From the Department of Ophthalmology, Bascom Palmer Eye Institute,University of Miami School of Medicine, Miami, Fla (Drs Cousins and Espinosa-Heidmann);and the National Eye Institute, National Institutes of Health, Bethesda, Md(Dr Csaky).
To evaluate the activation state of macrophage function in patientswith age-related macular degeneration (AMD) by quantifying the productionof the proinflammatory and angiogenic factor tumor necrosis factor α(TNF-α) and by correlating its expression with dry and wet AMD.
Circulating monocytes were obtained from the blood of patients withAMD or age-matched control subjects by gradient centrifugation. The monocyteswere then analyzed for either TNF-α release from cultured macrophagesin response to retinal pigment epithelium–derived blebs and cytokinesor TNF-α messenger RNA content by reverse transcriptase–polymerasechain reaction.
In human monocytes obtained from controls and AMD patients, TNF-αwas expressed by freshly isolated monocytes and produced by macrophages inculture after stimulation with retinal pigment epithelium–derived blebs.However, wide variability in TNF-α expression was observed among differentpatients. Patients with monocytes that expressed the greatest amount of TNF-αdemonstrated higher prevalence of choroidal neovascularization.
Both controls and AMD patients vary in the activation state (definedas TNF-α expression) of circulating monocytes. Partially active monocytes,defined as high TNF-α expression, may be a biomarker to identify patientsat risk for formation of choroidal neovascularization.
Early diagnostic testing may prove useful to detect those patients whowill progress to the more severe complications of the disease.
The concept that cells of the monocyte lineage play a role in both dryand wet age-related macular degeneration (AMD) has emerged as a major mechanismin AMD pathogenesis.1- 5 Monocytesand macrophages belong to the family of circulating bone marrow–derivedmononuclear leukocytes. In dry AMD, macrophages and dendritic cells have beenobserved to insert processes into drusen or other deposits. Human and animalmodel studies also demonstrate that macrophages are numerous in choroidalneovascularization (CNV) and that these cells express high levels of variousinflammatory and angiogenic cytokines.5 A reasonableassumption is that these macrophages (or other monocyte lineage cells) arerecruited to the choroid from circulating monocytes in the blood, althoughtheir function (harmful or beneficial) remains unknown.
The cytokine tumor necrosis factor α (TNF-α) is a multifunctionalprotein synthesized by macrophages that mediates many of the inflammatoryand tissue destructive functions of these cells, including angiogenesis.6- 10 Tumornecrosis factor α induces proangiogenic activity in many cell types(including fibroblasts, tumor lines, and others) and can induce angiogenesisin many standard assays.11,12 Tumornecrosis factor α receptors are expressed on many cell types in theretina, including retinal pigment epithelium (RPE), Müller glial cells,and choroidal vascular cells, and TNF-α stimulation of the RPE inducesmany proangiogenic responses.13- 15 Furthermore,TNF-α–deficient mice demonstrate reduced angiogenesis in a modelof hypoxia-induced retinal neovascularization.16 Onthe basis of these observations, we sought to characterize TNF-α productionby blood monocytes isolated from AMD patients and to correlate its expressionwith dry and wet AMD. Our results show that patients and control subjectsdemonstrate a wide range of expression of TNF-α cytokine in cultureor messenger RNA (mRNA) in isolated monocytes. Furthermore, high levels ofTNF-α mRNA correlated with a 5-fold risk of neovascular AMD. Analysisof monocyte TNF-α may serve as a biomarker for risk of CNV formation.
The study was conducted with the prior approval of the University ofMiami School of Medicine (Miami, Fla) Institutional Review Board. Ninety-ninepatients and subjects aged 21 to 90 years who were examined at the BascomPalmer Eye Institute of the Palm Beaches, Palm Beach Gardens, Fla, were enrolledin the study. Patients underwent a complete ophthalmic examination, includingfundus photography and fluorescein angiography if appropriate. Exclusion criteriaincluded a history of human immunodeficiency virus infection, treatment formalignancy, recent acute illness that required hospitalization within 6 months,or other known immunologic condition.
Patients were defined as having dry AMD on basis of the presence ofat least 5 drusen larger than 125 µm with or without hyperpigmentaryor hypopigmentary changes. Patients were classified as having neovascularAMD on the basis of standard findings by clinical examination and fluoresceinangiography. No distinction was made about the type or stage of CNV (ie, classic,occult, retinal angiomatous proliferation, or disciform scar formation). Patientsmay have undergone treatment for their CNV. Controls included patients whowere seen for routine eye examination with or without other ocular disordersbut without evidence of drusen, pigmentary changes, or CNV.
After informed consent was obtained, 50 mL of blood was obtained byphlebotomy with EDTA anticoagulation, and samples were immediately placedon ice. Patients were interviewed about their use of vitamins and smokinghistory.
As a result of extensive pilot work, density gradient centrifugationwas used to recover monocytes. Briefly, red blood cells were removed by centrifugation(550g for 20 minutes at room temperature). Then thebuffy coat was collected in a 50-mL polypropylene conical tube, 4 mL of OptiPrep(AXIS-SHIELD PoC AS, Oslo, Norway) solution was added, and the combinationwas gently mixed. A gradient solution (iodixanol–sodium chloride andHEPES) with a density of 1.088 g/mL was carefully overlaid on top of the OptiPrep–buffycoat mixture with a glass pipette. Without disturbing the layers, a secondgradient solution (iodixanol–sodium chloride and HEPES) with a densityof 1.078 g/mL was added on top of the 1.088-g/mL solution. Similarly, a thirdgradient solution (iodixanol–sodium chloride and HEPES) with a densityof 1.068 g/mL was added on top of the 1.078-g/mL solution. In addition, HEPES-bufferedsaline was carefully added on top of the 1.068-g/mL solution. Centrifugationwas performed at 200g for 10 minutes at room temperature.The top band of cells corresponding to the monocytes was collected and washedin HEPES-buffered saline.17 Flow cytometryusing anti-CD14 or anti-CD68 immunofluorescence confirmed that purity was85% to 95% monocytes. This technique of monocyte isolation resulted in lessartifactual activation compared with other techniques tested, such as adherence,elutriation, cell sorting, magnetic beads, and other density-gradient isolationtechniques. Fresh monocytes were either used for in vitro cell culture studies(in which case they were defined as macrophages) or immediately used for isolationof mRNA.
Extensive pilot experimentation determined that the following techniquewas most reproducible. Monocytes were immediately plated at 2 × 105 cells in triplicate onto 2% agarose-coated, 24-well plates in 10%fetal bovine serum. Agarose rather than plastic or collagen was chosen toavoid artifactual activation of cultured macrophages. Viability was approximately85% to 90% after 24 hours. These macrophages were then stimulated with 1 of4 conditions: medium alone, 5µM macrophage chemotactic protein 1 (MCP-1),RPE-derived cell membrane blebs, or both MCP-1 and blebs. A cytokine knownto be produced by injured RPE, MCP-1 is an activation factor for macrophages.18,19 Blebs are vesicles of cell membraneand cytoplasm that form after oxidant injury and may be a component of drusen.20 Blebs were generated by using the spontaneously transformedadult human RPE cell line (ARPE 19), which was genetically modified by retroviraltransduction with a construct containing green fluorescent protein–fernesylatedr-Ras, which anchors the fluorescent marker to the inner leaflet of the plasmamembrane.20 With the use of 10µM menadione,green fluorescent protein–modified (ie, green) blebs released into themedium were collected, washed, and concentrated by centrifugation and thencharacterized and quantified by flow cytometry. With the use of 3-µmfluorescent latex beads as size standards, blebs ranging in size from 0.3to 5 µm in diameter were collected. The protocol was to add medium orMCP-1 in 1% serum overnight followed by medium or 106 blebs permilliliter the next morning. Forty-eight hours later, media were collectedand TNF-α was measured by enzyme-linked immunosorbent assay (ELISA;R&D Systems, Minneapolis, Minn). In some experiments, TNF-α or glyceraldehyde-3-phosphatedehydrogenase (GAPDH) mRNA was recovered from cultured macrophages to determinetranscriptional regulation and was amplified by reverse transcriptase–polymerasechain reaction (RT-PCR) as described previously.21
Monocytes were recovered as described herein. Total RNA was extractedusing TRI-Reagent (Sigma-Aldrich Corp, St Louis, Mo). Tumor necrosis factor αprimers and probe were purchased ready to use from Applied Biosystems (FosterCity, Calif). Quantitative RT-PCRs were performed using the TaqMan One-stepRT-PCR Master Mix reagents kit and ABI Prism 7700 sequence detection system(Applied Biosystems) in a total volume of 50 µL of reaction mixtureaccording to manufacturer instructions. The TaqMan ribosomal RNA control reagentskit was used to detect the 18S ribosomal RNA gene, which represented an endogenouscontrol. Each sample was normalized to the 18S transcript content. The primerprobe mixture was purchased from Applied Biosystems and used as specifiedby the manufacturer's protocol. The standard curves and 18S were generatedusing serially diluted solutions (0.001-100 ng) of mRNA from the human monocyteline U937 stimulated with phorbol myristate acetate. The PCR assays were conductedin duplicate for each sample.
Statistical analysis was performed using a Statistica software package (Statsoft, Tulsa, Okla). Comparison of patient characteristicswas performed with the use of the unpaired, 2-tailed t testfor independent samples. Differences in median values of the ELISA and PCRdata were determined using the Mann-Whitney test. An arbitrary cutoff valuewas used to subdivide the ELISA data into high, low, and intermediate categories.High activity was defined as the upper 25th percentile of the spontaneousTNF-α production or a more than 3-fold increase after stimulation. Lowactivity was defined as the lower 25th percentile of the spontaneous productionor a less than 2-fold increase after stimulation. For PCR data, tertiles weredetermined using the 33rd and 67th percentile values of the control samples,and then those cutoffs were applied to the AMD data. Odd ratios and 95% confidenceintervals were calculated for comparisons of middle vs lowest tertile andhighest vs lowest tertile between the patients with dry and wet AMD. The Fisherexact test was used to determine statistical significance.
Table 1 gives the demographicfeatures of the patients with AMD and the controls. Although the controlswere slightly younger, none of the differences were statistically significant.
We evaluated the pattern of TNF-α production by freshly isolatedblood monocytes in culture when stimulated in cell culture with RPE-derivedcell membrane blebs and MCP-1. Macrophage samples from patients and controlsdemonstrated a similar range of quantitative concentration of cytokine producedand a qualitative pattern of TNF-α production. We found that monocytesamples from different patients demonstrated qualitatively different patternsand magnitudes of TNF-α production. Some patients demonstrated a patternof "responsiveness to stimulation" (Figure1A). Unstimulated cells demonstrated significant "spontaneous" productionof TNF-α protein (approximately 290 pg/mL) when simply placed in nonadherentculture. However, when MCP-1 and blebs were added, macrophages demonstratedeven greater "induced" production, demonstrating a more than 3-fold higherTNF-α protein secretion (921 pg/mL) and indicating transcriptional regulationof TNF-α production (Figure 1A,top).
Tumor necrosis factor α(TNF-α) produced by macrophages in culture. A, TNF-α messengerRNA (mRNA) expression (top) and TNF-α cytokine production (bottom) bya sample demonstrating the typical high-responder pattern. B, TNF-αmRNA expression (top) and TNF-α cytokine production (bottom) by a typicallow-responder pattern. GAPDH indicates glyceraldehyde-3-phosphate dehydrogenase;MCP-1, macrophage chemotactic protein 1. The bars represent mean values; thelimit lines, SEM.
However, we also noted that macrophages isolated from other patientsdemonstrated a different pattern of poor responsiveness to stimulation (Figure 1B). In this example, unstimulatedmacrophages produced less than 30 pg/mL. No significant increase was notedon stimulation with blebs or MCP-1 (compared with a 3-fold increase to 900pg/mL). Also, mRNA expression was consistent with the level of protein synthesis.These cells, however, could be stimulated with the classic inflammatory activationalstimuli, interferon-γ (10 µg/mL) and lipopolysaccharide (10 ng/mL),to produce TNF-α levels greater than 800 pg/mL (not shown). Repetitionof the experiment on a separate day produced similar results.
A comparison of samples from AMD patients and age-matched controls demonstrateda wide range in TNF-α production for both spontaneous (stimulated bymedium only) and induced (stimulated with both blebs and MCP-1) TNF-αproduction. The median spontaneous TNF-α production was 370 pg/mL (range,50-3260 pg/mL) for patients with AMD and 285 pg/mL (range, 45-1380 pg/mL)for age-matched controls. The median induced TNF-α production was 460pg/mL (range, 100-6750 pg/mL) for patients with AMD and 615 pg/mL (range,50-1390 pg/mL) for age-matched controls. Thus, among all samples, a 70-foldrange in (unstimulated) spontaneous production of TNF-α was observed(median, 365 pg/mL) and a 300-fold range in induced production (stimulatedwith both blebs and MCP-1) was observed (median, 625 pg/mL). The presenceof high heterogeneity among controls and AMD patients indicates that differencesin TNF-α production are an intrinsic function of the macrophage immuneresponse, not the result of AMD.
We have shown herein that blood-derived macrophages from patients withAMD produce TNF-α when cultured with RPE-derived cellular debris orblebs but that high intersubject variability in TNF-α production wasobserved among patients. If this property reflects a physiologically importantfunction of monocytes, then it is reasonable to assume that high or low productionin culture might reflect high or low cytokine mRNA content of circulatingmonocytes. Thus, we sought to determine whether circulating monocytes demonstratedwide intersubject variability in expression of TNF-α mRNA content toaccount for the observed variation of spontaneous and induced synthesis ofTNF-α in vitro.
The TNF-α mRNA content of freshly isolated monocytes was determinedby real-time RT-PCR. A 200-fold range in TNF-α mRNA content was detectedamong AMD patients and control subjects (Figure 2). Monocytes from the total population of AMD patients demonstrateda similar range of TNF-α mRNA content as those recovered from age-matchedcontrols. However, when AMD patients were subdivided into those with dry andwet AMD, the median value for patients with CNV was significantly greaterthan the value for patients with drusen (200 vs 124 arbitrary units, P = .002).
Scatterplot of messenger RNA (mRNA)levels from freshly isolated monocyte samples. AMD indicates age-related maculardegeneration.
We sought to test the hypothesis that high monocyte TNF-α productionmight be associated with CNV. We compared patients with early AMD (definedas drusen only) or neovascular AMD in terms of macrophage TNF-α productionin cell culture using criteria that reflected both quantitative differencesin production and the qualitative pattern of induced production after stimulation(see the "Methods" section). Table 2 givesthe odds ratios and confidence intervals for patients with dry and neovascularAMD. Although not statistically significant, a strong trend suggested thatAMD patients with monocytes classified as high responders demonstrated greaterrisk of CNV.
We also correlated TNF-α mRNA level in monocytes and AMD severity.Direct numerical cutoffs were used to subdivide PCR data into tertiles. Table 3 gives the the odds ratios and confidenceintervals for patients with dry and neovascular AMD. A total of 64% of patientswith CNV had monocyte mRNA levels in the upper tertile compared with 32% ofpatients with drusen. Conversely, 20% of patients with CNV compared with 50%of patients with drusen demonstrated low TNF-α mRNA. The odds ratiofor CNV in patients with high vs low TNF-α expression was 5.03 (P = .02).
The concept that inflammatory mechanisms play an important role in theprogression of many chronic degenerative diseases has emerged as a major paradigmshift in our understanding of disease pathogenesis. For example, both innateand antigen-specific immune responses have been implicated in the pathogenesisof atherosclerosis, Alzheimer disease, and renal disease.22- 27 Thus,similar mechanisms are likely to be involved in AMD pathogenesis. In thisstudy, we focus on the potential role for blood-derived macrophages.
Monocytes and macrophages belong to the family of circulating bone marrow–derivedmononuclear leukocytes. The monocyte (the circulating cell) and the macrophage(the tissue-infiltrating or cell culture equivalent) are important effectorsin tissue repair, cell injury, inflammation, and innate immunity. In blood,the circulating monocyte can be identified by its relatively large size (12-20µm), single-lobed indented nucleus, and the presence of cell surfacemarkers, including CD68, CD11b/CD18, CD11c, and CD14. Monocytes can be recruitedinto many normal tissues by specific chemotactic factors, where they becomemacrophages up to 40 µm in size. In general, at least 4 different subsetsof monocytes can be identified within normal tissues on the basis of function,morphologic features, and cell surface marker expression, including blood-derivedmacrophages (the major topic of this study), resident macrophages, microglia,and dendritic cells.28
Cells of the monocyte lineage have been observed in histologic specimensof AMD. Macrophages have been detected along the choriocapillaris side ofthe Bruch membrane, underlying areas of thick drusen, or other deposits. Someinvestigators29 have even observed choroidalmacrophages associated with sub-RPE basal laminar deposits. Processes fromchoroidal monocytes have been noted to insert into nodular drusen, presumablyfor the purpose of scavenging debris.29 Theidentity of these cells is uncertain, but they seem to lack typical phagocyticvacuoles and express HLA-DR, suggesting that most of the cells may representdendritic cells.
Macrophages also contribute to the severity of CNV.28,30- 32 Approximately60% of surgically excised CNV membranes contain macrophages.5 Thecausal role of macrophages in regulating severity in CNV has been recentlyshown in a mouse model of experimental CNV in which macrophage depletion decreasedseverity by approximately 50%.30 Our grouphas also recently shown that approximately 90% of the macrophages in experimentalCNV are blood derived, confirming that the analysis of circulating monocytesis relevant to AMD.31 The contribution (beneficialor harmful) of monocytes in AMD pathogenesis or progression is unknown. Theoretically,macrophages might mediate drusen resorption by scavenging and removing deposits.Conversely, macrophages may stimulate progression of drusen into CNV by releasingcytokines such as TNF-α or others into deposits and by releasing factorsthat regulate CNV growth and severity.
In this study, we analyzed the TNF-α expression by circulatingmonocytes isolated from AMD patients and observed 3 important findings. First,we demonstrate that RPE-derived blebs and MCP-1 (as surrogates for drusen)were a stimulus to induce low-grade macrophage activation. Using in vitrocell culture of monocytes from patients, we observed a 2- to 3-fold inductionof TNF-α in macrophages from many but not all patients, indicating thatstimuli relevant to drusen deposits might induce modest expression of cytokinesby choroidal macrophages. Preliminary unpublished data for interleukin 6,a proinflammatory cytokine, gave similar results, suggesting that the observationis not restricted to TNF-α.
Second, and more important, we observed a wide range of heterogeneityof TNF-α expression in macrophages in culture or by mRNA analysis offreshly isolated circulating monocytes. The heterogeneity was present in controlsand AMD patients, indicating it was a property of the macrophage immune responseand not disease specific. The heterogeneity among different patients spanneda 200-fold range, a magnitude that is likely to be more physiologically importantthan the 3-fold induction by RPE debris in culture. On the basis of this observation,we speculate that the preexisting activation state of the monocyte (in thiscase, defined as the amount of TNF-α production) is more biologicallyrelevant to AMD pathogenesis than is local induction of macrophage activationby drusen components in the eye. This is a novel finding that, if confirmed,will have implications about the regulatory mechanisms of macrophage activationbeyond AMD. For example, activated macrophages may also contribute to complicationsof atherosclerosis, possibly contributing to the epidemiological associationbetween AMD and vascular disease.33
Third, our data suggest that macrophage activation state, defined asTNF-α production, might serve as a predictor of risk for progression.The presence of inflammation in chronic diseases, irrespective of the cause,has been noted to serve as a risk factor for progression. For example, inatherosclerosis, a high serum C-reactive protein level is a strong predictorof myocardial infarction and is used as a biomarker to identify high-riskpatients.34 Similarly, in this study, we evaluatedthe possibility that the wide spectrum of intersubject variation of macrophageactivation in cell culture or in mRNA levels of fresh monocytes serve as apredictor of neovascular AMD. Our results indicate that AMD patients withblood monocytes that express high TNF-α mRNA levels demonstrate an almost5-fold increased prevalence of neovascular AMD. Tumor necrosis factor αsynthesis in culture also demonstrated a suggestive trend.
Several possible weaknesses are apparent in this small pilot study.The assays are technically complex, are susceptible to experimental artifact,and must be processed on site. For example, elutriation and flow cytometry,2 commonly used isolation techniques, induced unacceptable artifactual activation.In addition, we have not extensively retested the same patients on multipleoccasions to determine whether TNF-α levels are a consistent propertyor vary over time, but these experiments will be performed. Finally, evaluationof other macrophage-derived mediators relevant to AMD pathogenesis might beinformative. Preliminary data indicate wide heterogeneity in monocyte expressionof other factors relevant to neovascular AMD, including vascular endothelialgrowth factor.
In summary, these results suggest the hypothesis that the preexistingmacrophage activation state, defined as the level of cytokine or mediatorexpression of the circulating monocytes, might determine the negative or positiveconsequence of macrophage recruitment as a disease modifier. Macrophages withhigh expression of relevant mediators might contribute to disease progression.Furthermore, identification of AMD patients with high levels of TNF-αor other relevant mediators may serve as a biomarker of progression. We hopethat this approach can be perfected to identify high-risk patients in CNVprevention trials.
Correspondence: Scott W. Cousins, MD, William L. McKnight VisionResearch Center, Bascom Palmer Eye Institute, 1638 NW 10th Ave, Miami, FL33136 (firstname.lastname@example.org).
Submitted for publication July 16, 2003; final revision received December11, 2003; accepted December 17, 2003.
This study was supported by grant NEI EY/AI 13318 from the NationalEye Institute, Bethesda, Md.
Thank you for submitting a comment on this article. It will be reviewed by JAMA Ophthalmology editors. You will be notified when your comment has been published. Comments should not exceed 500 words of text and 10 references.
Do not submit personal medical questions or information that could identify a specific patient, questions about a particular case, or general inquiries to an author. Only content that has not been published, posted, or submitted elsewhere should be submitted. By submitting this Comment, you and any coauthors transfer copyright to the journal if your Comment is posted.
* = Required Field
Disclosure of Any Conflicts of Interest*
Indicate all relevant conflicts of interest of each author below, including all relevant financial interests, activities, and relationships within the past 3 years including, but not limited to, employment, affiliation, grants or funding, consultancies, honoraria or payment, speakers’ bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued. If all authors have none, check "No potential conflicts or relevant financial interests" in the box below. Please also indicate any funding received in support of this work. The information will be posted with your response.
Some tools below are only available to our subscribers or users with an online account.
Download citation file:
Web of Science® Times Cited: 59
Customize your page view by dragging & repositioning the boxes below.
Enter your username and email address. We'll send you a link to reset your password.
Enter your username and email address. We'll send instructions on how to reset your password to the email address we have on record.
Athens and Shibboleth are access management services that provide single sign-on to protected resources. They replace the multiple user names and passwords necessary to access subscription-based content with a single user name and password that can be entered once per session. It operates independently of a user's location or IP address. If your institution uses Athens or Shibboleth authentication, please contact your site administrator to receive your user name and password.