Article

Examining Nanoparticle Interactions on a CellularScale Using Correlative Light Electron Microscopy

Brian Caffrey

Department of Biochemistry andMolecular Biology, University of British Columbia, Vancouver, V6T 1Z3, Canada; brian.caffrey@ubc.ca; Tel.:+1-236-833-5738

Received: 15 January 2021; Accepted: 18 March2021; Published:25March 2021

Abstract: One key area of nanomedicine research isthe mechanism of action of nanoparticles within the body, it’s biodistributionand pharmacokinetic properties. However, insight into the nanoparticle-cellinterface is limited, due in part to the heterogeneous nature of nanoparticlesas a drug class and the numerous possible interactions between cell andnanoparticle which stem from this diversity such as surface modifications,particle composition and final target tissue. Knowledge of the specific eventswhich occur once a nanotherapeutic has reached its target are still very earlyin development, this is in part due to the difficulty of resolving interactionsin sufficient spatiotemporal resolutions for a concrete understanding of themechanism for nanoparticle-cellular uptake. However, this is cruciallyimportant for our understanding of the therapeutic capabilities ofnanoconstructs as a whole. Traditionally, thestudy of nanoparticle-cellinteractions has remained exclusively in either the light or electronmicroscopies, sacrificing either high resolution cellular information oraccurate identification of specific biomolecules through fluorescent markersrespectively. This study demonstrates a method for the study and categorizationof nanoparticle-cell interactions using Correlative Light Electron Microscopy(CLEM), towards monitoring different drug delivery methods and their effect ontarget tissues at high resolutions.

Keywords: CLEM; focused ion beam-scanning electronmicroscopy (FIB-SEM); confocal microscopy; nanoparticle-cell interactions;transferrin-mediated uptake

1. Introduction

Recently, we are witnessing asea-change in the design approach of novel therapeutics, we now recognize thatin order to create more potent and less hazardous pharmaceuticals in thefuture, a more sophisticated approach is needed beyond conventionalsmall-molecule therapeutics [1]. Medicalresearch is beginning to explore different avenues of enquiry to discover moretargeted drug delivery mechanisms. Accordingly, we need a modularized platformfor delivering therapeutics in a targeted and consistent manner [2]. One such platform attempts to perform such a taskthrough the careful design and presentation of therapeutics in 3-dimensions onthe nanoscale similar to viral particles and other cellular machinery. It seemslike a logical step to emulate biology’s own machinery for the purposes ofdefeating biological diseases [3,4].

Nanoparticles are aheterogeneous class of nanostructures with often very different physical andchemical properties from one another but are related through theircharacteristic size (<100 nm in at least 1 dimension) and the uniquebiophysical properties these size constraints illicit from these structures [5]. Unable to diffuse passively across cell membranesthey lend themselves amenable to exploiting more complex receptor targetingpathways like endogenous processes utilized by our cells daily to grow,communicate and die. This higher level of abstraction allows for a variety ofdifferent targets and targeting designs [6]. However, current nanoparticle-based therapeuticshave been very slow to make it to market and progress has been mired by thevery properties we wish to exploit i.e. nanoparticle’s size and it’s3-dimensional nature [7].

Currently, few therapies havebeen implemented clinically, therapies such as liposomes have had limitedsuccess and mainly perform an improved formulation role rather than an activeagent for targeting disease [8]. Questionsstill remain to be answered in detail such as, how do nanoparticles behave onand inside cells and how can their behavior be modified to increase cellspecificity, drug targeting, bio-availability or indeed how modification to onecan affect the others? Research into nanoparticle uptake and bioprocessingtends to focus on two main areas, either through the use of fluorescent labelsto track organelles/receptors [9,10] of interestrelative to nanoparticles of different compositions or through the use ofelectron microscopic methods to study the exact position, number and morphologyof organelles affected by nanoparticle uptake [11,12].

The aim of this study is to usea combination of both techniques, to create a symbiosis where the limitation ofone method is mitigated by the other and vice versa. This correlated technique allowsthe pin-pointing of nanoparticle receptor interactions down to a couple of nanometerswhile also identifying the substituents with the accuracy of immuno- andtransfected labels. This process can map the course of nanoparticle-cellinteractions from receptor binding and endocytosis to excretion anddegradation.

To develop a method forcorrelative imaging of nanoparticle-cell interactions; this study focused onthe endocytosis of the nanoparticles through the transferrin receptor.Endocytosis is characterized by the internalization of biomolecules intomembrane-bound compartments. Vesicular trafficking is divided into two maincategories, the clathrin-mediated endocytic pathway and the non-clathrinmediated endocytic pathway. The transferrin receptor (TfR) is a stereotypicalexample of the family of receptors internalized through the clathrin-mediatedpathway. Overexpression of the TfR is characteristic of several differentcancer cell populations [13], thustransferrin is a suitable candidate for drug delivery systems. Therefore,transferrin bound gold nanoparticles (Tf-AuNp) were used as a model particlefor this study because of its safety, selectivity, tumor-targeting capabilitiesand because of its interest in the nanoparticle field [14].

Correlative Light ElectronMicroscopy (CLEM) combines both light microscopic techniques such as confocalmicroscopy and electron microscopic techniques such as FIB-SEM, thesetechniques complement each other, merging the large sampling volume and broadselection of fluorescent labels and applications e.g. Förster Resonance EnergyTransfer (FRET) [15], Fluorescence Lifetime Imaging (FLIM) etc.with the high resolving power of the electron microscope. As nanoparticle-cellinteractions occur on the scale of the tens of nanometers, combining bothtechniques provide unique insights greater than the sum of their parts,leveraging the discriminating power of light microscopy with the high precisionand high spatial isotropic resolution of electron microscopy.

2. Materials and Methods

2.1. Subsection

A549 Human lung carcinomaepithelial cells obtained from American Type Culture Collection (ATCC, Manassas,VA, USA) were grown in cell culture flasks (Stratagene, San Diego, CA, USA)with 10ml Complete Minimum Essential Media (cMEM) which contained MinimalEssential Medium (Corning Incorporated, Corning, NY, USA) with 10% Fetal BovineSerum (FBS, Gibco, Ireland) and 1% Penicillin-Streptomycin (Pen-Stryp; Invitrogen,Carlsbad,CA, USA)in a humidified 37oC incubator with 5% CO2. Cells were passaged 2-3 times aweek depending on the confluency of the cultures when passaging i.e. between60-90% coverage of flask surface, the medium was removed from the flasks andwashed with Dulbecco’s Phosphate Buffered Saline (DPBS; Gibco,Carlsbad, CA, USA),the cells were incubated in 5ml of 0.05% trypsin-EDTA (Gibco) at 37^oCfor 3 min, the flasks were tapped gently to ensure complete detachment of cellsfrom the flask surface. 7 mL of cMEM was added to the cell suspension and thesolution was transferred to a sterile 50ml Falcon tube. The falcon tube wasspun down at 200 g for 3 min and the supernatant removed. The cell pellet wasresuspended in 10 mL cMEM and mixed thoroughly. 10 μL ofthe suspension was placed on a Neubauer haemocytometer and their concentrationdetermined. A volume containing 5 × 10⁵ cellswas mixed with 10ml cMEM in a fresh flask and incubated in an incubator at 37^oCand 5% CO₂. Cells were tested monthly for mycoplasma using theMycoAlert test kit (Lonza Inc., Morristown, NJ, USA).

2.2.Transferrin Absorbed Gold Nanoparticle Preparation

GoldNanoparticles (100 nm) (Cytodiagnostics, Burlington, ON, USA) were mixed withholo-Transferrin protein powder obtained from Sigma-Aldrich to a final proteinconcentration of 1 mg/mL and incubated for 1hr at 37^oC. Thesolutions were removed from the water bath and spun down three times at 400 gfor 30 min, after each spin the supernatant was removed and replaced with anadequate volume of DPBS. The final nanoparticle solution concentration wasmeasured using the Nanodrop Spectrometer 2000 and normalized to a concentrationof 1 × 10⁷nanoparticles per ml at peak wavelength 572 nm and using the molar extinctioncoefficients supplied by Cytodiagnostics (1.57 × 10¹¹ M^-1cm^-1). Particle stability, size and polydispersity was measuredusing the UNCLE Dynamic Light Scattering (DLS) instrument. The Z-average (Z)was calculated as the intensity-weighted mean hydrodynamic size of the particlesin solution, derived from a Non-Negative Least Squares (NNLS) analysis of themeasured correlation coefficient between the observed change in scatteringintensity over time and that of a theoretical solution [16]. The PolyDispersity Index (PDI) was calculated as thesquare of the standard deviation () divided bythe square of the Z-Average (Equation.1).

PDI= (/Z)²                                            (1)

PDI values < 0.05 areconsidered monodisperse i.e. of similar hydrodynamic diameter and those > 0.7are considered to be polydisperse i.e. a mixture of different particle sizes [17].

2.3. Tf-Gold Addition to A549 Cells

A549cells were split according to section 2.1 and 5 × 10⁴cells were mixed with 3 mL of CMEM and added to each 35 mm culture dish withGridded Coverslips (No. 1.5) (MatTek,Ashland, MA, USA). The dishes were incubated at 37^oCwith 5% CO₂ for 48 h. The media was removed from each dish and thecells were washed with 3ml of warm DPBS and 1ml of freshly prepared andnormalized Tf-Gold nanoparticle solution, as per section 2.2, was added to eachdish. The cells were incubated at 37^oC with 5% CO₂ for aperiod of time. The nanoparticle solution was removed and the cells were washedwith 3 mL of cold DPBS (4^oC) and a fixative solution of 1 mL of 0.2%(v/v) of glutaraldehyde (GA) and 4% paraformaldehyde (PFA) in DPBS was added,the dishes were placed in the fridge and incubated at 4^oC for 1hr.The fixative solution was mixed with 2 mL of cold DPBS and removed.

Thecells were washed once more with 3 mL cold DPBS. For permeabilizationexperiments, a permeabilization solution of 3% (w/v) BSA and 0.1% (w/v) saponinin DPBS was added to the fixed cell and incubated at room temperature for 1 h.The cells were treated according to section 2.4 depending on the experiment.

2.4. Antibody Receptor Staining

Anti-Transferrinreceptor rabbit IgG polyclonal antibody (1:100 in DPBS with 1% (w/v) BovineSerum Albumin) (BSA; Abcam, Cambridge, UK) was added to each gridded coverslipand incubated at 4^oC for 1 h or o/n. After incubation in primaryantibody, the dish was washed with cold DPBS w/1%BSA and Anti-Rabbit IgG AlexaFluor 647 antibody (1:200 in DPBS w/1%BSA) is added to the coverslip andincubated at 4^oC for 1 h or overnight.

Anti-Clathrinheavy chain mouse IgG polyclonal antibody (1:200 in DPBS with BSA) (ThermoFisher, Waltham, MA, USA) was added to each gridded coverslip and incubated at4^oC for 1 h or o/n. After incubation in primary antibody, the dishwas washed with cold DPBS w/1%BSA and Anti-Mouse IgG Alexa Fluor 568FluoroNanogoldÔ antibody (1:200 in DPBS w/1%BSA) (Nanoprobes, Yaphank, NY,USA) is added to the coverslip and incubated at 4^oC for 1 h orovernight.

Fordouble-stain experiments, the primary and secondary staining steps wereperformed sequentially with transferrin and anti-rabbit staining beingperformed first, as described above, followed by several washes of cold DPBSwith 1% BSA and subsequent addition of the anti-clathrin and anti-mouseantibodies as described above. Depending on the experiment,4′,6-diamidino-2-phenylindole (DAPI) (Thermo Fisher) counterstaining wasperformed by adding 100 mL of 10 mg/mL solution to the cell and incubated in the dark for 1min at room temperature and Phallodin Alexa Fluor 546 staining was performed byadding 5 mLof methanolic stock solution to 200 mL DPBS. This solution was added to the cells and incubatedin the dark at room temperature for 20 min. After incubation the dish waswashed with cold DPBS with 1% BSA, twice with cold (4^oC) DPBS andfinally suspended in cold DPBS. The dish was brought to the confocal microscopycore for imaging.

2.5. Confocal Imaging of Tf-Gold Incubated A549

Imagingwas carried out with an inverted confocal laser scanning microscope inreflection mode, LSM780 (Carl Zeiss, Jena, Germany) using both 63 × oilimmersion and 10 × air objectives.

Theargon laser line of wavelength 488 nm (Power: 25 mW /Pinhole diameter: 49.6 mm/ (1.1AU)(0.34 mmZ-slice) ) was used in conjunction with a dichroic mirror (MBS T80/R20, CarlZeiss) and a Main Beam splitter (MBS) MBS-405 to visualize the goldnanoparticles with a photomultiplier tube (PMT) detector (detection window: 463–507nm). Other laser lines of wavelength 561 nm (DPSSL561/ Power: 20 mW/ PinholeDiameter: 47.2 mm (0.34 mm Z-slice)/ Detection Window: 579–614 nm) and 633 nm(HeNe633) (Power: 5 mW / Pinhole diameter: 46.9 mm (0.34 mmZ-slice)/ Detection Window: 650–695nm) was used in conjunction with a Main BeamSplitter (MBS 488/561/633, Carl Zeiss) to excite and detect fluorescence fromAlexa Fluor 568 and Alexa Fluor 647 secondary antibodies respectively.Widefield mode was used to locate ROIs and a series of three stacks atmagnifications × 10, × 63 and a digital zoom × 126 (2 × 63) were typically madefor correlative work.

Thefirst image at × 10 (Image size: 512px × 512px / Pixel size: 1.66 mm ×1.66 mm (xy)was to locate a ROI on the grid in widefield mode, the second was a × 63 stack(Image size: 512px × 512px / Pixel size: 0.26 mm × 0.26 mm ×0.34 mm (xyz)of the ROI with widefield, 488 nm and 561 nm or 633 nm laser lines to image thenanoparticles, transferrin receptors and clathrin molecules respectively. Thethird and final image was a × 2 digital zoom of a × 63 magnification (Imagesize: 512px × 512px / Pixel size: 0.130 mm × 0.130 mm ×0.34 mm (xyz))on a cell of interest. Several locations on the grid were taken in this manner.

2.6. Fixing, Staining and Resin Embedding of Confocal ImagedA549 with Tf-Gold Nanoparticles for FIB-SEM

Thismethod was performed with some adaptations as previously described in [18]. Immediately after imaging the DPBS was removed andreplaced with 2% Glutaraldehyde in 0.1 M Cocodylate Buffer (Electron MicroscopySciences (EMS),Hatfield, PA, USA) at 4^oC, the cells wereincubated at 4^oC for 1 h.

ForSEM visualisation of the FluoroNanogoldÔ antibodies,the cells were incubated 3 times (×3) with 50 mM glycine in PBS for 5 min eachto remove trace aldehydes which inhibit gold enhancement. The cells were washedwith 1% BSA in PBS for 5 min (×3) and finally for 5 min (×3) in ddH₂O.Using the Gold EnhanceÔ EM kit (Nanoprobes), 10 mL each ofSolution A and B were mixed and incubated at room temperature for 5 min, 10 mLeach of Solution C and D were added. The 40 mL were addedto the cells and incubated for 5 min. The cells were washed with ddH₂O(×3).

Afterfixation and/or FluoroNanogoldÔ enhancement, the cells were washed with 0.1 M of coldcocodylate buffer (×3) and incubated in 2% v/v OsO₄ in 0.1 Mcocodylate buffer with 0.5% (w/v) Ferrocyanide (Thermo Fisher) for 1 h at roomtemperature (RT).The sample was washed with ddH₂O(×3) and incubated in cold-filtered 2% v/v Uranyl Acetate (UA) in ddH₂Oat 4^oC overnight. After o/n incubation the UA solution was removedand the sample washed with ddH₂O (×3) and replaced with a 50 : 50ethanol:water solution and incubated at RT for 5 min (×3). The ethanol : watersolution was replaced with 70 : 30 ethanol : water solution and incubated at RTfor 5mins (x3), this process was repeated for 90 : 10 ethanol : water (×3) and100% ethanol solutions for 10 min (×2). During the dehydration procedure withethanol the epoxy resin was prepared by creating “Solution A” by mixing DoDecenylSuccinicAnhydride (DDSA) with Embed-812 resin (EMS) to a ratio of 1 g : 0.76 g. Anothersolution, “Solution B” was also created by mixing resin hardener Methyl-5-Norbornene-2,3-dicarboxylicAnhydride (NMA) (EMS) with Embed-812 resin to a ratio of 0.87 g : 1 g.

10min before addition to the cells, Solution A and B were mixed to a ratio of 2 mL: 8 mL and mixed thoroughly, immediately before addition to the cells thesolution was mixed with 0.2 mL of accelerator Tris-(Dimethylaminomethyl) Phenol(DMP-30) (EMS) and heated to 60^oC for 5 min to remove bubbles.After incubation in 100% Ethanol, a 50:50 Resin : Ethanol mixture was added tothe cells and incubated for 1 h at RT and placed on an orbital shaker at < 5Hz. The resin : ethanol mixture was replaced with 70 : 30 resin : ethanolmixture and incubated for 1 h at RT on an orbital shaker. This was repeated for90 : 10 resin : ethanol and 100% resin solutions. After incubation in 100%resin for 1 h, a fresh batch of 100% resin solution was prepared as previouslydescribed and added to the cells. The cells were incubated at RT for 1 h andmoved to an oven to cure overnight at 60^oC.

2.7. FIB-SEM Imaging of Tf-Gold Incubated A549

Thesample was mounted on an aluminium stub using a double-sided adhesiveconductive carbon tab, and the sides painted with silver paint to preventcharge build-up, allowed to dry and then placed in a sputter coater(Cressington model 108), and coated with gold for 40 s at 30 mA.

Aftergold coating, the sample was placed into the sample chamber of the FIB-SEM.FIB-SEM imaging was performed using a Zeiss NVision 40 microscope, with the SEMoperated at 1.5 keV landing energy, a 60 μm aperture and backscatteredelectrons were recorded at an energy selective back-scattered electron (ESB)detector. The user interface employed ATLAS 3D from Carl Zeiss, consisting of adual 16-bit scan generator assembly to simultaneously control both the FIB andSEM beams and dual signal acquisition inputs, as well as the necessary softwareand firmware to control the system.

Theregion of interest was located using the SEM, and the instrument was brought toeucentric and coincidence point at a specimen tilt of 54^o, i.e. thespecimen height where the specimen does not move laterally with a change intilt and where the focal point of both FIB and SEM coincide. Once the exactmilling area was determined with reference to the microscope images, aprotective platinum pad was laid down on top of the area using a Gas InjectionSystem (GIS) of size 60 μm × 30 μm and 5 μm in thickness. Alignment marks wereetched into the platinum pad using an 80 pA FIB aperture to allow for automatedtracking of milling progress, SEM focus and stigmation during acquisition.After alignment etching, the platinum pad was covered with a carbon pad using theGIS to protect the etched marks from the milling process. After deposition ofthe carbon pad, a trench was dug using a 27 nA FIB aperture to allow forline-of-sight for the SEM ESB detector. After the trench was dug, the imagingface was polished using a 13 nA FIB aperture. The FIB aperture was changed to700 pA and SEM imaging area selected (Typical Image size: 4000 px × 2000 px/Pixel size: 15 × 15 × 15 nm³ (xyz) the automated acquisitionsoftware was set up and run until all the sample area was acquired.

2.8. Electron Dispersive X-Ray Spectroscopy

EDSdetection was performed on a FEI Helios 660 Nanolab FIB-SEM with a 5keV landingenergy, a 200pA aperture and a windowless Oxford XMax Xtreme detector. Pointsof interest were selected for analysis and spectra were acquired from 30 secexposures at a time. The data was graphed using DTSA-II software (version Jupiter,National Institute of Standards and Technology (NIST), Gaithersburg, MD, USA).

2.9. Image Processing and Segmentation

2.9.1. Confocal Image Processing

Confocalimage file (.lsm) was opened in ImageJ and the channels were split(Red:568/647/Green:488/Widefield). The coverslip bottom was found using thegreen laser line reflectance and everything below this Z-height was removed.Each image was set to a 10-bit grayvalue (images were acquired at 12-bit) forboth the Red and the Green channels. Both channels were merged together, savedas .tif and used for correlative alignment.

2.9.2. FIB-SEM Image Processing

AfterSEM acquisition the individual image files (.tif) were aligned using IMOD’stiltxcorr program and binned by 3 in the x and y planes to produce isotropicpixels of size 15 nm × 15 nm × 15 nm (xyz). The images were opened in ImageJand cropped to isolate ROIs, a representative slice was picked and an area ofsize 10 × 15 mm was selected and using the “Enhance Contrast” functionwas used to normalize and equalize the whole stack relative to the chosen area.The stack was binned by 2 in the x,y and z axes to produce a final voxel sizeof 30 mm³,the final stack was resliced down the Y axis to aid in manual alignment andregistration of the light and electron images.

2.9.3. Semi-Automated Registration, Alignment and Correlationof Light/Electron Images

BothConfocal and FIB-SEM stacks were moved to Icy software [19] where both stacks were opened in the eC-CLEM plugin [20] and gold nanoparticles in both images were correlated. TheeC-CLEM software aligned the stacks based on the coordinates of the sharedregistration marks. A non-rigid transformation was also performed on alignmentsthat didn’t correlate efficiently, this inefficient correlation is mainly dueto the contraction of cells after dehydration and resin embedding.

2.9.4. Segmentation and 3D representation of CellularStructures

The correlated data set allowedfor identification of nanoparticles of interest, i.e. nanoparticle/Transferrinreceptor co-localisation. These nanoparticles of interest were selected in ahigh-resolution image stack (Voxel size: 5 nm × 5 nm × 15 nm) and cropped in volumes of 2 mm³. These volumes were moved toSlicer-3D software [21] for segmentation. Cellular structures ofinterest i.e. cell membrane, vacuoles and nanoparticles were manually segmentedby using threshold values appropriate for the structures and using automatedsegmentation methods described in [22].

3. Results

3.1. Nanoparticle Characterisation andStability

In order to ensure a uniform particle preparation and the stability ofprepared nanoparticles in solution during nanoparticle uptake experiments, itwas necessary to analyse the particle size and distribution over time insolution.

Table 1demonstrates a shift in hydrodynamic size (Z-Average), measured by DynamicLight Scattering (DLS) of 100 nmgold particle upon Transferrin absorption. The polydispersity index is used asan indication of the relative distribution of particle sizes within themeasured population.

Table1. DynamicLight Scattering Measurements of Nanoparticle Protein Complex.

¹n=numberof technical replicates

Figure 1a, demonstrates the increase inhydrodynamic diameter upon addition of transferrin to the solution and thebroadening of the peak reflects the increase in the polydispersity of thesolution. Figure 1b, indicates that the preparedtransferrin particle solution remained stable up to 15 h. The gradual increasein Z-average and subsequent drop over the next 8hrs is most likely due to thegold nanoparticle agglomeration typical of nanoparticles of this size and notaggregation, nevertheless this graph shows that the particles are stable wellbeyond the timescale of nanoparticle passaging in this project (< 4 h) andno passage of gold nanoparticles was made 12 h after production.

Figure 1. DLS Analysis of Transferrin BoundNanoparticle Stability. a) DLS intensity measurement of 100 nm Goldparticles before (orange) and after (blue) transferrin absorption. b)DLS Z-Average stability measurement of Tf-absorbed gold nanoparticles at 37^oCover a 23 h period.

3.2.Characterization of Nanoparticles and Heavy-Metal Staining in FIB-SEM Images

Gold nanoparticles were chosenfor this FIB-SEM analysis, primarily due to their high electron-scatteringproperties which generated high-contrast objects, easily identifiable in theacquired FIB-SEM volume. However, due to the application of heavy metalcontrast agents there were a number of high contrast objects in our acquiredvolumes which required identification in order to ensure accuratecharacterization of gold nanoparticles.

Energy-Dispersive X-RaySpectroscopy (EDS)analysis allowed for the elemental characterisation of the three maincomponents of the sample with high electron-scattering intensities (circled)and allowed for the verification of gold nanoparticle size and shape in thecell sample, for future acquisitions. The three main elements of interest in Figure 2 are osmium, gold and uranium, characteristic peaks ofeach element in the EDS spectra were circled green, red and blue respectively.Elemental chlorine was found to varying degrees in all datasets due to thepotassium and sodium chloride solutions used in PBS and cell culture media,carbon and oxygen peaks were also present. Gallium peaks were present tovarying degrees in all datasets, due to the use of focused gallium ions toablate the surface of the sample in standard FIB-SEM acquisition.

Figure 2. Energy-Dispersive X-Ray Spectroscopy (EDS)Analysis of Resin Block. a) SEM image of typical central slice throughnanoparticle treated cell, with high-scattering ROI’s selected for EDS analysiscircled; b)Relativeabundances of key elements in EDS spectra. Scale bar: 1 mm.

The first object in the volumesto be analysed (Figure2a), circled blue, was foundto almost exclusively contain uranium (Figure 2b), this is a common feature of uranyl acetatesolutions which can develop aggregations such as the one present here duringspecimen preparation. The second object investigated in these volumes, theputative gold nanoparticle, circled red, was confirmed to indeed contain highconcentrations of elemental gold. The relatively low number of counts comparedto the other two spectra is due to the small nanoparticle size, which isquickly ablated by the intense electron beam. Finally, circled in green, wasfound to contain high concentrations of osmium with trace amounts of gold anduranium. The identification of osmium as the primary source of the highelectron-scattering found in these types of cellular vesicles, indicate thepresence of lipids, when coupled with their relatively large size (~500 nm)suggests these vesicles are lysosomes and not intracellular accumulations ofgold nanoparticles.

Energy-Dispersive X-Ray Spectroscopy (EDS) analysis allowed for theelemental characterisation of the three main components of the sample with highelectron-scattering intensities and provided for the verification of goldnanoparticle size and shape in the cell sample, for future acquisitions.

3.3. Determination of Fixation andPermeabilization Parameters for Retention of Ultrastructural Information fromLight to Electron Microscopy

During CLEManalysis it became apparent there was a need for optimisation of cellpreparations for electron microscopy. Paraformaldehyde fixation precedingpermeabilization was not sufficient to preserve cellular ultrastructure duringthe harsh dehydration and resin embedding steps necessary for EM acquisition (Figure 3a).

GA fixation wasavoided in early experiments due to its associated reduction of antigenicityand autofluorescence in Light Microscopy (LM), however after testing severalconditions it was determined that GA was necessary to stabilise cellularstructure enough to survive permeabilization and subsequent dehydration andembedding. A concentration of 0.2% v/v of GA, in combination with a reducedsaponin concentration (0.1% w/v) was determined to be optimal to allow forsufficient retention of high-resolution ultrastructural information in the EM (Figure 3b) while also minimisingundesirable effects in the LM.

3.4. Correlative Light Electron MicroscopyAnalysis of Nanoparticle-Cell Interactions

Figure 4 illustrates the steps requiredto move from light microscopy to electron microscopy with the help of griddedcoverslips. By co-localising nanoparticle signal with that of the transferrinreceptor it was possible to focus the ultrastructural analysis of nanoparticleuptake on specific receptor-mediated interactions, rather than analysingseveral of specific and non-specific uptake mechanisms.

Figure 4. Correlating Light and Electron Microscopy Volumes.a) 10× brightfield image of coverslip surface withA549 cells treated with transferrin nanoparticles with inset 63× (red) and 4× digital zoom (cyan).; b)Immuno-stained confocal image with co-localized signals (pink circle) ofnanoparticles (green) with transferrin receptor (red) and nucleus (blue).; c)SEM image of resin surface with same region of interest as in Figure 4a,b.; d)Maximum Intensity Projection (MIP) of FIB-SEM volume with the same co-localizednanoparticles as in Figure 4b (pink circle).; e) MIP ofCorrelatedLight Electron Volume. f) Correlative 3D projection of singlenanoparticle with colocalized light microscopy signal.

Figure 5.Segmented Images of Different Stages ofNanoparticle Uptake. (i) High resolution EM slices with manuallysegmented (Pixel size: 5 × 5 nm (xy); 5 × 15(xz) / Scale bar: 500 nm); (ii) High resolution EMwith correlated LM slice (Red: TfR / Green: AuNp / Cyan: Nucleus / Blue:Clathrin) (Pixel size: 5 × 5 nm (xy) / Scale bar: 500 nm); (iii)Segmented 3D volume of EM data (Pink: Cell Membrane/ Green: AuNp/ Blue: LowUA/OsO₄ labelling density (e.g. Vacuoles)/ Red: High UA/OsO₄labelling density (e.g. high lipid/protein content)/ Yellow: Lysposome/)(Voxel size: 5 × 5 ×15 nm³ (xyz) / Scale bar: 500 nm). a)Np-Tf Receptor (TfR) co-localised on cellular extension:Nanoparticlesbind to transferrin receptors on cellular protrusions.; b) Np-TfRco-localised at cell surface:Nanoparticles can also bind receptors onthe surface of cells, but it is difficult to determine whether this is a resultof the retraction of cellular protrusions or a separate cellular event. Someexamples have indicated the presence of microtubules consistent with theretraction of cilia from the surface of the cell, whilst others are absent.;c) Np transport in endocytic vesicle:This shows the a nanoparticlecompletely engulfed by a cell in an endocytic vesicle; d) Np-TfRco-localised engulfing of Np:The nanoparticle is engulfed in a pocketof cellular membrane at the base of a cilium before uptake is completed and isconnected to an early endosome.; e) Np-TfR co-localised in an earlyendosome:As the nanoparticle is transported deeper into the cytosol, itfuses with an early endosome for sorting and further trafficking.; f) Npin a Late Endosome/Multi-vesicular body (MVB):In later stages ofnanoparticle trafficking the nanoparticle is transported to a late endosome inpreparation for degradation and exocytosis.; g) Lysosomal fusion andformation of Autolysosome:The nanoparticle is degraded further in alysosomal vesicle within the autolysosome.; h) Particle Excretion: Aftertrafficking through the cell the nanoparticle is excretedfrom the cell.

Vesicles in EMvolumes were classified according to previous morphological studies ofintracellular compartments, as follows [23–25]:

Endocytic vesicles:Small spheroidvesicles (50–200 nm diameter) with densely labelled membranes and an innervacuole.

Early endosomes:Large irregularvesicles(200–400nm diameter) with densely labelled reticulated membranes and compartments withan inner vacuole.

Lateendosome/Multi-vesicular bodies: Large spherical vesicles (400–700nm diameter) containing severalsmaller (20–100 nm) vesicles.

Lysosome: Large (400–700 nm) very denselylabelled spherical vesicles with no visible inner compartments.

Autolysosome: Extremely large (0.7–1.5 mm) vesicle containing multiplecompartments of varying labelling density and containing at least one lysosome.

4. Discussion

Thestudy of nanotherapeutics is particularly suited to the use of correlativelight electron microscopic methods. This manuscript presents methods forassessing the cell’s behaviour and interactions which occur innanoparticle-cell interactions and demonstrates the ability to track multiplebiochemical tags of interest to high resolutions and determine the structuralenvironment nanoparticles encounter along their path through the cell.

Thisinvestigation generated a correlated 3D map of a model nanoparticle system’sinteraction throughout the cell, using both light and electron information to identifythe key checkpoints from initial binding to final excretion of thenanoparticle. It was noticed early on that sample preparation methods are oftendesigned explicitly for one mode of microscopy, in ways that are regularly tothe exclusion of another, e.g. fixation methods that work best for EM samplesoften degrade protein antigenicity, making fluorescent labelling difficult.Therefore, this method, which retains both the ultrastructural and antigeniccharacter of the specimens throughout the light and electron microscopyprocesses will be extremely useful in studying nanoparticle action in cells inthe future. Although this method relies on gold nanoparticles as both markersfor correlation (due to their reflective and opaque properties in both lightand electron microscopies respectively) and as subjects themselves for study,this method can be applied to a wide variety of nanoparticles with the use ofstandard fiducial markers to correlate the two microscopies.

Despiteseveral advantages of this methodology over conventional imaging in eitherlight or electron modes, there still remains a number of limitations to thistechnology. The principle limitation is that sample preparation in eitherimaging mode must not preclude the ability to image in the other, for mostapplications this can be avoided by careful experimental design, however thisdoes limit the number of conditions and types of experiments available to CLEMtechniques. Another common issue with CLEM methods is the limited quantitativeinterpretability of EM datasets [26],however this is a very active field in electron microscopy and computationalbiology and recent advances [22,27,28]are developing methods for both the accurate segmentation and quantitation ofEM datasets. Finally, probably the largest hurdle to the broad application ofthis technology to the wider scientific community, is the relatively highinstrument cost and maintenance fees of the FIB-SEM, not to mention theinfrastructure required to house these instruments and to handle/process thelarge files produced by these microscopes. However, just as the recentcryo-electron microscopy revolution [29,30]has led to a surge in the adoption of transmission electron microscopes andwitnessed a reduction in the cost of producing high quality protein structures [31,32].

Inconclusion, this technique can be applied to a wide variety of nanotherapeuticstudies and expanded to even higher-resolution techniques such as STtochasticOptical Reconstruction Microscopy (STORM) to identify and differentiate smallercellular structures in electron microscopy [33], often key in the accurate targeting and distribution oftherapeutic agents. This manuscript demonstrates a proof-of-concept, thatdefines the biomolecular makeup of nanoparticle-cell interactions on ananoparticle by nanoparticle scale.

Funding: None.

Acknowledgments: Thank you to theDawson and Subramaniam laboratories for helpful discussions and providingaccess to facilities.

Conflicts of Interest:The author declares noconflict of interest.

CopyrightStatement

References

1.Aggarwal BB, Sethi G,Baladandayuthapani V, Krishnan S, Shishodia S. Targeting cell signalingpathways for drug discovery: An old lock needs a new key. Journal ofCellular Biochemistry, 2007, 102: 580–592.

2.Rollings N, Barker C, VellacottR. A flexible, modular platform to disrupt the blockbuster manufacturing model.ONdrugDelivery, 2019, 103: 48–52.

3.Singh AP, Biswas A,Shukla A, Maiti P. Targeted therapy in chronic diseases usingnanomaterial-based drug delivery vehicles. Signal Transduction and TargetedTherapy, 2019, 4.

4.Patra JK, Das G, FracetoLF, Campos EVR, Rodriguez-Torres, MP, et al. Nano based drug delivery systems:Recent developments and future prospects. Journal of Nanobiotechnology,2018, 16: 71.

5.Salata OV. Applicationsof nanoparticles in biology and medicine. Journal ofNanobiotechnology, 2004, 2: 3.

6.Pearce AK, O’Reilly RK.Insights into Active Targeting of Nanoparticles in Drug Delivery: Advances inClinical Studies and Design Considerations for Cancer Nanomedicine. BioconjugateChemistry, 2019, 30: 2300–2311.

7.Pasut G. GrandChallenges in Nano-Based Drug Delivery. Frontiers in Medical Technology,2019, 1.

8.Sercombe L, Veerati T, MoheimaniF, Wu SY, Sood AK, et al. Advances and challenges of liposome assisted drugdelivery. Frontiers in Pharmacology, 2015, 6.

9.Varela JA, Åberg C,Simpson JC, Dawson KA. Trajectory-Based Co-Localization Measures forNanoparticle-Cell Interaction Studies. Small, 2015, 11: 2026–2031.

10.Kalimuthu K, Lubin BC, BazylevichA, Gellerman G, Shpilberg O, et al. Gold nanoparticles stabilizepeptide-drug-conjugates for sustained targeted drug delivery to cancer cells. Journalof Nanobiotechnology, 2018, 16: 34.

11.Guehrs E, Schneider M, GüntherCM, Hessing P, Heitz K, et al. Quantification of silver nanoparticleuptake and distribution within individual human macrophages by FIB/SEM sliceand view. Journal of Nanobiotechnology,2017 15: 21.

12.Wierzbicki R, Købler C, JensenMRB, Łopacińska J, Schmidt MS, et al. Mapping the Complex Morphology of CellInteractions with Nanowire Substrates Using FIB-SEM. PLoS One,2013,8.

13.Gatter KC, Brown G,Strowbridge I, Woolston RE, Mason DY. Transferrin receptors in human tissues:Their distribution and possible clinical relevance. Journal of ClinicalPathology, 1983, 36: 539–545.

14.Qian ZM, Li H, Sun H, HoK. Targeted drug delivery via the transferrin receptor-mediated endocytosispathway. Pharmacological Reviews, 2002, 54: 561–587.

15.Metskas LA, Briggs J.AG.Fluorescence-based detection of membrane fusion state on a cryo-EM grid usingcorrelated cryo-fluorescence and cryo-electron microscopy. Microscopy andMicroanalysis, 2019, 25: 942–949.

16.Stetefeld J, McKenna SA,Patel TR. Dynamic light scattering: a practical guide and applications inbiomedical sciences. Biophysical Reviews, 2016, 8: 409–427.

17.Danaei M, DehghankholdM, Ataei S, Hasanzadeh Davarani F, Javanmard R, et al. Impact ofparticle size and polydispersity index on the clinical applications of lipidicnanocarrier systems. Pharmaceutics, 2018, 10: 57.

18.Narayan K, Danielson CM,Lagarec K, Lowekamp BC, Coffman P, et al. Multi-resolution correlativefocused ion beam scanning electron microscopy: Applications to cell biology. Journalof Structural Biology, 2014, 185: 278–284.

19.De Chaumont F, DallongevilleS, Chenouard N, Hervé N, Pop S, et al. Icy: An open bioimage informaticsplatform for extended reproducible research. Nature Methods, 2012, 9:690–696.

20.Paul-Gilloteaux P, HeiligensteinX, Belle M, Domart MC, Larijani B, et al. EC-CLEM: Flexible multidimensionalregistration software for correlative microscopies. Nature Methods, 2017,14: 102–103.

21.Fedorov A, Beichel R, Kalpathy-CramerJ, Finet J, Fillion-Robin JC, et al. 3D Slicer as an image computing platformfor the Quantitative Imaging Network. Magnetic Resonance Imaging, 2012, 30:1323–1341.

22.Caffrey BJ, MaltsevAV, Gonzalez-Freire M, Hartnell LM, Ferrucci L, et al. Semi-automated 3Dsegmentation of human skeletal muscle using Focused Ion Beam-Scanning ElectronMicroscopic images. Journal of Structural Biology,2019, 207:1–11.

23.Klumperman J, Raposo G.The complex ultrastructure of the endolysosomal system. Cold Spring HarborPerspectives in Biology, 2014, 6: a016857.

24.Mukherjee S, Ghosh RN, MaxfieldFR. Endocytosis. Physiological Reviews, 1997, 77: 759–803.

25.Murk JLAN, Humbel BM, ZieseU, Griffith JM, Posthuma G, et al. Endosomal compartmentalization in threedimensions: Implications for membrane fusion. Proceedings of the NationalAcademy of Sciences, 2003, 100: 13332–13337.

26.Cortese K, Diaspro A,Tacchetti C. Advanced Correlative Light/Electron Microscopy: Current Methodsand New Developments Using Tokuyasu Cryosections. Journal of Histochemistry& Cytochemistry,2009, 57: 1103–1112.

27.Saladra D, Kopernik M.Qualitative and quantitative interpretation of SEM image using digital imageprocessing. Journal of Microscopy,2016,264: 102-124.

28.Scheffer LK, Xu CS, JanuszewskiM, Lu Z, Takemura SY, et al. A connectome and analysis of the adult drosophilacentral brain. Elife, 2020, 9: e57443.

29.Callaway E. Therevolution will not be crystallized: A new method sweeps through structuralbiology. Nature 2015, 52: 172–174.

30.Callaway E.Revolutionary cryo-EM is taking over structural biology. Nature, 2020, 578:201.

31.Merk A, Fukumura T, ZhuX, Darling JE, Grisshammer R, et al. 1.8 Å resolution structure ofβ-galactosidase with a 200 kV CRYO ARM electron microscope. IUCrJ, 2020,7: 639–643.

32.Hamdi F, Tüting C, SemchonokDA, Visscher KM, Kyrilis FL, et al. 2.7 Å cryo-EM structure of vitrified M.Musculus H-chain apoferritin from a compact 200 keV cryo-microscope. PLoSOne, 2020, 15: e0232540.

33.Bowler M, Kong D, Sun S,Nanjundappa R,Evans L, et al. High-resolution characterization of centriole distal appendagemorphology and dynamics by correlative STORM and electron microscopy. NatureCommunications, 2019, 10: 993.