|Home | About | Journals | Submit | Contact Us | Français|
Understanding tissue architecture and physical and chemical reciprocity between cells and their microenvironment provides vital insight into key events in cancer metastasis, such as cell migration through three-dimensional extracellular matrices. Yet many mechanistic details associated with metastasis remain elusive due to difficulty studying cancer cells in relevant three-dimensional microenvironments. Recently optical imaging has facilitated direct observation of single cells undertaking fundamental steps in the metastatic processes. As such, optical imaging is providing novel “optical biomarkers” with diagnostic potential that may be linked to cell motility pathways associated with metastasis, and can help guide new approaches in cancer diagnosis and therapy. Herein, we present recent advances in one subclass of optical imaging of particular promise for cellular imaging, multiphoton microscopy, that can be used to improve detection of malignant cells as well as advance our understanding of the cell biology of cancer metastasis.
The spread of cancer to nearby and distant sites involves several steps, termed the metastatic cascade: local invasion, entrance into the vasculature (intravasation), exit from the vasculature (extravasation), and growth in a distant tissue1,2. Despite advances in our understanding of cancer progression, key steps in the metastatic process are still poorly understood and seldom directly observed in vivo. We have now reached an understanding of several biochemical changes and gene profiles that accompany tumor progression. However, our understanding of the basic cell biology of metastasis will remain incomplete until we take this information forward to image molecular events within tumors. To accomplish this, advanced techniques are needed to visualize tumor cells in their relevant environment, and capture their behavior and molecular changes as they navigate the stages of metastasis.
Advanced imaging approaches are not only important for understanding the biology of metastasis on a fundamental level, but also are likely to impact patient outcome. Timely detection and accurate staging of human cancer is critical because the selected course of treatment is directly dependent on the location and pathological characterization of the tumor mass. Therefore, methods to accurately detect early stage cancer, and reliably diagnose and stage progressing cancers, can have a profound impact on treatment courses that may increase patient survival. In particular, imaging techniques to distinguish early hyperplastic lesions from normal tissue and malignant versus benign tumors are decidedly necessary. As such, optical techniques that identify changes in cellular behavior and/or the extracellular matrix (ECM) to produce consistent and quantifiable markers of cellular and molecular changes associated with cancer, i.e. “optical biomarkers”, are desired. Such optical biomarkers, when linked to molecular pathways, have the potential to provide guidance for specifically targeted therapeutic intervention. Several recent advances in optical imaging of tumor progression (reviewed below) are beginning to make this potential achievable.
To date, most clinical tools focus on identifying a tumor mass that is large enough to be detected with whole body/organ imaging technologies, such as ultrasound, magnetic resonance imaging, computed tomography, fluorescence-mediated tomography, and positron emission tomography (PET)3,4. PET in particular has promise due to its ability to selectively image tumors based on metabolism, proliferation, or specifically labeled receptor or antigen biomarkers5. Yet, while these tools are extremely valuable in clinical oncology, and are the primary means for non-invasively detecting a tumor mass, they are not capable of subcellular resolution3,5–7. As such, they are unlikely to provide a highly accurate indication of disease state by themselves since comprehensively understanding cancer progression requires the ability to visualize single cell phenotypes that are linked to defined molecular pathways. Therefore a great deal of work has been directed toward imaging single cells and identifying optical biomarkers that are reliably indicative of early cell transformation, such as changes in cell metabolism8–14 or stromal extracellular matrix (ECM) microenvironment architecture8,12,13,15,16. However, at the time of detection, it is common for tumors to have already progressed to an invasive stage. As such, information that helps stop or limit the spread of primary tumor cells to new sites in the body, or indicates treatment for secondary tumor risk, is of importance. Yet, detecting optical biomarkers that specify an invasive tumor is more complex than tumor mass detection. Moreover, in order for optical biomarkers associated with metastasis to reach their full potential, a detailed understanding of their coupled molecular mechanisms need to be determined in the context of the cell phenotype, initially in animal studies, and ultimately in patients.
Recent insights into the mechanisms of cell motility events and characterization of optical biomarkers associated with metastatic processes have been gained using multiphoton laser-scanning microscopy (MPLSM), allowing visualization of invasive cells within 3D microenvironments in vivo and in vitro. Here, we highlight these novel findings obtained from MPLSM imaging of cancerous cells within their native environments, with particular emphasis placed on the potential to link molecular mechanisms of motility during local invasion through the collagenous stroma with optical biomarkers indicative of disease state. Current limitations of optical imaging as a diagnostic tool mean that emerging technologies will increase the diagnostic potential of MPLSM.
Of course, while MPLSM provides particular advantages over many of the more conventional optical imaging technologies and is uniquely suited for high resolution imaging of the collagen matrix, it is important to note that recent advances in other imaging modalities - such as optical coherence tomography17–19, spinning disk confocal microscopy20,21, and photoacoustic tomography22,23 - are also capable of providing insight into single cancer cell behavior. However, discussion of these imaging modalities is beyond the scope of the current review and the reader is encouraged to examine the listed references for additional information. Moreover, since technologies such as these can often be combined with MPLSM for multimodal imaging approaches (e.g.17,24) they can provide an even greater depth of information to aid our understanding of tumor biology.
Multiphoton laser-scanning microscopy is an optical sectioning technique where the fluorescence emission is proportional to the molecular cross-section and has a quadratic dependence on the excitation laser power25–27 (Box 1; for detailed reviews of MPLSM and second harmonic generation see refs.7,27–30). Due to this quadratic dependence on excitation intensity, higher order processes are unlikely to occur outside the focal volume (i.e. the probability of two or more photons simultaneously exciting a fluorophore outside the focal volume is extremely low)26. As a result, improved signal to noise and background discrimination is achieved without the need for a confocal aperture, which allows the majority of emitted photons to be collected while not photobleaching the out-of-focus volumes of the tissue. Moreover, mode-locked lasers used for MPLSM excite at longer wavelengths (typically 650–1050 nm excitation; this broad tunability also has practical advantages for being able to excite a wide range of fluorophores) than conventional fluorescence imaging allowing for deeper penetration into tissues since longer wavelengths are more immune to light scattering and a more efficient collection of scattered photons emitted following deep tissue imaging is utilized with MPLSM. In addition, using pulsed lasers such that the mean power at the sample is moderate minimizes photo-damage. It has been demonstrated that for many cells and tissues the viability is improved when longer wavelengths are used due to circumventing some of the know mechanisms of phototoxicity (including cycle checkpoints that are UV light sensitive)31,32. Hence, MPLSM26 can image deeper26,33 into 3D structures, with improved signal-to-noise ratio26,29,33 and viability32, than more conventional optical imaging techniques, such as confocal microscopy33.
Multiphoton microscopy possesses several advantages over more conventional microscopy techniques for imaging cells within 3D matrices:
It should be noted, however, that the discussion of deep tissue imaging herein is in the context of optical imaging, where MPLSM provides an effective imaging depth that greatly exceeds more conventional optical imaging approaches, but is still significantly less than current clinical tools that can image through the human body. Standard MPLSM can effectively image to depths of 400–1000 m, depending on the tissue structure25,29, which currently precludes imaging through the skin into human organs, but does not exclude MPLSM as a potential diagnostic technology (see Box 2) or its utility for studying fundamental cell biology in context. With MPLSM, cell biology studies can now be performed on cells within relevant 3D matrices34–36 as well as in vivo with intravital imaging37–40. This allows detailed analysis of cell phenotype, and a quantitative description of cell behavior (i.e. migration speed, direction etc.) within biologically relevant environments. Moreover, MPLSM is particularly well suited to image quantum dots or cells expressing fluorescent protein(s) (while a detailed discussion of fluorescent proteins and biosensors is beyond the scope of this review, the reader is encouraged to examine recent reviews describing this rapidly growing and exciting field, such as41–47 and Lidke and Wilson in this issue). In addition, utilization of fluorescent protein technologies for localization studies or multiple fluorescent labels for protein-protein interaction/activation studies with Förster resonance energy transfer (FRET)42–44 provide novel insight into signal transduction in vivo (see review by Balla, this issue).
Although MPLSM has emerged as a powerful technology to study cell phenotype and associated signal transduction in 3D microenvironments, the clinical capability of MPLSM currently remains limited since the imaging depth is still not adequate for deep organ imaging through the skin. Yet, MPLSM still holds great promise as a diagnostic technology.
MPLSM is additionally advantageous because it generates harmonic signals, which are not fluorescent, but rather a polarization process with an emission wavelength equal to exactly half of the excitation wavelength48–51, and therefore can be easily separated from fluorescence signals. In the context of cell biology, this results in second and third harmonic generation (SHG and THG, respectively) signals from biological components, such as fibrillar collagen48–52, myosin, and tubulin30,48,53. This proves to be a very useful aspect that can be exploited, since biological materials such as collagen are strong harmonophores; providing a powerful means to simultaneously image collagen and selected fluorophores. This approach thus allows imaging of cells (and relevant intracellular components) within complex 3D collagen matrices to define molecular pathways representative of the in vivo condition, and can also provide diagnostic markers of the ECM that are linked to tumor progression15,34.
Multiphoton laser-scanning microscopy has been used to study cancer cell motility and local invasion in vivo and within 3D microenvironments in vitro to provide novel information about these processes. The studies presented here, while not a comprehensive review of the regulation of these processes in metastasis, demonstrate the link particular cell phenotypes with molecular pathways, providing a potential link between the fundamental cell biology and optical biomarkers.
Three primary cell phenotypes have been reported for cells migration through the ECM during local invasion: 1) amoeboid, 2) mesenchymal (also referred to as fibroblastoid), and 3) collective migration depending on tumor type, stage, and location relative to the primary tumor mass or vasculature (Figure 1). Fundamental work by Condeelis and colleagues54 using intravital multiphoton microscopy identified key differences between well-described motility phenotypes on 2D substrates and the mechanisms of cell motility during metastasis in vivo. Notably, they observed migrating cancer cells moving along collagen fibers toward and into blood vessels with rapid shape changes and no cell polarity, characterized as amoeboid motility37,54, which is distinct from fibroblastic migration where cells are elongated and form a well defined lamellapodia37. Furthermore, MPLSM imaging of live tumors and carcinoma cells in 3D culture show features of collective invasion (sheets, strands, and tubes) in regions of local invasion, as well as single cell mesenchymal and amoeboid phenotypes, in contact with aligned collagen fibers 8,15,34,36,55–58. Each of these phenotypes has been linked to regulation of specific signal transduction networks. A brief description of these findings follows.
Interestingly, amoeboid motility phenotypes were also identified during migration within 3D matrices in vitro in a protease-dependent manner56,59. In 3D collagen matrices, broadly inhibiting multiple protease families results in a switch from mesenchymal motility, which is characterized by strong β1-integrin localization at the cell-matrix interface56,60 co-localized with active MT1-MMP (MMP14)56, to an effective amoeboid-like migration, termed a mesenchymal-amoeboid transition56. Intravital multiphoton microscopy of fibrosarcoma cells treated with protease inhibitors and then injected into the mouse dermis supports conclusions from in vitro 3D matrix experiments, suggesting a mechanism for protease-independent amoeboid movement in vivo56. Moreover, MPLSM imaging of collagen fibrils in 3D reconstituted collagen matrices demonstrated that broad inhibition of proteases did not inhibit the ability of the cell to reorganize the matrix or invade into the matrix34,36. However, it is generally accepted that in vivo, before invasive cells can break free of the primary tumor mass and invade into the stroma, the physical ECM boundary at the tumor-stroma interface must be broken down and/or reorganized61,62. A need for protease activity in vivo makes intuitive sense when one observes the tumor boundary using SHG imaging, which shows that non-invasive regions of tumors are confined by collagen fibers that are arranged parallel to the tumor boundary8,15 (which is also an optical biomarker used when inspecting for the presence or absence of local invasion; see Box 3 and Figure 2). Thus, while multiple studies have presented strong evidence demonstrating that tumor cells can migrate through the collagen matrix in the absence of protease activity34,36,56,59, most of this work has been performed in 3D culture in vitro or on cells injected into mouse models.
Optical biomarkers, the detection of changes in cellular behavior, phenotype, or physical-chemical properties, and/or changes in the extracellular matrix composition and architecture by optical imaging technologies, that can consistently quantify cellular and molecular changes associated with cancer have diagnostic potential.
Stromal Collagen Markers:
Deposition of collagen surrounding tumors has been noted by pathologists, yet only recently has the functional significance of desmoplasia and collagen architecture been investigated in vivo. Significant changes in the collagen matrix occur with tumor initiation and progression, which can be used as optical biomarkers:
Additional Optical Biomarkers:
While collagen and the metabolites NADH and FAD have been the best characterized intrinsic fluorophores that are of relevance to cancer, there are many other intrinsic fluorophores that warrant further study as being of benefit to cancer research (and are perhaps future biomarkers) such as Protoporphyrin IX134,135 (component of the Heme pathway), lipofuscin91,136, elastin91, and tryptophan91,93.
In vivo, a defined matrix architecture that results from normal development is in place, and becomes stretched by the growing tumor, and is later reorganized and aligned, facilitating local invasion into the stroma15. In live tumors, MPLSM imaging reveals that this transition occurs over time (weeks to months in rapidly progressing mouse tumor models; which may indicate changes over months to years in human patients). Therefore, it seems likely that a remodeling of the collagen matrix may be required61,63,64 to aid matrix re-alignment that facilitates invasion. Moreover, recent work by Sabeh and co-workers61 suggests that part of the discrepancy between protease-dependent and protease-independent 3D migration may be influenced by the preparation of the collagen matrix; in particular, whether or not the ECM is crosslinked, and whether or not a path for the cell has already been created55,61,65, suggesting that more work needs to be performed in native environments in vivo. Combined, these reports advocate studies to understand proteolytic degradation of the ECM in vivo that will likely require imaging single cells at the invasion boundary, along with the tumor-associated cells of the microenvironment, and the collagen fibers adjacent to the invading cells in endogenous tumors over time in order to understand the process. However, in vivo studies also provide strong evidence that an amoeboid phenotype exists in tumors36,37,54,56, and suggests that the invading cells may not be susceptible to protease inhibition therapy. As such, the detection of ameoboid phenotypes in biopsy section may guide decisions regarding the therapeutic regime.
Sahai and co-workers identified a link between the mechanism for migration through 3D matrices and the Rho/ROCK signaling pathway59. In rounded, dynamically blebbing cells analogous to amoeboid phenotypes, blockade of Rho/ROCK signaling is significantly more effective at reducing 3D migration than in cell lines with elongated phenotypes59; and knockdown of Smurf1, a ubiquitin ligase that can target RhoA for degradation, increases RhoA activity and promotes tumor cell motility through a mesenchymal-amoeboid transition66, suggesting that in contrast to protease therapy, cells with amoeboid phenotypes may be susceptible to targeted inhibition of the Rho pathway. Importantly, it has recently been established using MPLSM that Rho/ROCK signaling and resulting acto-myosin regulated contractile force are necessary for local collagen matrix deformation/reorganization near the cell boundary34,36. This matrix reorganization near the cell boundary provides physical contact guidance cues that strongly influence cell migration through 3D matrices34,67,68. Interestingly, when the collagen matrix is engineered to mimic matrix architecture that most efficiently promotes 3D migration, Rho and ROCK become unnecessary for migration of invasive breast carcinoma cells through the collagen matrix34, suggesting that the role of Rho-mediated intracellular contractile force during 3D migration may be more related to organizing the microenvironment than the movement of single motile cells.
In regions of local invasion, MPLSM imaging has revealed that collective invasion takes place at regions of matrix re-alignment along with single motile cells progressing deep into the aligned stroma8,15,34. This may be especially true for a subpopulation of individual tumor cells that may be migrating particularly fast if they have been strongly stimulated by paracrine signaling from tumor-associated macrophages37,69–71 or provided with a migration promoting ECM architecture or paracrine stimulation by tumor-associated fibroblasts65,72. Chemotactic gradients can enhance the invasion process by stimulating intracellular signaling networks associated with cell motility. Microarray analysis of invasive MTLn3 rat adenocarcinoma cells in vivo (compared to the non-metastatic MTC cell line), which were used to identify amoeboid movement with MPLSM for cells migrating along collagen toward blood vessels54, identified the growth factor regulated (EGF and CSF-1) ‘minimum motility pathway’54,73,74. This pathway is characterized by regulation of the actin cytoskeleton and in particular three end-stage effectors, Arp2/3 complex, capping protein, and cofilin, that influence actin dynamics associated with cell motility73,75,76, and directly link an in vivo cell phenotype identified by MPLSM with molecular targets. Furthermore, Mena, an Ena/Vasp protein that antagonizes actin capping, is upregulated54,73,74, and intravital multiphoton imaging experiments showed that cells overexpressing EGFP-Mena in orthotopic xenografts are more motile77. Interestingly an isoform of Mena (termed MenaINV) enhances sensitivity EGF stimulation and promotes EGF-induced cell motility and invasion, suggesting that these cells are more likely to metastasize even in the presence of basal levels of EGF77. Hence, it is clear that invasive cells have multiple compensatory mechanisms to efficiently migrate through collagenous matrices, highlighting the need for a detailed understanding of the link between cell phenotype and active signal transduction for identifying cell populations that may or may not be susceptible to a particular therapeutic intervention.
In addition to chemoattraction, integrin-mediated adhesion is known to play a role in tumor progression to metastasis78. While matrix adhesions in motile cells in 3D environments are not well understood largely due to a current lack of high-resolution imaging studies of these structures relative to the ECM microenvironment in vivo, and are likely less developed than large focal adhesions that form on stiff 2D substrates, current evidence strongly suggests a role for integrins and focal adhesion proteins in regulating metastasis-associated motility in vivo. For instance, focal adhesion kinase (FAK) is a well-described regulator of focal adhesions and cell motility79, is overexpressed in numerous human cancers80, and its loss has been shown to block malignant conversion81,82. In addition, FAK has been shown to regulate invasion by influencing urokinase plasminogen activator (uPA) activity83, which may result in changes to the ECM that are detectable with MPLSM. Moreover, FAK deletion represses a number of transcripts associated with adhesion, motility, and the cytoskeleton (many of which are ‘minimum motility pathway’ genes identified by MPLSM imaging and simultaneous collection of invasive cells in vivo54,73), which suggests FAK regulates actin dynamics and cell protrusion and motility81. In addition, FAK-null cells injected into the vasculature cannot efficiently form protrusions across the vessel wall that are necessary for them to extravasate84. Hence, adhesion complexes and regulators of the actin cytoskeleton are emerging as viable targets for metastasis therapy. Yet, while FAK has been associated with regulating a subset of genes that are predictive of human patient outcome81, and a number of gene signatures have been compiled to predict patient outcome85–88, specific quantitative information linking cell and/or matrix phenotype to specific cells bearing these signatures is currently lacking. Therefore, studies that can identify optical biomarkers and signal transduction pathways/gene signatures within invading cells are of great value not only in increasing our basic understanding of tumor biology, but also providing information that can be used in a diagnostic capacity to guide therapeutic intervention.
In addition to the use of exogenous fluorescent markers, such as GFP, Quantum dots, etc., to track molecular events in cells, a set of “built-in” endogenous fluorophores exists that can be studied to further our understanding of normal and pathologic processes. These molecules have the advantage of being naturally present in cells and tissues of interest, are involved in key biological processes, and therefore can be imaged in live, unstained tissue. Not only are these endogenous fluorophores of great use in various cell culture and animal models, but will ultimately be of use as biomarkers of disease state in human samples where fluorescently-labeled proteins are not available.
Although it has long been known that glycolysis is increased in tumors89,90, specific metabolic intermediates are emerging as intrinsic optical biomarkers of these changes. In particular, the metabolites nicotinamide adenine dinucleotide (NADH; reduced form) and flavin adenine dinucleotide (FAD; oxidized form) have characteristic fluorescent properties and spectral emissions that have been well characterized by spectroscopy91–93, and are detectable in live, unstained cells10,11. Moreover, changes in the redox state of the cell can be determined by imaging the relative intensity of NADH and FAD12,94, which correlates linearly with glucose uptake in tumor models95.
In addition to intensity changes, fluorescent lifetime microscopy (FLIM; Box 1) can be exploited to detect changes in metabolic state that accompany tumor progression. The fluorescence lifetime of protein-bound vs. free NADH or of FAD differ, and can be distinguished by FLIM. This approach has been used to discern precancerous lesions from normal tissue11, and invasive mammary carcinoma cells from non-invasive cells8. Specifically, tumor cells have a higher NADH and FAD fluorescent intensity and a longer fluorescent lifetime than normal cells11. Moreover, the fluorescence lifetime for FAD is longer in invasive cells than cells in the primary tumor mass, suggesting changes that accompany carcinoma progression8. Unexpectedly, these fluorophores are preserved in fixed, paraffin-embedded tissue, where the fluorescence intensity and lifetime of NADH and FAD discerns tumor from normal mammary epithelium in mouse samples11. Thus, application of MPLSM/MPLSM-FLIM to human pathological samples or fresh biopsy tissues may allow the readout of metabolic states, and may help further differentiate subsets of patients, or suggest more targeted therapies. For example, as recently reviewed by Thompson and co-workers90 the phosphoinositide 3-kinase (PI3K) pathway (Akt, the tumor suppressor PTEN etc..), which is important in many human cancers and a target for therapeutic intervention96, can regulate cell proliferation and motility, as well as glucose metabolism. AKT, a downstream kinase of PI3K, can regulate glucose transporter expression and capture, even in cells that are not insulin dependent90,97. Findings such as these suggest that optical detection of changes in fluorescent lifetimes of endogenous fluorophores such as NADH and FAD, which indicate altered glycolysis, may indicate whether targeted therapy against the PI3K pathway is relevant. Moreover, oncogenes such as Myc and Ras, and tumor suppressors such as p53 either cause or are sensitive to changes in glucose or glutamine metabolism90. Hence, a real-time understanding of metabolic state in single cells can not only be an indicator of transformation or metastatic potential, but optical biomarkers of metabolism, particularly when combined with information about single cell phenotype, may be useful for identifying particular oncogenic signaling pathways important in human cancer.
For many years, the deposition of collagen surrounding tumors, termed “desmoplasia,” has been noted by pathologists, yet only recently has the functional significance of desmoplasia been investigated. Importantly, collagen is perhaps the most readily imaged endogenous molecule, as fibrillar collagen is a strong harmonophore48–52, and therefore changes in collagen within mammary15,98,99 and ovarian12 tumors with SHG can be detected.
As described above, visualization of collagen surrounding mammary tumors demonstrates a characteristic sequence in the intensity, organization and alignment of collagen that accompanies tumor progression15 (see Figure 2, Box 3). These “tumor associated collagen signatures,” or TACS, are potentially relevant optical biomarkers, which is consistent with general collagen changes in human breast cancer detected by SHG99. Because aligned collagen fibers in animal models and in 3D cell culture are associated with and facilitate local invasion, these studies link collagen organization to cellular behavior and specific signal transduction pathways, such as Rho/Rock-dependent matrix reorganization34; and suggest that inhibition of intracellular contractility in a non-invasive primary tumor mass may help block conversion to an invasive tumor. Moreover, the observation that locally invasive regions near the tumor mass have a collective migration phenotype suggests contractile and proteolytically active behavior15,34,55,60,61 that could be targeted with a treatment regime specific to the mesenchymal phenotype. Furthermore, single cells invading away from the primary tumor mass can show mesenchymal or amoeboid phenotypes that may depend on the matrix architecture61 and the mechanism of chemoattraction37,69, and therefore understanding the matrix architecture and the cell phenotype could direct therapeutic intervention as well (i.e. amoeboid motility may be targeted for pieces of the ‘minimum motility pathway’54,73,74,77). Hence, it is becoming clear that a real-time understanding of matrix architecture and the corresponding cell behavior in vivo or in situ with optical imaging is of great utility in cancer biology. Moreover, as our understanding of the molecular basis of invasion into 3D matrices increases, future therapies targeted at these events will hopefully emerge, and would be of particular benefit to patients with an “invasive” optical biomarker.
The further development of collagen as a biomarker of tumor progression, will benefit from additional visualization tools to make measurements of intensity and alignment readily quantified, such as algorithms to identify collagen structures in SHG images and color map their angle or scatter properties. In a broader sense, because collagen is the most abundant protein in the body of most vertebrates100, the potential to use collagen as a means to depict tissue architecture in the context of a wide variety of cell biological, developmental, and pathological investigations in model organisms is great.
Because optical biomarkers, such as NADH, FAD, and collagen, have relevance to carcinoma progression and potential to be linked to relevant signal transduction cascades they are potential read outs for high-throughput screens. For example, because collagen matrix alignment facilitates invasion8,15, and can be recapitulated in 3D culture34, SHG imaging and analysis of collagen alignment by cells can be used to find small molecules that may prevent this process. Already, we know that this is a Rho-mediated contractility event34, and molecules that inhibit the Rho pathway may be effective in preventing carcinoma progression. Similarly, if the tumor-associated switch to glycolysis is recapitulated in cells or tissue explants in culture, then imaging of NADH and FAD may be a useful end-point for screens of small molecules to prevent the glycolytic switch.
Quantitative techniques and computational tools to standardize and quantify complex datasets from 4D (x-y-z-time) imaging101, will greatly aid our ability to understand fundamental questions in cancer biology, but will also be extremely important for accurately identifying optical biomarkers associated with disease (see Swedlow & Eliceiri, this issue). Once an optical biomarker (or a consensus panel of biomarkers) is established for a particular human cancer, the link between biomarker and molecular mechanism may be elucidated, and with the appropriate detection technology, analytical tools, and drug screens, a personalized tumor-phenotype-specific treatment regime may be in hand. Therefore, the information we gain from optical imaging studies in animal and culture models of human cancer will not only provide insight into the mechanisms of disease, but hold great promise for improving disease outcome.
Because of its advantages for imaging in tissue, the use of MPLSM as a tool for the cell biologist to advance our understanding of cancer progression in animal models is likely to expand, as additional analysis approaches, instrumentation, and expertise become more commonly available. Already, MPLSM has assisted in our understanding of cell invasion and metastasis, and captured single cells as they invade in vivo. In addition to cancer biology, these imaging approaches are likely to illuminate many normal developmental and pathological processes in vivo. Moreover, MPSLM has great potential to become of use as a clinical diagnostic tool (see Box 2) to take advantage of optical biomarkers in fresh tissue biopsies even as a patient is undergoing surgery, or in classic histopathology samples.
We apologize to those authors whose work we were unable to cite because of space and date of publication limitations. We thank members of the Keely laboratory and the Laboratory for Optical and Computational Instrumentation (LOCI) for helpful discussions regarding this work. This work was supported by a NIH postdoctoral training grant (T32CA009681) to PPP, and grants from the DOD: W81XWH-04-1-042 (PPP), the Mary Kay Ash Foundation (PJK), DOD CDMRP BC074970 (PJK), Am. Cancer Society RSG-00-339CSM (PJK), NIH CA076537 (PJK), and NIH EB000184 (KWE).
Conflict of Interest
The authors declare no conflict of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.