Quantitative Particle Uptake by Cells as Analyzed by Different Methods

Abstract There is a large number of two‐dimensional static in vitro studies about the uptake of colloidal nano‐ and microparticles, which has been published in the last decade. In this Minireview, different methods used for such studies are summarized and critically discussed. Supplementary experimental data allow for a direct comparison of the different techniques. Emphasis is given on how quantitative parameters can be extracted from studies in which different experimental techniques have been used, with the goal of allowing better comparison.

II) Uptake studies of SNARF-loaded capsules by cells using fluorescence microscopy III) Flow cytometer analysis of SNARF-loaded capsules internalized by cells IV) Elemental analysis of gold nanoparticle (GNP)-loaded capsules internalized by cells V) Viability measurements of cells exposed to capsules and inhibitor VI) References Note that excerpts of the results shown in the graphs shown in the Supporting Information are also presented in a different representation/compilation in the main manuscript.

I) Particle tracking of SNARF-loaded capsules inside cells I.1) Materials, reagents, and equipment I.2) Synthesis and characterization of SNARF-loaded polyelectrolyte micro-(PEM) capsules I.3)
Uptake studies by confocal laser scanning microscope (CLSM) I.4) Particle tracking by CLSM, image processing, and data evaluation I.5) Fractal dimension and average end-to-end scaling exponent I.6) Results
A confocal laser scanning microscope (CLSM 510 Meta) from Zeiss was used for visualizing and live imaging of HeLa cells engulfing SNARF-loaded capsules. For image acquisition the fluorescence was excited at 543 nm using the helium neon laser of the CLSM and samples were observed through a 63X/1.40 oil-immersion DIC M27 objective. ImmersolTM 518F immersion oil (Zeiss) was used during imaging.

I.2) Synthesis and characterization of SNARF-loaded polyelectrolyte micro-(PEM) capsules
PEM capsules with encapsulated SNARF dextran inside their cavities were fabricated by coprecipitation, followed by layer by layer (LBL) assembly of oppositely charged polymeric layers [1] . Briefly, pH-sensitive PEM capsules having 2 and 2.5 bilayers of non-biodegradable polymers composed of PSS and PAH were fabricated based on LBL assembly of oppositely charged polymers around sacrificial template cores containing the pH sensitive dye SNARF-1 conjugated with dextran. For co-precipitation of SNARF-1 dextran with CaCO 3 microparticles, solutions of CaCl 2 and Na 2 CO 3 were mixed under vigorous stirring in the presence of SNARF-1 dextran at room temperature (RT) in aqueous media. In a glass vial, 4.2 mL SNARF-1 dextran (0.5 mg/mL) was added to 3 mL of 0.33 M CaCl 2 (0.33 M). Under vigorous magnetic stirring (1100 rpm) 3 mL of Na 2 CO 3 (0.33 M) solution was quickly mixed with the above mixture for 30 s, followed by keeping the reaction contents without agitation for 2 min. Calcium carbonate particles were washed three times with DDW and used for LBL assembly of oppositely charged polyelectrolytes (2 mg/mL 0.5 M NaCl). The alternating layers of negatively and positively charged polymers, i.e., PSS and PAH respectively were deposited around the charged sacrificial microparticle templates following a well established protocol [2] . Layer deposition was achieved by alternating immersion of microparticles inside the corresponding polymer solution (5 mL) for 13 min, followed by subsequent washing with DDW to remove excess polymers. Finally the cores of PEM capsules were dissolved by complexion of Ca 2+ ions with EDTA (5 mL, 0.2 M, pH 6.5) and particles were washed with DDW and stored at 4 °C for further use. In order to minimize the artifacts due to size variation of capsules (mean diameter ≈ 3.5-4 µm), the cores were manufactured on large scale, dried under vacuum after washing with acetone, and stored at 4 °C. Latter in all experiments the same cores were used for LBL deposition of polyelectrolyte shells.
SNARF is a ratiometric pH sensitive dye having pK a value of 7.5 [3] . After incorporation inside the cavity of capsules, the dye retained its pH sensitive fluorescence characteristics [1d] . Upon excitation at 543 nm it has two emission peaks at 580 nm (yellow fluorescence) and 640 nm (red fluorescence). When the ambient pH is low its emission maxima is at 580 nm, while at high pH the emission maximum is at 640 nm. This pH dependent shift in fluorescence intensity makes SNARF-bearing capsules ratiometric indicators for intracellular sensing without the need of additional reference fluorophores [1d, 4] . The pH dependent fluorescence response of the SNARF capsules as synthesized for the present work was monitored by adjusting the pH of the capsules by means of immersion in commercially available buffers (pH 3 -10), cf. Figure SI-I.2.1 [1d, 5] . The intensity of red (640 nm) and yellow (580 nm, displayed as "green" in false colors of the microscope)) fluorescence of SNARF was measured using confocal laser scanning microscopy (CLSM) and the ratio of the respective fluorescence intensities was plotted versus pH.  and zeta potential ζ ± Δζ as measured in water given as average and standard deviation. These data were derived from Figure SI-I.2.2.

I.3) Uptake studies by confocal laser scanning microscope (CLSM)
HeLa cells were grown in DMEM supplemented with 10% FBS, 1% P/S, and 1% Glutamax. 15,000 cells were seeded per well in 8-well µ-slides having growth area of 1 cm 2 per well. Each well was filled with 300 µL of growth medium. Cells were kept inside an incubator set to 37 °C and complemented with 5% CO 2 at constant rate. Capsule tracking and uptake experiments were performed in serum supplemented or in serum deprived media, in the presence or absence of inhibitors using CLSM. In order to track the internalization of PEM sensor capsules, the CLSM was equipped with a portable incubator in order to maintain the µ-slides at 37 °C with 5% CO 2 . Before starting the experiment the cells were provided fresh serum supplemented/ serum deprived media. In addition, the inhibitors (cytochalasin D, 300 nM & bafilomycin A1, 0.25 µM) were added 1 hour before the addition of PEMs and starting the experiments. PEMs were provided to the cells at a concentration of 5 capsules per seeded cell just at the time of start of the imaging using a Plan-Apochromat 63x/1.40 Oil DIC M27 objective. During imaging, time lapse image series were captured [1f] . In order to minimize photobleaching of SNARF during image acquisition, very low laser excitation power was used in order to detect fluorescence for long term studies. The fluorophore was excited at λ ex = 543 nm and its fluorescence emission in the yellow and in the red region was detected using a 560 -615 nm band pass filter and a 633 nm long pass filter, respectively. During imaging the lateral resolution (i.e. in the x-y-plane) was adjusted to 0.32 µm, whereas, 120 s temporal resolution was used. In order to acquire two images from the same lateral position, the zposition (i.e. the height of the focus with respect to the substrate) of the maximum scattered photons at the boundary of substrate/medium (µ-dish plate/cell medium) was detected, which helped in determining the absolute axial position of this boundary layer [1f] . Two imaging slices were acquired 3 and 2.2 µm above the substrate in order to resolve the extracellular (attached with the cells) and intracellular (internalized by the cells) capsules, respectively. For imaging of both slices the pinhole was adjusted to capture 2 µm thick image sections. A software based autofocus routine was always set to recover the imaging alignment before each time point of measurement [1f] . Image dimensions were further reduced after determining the projection along the z-axis of each slice at the end of image acquisition.

I.4) Particle tracking by CLSM, image processing, and data evaluation
The experimental protocol and description follows a previous study by Hartmann et al. [1f] . Large CLSM images were cut in small segments covering the whole internalization path of given capsules. A Matlab (Mathworks, USA) based particle analyzing toolbox written by Raimo Hartmann [1f] was used for visualization, identification, particle (capsules) segmentation, tracking, and automated image processing, as described previously [1f] . In this process, by means of median filtering photon shot noise was decreased in the red and yellow (green in false colors) channels of CLSM images and the background was entirely eliminated after defining a threshold. A modified Hough transformation was computed by using the method of Gonzalez, et al., for automatic identification, and masking the PEM capsules from CLSM images. This mask was later used for defining regions of interest (ROIs) around the regions of maximum fluorescence intensity [6] . The dimensions of PEM capsules resembled circles, that is why the centers of all circles were identified by checking the positions of individual pixels whether they match the origin of a circle of given radius. From the classical Hough transformation the radius of the circles cannot be determined. That is why the parameterization of a donut shaped structure (shell thickness ≈ 0.64 µm) was used in order to identify the positions and dimensions of individual capsules. The radius of this structure was varied for each pixel and the sum of the underlying pixel intensities was recorded as a function of this radius. The capsule radius was determined from the first maximum of this function. Individual capsules were trace as described in Hartmann et al [1f] . A mask for each capsule's ROIs (in the red and yellow fluorescence channel) was created by using their respective x-y coordinates and radii. The average red/yellow ratio designated as I r /I y corresponding to the ambient pH values was calculated from the noise-corrected raw data within the masked regions. Taking inspiration by the particle tracking algorithms developed by John Crocker et al [7] , which were provided by Danial Blair and Eric Dufresne for Matlab, a capsule tracking function was applied in order to get the progression of I r /I y versus time for individual capsules during their uptake and cellular trafficking (i.e., so called trajectories). A brief description of the workflow for the whole image processing is described in a previous study [1f] .
Some representative examples of trajectories of SNARF capsules having different surface charge and bilayers in serum-supplemented and serum-deprived media and in the presence and absence of inhibitors are provided in Figure  The trajectories of individual internalized capsules were used to determine their acidification (∆t A ) and processing (∆t P ; e.g., ∆t P10% and ∆t P50% ) times upon internalization [1f] . The acidification time ∆t A was determined from the sigmoidal curve as the time interval describing the duration of acidification, i.e. the duration of the transition from high to low pH. The processing time ∆t P50% is defined as the time interval from once a capsule attaches to the cell membrane until it becomes located in acidic endosomes/lysosomes [1f] . ∆t P10% is the time interval from the first contact of the PEM capsule with a cell until 10% of the final drop in pH have happened. The time after addition of the capsules to the cells until the first contact of the observed capsule with a cell is defined as time of first contact t C . Examples are shown in Figure From each fit 4 fit parameters were obtained: R a = (I r /I y ) max , the I r /I y ratio when the capsules are still in the cell medium, i.e. at neutral/slightly alkaline pH R b = (I r /I y ) min , the I r /I y ratio when the capsules are fully internalized ,i.e. at maximum acidic pH in endo/lysosomal compartments t 50% , the time at which upon internalization the pH has dropped, so that 50% of the acidification has been achieved ∆T, describing the length of the acidification period in which the pH drops From these fit parameters the following set of parameters is obtained. t 10% and t 90% are the times when upon internalization 10% and 90% of the pH drop have been achieved: In an analogous way one can calculate This leads to the definition of the acidification time ∆t A , which is the time interval in which the pH drop goes from 10% to 90%: ∆t A = t 90% -t 10% = ln(0.9)⋅∆T -ln(1/9)⋅∆T = 2⋅ln(9)⋅∆T (Equation SI-I. 4.

2)
Based on this also the processing time ∆t P10% and ∆t P50% are calculated: t 10% = ln(1/9)⋅∆T + t 50% t 90% = ln(9)⋅∆T + t 50% ⇒ ∆t P10% = t 10% -t C (Equation SI-I. 4.3) ∆t P50% = t 50% -t C These are the time intervals needed from the first contact of a capsule with a cell unto 10% and 90% of the pH drop in the locale capsule environment due to internalization has been achieved. t 50% in contrast refers to the start of incubation t 50% = t C + ∆t P50%. It refers to the time needed from the start of incubation until 50% acidification has been achieved. In addition, the maximum of the absolute slope of the pH response was determined as (|d(I r /I y )/dt|) max . Thus, from the fit of each position/pH trace the following 5 parameters were obtained: t C , t A , t P10% , t P50% , (|d(I r /I y )/dt|) max . Some examples are given in the following.      Note, that the bafilomycin A1 is lysomotropic reagent which alkalinizes the intra-lysosomal pH and inhibits the internalization of PEM capsules [5,8] . The experimental concentration of the reagent was selected that it (partly) inhibited capsule internalization, and the intra-lysosomal pH remained less than pH of the extracellular medium. That is the reason why the internalized capsules turn orange instead of yellow (i.e. less acidic environment than without the presence of bafilomycin A1). Moreover, it was hard to identify the point of inflection in the I r /I y traces, i.e. the parameter t 50% . Thus, only a limited number of experiments were performed with this reagent and the parameters (e.g. acidification, processing time, etc.) are not provided in results.

I.5) Fractal dimension and the average end-to-end scaling exponent
Similarly other parameters, such as fractal dimensions D and the average end-to-end scaling exponent ν, can be determined from this data set. These data are based on the particle trajectories in the CLSM images. The position of each capsule at time t is given by its x-and ycoordinates x(t) and y(t). At images were taken in time intervals of 2 min, discrete coordinates x i = x(t i ) and y i = y(t i ) were obtained with t i+1 -t i = 2 min.
To characterize the trajectories quantitatively one can make use of the fractal dimension D, which is a measure of self-similarity and, thus, remains unchanged when the scale of measurement is changed [9] . D is also a measure of spatial extent, i.e. the space filling properties, and self-affinity [9] . The correct determination of the fractal dimension of a trajectory is a nontrivial problem, several approaches exist in the literature [9][10] . Adopting the method by Sevcik et al. [9] the fractal dimensions D of the full capsule trajectories were approximated by the fractal dimensions D M of each capsule trajectory consisting of M sample points. In doing so, the trajectory (x,y)(t) in the two-dimensional plane of the M sample points (each at time t i , i = 1,…, M) was mapped into a unit square Where, x/y max/min is the maximum/minimum of x/y i . Then, the fractal dimension D of the trajectory was approximated by Where, L is the contour length of the trajectory in the unit square: The average end-to-end distance〈R(L)〉is a strong characteristic of the spatial structure of polymers [11] . Given a polymer of contour length L, the average end-to-end distance <R(L)> scales as The exponent ν depends on the dimension of the system, taking values from ν = 1 to ν = 0. Similar to this scaling behavior of polymers one can use the average end-to-end distance scaling exponent to characterize the trajectory of a capsule. The scaling exponent ν is then given by , where, x M and y M are the last sample points of the trajectories of each capsule along the x-and y-dimension. Note, that for the scaling exponent the trajectories were not mapped into a unit square, but the absolute values were used.

I.6) Results
The results from uptake data of capsules of almost similar sizes in terms of various surface charge, presence and absence of serum, and cytochalasin D inhibitor by determining various parameters from the trajectories of internalized particles are presented in the following. Each data set corresponds to at least 100 different trajectories that were evaluated, the exact number is given in Table SI-I.6.1. Two independent complete set of experiments with two batches of SNARF-loaded capsules were performed in order to validate the effect of cytochalasin D (300 nM, i.e. a potent inhibitor of particle internalization by disrupting actin polymerization [8] ) on the uptake and intracellular processing of these capsules. Results are listed individually for both batches. The extracted parameters t C , ∆t A , ∆t P50% , ∆t P10% , t P50% , and (|d(I r /I y )/dt|) max are enlisted in Table SI Table SI-I.6.1. Number of traces n that were evaluated for SNARF-loaded capsules added to HeLa cells with positive ("+", 2 bilayers) and negative ("-", 2.5 bilayers) charge, with serumsupplemented ("w") and serum deprived ("w/o") conditions, with ("w": two independent experiments "w 1 " and "w 2 ") and without ("w/o") the presence of 300 nM cytochalasin D, as determined from the trajectories of capsules while cellular uptake.  Contact time t C of SNARF-loaded capsules added to HeLa cells with positive ("+", 2 bilayers) and negative ("-", 2.5 bilayers) charge, with serum-supplemented ("w") and serum deprived ("w/o") conditions, with ("w": two independent experiments "w 1 " and "w 2 ") and without ("w/o") the presence of 300 nM cytochalasin D, as determined from the trajectories of capsules while cellular uptake. The data for the contact time is provided as median values t C , plus minus the confidence intervals ∆t C .

capsule charge serum cytochalasin D ∆t A [min] ∆∆t A [min]
Table SI-I.6.3. Acidification time ∆t A of SNARF-loaded capsules added to HeLa cells with positive ("+", 2 bilayers) and negative ("-", 2.5 bilayers) charge, with serum-supplemented ("w") and serum deprived ("w/o") conditions, with ("w": two independent experiments "w 1 " and "w 2 ") and without ("w/o") the presence of 300 nM cytochalasin D, as determined from the trajectories of capsules while cellular uptake. The data for the acidification time is provided as median values ∆t A , plus minus the confidence intervals ∆∆t A .  Processing time ∆t P50% of SNARF-loaded capsules added to HeLa cells with positive ("+", 2 bilayers) and negative ("-", 2.5 bilayers) charge, with serum-supplemented ("w") and serum deprived ("w/o") conditions, with ("w": two independent experiments "w 1 " and "w 2 ") and without ("w/o") the presence of 300 nM cytochalasin D, as determined from the trajectories of capsules while cellular uptake. The data for the 50% processing time is provided as median values ∆t P50% , plus minus the confidence intervals ∆∆t P50% . Table SI-I.6.5. Processing time ∆t P10% of SNARF-loaded capsules added to HeLa cells with positive ("+", 2 bilayers) and negative ("-", 2.5 bilayers) charge, with serum-supplemented ("w") and serum deprived ("w/o") conditions, with ("w": two independent experiments "w 1 " and "w 2 ") and without ("w/o") the presence of 300 nM cytochalasin D, as determined from the trajectories of capsules while cellular uptake. The data for the 10% processing time is provided as median values ∆t P10% , plus minus the confidence intervals ∆∆t P10% .  Table SI-I.6.6. Time t 50% until 50% acidification of SNARF-loaded capsules added to HeLa cells with positive ("+", 2 bilayers) and negative ("-", 2.5 bilayers) charge, with serum-supplemented ("w") and serum deprived ("w/o") conditions, with ("w": two independent experiments "w 1 " and "w 2 ") and without ("w/o") the presence of 300 nM cytochalasin D. The data is provided as median values t 50% , plus minus the confidence intervals ∆t 50% .  6.7. Maximum of the absolute first derivative of the I r /I y versus t traces of SNARFloaded capsules added to HeLa cells with positive ("+", 2 bilayers) and negative ("-", 2.5 bilayers) charge, with serum-supplemented ("w") and serum deprived ("w/o") conditions, with ("w": two independent experiments "w 1 " and "w 2 ") and without ("w/o") the presence of 300 nM cytochalasin D, as determined from the trajectories of capsules while cellular uptake. The data is provided as median values (|d(I r /I y )/dt|) max , plus minus the confidence intervals, i.e., ∆(|d(I r /I y )/dt|) max .  The fractal dimensions D and end-to-end distance scaling exponents ν were calculated for each trajectory for three different regions of the trajectory: before acidification starts (i.e. capsules are still outside cells "out"), during acidification ("uptake"), and after acidification (i.e. when the capsules are fully internalized "in"). Values are provided in Table SI   HeLa cells with positive ("+", 2 bilayers) and negative ("-", 2.5 bilayers) charge, with serumsupplemented ("w") and serum deprived ("w/o") conditions, with ("w": two independent experiments "w 1 " and "w 2 ") and without ("w/o") the presence of 300 nM cytochalasin D. Trajectories were subdivided in the parts before acidification (D out ), during acidification (D uptake ) and after acidification (D in  Average end-to-end distance scaling exponents ν of the trajectories of SNARFloaded capsules added to HeLa cells with positive ("+", 2 bilayers) and negative ("-", 2.5 bilayers) charge, with serum-supplemented ("w") and serum deprived ("w/o") conditions, with ("w": two independent experiments "w 1 " and "w 2 ") and without ("w/o") the presence of 300 nM cytochalasin D. Trajectories were subdivided in the parts before acidification (ν out ), during acidification (ν uptake ) and after acidification (ν in ). Results are presented as mean values ± standard deviations.  In the following possible differences in D and ν before, during, and after acidification are discussed, see ν decreasing from ν out towards ν uptake to ν in (large error in ν uptake ) -w/o w 2 D increasing from D out towards D uptake to D in (maybe not significant) ν decreasing from ν out towards ν uptake to ν in However, the scaling exponent ν differs between ~0.67 and 0.45, which is more significant, because it corresponds to a difference of almost 50%. Thus, if one takes into account the correlation between the fractal dimension D and the scaling exponent ν (in general one would expect that when D → 1 (straight line) then also v → 1 and vice versa, which is in general true for the experimental data), then one might be able to extract some information about the different phases and between the different experiments using the data analysis. In that case, one might be able to make the statement that in some cases the existence of the different phases can be predicted from D and ν. It is however open to discussion whether differences are statistically significant. Considering the fractal dimension without the scaling exponent seems to result in rather non significant statements. However, taking both parameters into account may provide a chance to distinguish between the different phases before, during, and after acidification.

II.2) Synthesis and characterization of SNARF-loaded polyelectrolyte micro-(PEM) capsules
The protocol for capsule fabrication is same as described in section §I.2. In addition to SNARFmodified dextran, TRITC-modified dextran was also used as fluorophore, due to the much lower price as compared to SNARF. TRITC-loaded capsules were prepared following the same methodology, by using TRITC-dextran instead of SNARF-dextran at the same concentration.
The hydrodynamic diameter and zeta potential measurement results of TRITC-loaded PEM capsules in water are provided in Figure    and zeta potential ζ ± Δζ as measured in water given as average and standard deviation. These data were derived from Figure  For SNARF-loaded capsules the uptake study was carried out at various time points. HeLa cells were seeded into 8 well µ-ibidi plates (surface area 1 cm 2 /well) in 0.3 mL of complete growth medium at a density of 20,000 cells per well. After 18 h the growth media was exchanged to fresh growth media. Capsules were then added to cells. Uptake experiments were performed with two types of growth media, either supplemented with 10% FBS or without serum.
Positively and negatively charged capsules were added with two different concentrations, (i) at 10 capsules added per seeded cell, and (ii) at 20 capsules added per seeded cell. Cells were incubated inside an incubator at 37 • C with 5% CO 2 supply. Fluorescence microscopic images were recorded after 2, 4, 6, 12, 24, 30, 36, 42, and 48 hours of incubation with the help of an Axiovert fluorescence microscope using a 63x oil immersion objective. The fluorescence of SNARF was excited at 540 nm and its emission was recorded at 580 nm and 640 nm to detect the yellow and red fluorescence, respectively. Per sample and time point 10 images were captured, covering an average of 80 -100 cells per condition. Experiments were performed in duplicate. A representative fluorescence microscope image of HeLa cells after 2 h incubation with SNARF-loaded capsules is shown in Figure SI-II.2.2a. Internalized capsules could be identified by their yellow fluorescence, whereas capsules remaining outside cells were fluorescent in red [12] .
In order to evaluate the internalization of TRITC-loaded capsules, HeLa cells were seeded on small cover slips placed inside 24-well plates (surface area 1.82 cm 2 /well) in 0.6 mL complete growth medium at a density of 20,000 cells per well. After 22 h when the cells were firmly attached to the cover slips the growth media was aspirated and fresh growth media containing TRITC-loaded capsules was added. Uptake experiments were performed with two types of growth media, either supplemented with 10% FBS or without serum. Cells were incubated with positively and negatively charged capsules at a concentration of 10 capsules added per seeded cell and were incubated for 24 h inside an incubator at 37 • C with 5% CO 2 supply. In a series of experiments Bafilomycin A1 (0.25 µM) was added in order to investigate its effect as inhibitor on capsule uptake [5] . Bafilomycin A1 was added to cells with fresh growth media 1 h before the addition of capsules. The capsules were subsequently immunostained (cellular cytoskeleton, nuclei, and lysosomes), so that the intracellular localization of TRITC capsules could be detected by means of CLSM images [13] . In this way internalized capsules surrounded by lysosomal membrane, could be distinguished from capsules remaining in the extracellular medium [13] . For the immunostaining procedure, the cells were washed with PBS and fixed with 4% paraformaldehyde solution in PBS (20 min incubation at room temperature), followed by washing with PBS. The cells were washed thrice with Hanks' balanced salt solution (HBSS; PBS would have served the same purpose), and permeabilized (addition of glycin 5 mg/mL and saponin 0.5 mg/mL in PBS, in the following referred to as permeabilization solution, 5 min incubation). Then the cells were incubated at 37 • C in an incubator under 5% CO 2 supply with blocking solution (20 mg/mL BSA in permeabilization solution) for 30 min. Next, the lysosomes were immunostained by LAMP 1 (primary antibody; 5 µg/mL blocking sloution, 1 h incubation at 37 • C followed by 3 times washing with blocking solution) and Dylight 649 (secondary antibody; 1.25 µg/mL PBS, 1 h incubation at 37 • C followed by 3 times washing with PBS). In order to save time the staining agents phalloidin labelled with oregon green 488 (20 µg/mL) and Hoechst reagent (for staining of the cytoskeleton and nuclear membranes, respectively) were added within the secondary antibody solution in PBS to cells. The immunostained cells were washed with PBS, followed by water, dried, and fixed on glass slides by means of fluoromount G. The samples were kept in dark at room temperature for 48 h before analyzing by CLSM. CLSM images were captured using a 63x/1.40 oil-immersion DIC M27 lens. For visualization of different stained cellular compartments and internalized capsules, the fixed samples were excited at 405, 488, 543 and 633 nm, respectively. Hoechst 33342 stained nuclei were excited at 405 nm, and emission of the dye was observed between 420 and 480 nm. The cytoskeleton stained with phalloidin, oregon green 488 was visualized by exciting the fluorophore at 488 nm, and emission was observed between 505 and 550 nm. The fluorescence of TRITC was excited at 543 nm, and emission was recorded using a 560 nm long pass filter. The antibody-labeled lysosomes were excited at 633 nm, and their emission was recorded using a 650 nm long pass filter. Per sample 25 images were captured, covering an average of 100 cells per condition. Experiments were performed in triplicate. A representative CLSM image of TRITC capsules after 24 h incubation with HeLa cells (after immunostaining) is shown in Figure SI-II.2.2b.

Figure SI-II.2.2:
Representative a) fluorescence microscopy and b) CLSM images of HeLa cells exposed to SNARF-loaded and TRITC-loaded capsules, respectively. The image in a) was captured after 2 h of incubation with SNARF-loaded negatively charged capsules at 10 capsules added per seeded cell in serum supplemented media. From the color change of the SNARFloaded capsules (red to yellow (shown in green false-colors)), the extracellular capsules can be discriminated from the intracellular capsules. The scale bar corresponds to 10 µm. In b) the HeLa cells were incubated with TRITC-loaded capsules, negatively charged capsules at 10 capsules added per seeded cell in serum supplemented media, and the image was taken after 24 h incubation. From the orthogonal view the presence of TRITC-loaded capsules inside the stained lysosomes can be seen. The scale bar corresponds to 10 µm.

II.3) Uptake studies based on capsule counting
In order to evaluate the uptake of capsules inside HeLa cells, first the microscopy images were exported and their format was changed from ".ZVI" to ".JPG". The number of internalized capsules per cell N caps/cell was then manually counted for each time point of incubation for each condition [14] . In case of SNARF-loaded capsules the color change from red to yellow fluorescence was used as indicator of internalization [15] . The internalization of TRITC-loaded capsules verified in the orthogonal CLSM images of immunostained samples by the presence of capsules within stained lysosomal compartments [8, Parakhonskiy, 2015 #32632] . For each condition by counting, a histogram in which the number of cells f(N caps/cell ) which had internalized N caps/cell capsules was made. From this, cumulative probability / cumulative distribution function (CDFs) p(N caps/cell ) was calculated [8,13] , see Figure SI-II.3.1: ; 0 ≤ p(N caps/cell ) ≤ 1 (Equation SI-II.3.1) p(N caps/cell ) is the probability that a cell has internalized not more than N caps/cell capsules per cell (i.e. less than N caps/cell + 1 capsule). 1 -p(N caps/cell ) is the probability that a cell has internalized more than N caps/cell capsules per cell (i.e. at least N caps/cell + 1 capsule per cell).

From the histograms the mean number of internalized capsules per cell <N caps/cell > (h) (t) at each time point t can be calculated by summing up the intensities of all capsules.
The normalized fluorescence intensity I inside one cell due to the fluorescence of capsules is proportional to <N caps/cell > (h) (t).
Alternatively, from the CDFs the mean number of capsules <N caps/cell > (p) (t) that were internalized with 50% probability (i.e. p(<N caps/cell > (p) (t)) = 0.5) after incubation time t was derived. The way to calculate this number was explained in Figure SI-III.4.1 from Zyuzin et al [16] .
For flow cytometry a BD LSRFortessa™ cell analyzer flow cytometer (Becton, Dickinson and Company, USA) was used. The fluorescence of the SNARF was excited at 561 nm and its emission was captured using 586 nm/15 nm and 670 nm /30 nm band pass filters for detecting the intensity of yellow and red fluorescence, respectively. Data was analyzed with FlowJo, single cell analysis software (Ashland, OR, USA).

III.2) Synthesis and calibration of SNARF-loaded polyelectrolyte micro-(PEM) capsules
The method for the fabrication of SNARF-loaded capsules was the same as described in section §I.2.
Calibration curves of capsules immersed in different pH were recorded with flow cytometry, in order to distinguish between populations of capsules surrounded by medium with different pH [15] . For this positively and negatively charged SNARF-loaded capsules were mixed with buffers having a range of pH values from 3 -10. Solutions were analyzed with flow cytometry and 10,000 events per sample were recorded by exciting SNARF at 561 nm and recording its emissions at 586 nm and 670 nm. Results are plotted in different presentations. In Figure SI

III.3) Uptake studies by flow cytometry
For investigation capsule uptake by cells, HeLa cells were seeded in 24 well plates (growth area 1.82 cm 2 /well) at 40,000 cells seeded per well in 0.6 mL complete cell growth media. Cells were kept inside an incubator set to 37 °C with 5% CO 2 . After 24 h the cells were provided fresh growth media (either serum supplemented or serum deprived) containing positively or negatively charged SNARF-loaded capsules at a density of 10 or 20 capsules added per seeded cell. Cells which were not exposed to capsules served as control. The cells were incubated with the capsules for different times, i.e., 2, 4, 6, 12, 24, 30, 36, 42, and 48 h. After incubation cells were washed with PBS, trypsinized, and resuspended in PBS. For viability assessment, DAPI (2 µL, 1.09 mM) was added to each sample. The samples were then investigated with flow cytometry. The forward (FSC-A) and sideward (SSC-A) scattering signals were used to gate events involving cells, i.e. only events with sufficient scattering signal were further regarded. This gating, which in Figure SI-III.3.1 is referred to as G1, excludes event due to cellular debris and free capsules, which have a much lower scattering signal than cells (compare with Figure  SI-III.2.1a). By means of a second gating (G2), cell doublets were removed and excluded from the analysis. This was done in the forward scatter plot in which the area (FSC-A) was plotted against width (FSC-W). In order to detect a single cell population, 5000 events/sample were recorded in G2 and were used for further data processing. Using both gates G1 and G2, 2dimensional density plots of the red and yellow fluorescence signals I r and I y were created. In these plots populations of cells with adherent and internalized SNARF-loaded capsules can be distinguished, as internalized capsules are located in acidic environment [15] . Events without sufficient fluorescence were attributed to cells which had neither capsules adherent to their outer wall, nor internalized capsules. In this way from each plot three different cell populations were identified: cells without associated capsules, cells with adherent capsules, and cells with internalized capsules. From the density plots the fractions of the respective populations were derived according to the amount of detected respective events, see Figure   In order to collect the single cell population, 5000 events/sample were recorded in G2, which were used for further data processing. c) 2-dimensional density plots of red and yellow fluorescence signals enable to distinguish between the populations of cells with adherent (N cells w caps(adh) /N cells ) and internalized capsules (N cells w caps(in) /N cells ), by integrating the events above and below the separation line, respectively [15] .

III.4) Results
In Figure SI-III.4.1, the standard uptake curve in which fluorescence intensity per cell due to internalized capsules is plotted versus time is shown. This curve does not used any gating strategy. From this curve the mean intensity <I y > (sat,c) per cell under saturation and the mean time t up(sat,c) until the mean fluorescence intensity <I y > (c) per cell has reached 50% of the saturation value can be determined, see
A UV-vis absorption spectrometer (8453 UV-visible spectrophotometer) from Agilent was used for obtaining the absorption values of the protein content in the Lowry tests. An inductively coupled plasma mass spectrometer (ICP-MS) from Agilent 7700 Series was used to determine the concentrations of elemental gold and hence the GNP concentrations.

IV.2) Synthesis and characterization of gold nanoparticle (GNP) -loaded polyelectrolyte micro-(PEM) capsules
In order to fabricate GNP-loaded PEM capsules comparable to SNARF-loaded capsules, the basic strategy for the synthesis was same as that for SNARF-loaded capsules (cf. § I.2). Briefly, amino dextran was co-precipitated with CaCO 3 particles, by mixing the solutions of CaCl 2 and Na 2 CO 3 under vigorous stirring in the presence of amino dextran at room temperature (RT). In a glass vial, 1.4 mL amino dextran (0.5 mg/mL) was added to 1 mL of 0.33 M CaCl 2 (0.33 M). Under vigorous magnetic stirring (1100 rpm) 1 mL of Na 2 CO 3 (0.33 M) solution was quickly mixed with the above mixture for 30 s followed by keeping the reaction contents intact for 2 min. The CaCO 3 particles were washed three times with DDW and used for LBL assembly of oppositely charged polyelectrolytes (2 mg/mL, 0.5 M NaCl). The alternating layers of negatively and positively charged polymers, i.e., PSS and PAH, respectively, were deposited around the charged sacrificial CaCO 3 particle templates. Layer-by-layer deposition was achieved by alternating immersion of the particles inside the respective polyelectrolyte solutions (3 mL) for 13 min, followed by subsequent washing with DDW to remove excess of polymers. Negatively charged GNPs (2.7 mL of the solution as purchased mixed with 0.3 mL 0.5 M NaCl just before layer deposition) were incorporated inside the capsules after the first bilayer (PSS/PAH), followed by the deposition of additional 1 (PSS/PAH) and 1.5 (PSS/PAH/PSS) bilayers for positively and negatively charged capsules, respectively. Finally the cores of the PEM capsules were dissolved by complexion of Ca 2+ ions with EDTA (3 mL, 0.2 M, pH 6.5) and the resulting capsules were washed with DDW and stored in water at 4 °C for further use.
Their hydrodynamic diameter and zeta potentials were recorded and are shown in Figure    and zeta potential ζ ± Δζ as measured in water given as average and standard deviation. These data were derived from Figure SI-IV.2.1.

IV.3) Uptake studies by inductively coupled plasma mass spectroscopy (ICP-MS)
HeLa cells were seeded in 6 well plates (8.95 cm 2 surface area/well, 3 mL medium) at a density of 100,000 cells per well. After 24 h incubation fresh growth media, either serum supplemented or serum free, containing GNP-loaded capsules was provided to the cells. Both types of capsules (positively and negatively charged) were added in two different concentrations, at 10 or 20 capsules added per cell. HeLa cells were incubated with capsules for different times (2,4,6,12,24,30,36,42, and 48 h) inside an incubator set at 37 • C with 5% CO 2 supply. Afterwards, the cells were washed with PBS and detached from the bottom of the plates by means of 0.05% trypsin-EDTA. The cells were centrifuged with growth media and the pellets were resuspended in PBS. The pellets of cells were washed again with PBS and the maximum of the supernatant was removed (note that remaining trypsin in solution would interfere with determining the protein concentration of the samples as described later). Then, 1x lysis buffer in water (100 µL per sample) was added to each cell pellet and the samples were incubated at room temperature for 30 minutes in order to complete cell lysis. Samples were sonicated and stored at -20 • C for further analysis, which involved determination of the amount of cells (by measuring the protein content) and the amount of capsules (by measuring the amount of elemental gold) in the cell lysates, as described in the following.
In order to determine the number of capsules per cell, the number of cells and the number of capsules in the cell lysate has to be determined. Cells were quantified by detecting their protein content by means of a commercial protein determination kit (Lowry assay; TPO300-KT). Lowry reagent solution was prepared by dissolving 2 g of Lowry reagent powder in 40 mL of water. The labeling solution was prepared by transferring 18 mL of Folin and Ciocalteu's phenol reagent into an amber glass, followed by the addition of 90 mL of water (in order to achieve the working concentration of labeling reagent). First, a calibration curve was plotted, relating the amount of the detected proteins to the number of cells. HeLa cells were detached from cell culture flasks by means of 0.05% trypsin-EDTA, followed twice by washing with PBS. The cells were then dispersed in a small volume of PBS and their number in terms of cells per volume of solution was determined by counting thrice with a haemocytometer. Then 1x lysis buffer was added. After cell lysis serial dilutions of the cell lysates in lysis buffer was performed in order to achieve samples with subsequently smaller cell concentrations. Lysis buffer was used for blank measurements. In sample tubes, 5 µL of cell lysates were added into 100 µL of Lowry reagent solution, mixed well, followed by waiting for 20 minutes for completion of complex formation between proteins of the cell lysates and the Lowry reagent solution. Later, 50 µL of labeling reagent solution was added to the above mixture to develop blue color depending on protein content. After waiting for 30 minutes the absorption of samples was immediately measured by means of UV-visible absorption spectrometry. Spectra were recorded from 550 -800 nm and from the spectra the absorption (A 750 ) at 750 nm was determined, cf. Figure   The amount of capsules in the cell lysates was determined in terms of elemental Au (from the GNPs) as measured with ICP-MS. The cell lysates containing the internalized GNP-loaded capsules were first digested with aqua regia. For this first the stock solution for each sample was prepared by adding 50 µL of sample suspension into freshly prepared 150 µL aqua regia (1 HNO 3 : 3 HCl) inside 6 mL perfluoroalkoxy alkane tubes (PFA), followed by mixing for at least 8 h under constant agitation. By doing this the GNP-loaded capsules and remaining organic cell fragments were digested and broken down into small molecular / atomic components. In the second step, 4.6 mL HNO 3 solution (2%) as low matrix was introduced to each digested sample to prevent the aqua regia from digesting the ICP-MS machinery, as well as to provide an ion stable environment with constant background conditions for all samples. During measurements, 5 repetitions/ sample, and 100 sweeps were performed and a peak pattern of 3 peaks was used. The diluted samples were introduced to the ICP-MS set-up through an integrated autosampler coupled to a peltier cooling spray chamber where the samples were nebulized and taken up by the argon gas flow at a speed of ½ m/s. The concentration determination was performed using a calibration curve for Au consisting of 9 measurement points (0 -2500 µg/L) of freshly prepared Au concentrations derived from gold standard solutions from Agilent (1000 mg/L). Results were obtained as the mean of five measurements in parts per billion or µg/L (ppb = µg/L). First, calibration was performed in which the Au content in the samples of GNP-loaded capsules of given number (as determined with a haemocytometer) was determined. From this the mass of elemental gold per capsule was determined. Then the amount of gold in the cell lysates of cells with incorporated GNP-loaded capsules was determined by ICP-MS. Using the calibration curve the number of capsules N caps in the lysate was determined. Finally, the number of internalized capsules per cell was determined as N caps/cell = N caps /N cells .

IV.4) Results
The data obtained for the determined number of capsules per cell N caps/cell is summarized in Figure SI- Figure SI-IV.4.1 it can be seen that the internalization of positively charged GNP-loaded capsules was higher than for their negatively charged counterparts. Uptake in serum deprived culture as compared to serum supplemented cell culture was also higher, which is in agreement with some previous findings [8,17] . In addition, the uptake was dose dependent, i.e., more capsules were found internalized when cells were incubated with 20 capsules per cell as compared to the addition of 10 capsules per cell, which is also in agreement with previous work [16] . Note that for interpretation of the results it is important to realize that the number of capsules added per cell refers to the number of cells seeded. However, the number of internalized capsules per cell, refers to the number of cells which have been found with the Lowry assay after incubation with capsules. After approximately 24 h HeLa cells start to proliferate, which complicates the quantitative analysis [18] . Cell proliferation results in increase in cell number after capsule uptake as compared to the number of seeded cells.

V.1) Materials and reagents
96 well assay plates (clear bottom, # 3603) were purchased from Corning. Resazurin solution (alamar blue; #TOX-8) was purchased from Sigma-Aldrich. All other reagents, chemicals, and consumables were the same as described in sections §I.1 and §II.1.
Fluorescence measurements were performed with a Fluorolog®-3 spectrofluorometer from HORIBA JOBIN YVON. For plate reading a Micromax 384 microwell-plate reader compatible with the Fluorolog® was used.

V.2) Viability measurements
Resazurin based cytotoxicity assays were performed to determine the impact of GNP-loaded PEM capsules and cytochalasin D (which was used in the studies presented in section §I) exposure on cell viability. The test is based on mitochondrial activity of living cells [19] . Active mitochondria of living cells cause bioreduction of the dye, i.e. they convert the nonfluorescent blue dye (resazurin) into its reduced form resorufin which fluoresces in red. In order to perform the studies, HeLa cells were seeded in 96 well transparent bottom plates (7500 cells/well, 0.32 cm 2 area/well) in 100 µL of complete growth media (DMEM supplemented with 10% FBS, 1% glutaMAX TM and 1% P/S) and were incubated for 24 h at 37 • C with a constant supply of CO 2 (5%). After 24 h, the old growth media was replaced by fresh growth media containing capsules or cytochalasin D at different concentrations, i.e. N caps/cell (added), and c(Cytochalasin D). Capsules were added at N caps/cell (added) 10 or 20 capsules per cell for each condition (positive and negative capsules in serum supplemented and serum deprived cell culture media). In case of cytochalasin D serial dilution was performed to examine its toxic effect for a broad range of concentrations (2400 -4.5 × 10 -06 µM). Experiments for each dose (for capsules and cytochalasin D) were performed in triplicate. As negative control in a few wells of the assay plates, fresh growth media was added to the cells without capsules or cytochalasin D. In case of capsules fresh serum supplemented and serum deprived media was provided to the cells as negative control for each sample/condition. Whereas, for cytochalasin D as negative control fresh serum supplemented cell growth media was provided to the cells in the absence of cytochalalsin D. Cells were incubated with capsules for different time points, i.e., 2, 4, 6, 12, 24, 30, 36, 42, 48 h, while 24 h the incubation time with cytochalasin D was used. After incubation for defined time points, the cells were washed with PBS and 10% resazurin solution in growth media was added (100 µL/well) to the cells. In some wells of the assay plate only resazurin solution (10%) was added (without cells) which served as blank and the assay plates were incubated for 4 h under the same conditions as described above. Then, the fluorescence spectra of each well of the assay plates were recorded by means of a microplate reader attached to a spectrofluorometer. Spectra were recorded using an excitation wavelength of 560 nm acquiring the emission spectra from 572 to 650 nm [20] . Background emission of the blank sample containing only resazurin solution was subtracted from all spectra. For getting the viability V of cells the mean of fluorescence value of each sample was normalized with respect to fluorescence of the control samples in which cells were not exposed to capsules or cytochalasin D [20] . Experiments were performed in triplicate and values are expressed as means of 3 independent experiments ± standard deviations.

V.3) Results
The viability (V) of HeLa cells upon exposure to cytochalasin D and PEM capsules with incorporated GNPs is presented in Figure SI