Hyperpolarized MRI with silicon micro and nanoparticles: Principles and applications

Abstract

Silicon-based micro and nanoparticles are ideally suited for use as biomedical imaging agents because of their biocompatibility, biodegradability, and simple surface chemistry that facilitates drug loading and targeting. A method to hyperpolarize silicon particles using dynamic nuclear polarization (DNP), which increases magnetic resonance (MR) imaging signals by several orders-of-magnitude through enhanced nuclear spin alignment, was developed to allow silicon particles to function as contrast agents for in vivo magnetic resonance imaging. In this review, we describe the application of the DNP technique to silicon particles and nanoparticles for background-free real-time molecular MR imaging. This review provides a summary of the state-of-the-science in silicon particle hyperpolarization with a detailed protocol for hyperpolarizing silicon particles. This information will foster awareness and spur interest in this emerging area of nanoimaging and provide a path to new developments and discoveries to further advance the field.

This article is categorized under:

• Diagnostic Tools > In Vivo Nanodiagnostics and Imaging

• Therapeutic Approaches and Drug Discovery > Nanomedicine for Oncologic Disease

• Therapeutic Approaches and Drug Discovery > Emerging Technologies

1 |. INTRODUCTION

Elemental silicon is a formulation component of many materials used daily in the modern world from cement to polymers, alloys to electronics, and drug delivery to disease diagnosis. In this review, we focus on the application of silicon particles and nanoparticles in biomedical imaging, particularly magnetic resonance imaging (MRI). The isotope ²⁹Si is 4.75% naturally abundant and is magnetic resonance (MR) active. However, the application of ²⁹Si has been limited in MRI because of lack of naturally occurring silicon in biological systems as well as its low sensitivity (low gyromagnetic ratio). This limitation can be overcome by employing a technique called hyperpolarization. Hyperpolarization is a process in which nuclear spins are redistributed in energy states to produce a significantly enhanced signal compared to that obtained in thermal (Boltzmann) state polarization. Hyperpolarization is achieved by techniques such as, dynamic nuclear polarization (DNP; Abragam & Goldman, 1978; Carver & Slichter, 1953; Lilly Thankamony et al., 2017; Nikolaou et al., 2015; Overhauser, 1953; Wind et al., 1985), spin exchange optical pumping (SEOP; Walker & Happer, 1997), parahydrogen induced polarization (PHIP; Eisenschmid et al., 1987), and signal amplification by reversible exchange (SABER; Adams et al., 2009; Nikolaou et al., 2015). In the DNP process, magnetization is transferred from unpaired electrons to the nucleus at very low temperatures such as 1–3 K in high magnetic fields (~3 T) with microwave irradiation. Resonant polarization transfer between electron and nuclear spin is achieved via microwave excitation. This process increases the signal by over 10,000-fold compared to thermal (Boltzmann) polarization signal at the given magnetic field (Ardenkjær-Larsen et al., 2003). In SEOP, optical cells filled with Noble gases (¹²⁹Xe or ³He), buffer gases (N₂ and/or ⁴He), and small amounts of alkali metal vapors are irradiated with circularly polarized light in an applied magnetic field aligned in the direction of polarized light. Circularly polarized light produces electronic spin polarization in alkali metal vapors, which in collision with noble gases (¹²⁹Xe or ³He), transfers polarization to nuclear spins in a process called spin-exchange (Bouchiat et al., 1960; Nikolaou et al., 2015). The polarization of ¹²⁹Xe was reported to be up to 65% with 0.3 standard liters per hour flow of xenon and other optimized conditions (Hersman et al., 2008). Complemented with more sophisticated instrumentation and optimization, it has been recently reported that 1013 cm³ of xenon molecules can be polarized at 100% (Norquay et al., 2018). Alternatively, polarization can be induced when para-hydrogen (pH₂) is added to a molecule faster than the proton relaxation rate and the added hydrogens are magnetically distinct (Bowers & Weitekamp, 1986; Eisenschmid et al., 1987). For imaging, the polarization is quickly transferred to a nucleus of interest (¹³C or ¹⁵N) in the molecule with radiofrequency transfer pulses or magnetic field cycling (Hövener et al., 2018). A similar technique of polarization can be performed using a metal complex that can facilitate the reversible interaction of para-hydrogen with a suitable organic substrate (Adams et al., 2009; Hövener et al., 2018). This technique, SABER is an extended version of PHIP, in which the substrates can be hyperpolarized without changing their chemical identity. In this version, para-hydrogen in the presence of metal catalyst transfers the polarization to the substrate through a scalar coupling network. The hyperpolarized substrate then dissociates from the catalyst and builds up in the mixture over time (Rayner & Duckett, 2018). Recently the SABER requirement of binding substrate to catalyst was overcome using the SABER-RELAY technique (Iali et al., 2018); this important advance may open SABER for biomedical imaging applications in near future.

The DNP method of hyperpolarization which is the workhorse for biomedical hyperpolarization of ¹³C compounds, has been used to hyperpolarize silicon particles. Silicon is an attractive material for use in medical imaging as hyperpolarized silicon is reported to have a longitudinal relaxation time, T₁ ~39 min using 1–2 μm diameter silicon microparticles (SiMPs) (Cassidy et al., 2013); by contrast, ¹³C hyperpolarized signal within small molecules like pyruvate lasts only minutes (Ardenkjær-Larsen et al., 2003). The long lasting hyperpolarized signal of silicon particles at low temperatures depends on the applied external magnetic field strength, and the concentration of dopants like 2,2,6,6,–tetramethylpiperidin-1-yl)oxidanyl (TEMPO) within the particles (Lee et al., 2011). The increased T₁ at higher magnetic field strengths has been attributed to electron-nuclear dipole–dipole interaction. This increased T₁ with increased particle size has been explained by the diffusion-mediated relaxation at the particle surface. Lee et al. (2011) investigated T₁ of hyperpolarized silicon at room temperature and ambient magnetic field strength, after the particles are removed from high field polarization environment, and found the T₁ to be bi-exponential. Specifically, the slow decay constant was independent of external magnetic field and the fast decay constant was modestly decreased with an external magnetic field. This finding is an extension of the previous spin diffusion model explaining the longer T₁ of hyperpolarized ²⁹Si signal with increasing size of particles at room temperature (Aptekar et al., 2009; Lee et al., 2011). As the particle size increases, it takes longer for the core ²⁹Si signal to relax via spin diffusion to the surface thereby increasing the T₁. These long relaxation times of silicon has allowed for the application of these materials in imaging and nanomedicine.

2 |. HYPERPOLARIZED 29-SILICON PARTICLES AS MR IMAGING AGENTS

The long T₁ relaxation time constant of hyperpolarized ²⁹Si is necessary to achieve in vivo targeted MRI (Cassidy et al., 2013) and molecular imaging, similar to that achieved by positron emission tomography (PET) and single-photon emission computerized tomography (SPECT) without the need for radioactive imaging agents. The potential use of hyperpolarized silicon particles, 1–25 μm in diameter, as MR imaging agents has been demonstrated by Aptekar et al. (2009). Surface functionalization of these silicon microparticles with compounds such as APTES (3-aminopropyltriethoxysilane) did not affect their hyperpolarization, nor their T₁ at room temperature. The first evidence supporting the use of this technique for in vivo imaging was demonstrated with phantom imaging of ²⁹Si hyperpolarized microparticles (Aptekar et al., 2009). The in vivo MR imaging in a mouse using hyperpolarized silicon particles of ~2 μm was achieved in later work by Cassidy et al. (2013) with a T₁ of 39 ± 3 min. The gastrointestinal (GI) tract and intraperito-neal regions of mice were imaged with hyperpolarized silicon particles via direct injection of particles into either area. In a related application, hyperpolarized silicon particles were attached to the tip of a medical grade mouse catheter and imaged via MR to track the tip’s position within biomimetic phantoms and the GI tract of live mice (Whiting, Hu, Shah, et al., 2015). The low temperature DNP-generated hyperpolarized ²⁹Si signal persisted for about 40 min allowing the trajectory of the catheter in the GI of a mouse to be imaged for up to 5 min using interval snap shots. Si-catheter tracking does not involve ionizing radiation, generates positive contrast, and has zero background signal therefore, making it a possible replacement for X-ray fluoroscopy or a supplement to it. Future experiments will explore whether vessel occlusion can be visualized by this technique, similar to radiocontrast agents in X-ray-based imaging techniques.

These demonstrations were significant findings; nonetheless no targeted molecular imaging of functionalized hyperpolarized silicon particles was shown. E-selectin thioaptamer (ESTA-1) (a thiophosphate-modified oligonucleotide aptamer)-conjugated SiMPs exhibited similar hyperpolarized signal relaxation times as bare SiMPs and SiMPs that were surface-functionalized with poly(ethylene glycol) (PEG) oligomers, which are resistant to protein adsorption (Whiting, Hu, Constantinou, et al., 2015). This was an exciting finding, as ESTA-1 binds to the glycoprotein E-selectin (Mann et al., 2011), which is overexpressed in some ovarian cancer tissues, creating the possibility for these particles to target cancer tissues (Whiting et al., 2016). In more recent work, we have demonstrated that SiMPs also may be surface-functionalized with antibodies, which retain their tertiary structure and binding affinity, even after being subjected to the harsh nonbiological conditions required for DNP. These results provide important early evidence that targeted molecular imaging of cancerous lesions is achievable, using hyperpolarized, antibody-conjugated SiMPs.

One of the crucial tests of the application of DNP in biomedical imaging is if it is possible to hyperpolarize smaller size silicon particles that are likely to be more useful in clinical imaging applications. The larger silicon particles are useful for luminal imaging like targeting precancerous polyps in colorectal cancer in the gastrointestinal tract but, the large size and lack of mobility make microscale particles poorly suited for targeted imaging, where intravascular delivery of contrast agents through the narrow vessels is often desired. Smaller nanoparticles (10–300 nm) that have dimensions comparable to biological functional units, can be taken up or extravasated in the cells and also have favorable clearance properties (Nune et al., 2009). One of the fundamental challenges in hyperpolarizing 10–300 nm silicon nanoparticles (SiNPs) by DNP is the substantial reduction in surface defects. The decrease in size would result in shorter T₁ relaxation times for SiNPs compared to SiMPs. However, Kwiatkowski et al. (2017) recently demonstrated successful hyperpolarization of SiNPs of size 55 ± 12 nm without the addition of external radicals. The maximum hyperpolarization observed was 12.6% with 1 W microwave power. Bare SiNPs exhibited 42.3 ± 0.1 min of T₁, whereas SiNPs that were surface-functionalized with APTES-PEG exhibited a shorter T₁ of 34.0 ± 0.3 min. In a parallel study, Hu et al. (2018) reported hyperpolarization of ~70 nm diameter SiNPs, treated with (2,2,6,6,–tetramethylpiperidin-1-yl) oxidanyl (TEMPO). TEMPO increases the hyperpolarization signal by providing a free radical source, as confirmed by electron spin resonance (ESR) measurements of bare and TEMPO-treated SiNPs. Conditions were optimized to obtain a maximum hyperpolarized signal and longer relaxation time as a function of TEMPO concentration, SiNP particle size, and DNP duration. The longitudinal relaxation time of hyperpolarized SiNPs was ~20 min for SiNPs of 70 nm average size with the addition of 30 mM TEMPO. Another report from Seo et al. (2018) describes mesoporous silicon nanoparticles (PSi) as theranostic agents that can be hyperpolarized for MR imaging and also have the potential to deliver therapeutics to the target. That report describes the synthesis of mesoporous silicon particles of various sizes through magnesium-mediated reduction of Stober silica nanospheres. As a proof-of-principal, 300 nm PSi was chosen for hyperpolarization. Despite electron paramagnetic resonance (EPR) measurements suggesting a relatively high level of surface electronic defects, unmodified PSi showed minimal hyperpolarization capacity, but this was increased to a maximum polarization of ~7% after treatment with 30 mM TEMPO and optimized DNP conditions. The T₁ of hyperpolarized signal of PSi at room temperature in 7 T MR magnet was ~25 min, within the range reported by Hu et al (Hu et al., 2018). Last, the authors again functionalized the PSi surfaces with APTES, which improved aqueous suspension of the particles, and enabled future coupling of aptamers or antibodies for targeted molecular imaging and delivery of therapeutics.

Technical advances in hyperpolarization of micro- and nanoscale particles is promising for future diagnostic and therapeutic applications. However, these advances have not yet been used to successfully achieve molecular imaging of malignant lesions in a clinical setting, but preclinical research is promising so far. As a proof-of-principle, targeted molecular imaging using hyperpolarized functionalized SiMPs was shown in vivo utilizing a HeyA8 orthotopic ovarian cancer mouse model with ESTA-1 functionalized SiMPs (Figure 1) (Whiting et al., 2016). Similar targeted molecular imaging with hyperpolarized SiMPs was achieved with subcutaneous injection of colorectal cancer cells in a mouse model. The target molecule chosen was the membrane glycoprotein, MUC1, which is overexpressed in many cancer systems, including colorectal, pancreas, breast and ovarian cancers (Constantinou et al., 2011; Taylor-Papadimitriou et al., 1999). MUC1 extends ~500 nm beyond the cell surface, and its sequence contains a large number of tandem repeat motifs, recognized by the 214D4 monoclonal antibody used in these studies (Golubović & Bojić-Trbojević, 2006). Consequently, MUC1 is an attractive target for antibody based molecular imaging (Danysh et al., 2012). SiMPs were first treated as before with APTES, followed by secondary conjugation with PEG oligomers and 214D4 antibodies across the particle surface. 214D4-functionalized SiMPs were found to have similar hyperpolarization signal and T₁ to unconjugated particles (T₁ ~19 min). In vitro experiments with colorectal cancer cells expressing MUC1 demonstrated that the hyperpolarization process did not affect the stability or functionality of the antibody (Whiting et al., 2017).

FIGURE 1.

Adopted from Whiting et al. (2016)—in vivo ²⁹Si MRI of ESTA-1 functionalized 2 μm SiMPs (100 mg; dissolved in 400 ml PBS) directly injected into the tumor volume of an orthotopic ovarian cancer mouse (HeyA8). Tumor periphery outlined in green. Single image taken 20 min postinjection, showing the SiMPs retain their enhanced signal while in the tumor volume

2.1 |. Targeted molecular imaging applications in vivo

Targeted molecular imaging in preclinical settings can be achieved in vivo with hyperpolarized silicon particles, using similar synthetic workflows as described in the previous section. Preliminary surface preparation of the silicon particles with APTES or similar silanes provides reactive handles for secondary covalent attachment of bivalent PEG oligomers, which resist nonspecific protein absorption, and enable tertiary coupling of a tumor-targeting moiety such as a protein, peptide, aptamer, or antibody to the particle periphery. For preclinical work to demonstrate efficacy, after conjugation, the targeted particle can be hyperpolarized, delivered to the tumor-bearing animal model and scanned in an MRI machine with the appropriate ²⁹Si/¹H MRI coil. If the hyperpolarized conjugated particles reach and are concentrated at the desired target, a “hotspot” at the location of the tumor or the lesion of interest can be observed in the MRI. As an example, mice bearing colorectal tumors expressing human MUC1 on their surfaces can be administered a bolus of 214D4-functionalized SiMPs/SiNPs via rectal enema (Whiting et al., 2017). After binding of hyperpolarized particles to the tumor surfaces, the tumors can be imaged and true signal separated from noise using thresholding techniques. This strategy opens the door to selective targeting and imaging of tumor in vivo by MRI (Figure 2), an application with broad utility in minimally invasive tumor detection.

FIGURE 2.

Schematics of the workflow showing the steps of preclinical targeted molecular imaging using hyperpolarized functionalized silicon particles. (a) Bare silicon particles, reactive poly(ethylene glycol) and antibody of choice, (b) silicon particle functionalized with antibody via PEG linker, (c) functionalized silicon particles subjected to DNP process, (d) appropriate murine model prepared for hyperpolarized (HP) silicon particle administration, (e) murine model prepared for the injection of HP silicon particles placed in MRI scanner, (f) hyperpolarized ²⁹Si image is captured in vivo demonstrating targeting which can then be overlaid with ¹H anatomical image

2.2 |. Delivery to target of interest

One of the challenges of employing silicon particles is how they will be delivered to an imaging target of interest in vivo. Depending on the size of the particles, different delivery systems could be employed. If the particles are in the nanometer range (70–200 nm) (Blanco et al., 2015; Rizvi & Saleh, 2018), they could be delivered through either intravenous (IV) or direct injection to the site of injury (i.e., tumor). For larger particles, in the micrometer range, there are several delivery options depending on the site chosen for imaging. One is through an oral route for imaging upper stomach and upper GI tract for targeting precancerous polyps and inflammation of the gut. The animal/patient would ingest a slurry containing the targeted, hyperpolarized particles, then the imaging would be performed minutes later after the particles have bound to their targets. Another delivery option, commonly used in our laboratory to image the lower GI tract, is through rectal administration of an enema containing the targeted, hyperpolarized particles. Another possible delivery route is by intrathecal administration. This would be used for the delivery of particles to reach the cerebrospinal fluid (CSF) and study central nervous system (CNS) diseases or metastatic brain cancers (Fowler et al., 2020; Householder et al., 2019).

2.3 |. Multifunctional silicon particles for multimodal imaging applications

Multi-modal imaging applications can be achieved with silicon micro and nanoparticles, beyond just MR-based imaging. For example, one strategy can be to conjugate a targeting moiety labeled with a fluorescent probe that can be observed in vitro using confocal microscopy or in vivo 3D optical tomography, in addition to targeted molecular imaging by silicon hyperpolarization. A library of multiple antibodies, targeting different proteins of interest, can also be achieved for multifunctional targeted molecular imaging in vivo with hyperpolarized ²⁹Si MR or by optical imaging (Y. Tang, Polydorides, et al., 2018).

2.4 |. Silicon particles as theranostic agents

Silicon micro and nanoparticles have potential for wide application in treatment in several diseases based on their biocompatible and biodegradable natures (Haidary et al., 2012; O’Farrell et al., 2006; Park et al., 2009; Serda et al., 2011). Nano theranostics is an emerging field where nanoparticles are employed to diagnose and treat diseases simultaneously. Silicon and silica particles are now attractive agents for theranostics use because of their surface chemistry, nontoxicity, biocompatibility, optics, and their ability to be hyperpolarized.

One such example is ocular neovascularization, a condition which may cause severe vision impairments and eventual vision loss. The disease has been traditionally diagnosed using fluorescein angiography that is associated with vision loss related complications. To address this issue, M. Tang, Ji, et al. (2018) developed an approach for simultaneous imaging and treatment of neovascularization with functionalized SiNPs. Fluorescent SiNPs are highly promising for biological and biomedical applications, due to favorable biocompatibility and low toxicity (Michalet et al., 2005; Zhou et al., 2018). In this example, fluorescence imaging of SiNPs was captured at excitation of 405 nm, and detection was achieved in window of 420–480 nm. In this study, nanoparticles (SiNPs <10 nm) were conjugated with a cyclic Arg-Gly-Asp-d-Tyr-Cys (c-[RGDyC]) peptide and administered in a mouse model of alkali-induced corneal neovascularization via intravenous injection. The mice were monitored for neovascularization for seven days after injection. A significant reduction in length of neovascularization was seen in mice treated with SiNPs-c-(RGDyC) compared with other mice which were administered saline, RGD and only unfunctionalized SiNPs. This interesting study demonstrated that simultaneous imaging of angiogenesis and treatment is possible with SiNPs conjugated with RGDyC (Figure 3).

FIGURE 3.

Adopted from Tang et al (M. Tang et al., 2018)—(a) schematic diagram of the process of inducing neovascularization in mouse cornea and treatment with functionalized SiNPs. (b) Representative images of corneal neovascularization induced by sodium hydroxide burn injury (scale 5 mm). (c) Fluorescence images of angiogenic blood vessels in the corneas taken after 4 h of circulation of intravenously injected unfunctionalized SiNPs (upper) or SiNPs-RGDyC (lower) (scale 25 μm)

In another interesting study, photo luminescent and biocompatible silicon nanocrystals were synthesized. These particles on irradiation with laser of 337 nm with 10 ns pulse generated photoluminescence. The photoluminescence was quenched by transferring the energy to dissolved oxygen to generate singlet oxygen (¹O₂) (Timoshenko et al., 2006). It was shown that NIH-3T3 cancer cells in presence of silicon nanocrystal suspension irradiated with light of 337 nm pulse showed 80% cell death compared to the control cell culture with silicon nanocrystals suspension kept in dark (Timoshenko et al., 2006). In separate studies, porous SiNPs of ~150 nm diameter were synthesized and were demonstrated to be a good photosensitizers and can be employed in cell treatment (Xiao et al., 2011). The photo toxicity of these porous SiNPs on HeLa and NIH-3T3 cells were investigated. The HeLa and NIH-3T3 cells shown 45% cell death by photo toxicity of SiNPs, whereas cell cultures without particles but light irradiation and with silicon particles without light shown only <10% cell death (Xiao et al., 2011).

Mesoporous silicon nanoparticles (PSi) conjugated with thiolated polyethylene glycol (PEG-SH) exhibited anticancer effect by photothermal ablation using 808 nm wavelength light. The PSi particles were injected directly at the site of the tumor in a xenograft model of colon cancer. This was followed by irradiation of 808 nm light for 2 min and 2 min interval for 20 min. The authors concluded the combination treatment with functionalized PSi and 808 nm NIR irradiation on xenografts selectively destroyed the cancer cells and no reoccurrence of tumor was observed for at least 3 months (Hong et al., 2011). In another study involving ovarian cancer treatment, PSi encapsulated with anti-micro-RNA-21 (miR-21) drug and a peptide (CGKRK) were coupled for targeted therapy (Bertucci et al., 2019). In this study, mice bearing subcutaneous ovarian cancer tumor xenografts were injected with miR-21 loaded CGKRK-pSi particles. The bare pSi were measured to be diameter 182 ± 6 nm in size. A decrease in tumor size was observed in mice treated with miR-21 loaded CGKRK-pSi compared to mice treated with only saline or pSi nanoparticles functionalized with a control peptide CRA (Cys-Arg-Ala). Osminkina et al. tested luminescent PSi particles of average size 100 nm for ultrasound-assisted therapy (Osminkina et al., 2015). The ultrasound therapy is achieved by physical disruption of cellular structures, hyperthermia, and cavitation process. In this study, efficient uptake of particles into the cells and reduction in cell proliferation was seen by luminescent confocal microscopy. In vivo experiments in mice bearing B16 melanoma tumors demonstrated a reduction in tumor growth when treated with ultrasound therapy and dextran coated SiNPs compared to control mice that were treated with either stand-alone silicon particles or only ultrasound therapy (Osminkina et al., 2015).

As a final example, spherical SiNPs of mean diameter ~60 nm were synthesized by laser ablation (Al-Kattan et al., 2020). These nanoparticles demonstrated broad 1-photon absorption, maximized near 450–550 nm, and corresponding two photon excitations (TPE) at 900 nm. As a result, MCF-7 breast cancer cells that were incubated with these SiNPs and irradiated with 900 nm TPE showed 45% cell death, while nonirradiated cultures showed negligible (<10%) cell death. Similar results were extended to other cancer cell lines, in 2D and 3D cultures. The long excitation wavelength for TPE suggests a potential utility in vivo, to enable good light penetration with reduced tissue scatter.

3 |. OUTLOOK

Future research should concentrate on improving maximum HP signal and longer relaxation times for smaller silicon particle sizes (10–300 nm), customizing physiological targets through tethered biorelevant moieties, and optimizing the methods and routes of particle delivery for in vivo imaging. If intravenous particle delivery is necessary, then circulation time, biodistribution, toxicity, off and on target concentration, and particle clearance rates need to be assessed. Another area for advancement in this research is optimizing pulse sequences for multi-slice silicon MR imaging, which is currently limited to single slice imaging. The quantification methods should be developed to correlate the silicon signal intensity to tumor burden in vivo. A final area with innovative potential is the retention of spin polarization on silicon particles, and transfer to a target site for assessment of other distal nuclei, in applications such as metabolomics. Nevertheless, the biocompatibility, flexible surface chemistry, and nonionizing imaging modalities of hyperpolarized silicon particles make them excellent potential candidates for targeted molecular imaging via MRI. Porous and nonporous silicon particles either in nano or micro size scales can be employed for multifunctional imaging and/or therapy for different disease states (Al-Kattan et al., 2020; Bertucci et al., 2019; Hong et al., 2011; Rytkönen et al., 2012; Whiting et al., 2016). Many of these particles offer potential therapeutic effect by drug loading, reactive oxygen species (ROS) sensitization and photo thermal therapy and can simultaneously serve as contrast agents for targeted molecular MRI. Furthermore, testing of silicon, silica, and mesoporous silicon synthesized by different methods of varying sizes for hyperpolarization, relaxation times and therapeutic delivery will help to choose optimized particles for theranostic imaging.

3.1 |. Protocol for hyperpolarizing silicon particles

Despite a growing interest in hyperpolarizing silicon materials, no detailed protocol is available on how to hyperpolarize silicon particles. Here we present a step-by-step protocol on hyperpolarization of silicon particles.

Sample preparation

• 1Prepare silicon particles (~60 mg) either functionalized or bare (unfunctionalized) ready for hyperpolarization. High purity silicon particles of varied dimensions are commercially available (Alfa Aesar, Haverhill, MA; Meliorum Technologies, Rochester, NY) or can also be synthesized in house. If the particles are functionalized with antibody (IgG), store it at −20 C° until the hyperpolarization experiment begins.

Prehyperpolarization setup

• 2First step in hyperpolarization experiment is to evacuate the cryostat for 24–48 hr. We have employed a laboratory built DNP setup for silicon hyperpolarization (Figure 4); however this protocol can be modified for other commercial or laboratory built DNP systems. Connect the turbo and scroll pump to cryostat evacuation chamber and start the evacuation process. We recommend performing this cryostat evacuation step once every week before experiments at a minimum.
• 3Stop the evacuation and disconnect the pumps from the cryostat.

FIGURE 4.

A photograph of the ²⁹Si DNP setup highlighting the individual components

Hyperpolarization

• 4Fill the polytetrafluroethylene (Teflon) tube of 2 × 0.3 cm length and inner diameter with ~60 mg of silicon sample. These tubes are transparent to microwaves, diamagnetic, and can withstand low temperatures as low as 1 K and are usually used in ESR experiments. If the particles are functionalized, then disperse 50 mg of particles in 125 μl of PBS buffer and fill in the tube and tighten the tube with a cap leaving some space to attach this to the insertion solid flexible Teflon tube.
• 5Connect the Teflon tube with sample to the solid flexible insertion Teflon tube and insert this into the cryostat and tighten the knob.
• 6Insert the helium transfer line into the helium dewar and then connect the one end of the transfer line to the cryostat and other end to the scroll pump.
• 7Start the scroll pump and allow it to run for 15 min and then open the knob on the transfer line to pump the helium to the cryostat.
• 8During the 15 min of scroll pump run—set up the microwave equipment. Switch on microwave generator, multimeter, wavefunction generator, and oscilloscope. All equipment should be on uninterruptible power supply (UPS) to avoid disruption in experiments due to power fluctuation (Figure 4).
• 9Microwave source is 100 mW power and is set to operate within frequency range of 80.3 to 80.9 MHz with 20 KHz oscillation with ramp wave to cover the ~60 MHz ESR line width of electron defects on silicon particles or added TEMPO radicals.
• 10The DNP signal build-up is monitored in real time using an in situ Nuclear Magnetic Resonance (NMR) circuit and observed using miniature Kea NMR board (Kea2; Magritek, NZ) employing saturation recovery pulse sequence for every hour during the run.
• 11The hyperpolarization process of silicon micro and nanoparticles obtained from Alfa Aesar and Meliorum Technologies, Inc. takes about approximately 18 h. The exact duration of hyperpolarization depends on the nature of the silicon particles and requires optimization for every particle size. In general, polarization buildup times in silicon particles are function of particle sizes; microparticles that have longer buildup times than nanoparticles. The microparticles have surface defects that act as a source of electron radicals. Whereas in smaller sized particles, one may need to add TEMPO radical for the DNP process. The consequences of adding TEMPO to the particles is a shorter relaxation time of the hyperpolarized signal and potential toxicity.
• 12Stop the scroll pump and open the helium transfer valve completely to take out the hyperpolarized sample from the cryostat.
• 13Once the sample is taken out, disconnect the helium dewar from pump and cryostat (after 15–20 min). Take out the helium transfer line from the dewar carefully by releasing the pressure. If more experiments are planned, then do not disconnect units, wait for 30 min and insert new sample for hyperpolarization. This 30-min wait period is important to remove the ice from the cryostat due to condensation.

MRI and MRS

• 14Setting up the MRI machine for ¹H anatomic and ²⁹Si imaging: The setup of the MRI scanner should start an hour before the hyperpolarization of silicon signal reaches steady state for animal or phantom imaging. All images are acquired with 35 mm (ID) dual tuned ²⁹Si and ¹H litz volume coil (Doty Scientific, SC) on 7 T MRI scanner (Bruker, BioSpin MRI GmbH, Ettingen, Germany).
• 15Optimization with phantom: The optimization should be carried out using 10 ml silicon oil phantom (Whiting, Hu, Constantinou, et al., 2015). The α = 90° single pulse ²⁹Si NMR spectrum acquired with 2048-time domain points, and 10,000 Hz spectral width. ¹H images are acquired with Rapid Acquisition with Refocused Echo Sequence (RARE) with TR = 1926.9 ms, TE = 9.5 ms, 64 mm² FOV, 256 × 256 matrix, 0.25 mm² resolution, 1 slice, and 1 scan.

Animal preparation for imaging

• 16All the protocols and procedures should be approved by the institutional animal ethical committee prior to experiments. Mice will be on 2% (v/v) isoflurane anesthesia throughout the imaging procedure using a nose cone and body temperature of mice will be maintained using veterinary water heating pads.
• 17The hyperpolarized silicon particles are administered via the route as outlined in the animal protocol. The tube containing the hyperpolarized particles is warmed in hand near the MRI scanner, then dispersed in 300 μl of phosphate buffer saline (PBS) and administered into the mouse. After administration of particles, mouse will be placed inside the scanner for imaging studies.
• 18Once the mouse is placed in the MRI scanner, to confirm the hyperpolarization signal intensity and for normalization, an initial ²⁹Si NMR spectrum will be acquired with a 10° pulse.
• 19After 15 min of administration of particles into the mouse, ²⁹Si 2D image will be acquired with the following sequence and parameters. Rapid acquisition of refocused echo T2 weighted imaging sequence is used. Single slice ²⁹Si image is acquired with TR = 59.9 ms; TE = 1.8 ms; single transient; 32 × 32 matrix of 64 mm² FOV for a 2 mm² resolution. RG was kept 101 for all silicon imaging experiments.
• 20Following immediately ²⁹Si image ¹H T2 weighted images are collected using RARE sequence with TR = 1926.9 ms, TE = 9.5 ms, 64 mm² FOV, 256 × 256 matrix, 0.25 mm² resolution, 2 slice, and 4 averages.

Postacquisition processing

• 21²⁹Si Images: We process all of our data in MATLAB (MathWorks, Inc., Natick, MA) using the following procedure. The k-space data 32 × 32 is zero filled to 256 × 256 before Fourier transformation. Images are reconstructed in ParaVision (Bruker Biospin MRI GmbH, Ettingen, Germany) and imported to MATLAB. Often, there are particles remaining in the coil from the injection outside the body of the animal. The remaining particles are a result of injection method and if included, would distort the relative signal intensity because they are highly concentrated. The hyperpolarized signal from these excess particles outside of the mouse body are therefore ignored. The threshold should be applied to each image to remove signal that is less than five times the background noise. The noise is calculated by taking the average from 32 × 32 sections in each corner of the silicon image. Next ²⁹Si image should be normalized. Each ²⁹Si image divided by respective NMR data acquired from a Silicon oil sample to account for variation of the MR scanner in different experiments. All ²⁹Si images are normalized by dividing each image by the largest ²⁹Si signal value image across. The resulting images are transferred into 8bit for display for comparison between studies.
• 22¹H image processing: The reconstructed images from ParaVision are imported to MATLAB. Relevant ¹H image slice is identified and the contrast of the ¹H images are increased by saturating the top and bottom signal to 1%. Using “Image Processing Toolbox” in MATLAB the ²⁹Si and ¹H image is overlaid.

Abbreviations:

APTES

3-(aminopropyl)triethoxysilane

CNS

central nervous system

CSF

cerebrospinal fluid

DNP

dynamic nuclear polarization

ESR

electron spin resonance

ESTA-1

E-selectin thioaptamer-1

FOV

field of view

GI

gastrointestinal

ID

internal diameter

IgG

immunoglobulin-G

INEPT

insensitive nuclei enhanced by polarization transfer

IV

intravenous

MR

magnetic resonance

MRI

magnetic resonance imaging

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide

MUC1

mucin–1

NIR

near infrared

NMR

nuclear magnetic resonance

PBS

phosphate-buffered saline

PEG

poly(ethylene glycol)

PEG-SH

thiolated poly(ethylene glycol)

PET

positron emission tomography

PHIP

para hydrogen induced polarization

Psi

porous silicon nanoparticles

RARE

rapid acquisition with refocused echo

ROS

reactive oxygen species

SABER

signal enhancement by reversible exchange

SEOP

spin exchange optical pumping

SiMPs

silicon microparticles

SiNPs

silicon nanoparticles

SPECT

single-photon emission computerized tomography

T₁

longitudinal relaxation time constant

TE

echo time

TEMPO

(2,2,6,6,–tetramethylpiperidin-1-yl)oxidanyl

TPE

two photon excitation

TR

repetition time

UPS

uninterruptible power supply

DATA AVAILABILITY STATEMENT

Data sharing is not applicable to this article as no new data were created or analyzed in this study.