injury is markedly attenuated in db/db mice. Arterioscler Thromb Vasc Biol. 2003; 23: 2027–2033.
Sata M, Maejima Y, Adachi F, Fukino K, Saiura A, Sugiura S, Aoyagi T, Imai Y, Kurihara H, Kimura K, Omata M, Makuuchi M, Hirata Y, Nagai R. A mouse model of vascular injury that induces rapid onset of medial cell apoptosis followed by reproducible neointimal hyperplasia. J Mol Cell Cardiol. 2000; 32: 2097–2104.
Muzzin P, Eisensmith RC, Copeland KC, Woo SL. Correction of obesity and diabetes in genetically obese mice by leptin gene therapy. Proc Natl Acad Sci U S A. 1996; 93: 14804–14808.
Sjoland H, Eitzman DT, Gordon D, Westrick R, Nabel EG, Ginsburg D. Atherosclerosis progression in LDL receptor-deficient and apolipoprotein E-deficient mice is independent of genetic alterations in plasminogen activator inhibitor-1. Arterioscler Thromb Vasc Biol. 2000; 20: 846–852.
Whitesall SE, Hoff JB, Vollmer AP, D’Alecy LG. Comparison of simultaneous measurement of mouse systolic arterial blood pressure by radiotelemetry and tail-cuff methods. Am J Physiol Heart Circ Physiol. 2004; 286: H2408–H2415.
Takahashi A, Taniguchi T, Ishikawa Y, Yokoyama M. Tranilast inhibits vascular smooth muscle cell growth and intimal hyperplasia by induction of p21(waf1/cip1/sdi1) and p53. Circ Res. 1999; 84: 543–550.
Bernal-Mizrachi C, Weng S, Li B, Nolte LA, Feng C, Coleman T, Holloszy JO, Semenkovich CF. Respiratory uncoupling lowers blood pressure through a leptin-dependent mechanism in genetically obese mice. Arterioscler Thromb Vasc Biol. 2002; 22: 961–968.
Rahmouni K, Haynes G. Leptin and the central neural mechanisms of obesity hypertension. Drugs Today (Barc). 2002; 38: 807–817.