Medical Biophysics

Robert F. Bonner, PhD, Head, Section on Medical Biophysics
Zigurts Majumdar, PhD, Research Fellow
Martin Ehler, PhD, Visiting Fellow
Eleanor Ory, BS, Postbaccalaureate Fellow
Tatiana Kisseleva, MD, Guest Researcher
Sanford Meyers, MD, Guest Researcher
Mikhail Ostrovsky, PhD, DSc, Guest Researcher¹

Currently, we are developing optical technologies to characterize early stages of disease and to monitor responses to therapy in cancer and age-related macular degeneration (AMD). By integrating multispectral, non-invasive, clinical retinal autofluorescence imaging with automated image analysis, we seek to map the distribution of several intrinsic photochemicals implicated in health preservation and early disease. In pilot clinical studies, we are applying our new technology to understanding how changes in retinal spectral irradiance affect the balance between retinal photochemical pathways that, we hypothesize, drive early age-related maculopathy. Our broader goals are to quantify earlier stages of local molecular imbalance throughout the retina and to develop reliable means for classifying and quantifying early disease in order to evaluate the effectiveness of strategies to prevent disease progression. We are further developing our inventions of expression microdissection and laser capture microdissection for better integration with multiplex molecular analyses of specific cells and organelles extracted from complex tissue.

Laser microdissection and molecular diagnostics technology development

Majumdar, Ory, Bonner; in collaboration with Emmert-Buck, Pohida

Integrative molecular biology requires an understanding of interactions of large numbers of pathways. Similarly, molecular medicine increasingly relies on complex macromolecular diagnostics to guide therapeutic choices. A fundamental argument for laser capture microdissection (LCM) of tissues is that, without separation of specific cell populations from complex tissues, we will miss critical control functions of thousands of regulated transcription factors, cell regulators, and receptors that are expressed at low copy number. By detecting changes in these critical effectors, we expect to improve our integrative understanding of tissue function and pathology. In complex tissues—particularly among pathological variations—it is exceptionally difficult to measure the majority of molecules that are at low copy numbers per cell without first isolating specific cell populations. With our newest inventions, we are increasing the microdissection resolution to the organelle level.

The LCM techniques that we started developing 12 years ago are now widely used in molecular analysis of genetics and gene expression changes within target cells within complex tissues. However, in global proteomic and lipid studies without molecular amplification methods, the quantity of isolated cells sufficient to perform accurate characterization of less abundant species is problematic because the microscopic visualization, targeting, and isolation in laser microdissection has a maximal rate of 1–20 cells per second, depending on the cells’ microscopic distribution within the tissues.

Recently, in collaboration with NCI and CIT, we invented and are now refining an automatic “targetdirected microtransfer” technique based on macromolecule-specific staining of cells. The technique (patent pending) is built on our understanding of the physics of thermoplastic microtransfer and uses a much simpler device and transfer films than do commercial laser microdissection microscopes. Given that the new microbonding process does not require microscope-based visualization and targeting, it is capable of much higher throughput rates. Our current prototype is capable of isolating all specifically immunolabeled cells within a 1 cm2 region of an immunostained tissue section in about 5 seconds, which corresponds to specific separation from approximately 100,000 cells per second. This technique allows cell separation rates exceeding fluorescence-activated cell sorting (FACS) of labeled cell suspensions while preserving our ability to harvest specifically stained targets directly from sections of complex tissues. This rapid, automated microtransfer method has a spatial resolution determined by stain localization (less than 1 µ), not by optical diffraction. Consequently, it is uniquely suited to isolating highly dispersed, specific cell populations (e.g., stem cells or only those neurons in the supra-optic nucleus that express vasopressin) or specific organelles (e.g., neuronal nuclei in the brain). The spatial relationships (morphology) among the specific cells in the tissue are preserved on the transfer film. As the technology becomes more robust, we will integrate the automated microtransfer with molecular profiling of specific cells within tissues in our NIH LCM Core Facility.

United States Patents

Liotta LA, Buck MF, Weiss RA, Zhuang Z, Bonner RF. Isolation of cellular material under microscopic visualization. #6,569,639; May 27, 2003.
Bonner RF, Goldstein SR, Smith PD, Pohida TJ. Precision laser capture microdissection utilizing short pulse length. #6,420,132; July 16, 2002.
Bonner RF, Liotta L, Buck M, Krizman DB, Chuaqui R, Linehan WM, Trent JM, Goldstein SR, Smith PD, Peterson JI. Isolation of cellular material under microscopic visualization. #6,251,516; June 26, 2001.
Bonner RF, Pohida T, Buck M, Tangrea M, Chuaqui R. Target-directed microtransfer; patent pending.

Gene expression during normal development and pathology progression

Majumdar, Ehler, Ory, Bonner; in collaboration with Emmert-Buck, Brooks, McQueen

If microdissection and molecular analysis can be made clinically practical, the expression levels of sets of approximately 20–100 critical, stage-specific disease markers within a selected cell population might provide reliable diagnosis and intermediate endpoints of response to molecular therapies in individual patients. Our analysis of large gene expression and protein databases suggests that a significant fraction of all genes is expressed in any specific cell type and that the levels of gene products universally exhibit a highly skewed power—law distribution similar to those characterizing many other complex systems (Kuznetsov VA, Signal Processsing 2003;83:889). We have developed mathematical models for the evolution of such distributions—models that predict the observed distributions of genes, protein domains, and gene expression in species of increasing biological complexity (Kuznetsov et al., Genetics 2002;161:1321). In collaboration with Jacob Brown and Brian Brooks, we devised a method for using LCM to integrate small regions from the developing eye around the time of topological closure in the mouse embryo for comprehensive gene expression analysis in order to identify candidate genes associated with coloboma, a common human development defect. We are applying dimensionality reduction algorithms to identify temporal classes within Affymetrix® gene expression data from LCM-dissected target tissue for eight distinct time points around the time of normal retina closure. We will then apply such analysis to databases of gene pathway, transcription factor–related expression, and so forth in order to develop a more integrative understanding of the normal developmental program and extend the list of candidate genes that may be associated with the occurrence of coloboma.

To allow more routine and simpler multiplex molecular diagnostics, we are attempting to develop new approaches for better integration of our thermoplastic microtransfer methods of microdissection with downstream macromolecular analysis. Using a variety of confocal microscopy techniques, we are systematically measuring the complex physical interactions among scanned laser dosimetry, tissue stain density, thermoplastic polymer film, and underlying tissue sections. Using finiteelement modeling of thermal and fluid flow to fit the observed microbonding, we are refining our understanding of expression microdissection physics in order to automatically predict and optimize the process for any given tissue sample. Coupling the robust and simple automatic microdissection with rapid purification and detection of species might provide unique abilities for following macromolecular changes in normal tissue development and in pathologies such as cancer progression within prostate, colon, breast, lung, and ovary tissues. In continuing collaborations with NCI, we have developed standard procedures for isolating normal and pathological cells from clinical specimens. We have used our models of the statistics of expression levels in cell populations to identify genes differentially expressed in temporal patterns such as cancer progression. To date, our analysis points to a critical role for many less abundantly expressed genes at a critical stage of ovarian cancer progression, suggesting that, for most cancers, critical diagnostic marker sets should include such low-abundance transcripts. This notion is guiding our research in statistics of less abundant gene products and suitable detection methods. We foresee an evolution of molecular diagnosis from one based on qualitative or quantitative analysis of a few key macromolecules to one in which dimensionality reduction and classification algorithms analyze complex multivariate databases. Such analyses should allow a more complete identification of highly correlated clinical cases and the characterization of their response to molecular therapies specifically designed to prevent progression.

Prevention of progression of age-related macular degeneration through photoprotection

Majumdar, Ory, Meyers, Kisseleva, Ostrovsky, Bonner; in collaboration with de Monasterio, Cunningham, Chew, Wong, Avetisov

Our clinical studies are designed to test our biophysical model and clinical hypothesis that spectral imbalances of light reaching the retina lead to increased levels of toxic photochemicals within the retinal pigment epithelium, thereby driving early stages of AMD. If our hypothesis is correct, we believe that, during the earliest stages of retinal pathology, this process may be reversible. Therefore, progression to more advanced AMD might be prevented with the use of spectral sunglasses in bright daylight or by otherwise altering the spectrum of ambient light reaching the retina. Using our specifically designed bicolored sunglasses, we have begun non-invasive clinical studies in subjects who have undergone bilateral cataract surgery. Our aim is to image expected changes in the distribution of toxic photochemicals within the human retina. We are improving the autofluorescence imaging protocols and multispectral image analysis in order to provide high-resolution, molecularly specific images of early microscopic changes, thereby allowing us to follow the lesions’ changes over time in relation to photochemicals present in various retinal cell layers. In collaboration with the University of Maryland’s experts in the mathematical analysis of hyperspectral images and clinicians in NEI’s Clinical Branch, we are developing and testing new quantitative tools for following non-invasively the earliest changes in macular pathology in AMD and other retinal diseases. Our goal is to provide better early diagnosis and methods to evaluate early intervention and disease prevention strategies.

In our clinical collaborations at the NIH and with the Institute of Eye Diseases in Moscow, we are seeking to apply non-invasive spectral imaging methods to study early photochemical injury and the means of limiting it. Lipofuscin (including undigestible fluorescent photoproducts of the A2E pathway) accumulates with age in the RPE and co-localizes with the acute injury caused by photosensitization of reactive oxygen intermediates (ROI) in the primate retina when exposed to unnaturally high doses of blue light. This observation along with a variety of epidemiologicalstudies has led to a widely held hypothesis that cumulative blue-light photochemistry is harmful and plays a role in AMD progression. We have developed an alternative biophysical model of (1) the normal accumulation during aging of potentially damaging photoproducts in the RPE and (2) changes in this “natural” aging associated with external spectral filtering (sunglasses) or cataract surgery with intra-ocular lens implantation. The precursors (principally A2PEH2) of A2E and its oxidation products that segregate within lipofuscin granules in the RPE originate from reactions of all-trans-retinal within the photoreceptor outer segments (ROS) when in bright daylight. Although RPE lysosomal processing enzymatically digests over 99 percent of shed ROS contents, A2E and related fluorophores are not broken down but rather concentrate into lipofuscin granules. By age 60, the average steady-state concentrations within the RPE reach approximately 200 µM in normal eyes. However, A2E is toxic to cellular membranes at much lower concentrations.

Our novel hypothesis (Meyers et al., Trans Am Ophthalmol Soc 2004;102:83) posits that blue light photosensitizes singlet oxygen generation within lipofuscin granules. The singlet oxygen efficiently oxidizes A2E within the lipofuscin granule in which it was generated and is the principal means of controlling steady-state levels of A2E in the RPE. As short-wavelength macular irradiance decreases with age, the rate of A2E photo-oxidation falls by approximately up to 20-fold, causing the syeady-state concentration ([A2E]_ss) in the normal phakic eye to increase even as rod bleaching and A2E production decrease. Our theoretical model of macular aging reproduces the normal age dependence of lipofuscin and A2E. It provides a primary cytotoxic mechanism in which, once A2E reaches a threshold concentration in the RPE cell, A2E redistribution into critical membranes causes damage with or without additional photo-activation. The model also predicts that, in normal RPE, nearly constant levels of A2E are maintained at a given age and lens color, irrespective of total ambient light exposure. It is primarily the yellowing of the lens with age that distorts the original spectral balance between the rate of production and rate of photo-oxidation found in youth, allowing [A2E]_ss to rise with age. If our model is correct, then restoring or optimizing the spectral balance with external spectrally selective sunglasses could significantly lower A2E levels and may prevent associated macular degeneration. Our clinical studies using non-invasive, multispectral autofluorescence imaging to map the levels of A2E and other retinal photochemicals are designed to test the predictions of our model and the benefits of specific spectral photoprotective filters.

¹ Russian Academy of Sciences, Moscow, Russia

Collaborators

Sergei Avetisov, MD, Institute of Eye Diseases, Moscow, Russia
David Berler, MD, Washington Eye Physicians & Surgeons, Chevy Chase, MD
Brian Brooks, MD, PhD, Ophthalmic Genetics and Visual Function Branch, NEI, Bethesda, MD
Jacob Brown, MD, PhD, Ophthalmic Genetics and Visual Function Branch, NEI, Bethesda, MD
Emily Chew, MD, Clinical Branch, NEI, Bethesda, MD
Denise Cunningham, CRA, RBP, MEd, Clinical Branch, NEI, Bethesda, MD
Wojciech Czaja, PhD, Norbert Weiner Center, University of Maryland, College Park, MD
Francisco de Monasterio, MD, PhD, Clinical Branch, NEI, Bethesda, MD
Michael R. Emmert-Buck, MD, PhD, Laboratory of Pathology, NCI, Bethesda, MD
Philip G. McQueen, PhD, Mathematical and Statistical Computing Laboratory, CIT, Bethesda, MD
Sanford Meyers, MD, Retina Consultants, Des Plaines, IL
Tom Pohida, MSE, Computational Biology and Electronics Laboratory, CIT, Bethesda, MD
Jaime Rodriguez-Canales, MD, Laboratory of Pathology, NCI, Bethesda, MD
Wai T. Wong, MD, PhD, Ophthalmic Genetics and Visual Function Branch, NEI, Bethesda, MD

For further information, contact bonnerr@mail.nih.gov.