Semisynthetic Ferritin Nanocages for Flexible, Site-Specific Targeting and Ligand-Free Activation of Membrane Receptors

Abstract

Homopolymerization and cluster formation of cellular membrane receptors (MRs) are increasingly recognized as an essential facet of cell signaling and modulator of physiological responses. Yet, there is a lack of tools that can mediate precise stimulation to better understand the mechanisms and effects of clustering. Here, we designed fluorescent semisynthetic nanoparticles (NPs) based on the iron-storage protein ferritin and Staphylococcus aureus protein A to specifically target and activate distinct MRs, without causing side-effects. The NP exhibits high monodispersity and is readily equipped with a variety of antibodies with a K _D value below 5 nM. Specificity of the NP antigen recognition was evaluated for cells expressing transferrin receptor 1 (TfR1) or the death receptor CD95, both of which displayed NP-mediated cluster formation. Finally, our engineered NP acts as a natural ligand for TfR1 and induces apoptosis signaling solely by CD95 cluster formation in a ligand-independent manner.

Living cells constantly sense their environment and process signals effectively via membrane receptors (MRs), which, when activated, trigger intricate intracellular signaling networks and ultimately a physiological response. MRs are typically activated via the specific binding of extracellular ligands, whereafter the MRs form pleomorphic assemblies that evolve from freely diffusing monomers to small receptor oligomers, or further toward higher-order signaling clusters. Hitherto, the characteristics of these MR assemblies are little understood, despite their important role for the outcome of different physiological processes. − For example, the clustering of MRs is crucial for segregating proteins from undesired interaction partners and enhances the binding avidity between components at the plasma membrane. Moreover, studies of ligand topology, mobility, and density show that cluster formation alters the MR assembly and modulates the cellular response. More recently, the ‘receptor–ligand engagement’ paradigm as a mandatory prerequisite for signal initiation has been challenged, introducing ligand-free receptor activation. − Thus, further investigation of such mechanisms is important to understand the regulation of signaling pathways − and any malfunction in signaling activation that is related to various diseases. −

Powerful tools to uncover effects of MR cluster formation and ligand dependencies comprise DNA signaling platforms, photoactivatable reagents, , magnetogenetics, − and optogenetics, − as they allow locally controlling protein activity and learning about the signal initiation mechanism. While these approaches are unique in their active and dynamic modulation capability, they often require genetic modification or the application of sophisticated stimuli.

Here, we designed a versatile nanoparticle (NP) platform based on monodisperse semisynthetic ferritin nanocages. They can be altered in a controlled manner and enable the site-specific targeting and clustering of MRs. Iron-storage proteins of the ferritin family are popular templates for bio- and nanotechnological applications, as they feature spherical, hollow cages which can be used as cargo carriers. − Further, they consist of protein subunits that can be equipped with different tags to create multifunctional NPs. Human ferritin has a diameter of 12 nm, with an inner cavity of 8 nm, and is the most commonly used member of the ferritin family, due to low immunogenicity and biocompatibility. The protein cage consists of 24 subunits of an arbitrary combination of either liver/light chain ferritin (LCF) or heart/heavy chain ferritin (HCF). − The main difference between the two subunits is HCF’s ferroxidase domain, which oxidizes toxic ferrous iron to insoluble ferric iron. ,

In this study, we fused the HCF subunit to protein A (SpA) from Staphylococcus aureus, as it is capable of binding the most common antibodies (Abs), immunoglobulin G (IgGs). IgGs consist of two light and two heavy chains that form the characteristic Y-shape with a molecular mass of ∼150 kDa (see Figure A). , The protein is divided into the fragment antigen binding (Fab), where two target binding sites are located, and the fragment crystallizable (Fc), the protein’s backbone.

1.

(A) Structure of an IgG antibody with two light (light purple) and two heavy (dark purple) chains, fragment antigen binding (Fab), and fragment crystallizable (Fc) region. (B) Basic principle of protA-Ft activation via antibody coupling. Activated protA-Ft is able to specifically couple to the antibody’s target. (C) Scheme of the synthesis of ferritin nanoagents. Two constructs were genetically engineered: 6xHis-mEGFP-HCF and protA-HCF (see Figures S1 and S2). Plasmid DNA was expressed in E. coli and subsequently purified by affinity chromatography to obtain both forms, GFP-Ft and the hybrid protA-Ft (both diameters ∼ 16 nm).

SpA is a 42 kDa protein that specifically targets the Fc region of IgGs through one of its five antibody binding domains (Figure B). − One of these domains was modified to give rise to the 58 amino acid Z-domain (hereafter named protA), which couples to various IgGs with a K _D value of 20 nM. Coupling IgG to an NP via their Fc domain is crucial to retain their antigen-binding capabilities, since this leaves the Fab functional and free of steric hindrance. Hence, fusing protA to ferritin’s subunits enables quick and oriented decoration of the ferritin NP with IgG. ,

In order to fluorescently detect our NPs and to enable affinity purification, we further genetically fused monomeric enhanced green fluorescent protein (mEGFP) with a 6xHisTag to the HCF subunit. − Hence, two ferritin-based plasmids (Figure C) with the following inserts were created: (1) HCF coupled to mEGFP with 6xHisTag, and (2) HCF coupled to protA (see Figures S1 and S2). The nature of ferritin’s self-assembling properties then allowed to reproduce a well-established ferritin cage decorated with mEGFP (hereafter termed GFP-Ft) − and produce a novel hybrid cage combining both constructs, resulting in an mEGFP-protA-ferritin cage (termed protA-Ft, Figure C). In both cases, the ferritin cage consisted only of HCF subunits, which was previously shown to be beneficial, in case the cavity should be further exploited for the storage of iron compounds. − The hybrid protA-Ft is easily equipped with a matching antibody to direct it to any specific target on or inside a cell and thereby easily trackable via its fluorescent subunits (Figure C). GFP-Ft not only serves as a control for protA-Ft but also can shuttle substances into cells expressing transferrin receptor 1 (TfR1), as outlined in the following.

Tools that specifically target selected MRs to trigger cell signaling or to shuttle cargo are of high biological and medical interest. For this reason, we here chose two important MRs of cancer research, for which mechanisms of molecular clustering have been proposed: on one hand, cluster of differentiation 95 (CD95) is a death receptor of the tumor necrosis receptor family (TNFR) and initiates apoptosis, programmed cell death. , While excessive apoptosis has been linked to diseases such as Alzheimer’s and Parkinson’s, insufficient apoptosis can lead to excessive cell growth and, thus, cancer. , Prior to the initiation of apoptosis, some CD95 may form small oligomers and lipid rafts containing CD95 that were shown to promote receptor clustering. Upon binding its ligand CD95L, CD95 oligomerizes at least to dimers and trimers at characteristic distances to trigger apoptosis. ,,, Intriguingly, it was also suggested that the preligand assembly domain of TNFRs, which promotes their dimerization, can also initiate signaling in the absence of the ligand.

On the other hand, TfR1 was shown to be overexpressed in cancer cells, making it a suitable marker and target during cancer treatment. Naturally, TfR1 is responsible for cellular iron uptake and iron homeostasis by internalizing the iron transporter transferrin (K _D transferrin-TfR1: 1.1 nM), , but has also been shown to be connected to tumor resistance against DNA damage. By upregulation of iron import, cancer cells meet the increasing demand for gene regulation, metabolic catalysis, or bioenergetics. Interestingly, ferritin was suggested to be a natural ligand to TfR1, as HCF-mediated interaction with TfR1 was shown (K _D HCF-TfR1: 7.1 nM). , Previously reported K _D values of TfR1 ligands suggest that ferritin would be able to occupy a fraction of TfR1 (depending on concentrations of all involved entities). Hence, ferritin can enter the cell even in the presence of transferrin, an interesting aspect for in vivo studies. To further investigate the molecular interaction during signal initiation, we here probe the potential of CD95 and TfR1 targeting via ferritin.

The two ferritin constructs GFP-Ft and protA-Ft were expressed in E. coli and purified using IMAC (Figure ). In order to characterize the particles’ physical properties, transmission electron microscopy (TEM), dynamic light scattering (DLS), and SDS-PAGE were used (Figures , S3 and Table ). Most notably, all three methods demonstrated the monodispersity of ferritin samples. The 50 kDa (GFP-HCF) and 27 kDa (protA-HCF) molecular weights of the monomers determined by SDS-PAGE using molecular weight standards matched our expectations for the computed values (49.8 kDa and 28.5 kDa, see also Figures S1, S2, and S4). TEM and DLS data revealed intact cage structures consistent with previously reported (recombinant) ferritins. − , The observed size difference between TEM and DLS arises from the differences in the techniques, since TEM captures the dry protein shell, whereas DLS measures the particle’s hydrodynamic diameter D _H. Furthermore, all data revealed no significant changes in cage size and morphology between homo- and heteropolymeric ferritin. Of note, the successful synthesis of heteropolymeric ferritin is evident from the two bands that appeared in the SDS-PAGE, representing GFP-HCF and protA-HCF (Figure C).

2.

Characterization of ferritin NPs. (A) and (B) show TEM images of GFP-Ft and protA-Ft with an inset showing an enlarged single cage. (C) SDS-PAGE of both constructs. The first lane shows a protein standard ladder with components of known size (shown in kDa). Background of the image was subtracted using a sliding paraboloid with a radius of 50 pixels. Further analysis and lane intensities are shown in Figure S4. (D) Size distribution of NPs from TEM images. (E) Principle of FLISA assay to determine binding behavior between antibody and protA-Ft; for details see the SI. (F) Results from FLISA experiments showing K _D values of three different types of antibodies binding to protA-Ft.

1. Composition and Properties of Ferritin Constructs Used in This Study .

Name	Monomer	MW (kDa)		D TEM (nm)	D H (nm)	PdI	DoL
		Monomer	Cage
GFP-Ft	His6-mEGFP-HCF	49.8	1195	14.6 ± 2.7	16.1 ± 0.2	0.13 ± 0.01	11.2 ± 0.5
protA-Ft	His6-mEGFP-HCF	49.8	∼900	14.4 ± 1.2	15.4 ± 1.1	0.11 ± 0.04	3.0 ± 0.0
	protA-HCF	28.5

^aSince the exact composition of protA-Ft is unknown, the molecular weight can only be estimated. (MW: molecular weight, D _TEM: diameter obtained from TEM imaging, D _H: hydrodynamic diameter, PdI: polydispersity index (both from DLS), and DoL: degree of labeling). For further details see also Figures S1 and S2.

We further determined the degree of labeling (DoL)the number of fluorescent species coupled to an NPspectroscopically. The DoL of ferritin and mEGFP can be calculated using Beer–Lambert’s law:

$$DoL=\frac{{\epsilon }_{280,Ft}}{{\epsilon }_{488,GFP}}\frac{A_{488}}{(A_{280}-A_{488}·{CF}_{GFP})}$$

with absorbance A, molar extinction coefficients ε (both at the specified wavelength), and the correction factor CF_GFP:

$${CF}_{GFP}=\frac{A_{280,GFP}}{A_{488,GFP}}$$

The DoL for GFP-Ft was 11.2 ± 0.5 GFP/Ft (Figure S5). This value strongly deviates from the theoretically possible 24 GFP/Ft and shows that effectively only about half of the GFPs are actively fluorescent. A decreased number of active GFPs is commonly known, since GFP exhibits a typical maturation efficiency of ∼80%. , In our case, additional sources of GFP inactivation include (1) steric hindrance on the ferritin’s surface area that disturbs GFP’s structure, (2) HOMO-FRET events that decrease the absolute absorbance values, or (3) intrinsic folding and activity deficits when expressing GFP homologs in E. coli.

Nonetheless, a distinct difference was detectable between GFP-Ft and protA-Ft, where the latter exhibited a DoL of 3.0 ± 0.0 GFP/Ft (Figure S5). This difference suggests a stoichiometric ratio of 1 GFP-HCF to 3–4 protA-HCF on the heteropolymeric protA-Ft, which is further supported by SDS-PAGE, which shows a ratio of about 1 GFP-HCF to 4 protA-HCF (Figure S4). We hence conclude that the hybrid protA-Ft cage consists of a notable excess of protA-HCF, giving rise to some 12 to 18 antibody-binding sites. In the future, applying cryogenic electron microscopy could prove as a useful addition to further resolve our NP’s architecture or DoL as demonstrated previously for ferritin variants. ,

We then tested the versatility of protA-Ft functionalization by characterizing its coupling to various control antibodies: human IgG1, murine IgG2a, and a fluorescently labeled IgG1-APC. To this end, a fluorescence-linked immunosorbent assay (FLISA) was performed, as it sensitively determines binding affinities by evaluating the fluorescence signal of coupled molecules (Figure E). All antibodies bound to protA-Ft with high affinity, exhibiting K _D values between 1.6 and 3.7 nM (Figure F). These values undercut previously reported K _D values for protA by almost one order of magnitude and likely arise from the increased avidity: the antibody can be chelated by two protA, since multiple protA subunits per protA-Ft cage exist and since each of the two heavy chains in the Fc region of the antibody can bind one protA (Figure ). Whereas each of the tested antibody species bound to protA-Ft, no binding was observed for GFP-Ft under the same test conditions (data not shown). Hence, the hybrid protA-Ft is easily equipped with any IgG as long as the antibody has sufficient affinity toward SpA.

After verifying the high integrity and monodispersity of both ferritins, we tested the natural bonding between ferritin and TfR1. To this end, a microshower system to incubate single cells with protA-Ft was implemented, to precisely monitor the NP coupling to the cell membrane along with changes in the cell response (Figure A).

3.

Targeting of TfR1 using protA-Ft. (A) Schematic experimental setup. Cos7 cells were showered with a microcapillary filled with protA-Ft for 15 min. Binding to cells occurs on cells that are overexpressing TfR1. (B) Microscopy data of exemplary cell before and after protA-Ft microshower. Shown are phase contrast image, red channel (TfR1 – pHuji, contrast enhancement of 1%), and green channel (protA-Ft – mEGFP). (C) Intensity increase ΔI in protA-Ft’s fluorescence channel after microshower for 15 min in wild-type cells (‘no TfR1’, see also Figure S7B) and cells overexpressing TfR1 (‘TfR1’). A one-way ANOVA was performed to test for differences between groups (*** = p < 0.001). (D) Scatter plot of pixel values from TfR1 channel vs protA-Ft channel before and after protA-Ft microshower, both normalized.

After the microshower, protA-Ft was clearly bound to the membrane of TfR1-overexpressing cells (Figure B), corroborated by a substantial fluorescence intensity increase of 46 ± 16% in the ferritin channel. In the case of nontransfected, wild-type cells, negligible coupling of protA-Ft to the plasma membrane was detected. Accordingly, fluorescence intensities rose only by 2 ± 5% (Figures C and S7B). A direct, pixel-wise comparison of intensities in the TfR1 and protA-Ft channels after the microshower revealed their linear dependency at lower intensities, corresponding to a high level of colocalization of both proteins (Figure D). At higher intensities, this linear relation is not maintained, since the epifluorescent microscope integrates TfR1 intensities across the cell, which are naturally higher toward the cell center. Intensities in the protA-Ft channel, on the other hand, maintain a homogeneous distribution of intensities (I _{pixel, protA‑Ft} ≈ 0.2–0.8 and Figure B), supporting the exclusive localization of protA-Ft to the cell plasma membrane. Before the microshower, a homogeneous, low-autofluorescence signal is detected in the protA-Ft channel, which is independent of the TfR1 intensities.

Furthermore, FLISA analysis showed strong binding between protA-Ft and TfR1 with a K _D value of 2.6 nM (Figure S6). Also, we could even observe similar coupling of GFP-Ft to TfR1 on cells despite the higher amount of large mEGFP subunits attached to the cage (Figure S7A). Interestingly, passivating protA-Ft with polyethylene glycol suppressed coupling to TfR1 (data not shown). To conclude, we here provide evidence of the natural HCF-TfR1 interaction hypothesis, , since ferritin coupling requires the presence of TfR1 and seems to be impeded in the case of extensive ferritin surface modification.

We next tested if protA-Ft after MR targeting can induce a specific cellular response. As outlined above, the death receptor CD95 was chosen to challenge the hypotheses that (1) CD95 signaling is enhanced in the case of receptor cluster formation and (2) signal initiation is possible in the absence of the natural ligand. We first probed the specific binding of protA-Ft to the antibody αCD95, which in turn coupled to CD95 on the cell membrane, by different incubation experiments (Figure A): First, cells were incubated with αCD95 for 15 min, followed by a microshower with protA-Ft. Cells without an antibody or an unspecific antibody, and irrespective of the presence of CD95, showed no binding of protA-Ft (Figures and S7B–D). When showering antibody-treated wild-type cells with protA-Ft, a slight increase in the fluorescence intensity in the protA-Ft channel of 7 ± 8% occurred. This small increase can be attributed to the natural low-level expression of CD95.

4.

Targeting of CD95 using protA-Ft in combination with αCD95 antibody. (A) Schematic experimental setup. Cos7 cells were incubated with αCD95 for 15 min and subsequently showered with a microcapillary filled with protA-Ft for 15 min. Binding to cells occurs when cells are overexpressing CD95 and only in the presence of the antibody. (B) Microscopy data of an exemplary antibody-treated cell before and after a protA-Ft microshower. Shown are phase contrast image, red channel (CD95 – mCherry), far-red channel (αCD95 – APC), and green channel (protA-Ft – mEGFP). (C) Intensity increase in protA-Ft’s fluorescence channel after a microshower for 15 min in wild-type cells (‘no CD95’, see also Figure S7C) and CD95-overexpressing cells (‘CD95’) with and without previous antibody-treatment (Ab, no Ab). A one-way ANOVA was performed to test for differences between groups (*** = p < 0.001). (D) Scatter plot of pixel values from the CD95 channel vs protA-Ft channel before and after a protA-Ft microshower, both normalized.

Yet, cells overexpressing CD95, treated with antibody, and showered with protA-Ft displayed a large rise of the protA-Ft fluorescence intensity by 46 ± 12% (Figure C). The high specificity in the protA-Ft−αCD95–CD95 coupling was further corroborated by comparing the pixel-by-pixel intensities in the protA-Ft and CD95 channels (Figure D). As for TfR1, the two intensities exhibit a linear dependency at low values, reflecting strong colocalization. At higher intensities, intracellular CD95 also contributes to the signal, which impedes the colocalization interpretation. Nevertheless, protA-Ft again exhibits a primary localization to the cell plasma membrane, since the intensity distribution is sufficiently homogeneous and lacks contributions from locations of apparently intracellular CD95 (Figure B). As for TfR1 experiments, no binding was observed with PEGylated protA-Ft, regardless of the presence or absence of the antibody (data not shown).

Having verified that protA-Ft specifically targets TfR1 directly and CD95 through antibody-mediated binding, we explored the subsequent cell response. Figure shows two examples of cells overexpressing TfR1 or CD95, respectively, before and after microshowering with protA-Ft. In TfR1 as well as in CD95 overexpressing cells, the presence of protA-Ft led to a rapid formation of colocalized clusters of the targeted protein and protA-Ft, as indicated by the transition of the homogeneous fluorescence signal to a granular image directly after the shower. Within 2 h, the protein clusters forming in the periphery accumulated in a fluorescence front, presumably arising from internalization, which then redistributed toward the nucleus. In contrast, protA-Ft displayed no specific binding toward nontransfected cells (see also Figure S7). Also, no distinct granular structure occurred in CD95 overexpressing cells without the addition of protA-Ft or antibody over the course of 3 h (see Video S3).

5.

Exemplary Cos7 cells (A) overexpressing TfR1 (red) and (B) overexpressing CD95 (red) and treated with αCD95 antibody. Images were taken before or at denoted time points after a 15 min microshower with protA-Ft (green). Solid arrows are pointing to formed clusters. Dotted arrows indicate movement of protA-Ft and the respective receptor toward the cell nucleus. Full videos are shown in Video S1 and S2. (C) CTB viability assay of HeLa cells overexpressing CD95 (‘CD95’) or with CD95 being knocked out (‘no CD95’). Cells were incubated overnight with or without αCD95 antibody and with or without protA-Ft or GFP-Ft. CD95L served as positive control, which triggers apoptosis upon contact with CD95. A one-way ANOVA was performed to test for differences between groups (ns = not significant, *** = p < 0.001).

We then explored whether CD95 was activated after protA-Ft incubation using cell viability assays. To test a broad range of conditions, one cell line stably overexpressing CD95 and one knockout cell line without CD95 were incubated with different combinations of protA-Ft, GFP-Ft, or no Ft in the presence or absence of αCD95. As a positive control, CD95L was used, which naturally induces apoptosis in the presence of CD95. Only the combination of CD95 either with CD95L or with protA-Ft and αCD95 was able to induce apoptosis (Figure C). Here, the killing efficiencies amounted to 99 ± 1% (CD95L) and 65 ± 13% (protA-Ft + αCD95), respectively. For all other conditions, the killing efficiencies were 0% or negative, indicating cell proliferation. While the killing efficiency is higher in the case of CD95L in comparison to protA-Ft + αCD95, this assay unambiguously shows that our engineered protA-Ft initiates apoptosis in the absence of the natural ligand. It further reveals that receptor accumulation or oligomerization is necessary since the incubation with monomeric αCD95 does not trigger the signaling pathway. The difference in apoptosis activation between the two conditions might stem from the less precisely defined oligomerization state of CD95 when activated with protA-Ft, resulting in reduced signal transduction. Overall, our results contribute to the paradigm change, that cell signal formation does not require a specific ligand and that physicochemical aspects such as the local concentration of receptors are essential determinants.

In this comprehensive study, we present a genetically modified ferritin, which serves as an excellent nanosystem for site-specific cellular targeting with potential for in vivo applications, due to its biocompatibility and cargo shielding properties. Multifunctionality is readily introduced to this NP by creating a hybrid ferritin cage, where mEGFP- and protA-labeled subunits are combined into one cage. The implementation of protA further enables an easy adaptation of this system to various questions of MR activation, by simple incubation with the respective antibodies. The targeting capabilities of this hybrid protA-Ft were demonstrated by site-specific coupling to TfR1 or CD95, two cancer-related MRs. Through both targeting routes, the multivalent ferritin led to rapid receptor–complex formation and MR accumulations. In case of TfR1, we provide evidence that ferritin naturally binds to TfR1. In the case of CD95, we uncovered a ligand-free route to initiate apoptosis, which required CD95 clustering via antibody-mediated binding of protA-Ft.

Perspectively, the hollow cage structure of ferritin and the internalization of protA-Ft following MR targeting make protA-Ft a potent carrier for targeted drug delivery. , Previously, it was already shown that small molecules, such as drugs and specifically anticancer drugs, ,, could be loaded into ferritin. However, the site-selective release of cargo remains challenging. In this context, modified ferritin variants have been shown to disassemble and release loaded siRNA upon entering acidic environments such as lysosomes. Other applications include the delivery of stimuli (such as hyperthermia or the spatial manipulation with magnetic fields) on the nanoscale, for which, for example, magnetic cores were successfully synthesized into the ferritin cage before. ,− Further core modifications include the encapsulation of gold nanoparticles or quantum dots into the cage. −

Conclusively, our manufactured hybrid protein cage is an easily operatable, highly flexible nanosystem that (1) can target arbitrarily chosen MRs, (2) is easily modified for nanotechnological or cell biological applications, and (3) can be used to uncover various mechanisms underlying cell signal initiation.

Glossary

ABBREVIATIONS

MR

membrane receptor

SpA

protein A from S. aureus

protA

engineered Z-domain of S. aureus protein A

mEGFP

monomeric enhanced green fluorescent protein

Ft

ferritin

protA-Ft

ferritin with mEGFP and protA subunits fused to its cage

GFP-Ft

ferritin with mEGFP subunits fused to its cage

HCF

heavy chain ferritin

LCF

light chain ferritin

GFP-HCF

HCF subunit fused to mEGFP

protA-HCF

HCF subunit fused to protA

TEM

transmission electron microscopy

DLS

dynamic light scattering

SDS-PAGE

sodium dodecyl sulfate-polyacrylamide gel electrophoresis

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.nanolett.5c04582.

• Detailed description of materials and methods for protein expression, TEM imaging, FLISA, DLS, spectroscopy and degree of labeling, SDS-PAGE, mammalian cell culture, fluorescence microscopy, and viability assay; DNA and protein sequences of designed ferritin constructs; results of control experiments of DLS for cage size distribution, SDS-PAGE for cage composition, spectroscopy for degree of labeling, FLISA for interaction between TfR1 and ferritin, microscopic images of controls regarding construct coupling to membrane receptors (PDF)

• Video S1: Cos7 cell transfected with TfR1 and microshowered with protA-Ft for 15 min. The video starts after the microshower. This video corresponds to the data shown in Figure 5A (AVI)

• Video S2: Cos7 cell transfected with CD95, incubated with αCD95 antibody for 15 min and subsequently microshowered with protA-Ft for 15 min. The video starts after the microshower. This video corresponds to the data shown in Figure 5B (AVI)

• Video S3: Cos7 cell transfected with CD95. Cell was imaged for ∼3 h as a control for Video S2 (AVI)

■.

2nd Institute of Physics, University of Stuttgart, Pfaffenwaldring 57, 70569 Stuttgart, Germany

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Conceptualization: CM. Methodology: AN. Software: AN. Validation: AN, CM. Formal analysis: AN. Investigation: AN, CS, LZ, ACF. Resources: CM. Data Curation: AN. Writing–original draft: AN. Writing–review and editing: AN, CS, LZ, ACF, CM. Visualization: AN. Supervision: CM. Project administration: CM. Funding acquisition: CM.

The authors declare no competing financial interest.