Dynamic Nuclear Polarization NMR in Human Cells Using Fluorescent Polarizing Agents

Abstract

Solid-state nuclear magnetic resonance (NMR) enables atomic resolution characterization of molecular structure and dynamics within complex heterogeneous samples, but it is typically insensitive. Dynamic nuclear polarization (DNP) increases NMR signal intensity by orders of magnitude and can be performed in combination with magic angle spinning (MAS) for sensitive, high-resolution spectroscopy. Here we report MAS DNP experiments, for the first time, within intact human cells with >40-fold DNP enhancement and a sample temperature below 6 K. In addition to cryogenic MAS results below 6 K, we also show in-cell DNP enhancements of 57-fold at 90 K. In-cell DNP is demonstrated using biradicals and sterically-shielded monoradicals as polarizing agents. A novel trimodal polarizing agent is introduced for DNP, which contains a nitroxide biradical, a targeting peptide for cell penetration, and a fluorophore for subcellular localization with confocal microscopy. The fluorescent polarizing agent provides in-cell DNP enhancements of 63-fold at a concentration of 2.7 mM. These experiments pave the way for structural characterization of biomolecules in an endogenous cellular context.

Graphical Abstract

INTRODUCTION:

Solid state nuclear magnetic resonance (NMR) is exquisitely suited to characterize molecular structure within endogenous environments, including in cells^1–5. However, the concentration of target biomolecules in these complex preparations is much lower compared to highly purified samples, challenging the sensitivity limit of NMR spectroscopy. Strategies to increase NMR sensitivity are, therefore, an important aspect of improving solid state NMR and in-cell NMR spectroscopy. In-cell NMR experiments typically involve exogenous protein expression in cells or otherwise increasing target protein concentrations beyond endogenous levels and can result in sufficient NMR sensitivity^6–9. However, such perturbations alter endogenous biomolecular interactions and their associated pathways. Alternatively, increasing NMR sensitivity by multiple orders of magnitude will permit in-cell structural characterization of biomolecules at endogenous concentrations.

Strategies to increase NMR sensitivity include access to high static magnetic fields^10,11, cryogenic sample temperatures¹², more sensitive detection schemes¹³, and spin polarization transfers¹⁴. Such polarization transfer mechanisms most effectively increase NMR sensitivity when the transferring spin has the highest spin polarization. Dynamic nuclear polarization (DNP) utilizes electron spins as the originating polarization source^15,16. Cryogenic operation (<6 K), high static magnetic fields (>7 T), and efficient DNP transfers can result in high NMR polarization and >10,000-fold increases in NMR signals¹⁷. DNP has been successful at polarizing the outer membrane and cell wall of bacteria^18–20, but has not been demonstrated within bacteria, most likely due to the poor bacteria-cell permeability of DNP polarizing agents. Furthermore, magic angle spinning (MAS) DNP has yet to be extended to studies of human cells. Performing DNP in situ within intact cells has many advantages, including repeating experiments quickly for NMR signal averaging and determining the subcellular localization of NMR signals. Here we demonstrate in-cell DNP NMR in intact human cells and characterize the signal enhancements obtained using biradicals, sterically-shielded monoradicals, and a novel trimodal fluorescent polarizing agent. The fluorescent DNP polarizing agent is comprised of a nitroxide biradical, a targeting peptide for cell penetration, and a fluorophore for subcellular localization with confocal microscopy.

MATERIALS AND METHODS:

40 million intact human embryonic kidney cells (HEK293F), grown in Dulbecco’s modified eagle’s medium (DMEM) supplemented with [U-¹³C] glucose, were collected from cell culture and spun down at 300*g for 2 min. Cells were washed with 4 mL natural abundance minimal essential media (MEM) and spun down at 800*g for 1 min. The resulting cell pellet was suspended with 100 μL of natural abundance MEM (containing 10% DMSO) with DNP radical (20 mM, 2.7 mM AMUPol, 2.7 mM TotaFAM, 40 mM nitroxide monoradicals²¹). Of the suspended cell volume, 80 μL was aliquoted for DNP NMR, 40 μL for EPR characterization, and 2 μL for microscopy. Finally, samples were spun, at 800*g for 30 seconds, directly into NMR rotors or EPR tubes. The supernatant from samples was removed and the samples were frozen immediately in a liquid nitrogen bath. Total time taken for sample packing, defined as time from when radicals were added to cell suspension to when samples were frozen, ranged from 1 to 2 min. Sample masses packed into NMR rotors fell in a range of 32 to 35 mg, containing 15–20 million cells. Samples were stored at −80 °C (see further experimental details in Supplementary Section 4).

DNP experiments at B₀ = 7.05 T (300.184 MHz ¹H) were performed with a custom-built MAS NMR probe housing 3.2 mm outside diameter rotors²² and a frequency-agile gyrotron²³. Spectra were recorded with a rotor-synchronized, echo-detected cross polarization (CP) MAS pulse sequence (Supplementary Figure S1). In order to destroy residual polarization, saturation trains (SAT) were implemented before a DNP polarization time (τ_pol) and the CPMAS sequence, resulting in an overall sequence of SAT-τ_pol-CPMAS. A microwave irradiation frequency of 197.674 GHz was used for cross effect polarization transfer from nitroxide radicals. See Supplemental Information for more experimental details.

RESULTS AND DISCUSSION:

In-cell NMR with biradicals and monoradicals.

Magic angle spinning (MAS) NMR spectra of intact human embryonic kidney cells (HEK293F) were enhanced through DNP using both biradicals and monoradicals at sample temperatures below 6 K, and also at 90 K. We found that nitroxide biradicals and monoradicals effectively enhanced HEK293F carbon signals through DNP and cross polarization. The nitroxide biradical, AMUPol (Figure 1G)²⁴, yielded an enhancement of 46 within intact human cells (Figure 1A). Note, preliminary experiments which included a washing step after incubation of the polarizing agent did not show significant DNP enhancements.

Figure 1.

Characterization of nitroxide radicals in HEK293F cells. DNP CPMAS NMR spectra below 6 K using (A) 20 mM AMUPol, (B) 40 mM sterically-shielded nitroxide monoradical, and (C) 2.7 mM TotaFAM. Black spectra represent no microwave irradiation while red spectra are recorded with microwave irradiation. Asterisks (*) denote spinning side bands. The ¹³C resonances in the 50–100 ppm chemical shift range are attributed to sugars (Supplementary Figure S9). Intensity of integrated EPR spectra versus time of cellular samples prepared with (D) AMUPol, (E) sterically-shielded nitroxide monoradical, and (F) TotaFAM. Polarizing agents (G) AMUPol, (H) sterically-shielded nitroxide monoradical, and (I) trimodal polarizing agent, TotaFAM. The three moieties of TotaFAM include a TOTAPOL nitroxide radical (red), 11 residues of the HIV-1 Tat protein (blue), and a 6-FAM fluorophore (green).

Sterically-shielded nitroxide monoradicals (Figure 1H), which are less prone to radical reduction in cellular environments²¹, also provided a significant DNP enhancement of 31 (Figure 1B). To compare enhancements from nitroxides, concentrations were kept constant at 20 mM of biradicals and 40 mM of monoradicals. We note that the proton longitudinal relaxation time under microwave irradiation (T_{1_DNP}) was less than 3 s for HEK293F samples prepared with nitroxide radicals (Supplementary Figure S11). This permitted fast repetition of scans and reduced time needed for signal averaging. DNP enhancements of 57 from intact human cells were also observed at 90 K (See Figure 2).

Figure 2.

DNP CPMAS NMR of HEK293F cells at 90K. 13C DNP spectra recorded at 90 K of cells washed with phosphate buffered saline (PBS) and suspended in PBS with 20 mM AMUPol. Black spectra represent no microwave irradiation while red spectra are recorded with microwave irradiation.

Electron paramagnetic resonance (EPR) spectroscopy of nitroxide biradicals at room temperature, after incubation with HEK293F cells, was performed to monitor radical reduction in the cellular environment (Figure 1 D–F). After 30 minutes, 30% loss in signal intensity was observed for the AMUPol nitroxide biradical (Figure 1D), while the sterically-shielded monoradical signal loss was only 2.1% (Figure 1E). Combined with the successful DNP enhancements of >30, these results indicate radical reduction does not prevent in-cell applications of DNP. Nonetheless, our sample preparation protocol minimizes the exposure of radicals to the reducing environment of cells by quickly (<3 minutes) freeze-quenching the cellular preparations after addition of radicals. EPR measurements showed less than 2% loss of the biradicals before the samples were frozen.

Nitroxide radicals showed mostly uniform enhancement over all resonances. However, ¹³C resonances between 100–150 ppm exhibited particularly attenuated DNP enhancement (Supplementary Figure S7). Additionally, differences in the signal intensities between the spectra shown in Figure 1A and 1B using different radicals suggest a heterogenous cellular distribution of polarizing agents. To determine the distribution of the radicals, we synthesized a novel polarizing agent which includes a fluorophore to enable optical localization.

In-cell NMR with trimodal DNP polarizing agents.

We designed a trimodal fluorescent DNP polarizing agent, coined TotaFAM, containing a binitroxide radical for cross effect DNP, a Tat peptide for intracellular targeting, and a fluorophore for optical localization (Figure 1I). A maleimide-derived TOTAPOL²⁵ was conjugated to a N-terminal cysteine residue of the targeting cell-penetrating peptide, residues 47–57 of the HIV-1 Tat protein^26,27. The lysine side-chain on the C-terminus of the Tat peptide was covalently linked to fluorescein. In addition to cell penetration, the Tat peptide linker limits the proximity of fluorophore to radicals to mitigate fluorescence quenching, while retaining molecular connectivity to enable optical localization of DNP polarizing agents. The TOTAPOL biradical moiety²⁸ on TotaFAM provided DNP enhancements of 63 within HEK293F cells (Figure 1C). Similar to the AMUPol degradation rate, EPR spectroscopy of the biradicals on TotaFAM indicate 43% signal loss after 30 minutes within cells, and less than 3% loss during the sample preparation time of <3 minutes (Figure 1F).

The stability of the chemical linkages employed in TotaFAM within cellular environments^29,30, in addition to the short time TotaFAM was exposed to degradation before freeze-quenching (<3 minutes), suggests that the biradical moiety remains covalently linked to the fluorophore. This new class of fluorescent DNP polarizing agents enables the correlation of subcellular localization determined with optical microscopy, to chemical and structural information determined with DNP-enhanced in-cell NMR.

Confocal microscopy was used to confirm cellular uptake through fluorescence detection. Comparison of the differential interference contrast image (DIC, Figure 3A) with the fluorescent image (Figure 3B) demonstrates cellular uptake of TotaFAM. Additionally, the fluorescent image shows that TotaFAM was not observed between cells.

Figure 3.

Confocal microscopy of a subset of cellular culture which was used for DNP NMR experiments. Images confirm cellular uptake of TotaFAM through comparison of DIC image (A) and fluorescent image (B). Scale bar = 25 μm.

To confirm the intracellular structures in which the TotaFAM accumulates, we performed additional microscopy using a nuclear stain, DAPI. Figure 4 shows DIC and fluorescent micrographs of HEK293F cells post incubation with TotaFAM, fixed with formaldehyde and subsequently stained with DAPI. Nucleoli have a low concentration of DNA and therefore bind DAPI with much lower affinity, leaving dark regions within the nucleus³¹. TotaFAM colocalizes with the dark regions of the DAPI stain (Figure 4B), indicating that TotaFAM accumulates within nucleoli (Figure 4C,D). Tat peptides conjugated to low molecular weight cargo have previously been demonstrated to localize to nucleoli within minutes^32,33. TotaFAM accumulates in nucleoli, but is also observed throughout the cell (Figure 4C).

Figure 4.

TotaFAM localizes to nucleoli of HEK293F cells. Confocal microscopy shows (A) DIC image, (B) staining of nuclei with DAPI, (C) uptake of trimodal polarizing agent, TotaFAM, and (D) overlay of (A), (B), and (C). White arrows indicate nucleoli. Scale bar = 25 μm.

Comparison of ¹³C DNP NMR spectra and T_{1_DNP} times to the TotaFAM enhanced data indicate cellular uptake of both AMUPol and sterically-shielded monoradical. The short magnetization recovery time of the samples (T_{1_DNP} < 3 s, see Supplementary Figure S11) indicate a close proximity of radicals to the observed carbon nuclei and excludes the possibility that radicals polarize the carbon within the cells via long-range proton spin diffusion mediated polarization from radicals not uptaken by cells. Supplementary figure S10 shows virtually identical spectra of HEK293F cells in carbon free phosphate buffered saline (PBS) and the standard MEM media containing carbon, confirming the observed carbon resonances originate from the HEK293F cells. We note small differences in the 13C spectra recorded with the nitroxide monoradical compared to the biradicals. We attribute these minor differences to the amphiphilic nature of the nitroxide monoradical and expect an altered distribution of radical to both the cytosolic and membranous regions. Therefore, we conclude both AMUPol and the sterically-shielded nitroxide monoradical are within the cells, yet cannot currently determine their intracellular distribution.

To further characterize the performance of polarizing agents for in-cell DNP, we determined DNP enhancements using both AMUPol and TotaFAM at 2.7 mM. TotaFAM DNP enhancement was nearly twice as large as AMUPol enhancement (63 compared to 35, Figure 5). The AMUPol concentration must be increased to 20 mM to yield similar enhancements. We note that AMUPol has previously been demonstrated to have more extensive depolarization than Totapol, which decreases the intensity of the NMR signal recorded without microwaves on³⁴. Therefore, the performance of TotaFAM compared to AMUPol is even better than indicated by the microwave on/off DNP enhancements we report (see Supplementary Figure S8). We also note an observed lengthening of ¹H T₁ to 6.7 s from 2.7 s when TotaFAM radical concentration is reduced from 5.2 to 4.8 mM (see Supplementary Figure S12). TotaFAM is therefore an effective polarizing agent for in-cell NMR, providing high DNP signal enhancements at relatively low concentrations. Lower concentrations of such targeted DNP polarizing agents³⁵ can be advantageous to reduce associated cell toxicity³⁶.

Figure 5.

A comparison between DNP-enhanced spectra of HEK293F cells below 6 K with TotaFAM at 2.7 mM (red), AMUPol at 2.7 mM (black), and AMUPol at 20 mM (blue). Experimental details can be found in Supplementary Table S1. Asterisks (*) denote spinning side bands.

CONCLUSIONS:

We present MAS DNP employing trimodal polarizing agents to achieve significant DNP NMR signal enhancements within intact human cells. A novel trimodal fluorescent polarizing agent penetrates human cells, enables subcellular localization by fluorescence microscopy and yields excellent DNP performance. DNP enhancements of 63 in human cells significantly increase the sensitivity of in-cell NMR, while access to sample temperatures below 6 K provides an additional 41-fold gain in NMR signal intensity compared to room temperature. Importantly, experiments can be repeated using a short delay (<3 s) for signal averaging. Together, these improvements in NMR sensitivity will reduce required signal averaging time by multiple orders of magnitude, both for in-cell DNP and also in vitro sample preparations. DNP enhancements of >50 are observed below 6 K, and also at 90 K, a temperature regime which is readily accessible on commercially available DNP spectrometers. These promising results lay the foundation for developing polarizing agents and methodology required to further improve DNP-enhanced in- cell NMR spectroscopy.

Trimodal polarizing agents provide a powerful localization method to study molecular structures at atomic resolution within targeted cellular structures. Further customization of the three moieties of trimodal polarizing agents will maximize their utility. For example, peptide specificity can be tuned for further targeting of subcellular localization of DNP polarizing agents within cells³⁷. Furthermore, sterically-shielded radicals will slow reduction inside cellular environments and allow for time intensive studies^21,38. Employing sterically-shielded radicals as DNP polarizing agents will also reduce cellular toxicity. Lastly, utilizing specialized fluorophores for super-resolution will enhance optical resolution and provide more accurate subcellular localization³⁹.

MAS DNP NMR at temperatures below 6 K greatly improves the capability of NMR to track the fate of metabolites, with important applications to metabolic flux analysis^40,41. Not only are NMR signals increased by orders of magnitude, but also the subcellular localization of the enhanced NMR signals can be determined using fluorescent polarizing agents. Therefore, through the advancements in DNP methodology and polarizing agents, we have established the experimental framework required to track cellular trafficking and chemical modifications of isotopically enriched metabolites. The advances demonstrated herein provide an impactful platform for future studies in endogenous environments and improved in-cell DNP.