|Home | About | Journals | Submit | Contact Us | Français|
The science of ‘theranostics’ plays a crucial role in personalized medicine, which represents the future of patient management. Over the last decade an increasing research effort has focused on the development of nanoparticle-based molecular-imaging and drug-delivery approaches, emerging as a multidisciplinary field that shows promise in understanding the components, processes, dynamics and therapies of a disease at a molecular level. The potential of nanometer-sized agents for early detection, diagnosis and personalized treatment of diseases is extraordinary. They have found applications in almost all clinically relevant biomedical imaging modality. In this review, a number of these approaches will be presented with a particular emphasis on MRI and optical imaging-based techniques. We have discussed both established molecular-imaging approaches and recently developed innovative strategies, highlighting the seminal studies and a number of successful examples of theranostic nanomedicine, especially in the areas of cardiovascular and cancer therapy.
Nanotechnology is starting to invade different areas of science and ‘theranostic’ biomedical science is no exception [1–4]. The science of theranostics plays a critical role in personalized medicine, which represents the future of patient management. Nanoparticle-based medicinal approaches have emerged as an interdisciplinary area, that shows promise in understanding the components, processes, dynamics and therapies of disease at a molecular level. The unprecedented potential of nanoplatforms for early detection, diagnosis and personalized treatment of diseases have found application in every biomedical imaging modality.
MRI is a noninvasive diagnostic technique based on the interaction of nuclei with each other and the surrounding molecules in a tissue. The sensitivity of magnetic resonance is low in comparison to nuclear and optical modalities; however, the absence of radiation (transmitted or injected) and high spatial resolution (e.g., sub-millimeter) makes it advantageous over techniques involving radioisotopes. The introduction of higher magnetic fields (4.7–14 T) increases the signal-to-noise ratio, permitting higher resolution or faster scanning. The emerging field of hyperpolarized magnetic resonance [12–14] may improve the low sensitivity of the desired nuclei (e.g., 13C) and offer the use of stable isotope precursors for quantitative in vivo imaging and real-time metabolic profiling.
Probes for optical imaging that are excitable in the near-infrared (NIR) range are preferable for both in vitro and in vivo imaging. The ‘optical transmission window’ of biological tissues falls within the NIR range (λ = 650–900 nm). Investigation within this range allows for deeper light penetration and reduced light scattering, thus producing increased image contrast with excellent sensitivity of detection. In this review we will particularly emphasize advanced imaging methods and targeted nano-sized contrast agents for MRI and optical imaging modalities.
An understanding of magnetic resonance contrast agents is founded upon a rudimentary appreciation of MRI and the NMR phenomenon. The basic principles of NMR state that the intrinsic angular momentum or spins of protons (i.e., hydrogen nuclei) and electrons [10–14] when placed in a strong external magnetic field (B0) orientate themselves either parallel (i.e., spin-up) or antiparallel (i.e., spin-down) to B0. The overall impact, which is a function of B0, is minute, about 0.01–0.1 eV or approximately 10−6–10−7 more spin-up than spin-down states per voxel. Because tissues are predominantly water, this trivial distribution imbalance is perceptible! The ensemble of many spins exhibits a net magnetization that can be ‘tilted’ by magnetic gradients away from the direction of the main magnetic field after absorption of radiofrequency excitation energy. The transition from this excited state (tilted) back to the ground state is known as relaxation. magnetic resonance contrast is defined by the two-principle NMR processes of spin relaxation: T1 (spin-lattice or longitudinal relaxation time constant) and T2 (spin-spin or transverse relaxation time constant); relation rates are the inverse of the relaxation times (i.e., R1 = 1/T1, R2 = 1/T2) [10–12].
Magnetic resonance contrast agents accelerate the rate of T1 and T2 relaxation. Paramagnetic agents principally accelerate longitudinal T1 relaxation, producing ‘bright’ contrast in T1-weighted images (e.g., gadolinium based). Superparamagnetic agents primarily increase the rate of dephasing or transverse T2 relaxation and create ‘dark’ or negative contrast effects (e.g., iron oxide-based agents). T1 contrast agents directly influence protons proximate to themselves and are highly dependent on local water flux; whereas T2 contrast agents disturb the magnetic field beyond and independent of their immediate environment. The contrast impact of T2 agents extends well beyond their immediate surroundings, while T1 contrast agents have only very local influence. Thus, paramagnetic metals of T1 agents must be ideally exposed to water with fast exchange rates within an imaging voxel. In contrat, superparamagnetic metals can be sequestered anywhere within a supporting matrix and still elicit T2 contrast.
The term relaxivity (r1 or r2) refers to the change in relaxation rate (R1 or R2) as a function of contrast metal ion concentration and is expressed in mM−1s−1 (at varying temperatures); this is also referred to as ‘ionic’ relaxivity. However, for nanoconstructs with many metals associated with each particle, the moiety is considered as one unit, in which case the terms ‘molecular’ or ‘particulate’ relaxivity are often applied [15–16].
Early attempts to create targeted paramagnetic contrast agents failed, due to meager or poorly effective metal payloads (i.e., exposed to fast water exchange) per homing unit (e.g., antibody). In the late 1990s, the development of targeted paramagnetic contrast agents was considered infeasible by the magnetic resonance community. However, in 1998, Sipkins demonstrated in vivo imaging of angiogenesis with paramagnetic polymerized liposomes in the Vx2 tumor model  and, the same year, Lanza et al. reported fibrin imaging with paramagnetic per-fluorocarbon nanoparticles ; the prevailing dogma was proved wrong. Subsequently, alternative nonparticulate approaches to magnetic resonance molecular imaging have found success against abundantly expressed epitopes such as HER-2/neu receptors, using an avidin conjugated to gadolinium-DTPA (12.5 Gd/avidin) and a thrombus-avid, fibrin-binding peptide derivatized with gadolinium 19. A few years later, integrin-targeted liposome constructs, similar to polymerized liposome particles of Sipkins were reported by Mulder et al. for angiogenesis imaging in rodent cancer models [20–22]. Alternative approaches with paramagnetic lipoprotein mimics were similarly reported for macrophage imaging of atherosclerotic plaque [23–24].
Ligand-targeted paramagnetic liquid per-fluorocarbon (PFC) nanoparticles have been extensively demonstrated in vivo for molecular imaging and targeted drug delivery directed against fibrin in ruptured atherosclerotic plaques , as well as integrins of angiogenic endothelium in atherosclerosis [4,26–28] and nascent cancers [29–32]. For PFC paramagnetic nanoparticles, the ‘ionic relaxivity’ r1 is 35 (mM [Gd+3] s)−1 and molecular r1 relaxivity is > 2000,000 (mM[paramagnetic particles]s)−1 [15–16]. This extraordinarily high molecular relaxivity of targeted nanoparticles permits a voxel containing fewer than 100pM of the agent to be detected conspicuously, with a contrast-to-noise ratio of 5 at 1.5 and 3.0 T .
For the PFC nanoparticles, the high surface-to-volume ratio of the nanoparticle accommodated 100,000 chelated gadolinium atoms per particle, resulting in outstanding T1 signal amplification (bright) per particle bound to a receptor. However, an alternative approach to achieve these T1 signal amplification with dendrimers utilizes effective cellular internalization and concentration of the agent to build signal. This approach involves clever targeting of a cell membrane receptor specific system to mediate intracellular concentration of the imaging drug delivery agent. Recycling of the receptor between the membrane and the cytoplasm serves to continually amplify the target cell-specific concentration of contrast as described below.
Polyamidoamine dendrimers are an ideal platform for this type of targeted cell uptake and delivery due to their small size, monodispersed population and numerous surface amines to conjugate targeting and imaging molecules. As required, in order to escape the vasculature through vascular pores, dendrimers are less than 50 nm in diameter; this is well below the typical ceiling (150 nm) for endosomal particle internalization in nonphagocytic cells. Two versions of this approach have been used successfully in dendrimers to mark specific cells in vivo.
Bulte et al. explored the use of magnetic nanoparticle contrast agents as a means of following stem cell dispersal after grafting . Bulte’s research showed that it was possible to label and MRI-track stem and progenitor cells for up to 60 days after grafting them into a rat model. Bulte et al. utilized amine-terminated dendrimers that encased iron-oxide particles to promote highly enhanced nonspecific endocytosis by cells. However, these cationic particles can elicit deleterious biological effects, such as the activation of complement and clotting pathways. Despite the uniqueness of this work, the intent of the research is ex vivo cell labeling and neither were expected to work in vivo.
A second dendrimer-based approach, derived from the seminal work of Wiener et al. , was extended by Baker and his colleagues. For these works, the dendrimer surface was modified with different chemistries to achieve high cell avidity for monomeric targeting ligands by attaching multiples of these molecules . The highly specific delivery of drugs to tumor cells overexpressing folate receptors was illustrated in vivo with negligible nonspecific binding or toxicity. The Baker group employed this platform to deliver Fe3O4 [36,37] and gold , complexed within the interior of the dendrimer, as well as Gd+3 chelated to the surface to tumor cells with impressive specificity and uptake. MRI imaging demonstrated the specific marker (in this case the folate receptor) on the surface of tumor cells  and also provided confirmation of methotrexate drug delivery .
Beyond folate targeting, dendrimeric particles offer many opportunities to combine imaging and drug delivery for various cell and tumor markers. Other examples include the use of an anti-HER2 F(ab) coupled to dendrimers that identified HER2-expressing tumor cells, and a similar approach with anti-EGF binder to delineate EGF-overexpressing tumor cells . At this juncture, almost any cell with a specific membrane marker can be imaged and similarly treated by targeted dendrimeric contrast agents.
Neovessel formation (i.e., angiogenesis) is an important biosignature of cancer. One molecular signature, αvβ3-integrin, has recieved prominent attention for angiogenic-targeting applications, because it is expressed on the luminal surface of activated endothelial cells but not on mature quiescent cells. The αvβ3-integrin, a heterodimeric transmembrane glycoprotein, is expressed by numerous cell types, including endothelial cells [41–42], macrophages , platelets , lymphocytes , smooth muscle cells  and tumor cells [46,47]. Fortunately, the steric constraint of perfluorocarbon nanoparticles to the vasculature precludes significant interaction with nonendothelial integrin-expressing cells, which greatly enhances neovascular target specificity [48,49].
αvβ3-integrin-targeted paramagnetic nanoparticles sensitively detected histologically corroborated angiogenic endothelium at 1.5 T in New Zealand White rabbits bearing Vx2 tumors (<1.0 cm) implanted into the hind limb 12 days previously . In vivo competition studies in that report demonstrated that the ligand-directed homing was specific for the αvβ3-targeted nanoparticles. Moreover, the T1-weighted images obtained with αvβ3-targeted nanoparticles differentiated growing tumors from the inflammatory remnants of host-rejected cancer . These tumor remnants were not differentiated from growing cancers using standard T2-weighted MRI, nor would 18F deoxyglucose PET imaging be expected to discern the high oxygen utilization of macrophage metabolism from proliferating Vx2 tumor cell demand. This simple example illustrated how magnetic resonance molecular imaging with paramagnetic nanoparticles provides fundamental data that could support medical decisions to biopsy, treat or observe anomalous pathology or to interrogate early lesion response to chemotherapy or radiation treatment.
A follow-up study in athymic mice demonstrated that angiogenesis induced by very minute tumors, less than 40 mm3, could be detected by magnetic resonance within 0.5 h following IV administration of αvβ3-integrin-targeted PFC particles; this contrast signal continued to strengthen over the next 2 h (Figure 1) . Again, in vivo competition studies demonstrated the high specificity of ligand-directed targeting, which was further corroborated by αvβ3-targeted bimodal particles (fluorescent and MR) and immune fluorescent microscopy .
Anti-angiogenesis therapy in conjunction with chemotherapy or radiation therapy has become a well-established treatment for lung, colon and breast cancer [50–52]. However, only a limited subset of patients achieve optimal effectiveness from anti-angiogenic pretreatment; moreover, the clinical timing of the treatment and its duration are not individualized to acute therapeutic response. PFC paramagnetic nanoagents offer a sensitive and personalized approach for selecting patients for treatment and even delivering the anti-angiogenic therapy itself. In addition to ultrahigh payloads of paramagnetic chelates, enabling diagnostic MRI, αvβ3-targeted nanoparticles can incorporate therapeutic agents for effective targeted drug delivery [1,4,27,28,31,53–56]. Such dual-function agents are referred to as ‘theranostics’. Chemotherapeutics, such as paclitaxel, doxorubicin, rapamycin and fumagillin, as well as toxic peptides, such as mellitin and thrombolytic enzymes, have been effectively incorporated into the surfactant of PFC nanoparticles [2,4,28,31,54–55]. Theranostic agents can noninvasively quantify the delivery of potent chemotherapeutic agents, such as fumagillin, as well as provide quantitative measures of response for improved longitudinal medical management. Moreover, such agents vastly reduce total drug exposure with an anticipated benefit of improved safety.
Fumagillin is a mycotoxin produced by Aspergillus fumigatus, which suppresses angiogenesis by inhibition of methionine amino-peptidase 2 (MetAP2) [57–58]. TNP-470, a water-soluble functional analogue of fumagillin, selectively inhibits proliferating endothelial cells (i.e., angiogenesis). The anti-tumor efficacy of TNP-470 was widely demonstrated in rodents[59–62] and later studied in human clinical trials [63–67]. Unfortunately, at dosages required for therapeutic effects, TNP-470 elicited sudden moderately severe symptoms of neurotoxicity, including weakness, nystagmus, diplopia and ataxia [63,64,67]
Inclusion of fumagillin into the surfactant of the αvβ3-targeted nanoparticles enables the delivery of the drug into proliferating endothelial cells via ‘contact-facilitated transport’, which is promoted by the ligand-based tethering of the nanoparticle to the target cell surface and the spontaneous exchange of the lipid surfactant components with similar membrane lipids through hemifusion complexes (Figure 2) [1,53,56].
The anti-angiogenic effectiveness of αvβ3-targeted fumagillin nanoparticles was studied in the syngeneic Vx2 adenocarcinoma rabbit model using a minimal fraction of the TNP-470 dosages previously used in preclinical animal models . The Vx2 tumor volume was reduced by one-half to two-thirds among rabbits receiving αvβ3-targeted fumagillin nanoparticles, when compared with animals given nontargeted fuma-gillin nanoparticles, αvβ3-targeted nanoparticles without drug or saline. Moreover, 3D reconstruction and mapping of the contrast enhanced voxels over a chain model of the tumor surface demonstrated a coherent asymmetric peripheral distribution characterized by dense neovessel regions interspersed with a finer, reticular pattern (Figure 3).
19F presents an excellent probe for quantitative MRI, which is highly enriched in perfluorocarbon nanoparticles. 19F has 100% natural abundance, a spin of 1/2 and a gyromagnetic ratio of 40.08 MHz/T, close to the that of 1H (42.58 MHz/T), resulting in 83% of the sensitivity of 1H . In addition, the chemical shift of 19F, due to its seven outer-shell electrons, is sensitive to the molecular environment of the nucleus, including oxygen tension. The 19F spectroscopic signature manifests a range of >200 ppm [69–70], which permits unambiguous identification of distinctive 19F-containing compounds, even at low field strengths. Moreover, no background exists for the 19F signal in vivo, providing a unique spectroscopic signature for quantitative MRI.
Tissue oxygenation is a major therapeutic concern in many pathologic lesions, particularly in the treatment of cancers . The rapid growth of malignant tumors, combined with formation of dysfunctional neovasculature, results in regional hypoxia, which reduces susceptibility to chemotherapy and radiation therapy . Quantitative 19F MRI of PFCs were explored as an MRI method for in vivo mapping tumor hypoxia [73–75]. Although blood oxygen level dependent (BOLD) has been used in MRI of oxygen utilization and hypoxia in neurology, psychology and oncology [76–81], its qualitative information may be less preferable compared with the quantitative data achievable with PFCs . At a given temperature, the partial pressure of dissolved paramagnetic O2 (pO2) in PFCs directly correlated to the relaxation rate of 19F or R1, which can be estimated as a measure of local in vivo tissue oxygenation. Since some PFCs have lower sensitivity to physiological temperature variations (30–42°C), such as hexafluorobenzene (HFB), local 19F R1 can estimate tissue oxygenation using a priori calibrated standard R1–pO2 standard curves . The precision of the 19F MRI method can reach 1–3 mmHg in hypoxic region , comparable to fine electrode measurements . Moreover, fast 19F MRI techniques, such as fluorocarbon relaxometry using echo planar imaging for dynamic oxygen mapping (FREDOM), have been developed for dynamic mapping of tumor pO2 with high precision .
The unique capability of 19F MRI to directly determine the absolute quantity of 19F atoms has been largely unexploited until the recent advent of PFC nanoparticle-based molecular imaging, beginning in 1996 . Although these targeted nanoparticles were initially used to deliver high payloads of surface gadolinium to enhance T1-weighted 1H MRI, as discussed earlier, they inherently resulted in regional deposition of 19F atoms adequate for quantitative MRI. Using fibrin-targeted PFC nanoparticles, the particles bound on the surface of fibrin clots provided enough 19F atoms for MRI at 4.7T field strength [87,88]; later, the concept was extended to a clinical scanner at 1.5T . The linear correlation between the quantity of bound 19F atoms and the measured 19F magnetic resonance signal from functionalized PFC nanoparticles has been quantitatively mapped in clot and tissue phantoms . PFOB and perfluoro-15-crown-5-ether (PFCE) nanoparticles targeted to the same biological specimen were simultaneously or selectively assessed with 19F MRI at 1.5T, illustrating the unique spectral opportunity for phenotypic characterization achievable by delineation of multiple pathological biosignatures simultaneously . Subsequently, Neubauer et al. further demonstrated that paramagnetic Gd3+, closely bound to the surfactant surface of PFC nanoparticles, increased the 19F R1 fourfold, improving 19F signal intensity 125% at 1.5 T .
Examples of 19F imaging of sparse epitopes, as presented by biological tissues, has included detection of atherosclerotic valve angiogenesis . Angiogenic valve leaflets in hyperlipidemic rabbits following αvβ3-integrin targeted PFC nanoparticles exhibited approximately three-times higher 19F signal ex vivo than similar valve leaflets in atherosclerotic animals treated with untargeted nanoparticles . Similarly, molecular imaging with VCAM-1 targeted PFC nanoparticles in the kidneys of the ApoE mouse showed the potential assessing vascular inflammation . Potentially, in vivo quantitative 19F MRI, using site-targeted PFC nanoparticles, can be complicated by the background signal from unbound particles circulating in the blood pool depending on the dose, timing of imaging and field strength (particularly above 4.7 T); however, the use of diffusion-weighted 19F MRI techniques to selectively suppress the 19F signal from circulating nanoparticles versus the targeted bound particles eliminates this background problem. 
Regenerative therapy using stem cells offers great promise for the treatment of many diseases. To date, 1H MRI using iron oxides has served as a primary method for monitoring stem cell trafficking . Typically, in vitro cultured stem cells are incubated with 1H MRI contrast agents such as superparamagnetic iron oxide (discussed later) or paramagnetic Gd-DTPA [95,96], which results in uptake by endocytosis and modification of the cells as imaging agents. Labeled stem cells can be detected by 1H MRI based on the negative (i.e. dark) or ‘positive’ (i.e. bright) contrast effects, dependent on the labeling approach applied. Moreover, such 1H MRI techniques appear sensitive to even a single cell under certain circumstances .
19F MRI can be used as an alternative method for quantitative trafficking of stem cells in vivo . Ahrens et al. demonstrated that PFCE nanoparticles could be effectively internalized by dendritic stem cells with the help of a cationic transfection agent and specifically detected in vivo by 19F MRI at 11.7 T . Subsequently, phospholipid-encapsulated PFC nanoparticles were shown to internalize into stem/progenitor cells without the need of a transfection agent and be detected by 19F MRI at both clinical (1.5 T) and research (11.7 T) field strengths . Ruiz-Cabello et al. recently showed that PFCE nanoparticles with cationic surface charge were also effectively internalized by neural stem cells . In another example, in vivo monitoring of the migration of PFCE nanoparticle labeled T-cells 48 h into the pancreas of diabetic mice after intraperitoneal injection . Currently, a minimum of 2000 labeled cells is required for 19F cell-trafficking detectability, versus the single cell possibility with iron oxides .
One of the earliest applications of nanotechnology in MRI involved the use of paramagnetic iron-oxide particles. Iron-oxide crystals have long been used as superparamagnetic T2* contrast agents for MRI [103–106]. Superparamagnetic iron oxide (SPIO, particle diameter > 50 nm) and ultrasmall super-paramagnetic oxide (USPIO, particle diameter < 50 nm) particles have nonstoichiometric microcrystalline magnetite core(s) and are typically coated with dextran (e.g., ferumoxide) or siloxane (e.g., ferumoxsil) . While at very low doses, circulating iron oxides can decrease the T1 time of blood for positive contrast enhancement with magnetic resonance angiography, at the usual doses used for molecular imaging, the T2* effects predominate resulting in short T2 relaxation with marked signal loss [105,108–109]. Unfortunately, for molecular-imaging applications, persistent T2* effects from circulating iron oxides nanoparticles delays MRI by 24–72 h postinjection [110–112], complicating clinical implementation of these techniques. Moreover, iron oxides concentrated at a target site generate magnetic susceptibility artifacts (dark or negative contrast) with a marked blooming effect, which smear into adjacent nontargeted voxels.
Spontaneous phagocytic uptake of SPIO and USPIOs by macrophages in atherosclerotic plaques was recognized and demonstrated in 2000 and 2001 by Schmitz et al. and Ruehm et al. in hereditary or diet-induced hyper-lipidemic rabbits [113–116]. In 2003, this finding was extended to include human plaque . Systematic evaluation of USPIO-enhanced MRI contrast in carotid atheroma confirmed that the optimal signal intensity was achieved 24–36 h after administration. Subsequently, the USPIO compound ferumoxytol was compared with ferumoxtran-10 as a marker of macrophage activity in atherosclerotic plaques. While both were reported to be effective, ferumoxytol had optimal luminal signal intensity 3 days post-treatment and ferumoxytol-treated rabbits had peak measurements 5 days after injection [11,8]. Recently, new magnetic resonance pulse sequences and image postprocessing techniques have been developed to reverse the dark contrast appearance into a bright positive contrast effect [119–126]. For example, inversion-recovery with ON-resonant water suppression (IRON) allowed the dark magnetic distortion artifacts associated with T2* USPIO imaging to be viewed as positive contrast that correlated with macrophage counts in hyperlipidemic rabbits best approximately 72 h post-treatment .
The development of monocrystalline iron oxide nanoparticles (MION) helped to extend magnetic resonance iron oxide imaging beyond the limitations of passive targeting through tissue accumulation and particle phagocytosis to ligand-directed or active targeting. MIONs have an average core diameter of 3 nm and can be directly coupled to homing ligands that specifically target epitopes in the tissue of interest. Dextran-coated MION, coupled to human holo-transferrin (Tf-MION), was used to visualize transgene expression in a gliosarcoma mouse model in vivo . In those experiments, the cellular uptake of MION increased approximately 500% relative to control cells following overexpression of engineered transferrin receptor. In other experiments, MION was used to estimate blood volume distribution and indirectly assess angiogenesis in brain tumors . The targeting efficiency of iron-oxide particles improved further with the development of dextran cross-linked iron oxide (CLIO) particles . CLIO has been used with a variety of ligands, including E-selectin , a peptide sequence from the transactivator protein (Tat) of HIV-1 [131–134], annexin V  and VCAM-1 . Although these particles may be demonstrated with histology to target tissue specifically soon after injection, noninvasive MRI remains delayed and the combination of particle clearance into macrophages and nonspecific particle diffusion within tissue convolves the imaging result.
One recent approach to address the marked delay between targeted iron-oxide administration and imaging has been a return to using large microparticulate iron oxides. In high-field (9.4 T) rodent experiments, iron microparticles were targeted to the gpIIbIIIa receptor of platelets , and a leukocyte-mimicing, dual-epitope-targeting approach directed against vascular inflammatory biomarkers was imaged following intracardiac injection . These results suggest a potential role for large particle approaches, although previous issues of intra-vascular agglomeration and rapid pulmonary clearance when administered IV may still need to be addressed.
Colloidal iron oxide nanoparticles (CIONs) are a vascular constrained T1w molecular-imaging agent that avoids typical magnetic bloom artifacts, permits rapid in vivo molecular imaging without blood pool magnetization interference and supports targeted drug delivery [138,139]. A CION is comprised of iron oxide suspended within a hydrophobic matrix and encapsulated within a partially crosslinked phospholipid capsule, which decreases T2 effects more than T1, an entirely unexpected result. A CION has rapid clearance of circulating interference on T1 contrast (<60 min), while blood T2 shortening persists well over 2 h, as expected for superparamagnetic agents. Moreover, CION is designed for therapeutic drug delivery (e.g., fumagillin) via a unique mechanism termed ‘contact-facilitated drug delivery’.
Manganese-doped superparamagnetic iron oxide (Mn-SPIO) nanoparticles were studied to create ultrasensitive MRI contrast agents for liver imaging. Hydrophobic Mn-SPIO nanoparticles encapsulated in block copolymer mPEG-b-PCL micelles form self-assembled small clusters (mean diameter: ~80 nm). Mn-SPIO nanoparticles within the micelles decrease are superparamagnetic at room temperature. At the magnetic field of 1.5 T, Mn-SPIO nanoparticle-clustering micelles have a T2 relaxivity of 270 (Mn + Fe) mM−1s−1, which is much higher than a single Mn-SPIO nanoparticle in a lipid-PEG micelle. This clustered nanocomposite approach has yielded significant liver contrast with signal intensity decreases of approximately 80% in 5 min after intravenous administration. The time window for enhanced MRI is highly prolonged and the contrast in liver images is marked, offering potential to identify the smallest liver lesions and differential diagnosis of other liver diseases . Similar particles have been reported in combination with chemotherapeutic agents for targeted magnetic resonance detection and treatment of breast cancer [1,14].
Manganese was one of the first reported examples of paramagnetic contrast material studied in cardiac and hepatic MRI, because of its efficient R1 enhancement. Similar to Ca2+ and unlike the lanthanides, manganese is a natural cellular constituent and often a cofactor for enzymes and receptors. Manganese blood-pool agents, such as mangafodipir trisodium, have been approved as a hepatocyte-specific contrast agent but transient side effects due to dechelation of manganese from the linear chelate has reduced their use [142.143].
Aime et al. reported Mn(II) complexes bearing benzyloxymethyl functionalities that displayed relaxivity values only slightly smaller than those shown by the most clinically used contrast agents . In these Mn(II) chelates, the exchange rate of the coordinated water was much higher than those reported for Gd(III) complexes with octadentate ligands. This group exploited these rapid coordinated water exchange rates to develop macromolecular adducts with human serum albumin. The albumin-Mn(II) binding attained relaxivity values analogous Gd(III) systems. These results strongly supported the view that Mn(II) complexes, in spite of the lower effective magnetic moment, could be considered viable alternatives to the currently used Gd(III) complexes as contrast agents for MRI applications.
Similarly, Caravan et al. designed, synthesized and demonstrated a novel Mn(II) complex based on EDTA, which was functionalized with a bis-phenyl moiety to noncovalently complex with serum albumin, analogous to the gadolinium-based contrast agent MS-325. Albumin-binding measurements revealed that the complex adhered to plasma proteins (93–96%) primarily via serum albumin (rabbit: 89–98%). Relaxivities of the albumin-Mn-EDTA complex were: r1 = 5.8 mM−1s−1 (buffer); 51 mM−1s−1 (rabbit plasma); and 46 mM−1s−1 (human plasma). Variable-temperature NMR diffusion profiles indicated that the high relaxivity was due to slow tumbling of the albumin-bound complex and fast exchange of the inner sphere water. Imaging in an injured rabbit carotid artery model clearly showed that the agent could delineate both arteries from veins and differentiate healthy from damaged vessel wall. Again, suggesting that Mn-based agents could be used for T1w imaging in lieu of gadolinium .
MRI based on MnO nanoparticles has been used to elicit bright signal enhancement and fine anatomic detail in the T1-weighted (T1w) magnetic resonance image of a mouse brain. The MnO nanoparticles were targeted to breast cancer and selectively detected breast cancer cells in a metastatic tumor in brain . Subsequently, this Korean group greatly improved the relaxivity of MnO nanoparticles by developing MRI jack-o′-lanterns. These hollow manganese oxide nanoparticles, while retaining efficient cellular uptake, illustrate the potential of this bifunctional theranostic approach .
Another unique Mn-based agent was reported as manganese (III)-labeled nanobialys, which were shown to have potential as targeted magnetic resonance theranostic nanoparticles . The nanobialys produced by molecular selfassembly of amphiphilic-branched polyethyl-enimine assumed a toroidal, biconcave shape, which was tunable with regard to particle size and exhibited low polydispersity. The bialys presented Mn(III) in a kinetically stable, porphyrin-coupled complex directly exposed to the surrounding water. The vascular constrained nanobialys (180–200 nm) have ionic r1 and r2 relaxivities of 3.7 ± 1.1 (s · mmol [Mn])−1 and 5.2 ± 1.1 (s·mmol [Mn])−1, respectively and particulate relaxivities of 612,307 ± 7,213 (s·mmol [nanobialy])−1 and 866,989 ± 10704 (s·mmol [nanobialy])−1, respectively. The concept of Mn(III) nanobialy efficacy was illustrated in vitro using antifibrin monoclonal targeting to fibrin-rich clots.
Subsequently, we advanced the concept of ‘soft-type’ nanocolloids with the development of manganese oxide and manganese oleate nanocolloids, which incorporate divalent manganese . These nanocolloids, encapsulated by phospholipids, are designed for vascular targeting (>120 nm) similar to CION and PFC nanoparticle formulations. However, unlike the paramagnetic gadolinium PFC nanoparticles, the manganese was uniquely entrapped within the core matrix. Manganese nanocolloids presented a multiplicity of specific surface-homing ligands for high avidity and sensitivity for molecular imaging with MRI (excellent ‘stick and stay’ quality), such as antifibrin antibodies and more recently αvβ3-intergrin antangonists.
Magnetic resonance experiments demonstrated the high-resolution T1w molecular imaging in suspension and against fibrin with ManOC and ManOL. The ionic r1 relaxivities of ManOC and ManOL were 4.1 ± 0.9 (s·mmol [Mn])−1 and 20.4 ± 1.1 (s·mmol [Mn])−1, respectively, while the particulate relaxivities were 85,099 (s·mmol [ManOC])−1 and 631,208 (s·mmol [ManOL])−1, respectively. The ionic r2 relaxivities of ManOC and ManOL were 18.9 ± 1.1 (s·mmol [Mn])−1 and 65.6 ± 0.9 (s·mmol [Mn])−1, respectively and the r2relaxivities were 395,410 (s·mmol [ManOC])−1 and 2,028,925 (s·mmol [ManOL])−1, respectively. The specific relaxivities, obtained by measuring the relaxation rate as a function of the concentration of the contrast agent were also found to be markedly and unexpectedly increased for the ManOL, compared with the ManOC. Fibrin-rich clots were targeted in vitro with fibrin-specific monoclonal antibodies, revealing markedly increased T1w images of the targeted specimens, yielding signal intensities (75 ± 20 and 95 ± 19 au, respectively) verus the control nanocolloid (32 ± 07 au) and background air (7 ± 4 a.u.).
Optical imaging is a noninvasive, relatively low-cost technology that uses light to probe cellular and molecular function in the living body. The technique has gained tremendous attention due to its high sensitivity and high suitability for small-animal studies. In fluorescence imaging, the subject of interest is typically illuminated by excitation light and at a shifted wavelength emitted light is collected. The contrast is derived either from the use of exogenous agents or from endogenous molecules with inherent optical properties. Generally, a spatial scale of up to centimeters can be resolved by the optical-based techniques. However, in most cases, tissue auto-fluorescence generates significant background signal. Therefore, the information that is typically derived is not quantitative in nature and image information is mostly surface-weighed due to excessive tissue absorption. With the advent of recently developed advanced techniques such as fluorescence molecular tomography  and photoacoustic tomography [151–152], it is expected that optical-based techniques will gain more attention.
The NIR spectral window (700–1000 nm) offers unique opportunities for deep tissue fluorescence imaging . Light in this range can penetrate deep into living tissue, which is not possible below 700 nm. In the visible range (<700 nm) the scattering is high and the presence of strong exogenous absorbers (e.g., water, hemoglobin, deoxyhemoglobin and lipid) does not permit the light to penetrate deep into the tissue. Detection of contrast from the externally administered exogenous contrast is also difficult owing to the high characteristic auto-fluorescence from tissue and cellular components. The penetration of light at this imaging window is at maximum with minimum loss due to hemoglobin and water absorption. Therefore, the feasibility of deep imaging of the cells can be envisioned within this window . The immense potential offered by NIR imaging may enable us to detect physiological, metabolic and molecular function in vivo. In the past few years, a considerable amount of progress has been made in the areas of optical instrumentation and reconstruction algorithm, towards developing smart probes for NIR imaging for various diseases.
Numerous approaches have been proposed for optical imaging in terms of using fluorescent dyes. Some of the prerequisites for these dyes are:
The nature of the materials allows optical contrast agents to be classified into three main categories: organic, inorganic or hybrid in nature (Figure 4). The class of inorganic contrast agents is dominated by quantum dot (QD) particles . Other examples of this class are carbon dots and gold nanoparticles . Gold nanoparticles have been prepared in different shapes and sizes (e.g., spheres, rods or hexagonal) . These exogenous agents can be categorized in two major types: nonspecific and specific to cellular targets . BODIPY  and lanthanide chelates  are examples of hybrid optical agents. Cyanine dyes are an important class of optical agents, which have gained tremendous attention due to their favorable NIR optical properties (Figure 5) . Tetrapyrroles such as porphyrins, chlorins and phthalocyanines are another example of organic small-molecule dyes . Depending on the nature and mode of their activation, these dyes can be nonspecific, targeted or activatable in nature.
Quantum dots are as among the most widely studied optical contrast agents. They are inorganic fluorophores based on cadmimun sulphide- and cadmium selenide-type semiconductor crystals. QDs offer many advantages over traditional organic dyes, such as high photo stability, high quantum yield, continuous emission spectra spanning UV to NIR window, narrow emission spectral band width (20–30 nm) and long fluorescence lifetimes (10 ns or higher). Simulation results suggested two spectral windows for quantum dots (i.e. 700–900 and 1200–1600 nm). The excited state lifetimes (10–50 ns) of QDs are almost one order of magnitude higher than that of small-molecule organic dyes (2–5 ns), which makes them an interesting choice for fluorescence lifetime imaging of cells, tissue specimens and living animals. Since their conception, QDs have found wide applications in cells and in in vivo optical imaging, including cell trafficking, sentinel lymph-node imaging, neuro imaging, for example. QDs have been used as a stable fluorescent tracer for nonspecific uptake studies and lymph-node mapping in living animals [162,163]. Earlier in 2002, Akerman et al. demonstrated the feasibility of targeted ex vivo imaging with QDs . In this seminal work, ZnS-capped CdSe QDs (10 nm) were coated with a lung-targeting peptide that accumulated in the lungs of mice after intravenous injection, whereas two other peptides specifically directed QDs to blood vessels or lymphatic vessels in tumors. They also demonstrated that a polyethylenglycol coating of the QDs prevented their nonselective accumulation in reticuloendothelial tissues. In another work, QDs have been encapsulated in phospholipid micelles and injected directly into frog oocyte cells for real-time tracking of embryonic development .
αvβ3-integrin remains an attractive biochemical epitope that is highly expressed on activated neovascular endothelial cells and which is essentially absent on mature quiescent cells. RGD tripeptide-functionalized QDs (λem = 705 nm) exhibited high-affinity integrin-specific binding in cell culture and ex vivo. NIR fluorescence imaging was also demonstrated in nude mice bearing subcutaneous implanted integrin-positive U87MG human glioblastoma tumors . Real-time imaging and tracking of single-receptor molecules has been made possible with antibody-conjugated QDs on the surface of living cells [167,168].
Phospholipid encapsulated QDs encountered some fluorescence loss and stability issues [164,169], In another important report, Shuming and coworkers designed a polymer-encapsulated QD specific for prostate cancer cells in mice. ABC triblock copolymer uniformly coated the QDs and prevented particle aggregation and fluorescence loss by forming a stable hydrophobic protective layer around single QDs. Prostate specific membrane antigen antibody-coated nanoparticles were highly stable in vivo and demonstrated spectral imaging in live animals harboring C4-2 tumor xenografts. The results were supported by histological and immunohistochemical evidence .
Very recently, synthesis of a QD linked to α-fetoprotein (AFP) antibody has been reported for studying specific binding AFP, an important marker for hepatocellular carcinoma cell lines. QD-anti-AFPs were found to target the tumors in vivo and the authors investigated the inhomogeneous distribution of the probes in the tumor by using a site-by-site measurement method. The technique successfully demonstrated their ability to detect the distribution of the probes within cancer cells 
Another interesting example is NIR type II quantum dots. Frangioni and co-workers synthesized 10-nm-sized NIR QDs encapsulated by an oligomeric phosphine coating . This poly-dentate phosphine coating rendered the parent particles soluble, dispersed and stable in serum. When these particles were injected (400 pmol) in mice, sentinel lymph nodes were detected 1-cm deep in real time, using excitation fluence rates of only 5 mW/cm2. This approach was considered important for real-time, operating room sentinel lymph node resection, a common procedure in breast cancer surgery.
Typically, semiconductor QDs are prepared in organic solvents under high-temperature condition. Prior to their in vivo application, they are rendered water soluble by biocompatible, hydrophilic entrapment. This encapsulation, often challenging, can generate aggregated particles with poor shelf life and in vivo stability. Due to their agglomerated size and short circulation half lives, the in vivo targeted imaging would be very challenging with these types of particles and very few successful examples have been reported so far.
Quantum dots include heavy metal crystals that could elicit toxicity in their soluble form . Given the known toxicity potential of Cd, cadmium-free examples of QDs, including InAs and CuInSe, have been proposed . Bawendi and co-workers report the development of luminescent QD materials to ternary I-III-VI semiconductor systems by following a modular hot injection synthetic method . Strongly luminescent CuInS2/ZnS core/shell nanocrystals were synthesized by the same group from copper iodide, indium acetate, zinc stearate and dihydrolipoic acid as a surface ligand. CIS/ZnS NCs are the first example of lower toxicity QDs used for in vivo imaging.
The development of quantum-sized carbon analogues has been reported very recently by Ya-Ping Sun and co-workers . The carbon dots were produced via laser ablation of a carbon target in the presence of water vapor with argon as carrier gas and the surface of the particles was passivated by a diamine-terminated oligomeric PEG H2NCH2(CH2CH2O)nCH2CH2CH2NH2 (average n = 35, PEG1500N). Interestingly, the passivated carbon dots with organic moieties attached to the surface were strongly photo-luminescent in suspension and the emissions cover both the visible and NIR wavelengths. Subsequently, the same group of authors demonstrated that carbon dots injected into mice provided strong fluorescence in vivo, which, combined with their biocompatibility and non-toxic characteristics, suggested great potential for optical molecular imaging . Another interesting example of fluorescent carbon nanoparticles (CNPs: 2–6 nm in size) with a quantum yield of approximately 3% was synthesized via nitric acid oxidation of carbon soot .
Flurorescein and dyes with related structures (i.e., xanthane dyes) [178,179], have high quantum yields (fluorescein = ~0.8) and have been widely used in optical imaging studies. Numerous fluorescein intermediates and derivatives are commercially available. These probes typically absorb and emit in the visible wavelength and are generally only suitable for superficial imaging. As a result, primary applications have been in laboratory instruments operating at visible wavelengths with limited pharmaceutical development activity to date.
Cyanine dyes have found numerous applications as nonlinear optical materials  and in biomolecular labeling [181–183], DNA sequencing and in vivo imaging . They have been employed as labels in fluorescence imaging studies of biological mechanistic pathways. NIR cyanine dyes have an advantage, in that light at their emission and absorption maxima are perfectly tuned in the 650–900-nm range where absorbance by biomolecules is relatively poor and autofluorescence is less. They have found wide interest among synthetic chemists due to their straightforward synthesis [185–186], broad wavelength tunability and large molar absorptivities. Cyanine dyes can be classified into two broad categories based on structural connectivity: basic and adaptable. In basic cyanine dyes, two aromatic rings are connected by a polymethine chain with conjugated C-C double bonds. Adaptable cyanine dyes consist of reactive -Cl or -NH2 groups in the central meso position for postsynthetic modifications via SRN1 reactions with nucleophiles (e.g., amines  or thiols ).
Targeted cyanine dyes are meant to achieve higher target-to-background signal ratios and may provide more specific information from the diseased tissue of interest than nontargeted dyes. Typically, cyanine dyes are developed carrying functional groups for conjugation to macromolecules (e.g., antibodies, peptides and nanoparticles and polymers). The dye-to-macromolecule ratio needs to be properly optimized to positively impact the fluorescence properties of the conjugates .
Novel VCAM-1-targeted MRI and optical-based multimodal imaging agents were synthesized recently. A phage display-derived peptide sequence, containing the VHSPNKK motif, was shown to bind VCAM-1. The peptide sequence and a cyanine 5.5 dye were conjugated to an iron oxide-based (CLIO) nanoparticle. These particles were injected into a cholesterol-fed apoE−/− mice to determine whether VCAM-1 expression could be detected in vivo in atherosclerotic lesions. Typically, extensive neovascularization and VCAM-1 expression are seen in this model. Extensive decrease in signal intensity associated with iron oxide NP accumulation in atherosclerotic lesions was observed in magnetic resonance and the results were further corroborated by macroscopic epifluorescence imaging of Cy 5.5 . Cy5.5-labeled fluorescent probes for targeting thrombin and activated-platelets were developed based on the glycoprotein IIb/IIIa (GP-IIb/IIIa)-binding sequence, Pro-Ser-Pro-Gly-Asp-Trp. Both linear and branched sequences were synthesized, and provided at least five-times greater signal enhancement than the control (irrelevant peptide) when targeted to in vitro plasma clots. 
Weissleder and Ching-Tung synthesized a new class of stable, monocarboxylate functionalized cyanine dyes. The synthetic strategy was based on the nucleophilic attack of alkyl-thiols on cyanine dyes bearing chlorosubstituted polymethinic linkers . Similar results were obtained with monocarboxylate-derivatized fluorochromes (CyTE dyes). NIR fluorescence microscopy was able to monitor the endothelial cell internalization of the CyTE dye labeled VCAM-1 targeting peptide .
Licha and co-workers synthesized a group of glucamine and gluosamine-substituted cyanine dyes structurally related to indocyanine green . In vivo results in rats indicated that, in order to overcome the limitations in spatial resolution and sensitivity, the strategy might facilitate optical-imaging applications where intrinsic contrast between tumor and healthy tissue needed to be amplified. The direct labeling of phage displaying a VCAM-1-targeting peptide was demonstrated using these fluorophores and the endothelial cell internalization of the VCAM-1-targeted phage was monitored via NIR fluorescence microscopy.
Renard and co-workers reported the synthesis of two novel water-soluble NIR cyanine dyes. When these dyes were connected within a peptide, mutual fluorescence quenching between the dyes was observed at both 705 and 798 nm. Based on this property, an internally quenched caspase-3-sensitive NIR fluorescent probe was prepared . A recent review on targetable optical-imaging probes described the development of these dyes in detail .
Activatable ‘smart’ probes have been designed to morph their physical properties after a molecular interaction. They have been specifically synthesized for reporting activity of proteases such as matrix metalloproteinases, thrombin and cathepsins. The probes are optically nonresponsive in their quenched state and become fluorescent after enzyme-mediated release of the dye. This results in at least 100-fold signal enhancement. The design of the probes allowed us to combine a carrier particle or macromolecule with a NIR dye, via enzyme specific peptide substrates. For an excellent review on these activatable probes, please refer to Mahmood et al. .
One of the most valuable attributes of optical imaging is its favorable safety profile. However, a major drawback is poor deep-tissue penetration. Photoacoustic imaging is a new nonionizing imaging technique that brings together the advantages of US and optical imaging [197–199].
Conventional exogenous imaging agents, including organic small-molecule fluorophores and lanthanide chelates, are generally susceptible to photobleaching, poor quantum yields and a broad emission window [200,201]. Lanthanide chelates have a tendency to distribute nonselectively into extravascular space .
Gold nanoparticles, on the other hand, offer unique optical properties and much better compatibility with the cellular environment, in comparison to other nanoparticles. Electromagnetic radiation irradiates the electrons in the outer shell of a gold nanoparticle to oscillate. This phenomenon is known as surface plasmon resonance (SPR). When excited, the SPR of gold particles scatter and absorb light in the visible or the NIR spectrum. The SPR can be controlled by tuning the size, shape and coating of gold nanoparticles. The optical properties of gold nanoparticle-treated cells have been controlled by size [203–205], morphology [206–208] and surface coating [209,210].
Gold nanoparticles (15 nm) have been utilized in a photoacoustic experiment to augment cell contrast upon irradiation by a short pulse laser. The acoustic emissions created by these particles were collected by ultrasonic array to image the target cells. Agarwal and co-workers have utilized these particles for the targeted photo acoustic detection of LnCAP prostate cancer cells .
In 2009, in an interesting novel finding, Pan and colleagues reported the development of gold nanobeacons (GNBs) . GNBs represent a novel class of compliant ‘soft’ metal nanocolloids, which incorporate multiple tiny gold nanoparticles (2–3 nm) within a bigger vascularly constrained nanoparticle. The final formulation was encapsulated by phospholipids to produce 120–130-nm-sized GNBs that incorporate 103 gold atoms/particle. GNBs provided significant signal enhancement within the NIR window (740–820 nm) relative to blood (Figure 6c). These agents were developed for targeted imaging applications; however, at three-times the expected molecular imaging dose, a strong blood pool signal was seen in rats. Using in vitro clot phantoms, fibrin-targeted GNBs provided several-fold signal increase in comparison with the control (no gold) and nontargeted GNB particles (Figure 6A–B).
Kim et al. reported the development of a golden carbon nanotube for the photoacoustic detection of sentinel lymph nodes in rats . In this very recent report, gold-coated carbon nanotubes were used for sensitive detection of lymphatic endothelial cells and cancer metastasis within sentinel lymph nodes. Golden nanotubes may present a more efficient in vivo biosensor compared with other nanocrystals (e.g., QDs) or fluorescent labels.
Molecular-imaging approaches, particularly those based on nanotechnologies, are expanding as more scientists with new expertises enter the field. In MRI, T2* imaging with iron oxide dominated the field 20 years ago. For the iron oxide-based nanoparticle platforms, the superparamagnetic effects induce magnetic susceptibility artifacts, creating dark signal dropouts or negative contrast. Today, there are numerous clinically practical T1w imaging approaches based on iron, manganese and gadolinium. Ligand-directed phospholipid-encapsulated PFC nanoparticles are vascularly constrained unique platform technology, which may be applied to clinically relevant modalities including fluorine and proton-based MRI techniques. A clever targeting mechanism of a cell membrane receptor specific system to mediate intracellular concentration of the imaging drug delivery agent is realized with dendrimer-based agents. This alternative approach to achieving these molecular-imaging goals with dendrimers utilizes effective cellular internalization and concentration of the agent to build signal. However, the recent discovery of nephrogenic systemic fibrosis (NSF) associated with gadolinum-based MRI agents have prompted the scientists to explore other paramagnetic metals. With the introduction of colloidal iron oxide-based nanoparticles, T1w MRI with iron oxide based agents can now be envisioned.
In optical imaging, where originally small-molecule dyes coupled to homing ligands were common, now nanoparticle concepts incorporating numerous fluorescent and/or NIR dyes into one particle are common. Solid-gold particles originally targeted individually are packaged within nanoparticles for markedly greater signal amplification. Moreover, beyond the imaging opportunities, many nanoplatform technologies are compatible with targeted drug delivery, offering further unique strategies to personalize the diagnosis and management of medical problems. With nanomedicine now at or closely approaching the clinic, an accelerated era of learning how to better design and use these new tools has begun.
For reprint orders, please contact moc.ecneics-erutuf@stnirper
Financial & competing interests disclosure
Funding was provided by the NCI, NIH and AHA for Samuel Wickline, Gregory Lanza, Junjie Chen and Dipanjan Pan. Samuel Wickline and Gregory Lanza receive medical imaging equipment support from Philips Healthcare. Angana Sen Pan, Anne Schmieder and Shelton Caruthers have no conflicts to report. The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.
No writing assistance was utilized in the production of this manuscript.