Cargo binding promotes KDEL receptor clustering at the mammalian cell surface
Bjorn Becker, M. Reza Shaebani, Domenik Rammo, Tobias Bubel, Ludger Santen, Manfred J. Schmitt
aa r X i v : . [ q - b i o . S C ] D ec Cargo binding promotes KDEL receptor clustering atthe mammalian cell surface
Bj ¨orn Becker , M. Reza Shaebani , Domenik Rammo , Tobias Bubel , Ludger Santen ,and Manfred J. Schmitt Molecular and Cell Biology, Department of Biosciences and Center of Human and Molecular Biology (ZHMB),Saarland University, D-66041 Saarbr ¨ucken, Germany Department of Theoretical Physics, Saarland University, D-66041 Saarbr ¨ucken, Germany * B.B. and M.R.S. contributed equally to this work. + Correspondence should be addressed to [email protected]
ABSTRACT
Transmembrane receptor clustering is a ubiquitous phenomenon in pro- and eukaryotic cells to physically sense recep-tor/ligand interactions and subsequently translate an exogenous signal into a cellular response. Despite that receptor clusterformation has been described for a wide variety of receptors, ranging from chemotactic receptors in bacteria to growth factorand neurotransmitter receptors in mammalian cells, a mechanistic understanding of the underlying molecular processes isstill puzzling. In an attempt to fill this gap we followed a combined experimental and theoretical approach by dissecting andmodulating cargo binding, internalization and cellular response mediated by KDEL receptors (KDELRs) at the mammaliancell surface after interaction with a model cargo/ligand. Using a fluorescent variant of ricin toxin A chain as KDELR-ligand( eGFP-RTA
H/KDEL ), we demonstrate that cargo binding induces dose-dependent receptor cluster formation at and subsequentinternalization from the membrane which is associated and counteracted by anterograde and microtubule-assisted receptortransport to preferred docking sites at the plasma membrane. By means of analytical arguments and extensive numericalsimulations we show that cargo-synchronized receptor transport from and to the membrane is causative for KDELR/cargocluster formation at the mammalian cell surface. S ensing of and responding to extracellular stimuli is an intrinsic property of eukaryotic cells to tightly regulate essential basicprocesses such as proliferation, migration, neurotransmission, or even immune defense. In particular plasma membrane(PM) receptors, e.g. G-protein coupled receptors (GPCRs), play an important role in recognizing extracellular ligands, suchas peptide hormones or drugs, and subsequently transducing the exogenous signal into a cellular response. In this context,a series of cell surface receptors, including EGF and T-cell receptors as well as receptors that are parasitized by certain A/Btoxins or viruses for endocytic internalization, are known to cluster in dynamic membrane nano-domains allowing cells to tunesignaling efficiency and ligand sensitivity, or control protein interactions.
Since various human diseases are directly linkedto abnormalities in membrane-receptor distribution and/or activation, it is important to understand the underlying mechanisticprinciples responsible for receptor clustering and dynamic reorganization to develop potential strategies for a therapeutictreatment.
To address such essential biophysical aspects in receptor biology, we focused on mammalian KDEL receptors (KDELRs)at the cell surface that we and others have shown to be responsible for the sensing and binding of KDEL-cargo and KDEL-bearing A/B toxins.
Besides having a central function in the retrieval of luminal proteins of the endoplasmic reticulum(ER) and in KDEL-cargo uptake from the cell surface, KDELRs are also known to act as GPCRs in the regulation of geneexpression. The loss of KDELR1 has been recently demonstrated to cause lymphopenia and a failure in controlling chronicviral infections.
Because of the biomedical importance of KDELRs at the mammalian cell surface we addressed thisaspect in more detail and aimed to answer the following questions: (i) How are KDELRs distributed in the PM and how doescargo binding affect receptor dynamics and distribution at the cell surface? (ii) How do cells respond to cargo binding andwhat is the underlying cellular mechanism? In contrast to the majority of studies on receptor clustering that either focused onbiological or on theoretical aspects, we here followed a combined experimental, computational, and theoretical approach todissect and modulate cargo binding, internalization and cellular response mediated by KDELRs at the mammalian cell surface.We thereby demonstrate that cargo binding induces dose- and temperature-dependent receptor clustering at and internalizationfrom the PM that is accompanied and counteracted by microtubule-assisted anterograde receptor transport to distinct dockingsites at the membrane. Based on the results of extensive Monte Carlo simulations and analytical arguments we disentangle theeffects of surface dynamic processes from those of cargo-synchronized anterograde KDELR transport along the microtubulenetwork towards and from the PM, and verify that the statistical properties and temporal evolution of the receptor cluster-sizeistribution is mainly induced and controlled by the later process.
Results
KDELRs represent transmembrane proteins which recognize and bind soluble residents of the ER containing a C-terminal re-tention motif (KDEL or KDEL-like) to prevent escape from the secretory pathway.
Recent studies however demonstratedthat KDELRs are not restricted to ER and Golgi compartments but also localize in the PM where they bind KDEL-cargo suchas mesencephalic astrocyte-derived neurotrophic factor (MANF) and internalize microbial A/B toxins such as the HDEL-bearing K28 virus toxin. Until now, however, it is unknown what mechanistically happens after a potential H/KDEL-cargohas bound to the pool of PM localized KDELRs. In addition to the equilibrium between anterograde receptor delivery to andinternalization from the plasma membrane, receptor clustering as well as lateral membrane diffusion in response to ligandbinding could play a key role in determining the total amount of KDELRs at the cell surface, similar to how EGFR (epidermalgrowth factor receptor) and AChR (acetylcholine receptor) control ligand sensitivity and activate signaling pathways.
Design and biological activity of a model KDELR cargo
KDELR cluster formation at the mammalian cell surface in response to cargo binding was analyzed and visualized on a modelcargo by using a GFP-tagged variant of the cytotoxic A-subunit of ricin (RTA) extended by a C-terminal H/KDEL motif(Fig. 1A). Using cell surface biotinylation, we were able to detect KDELR1 at the cell surface of mammalian cells (Fig. 1B),in agreement with recent studies in which a pool of KDELR1 was observed at the PM of neuroblastoma cells. Cell surfacelocalization of KDELR1 was also analyzed by adapting an imaging assay originally designed to confirm PM-localization ofAMPA receptors through binding of the snake venom α -bungarotoxin, Btx. For imaging analysis at the cell surface, aBtx binding site (BBS) was inserted into an extracellular loop of mammalian Erd2.1 (KDELR1) and subsequently used tovisualize physical Btx/KDELR interaction at the PM (Fig. 1C, bottom). Extracellular binding of Alexa488-labeled Btx to themodified and in vivo functional KDELR1 variant containing an Btx binding motif in an extracellular loop of the receptorfurther supported the biotinylation data and likewise demonstrated that a minor but significant number of receptors localizesin clusters at the PM of HeLa cells (Fig. 1C, top). The recombinantly expressed and purified cargo protein eGFP-RTA
H/KDEL remained toxic to HeLa cells (see Fig. 1D and Supplementary Fig. S1), confirming earlier observations that the addition of aC-terminal H/KDEL motif to RTA enhances its in vivo toxicity.
In contrast to the negative control of eGFP-RTA lackinga C-terminal KDELR binding site, eGFP-RTA
H/KDEL rapidly bound to the cell surface within seconds to minutes, indicating thata fraction of KDELRs at the PM is responsible for ligand binding (Figure 1E and Supplementary Fig. S2). This is furthersupported by the similar behavior seen in cell surface cargo clustering in response to eGFP-RTA
HDEL addition and KDELR1pattern formation after Btx binding (Fig. 1C). As analogous KDELR1/cargo clustering was also observed in different celllines (such as HEK-293T and RAW-Blue), KDELR-mediated cargo binding at the PM is not restricted to just a single cell typebut rather seems a general phenomenon in mammalian cells (Supplementary Fig. S3). Immunostaining of non-permeabilizedHeLa cells as well as binding studies at 4 ◦ C further demonstrated that eGFP-RTA
HDEL signals are indeed present at the plasmamembrane and, thus, not restricted to signals of internalized KDELR/cargo complexes (Supplementary Fig. S2B and Fig. S4).In addition, increased toxicity of eGFP-RTA
H/KDEL , visible intracellular fluorescent signals, especially after longer incubation( > eGFP-RTA H/KDEL is indeed internalized from the mammalian cell surface(see Fig. 1D, 1G, SupplementaryFig. S5 and Supplementary Movies S1 and S2).Furthermore, live cell imaging of cells loaded with eGFP-RTA
HDEL (see Fig. 1F) identified a strict time-dependent accumu-lation of fluorescent cargo signals at the PM which was absent in control cells treated with eGFP-RTA lacking a KDELRbinding motif (Fig. 1H). Interestingly, the development of fluorescent signals/clusters of eGFP-RTA
HDEL at the cell surface oc-curred in distinct phases: Initially, the system remained relatively inactive for a short time ( t <
20 min). After this transientregime, an exponential growth was observed, which eventually saturated at t >
80 min. The observed huge fluctuations of theaccumulated receptor density at the PM is a signature of the stochasticity of the underlying nonequilibrium process, where thesystem ultimately reaches a balance between the loss of surface receptors due to endocytosis and gain by recycling them.
Adaptive Monte Carlo simulations of KDELR/cargo dynamics at the cell surface
Aiming at better understanding the in vivo observed KDEL/cargo interactions at the cell surface, we performed extensiveMonte Carlo simulations which shed light on the underlying mechanisms of receptor clustering at the cell membrane. Wemodeled the cell membrane as a lattice with a spacing of the size of receptors ( ∼
10 nm) and periodic boundary conditions(Fig. 2A). Each lattice site can be occupied by at most one receptor which is either liganded or unliganded. The membrane size(4 × lattice sites) is comparable to that of a typical HeLa cell. Assuming that the frequency, spatial extent, and target regionof endocytosis and recycling of receptors are independent stochastic events, we introduced asymmetric rates of endocytosis igure 1. H/KDEL-cargo binding to the mammalian cell surface induces receptor cluster formation. (A) (top) Schematicoutline of the fluorescent model cargo eGFP-RTA
HDEL consisting of the cytotoxic A-subunit of ricin (RTA), mammalianenhanced GFP (eGFP) and a C-terminal (His) -Tag for purification. (bottom) eGFP-RTA lacking a KDELR binding motifserved as negative control. (B) Cell surface biotinylation of mammalian KDELR1. HeLa cells were transiently transfectedwith KDELR1 (Erd2.1-V5 (+)) or an empty vector (-) and cultivated for 48 h. Cell surface proteins were biotinylated bytreatment with (+) or without (-) Sulfo-NHS-SS-Biotin and purified with streptavidin beads. Whole cell lysates (input)served as control to determine the total amount of Erd2.1-V5 (detected with anti-V5), while β -actin and Na+/K+ ATPaseserved as cytosolic and plasma membrane marker proteins, respectively. Membrane fraction (surface) illustrates the totalfraction of proteins at the cell surface. (C) (bottom) Schematic outline of α -bungarotoxin (Btx) cell surface binding. HeLacells expressing a KDELR variant in which a Btx binding site (BBS) was inserted between positions T114 and P115 ofc-myc-tagged KDELR1 (Erd2.1) were treated with Alexa488-labeled α -Btx. As Btx is incapable to cross the mammalianPM, any physical interaction between Btx and BBS can only occur if KDELR1 is present in the PM. (top) Confocal laserscanning microscopy of HeLa cells transfected with pERD2.1-BBS-cmyc or an empty vector control and treated with 10 µ g/ml Alexa488-labeled α -Btx. (D) In vivo toxicity of eGFP-RTA H/KDEL against HeLa cells. Cell viability (N=3, n=5 replicates)was determined after 48 h in the presence or absence of 160 µ g/ml of the indicated RTA variant (Mock, PBS buffer). Meanvalues and standard deviations are displayed ( ∗∗∗ , P < . µ g/ml eGFP-RTA H/KDEL or eGFP-RTA for 5 min and cargo binding was analyzed after 10washing steps. (F) Live cell imaging (45 frames/h) of HeLa cells treated with 160 µ g/ml eGFP-RTA HDEL . Three representativetime points (0, 30, 60 min) are shown. (G) Temporal evolution of the density of cargo signals at the surface of HeLa cells.The accumulation of fluorescent signals is shown after treatment with 160 µ g/ml eGFP-RTA HDEL or eGFP-RTA . The symbolsrepresent the optimal signal-to-noise ratio in image analysis. The error bars reflect the variation range of signal intensity fordifferent threshold values of image analysis parameters. The functional form only weakly depends on the choice of thethreshold values. igure 2. (A) Schematic representation of (left) the minimal model of receptor cycle between the PM and endosomes, and(right) the simulation method. An example of a randomly chosen area for endocytosis (vesicle arrival) is marked in red (blue).Possible scenarios for the evolution of the surface receptor population during the next simulation step are shown. (B) Timeevolution of the density of accumulated cargo at the cell surface. A comparison is made between experimental data,simulation results (a single realization), and the analytical expressions Eqs. 2 and 3. The dotted line indicates the analyticalprediction via Eq. 2 for α gain = α loss = . × − s − . The starting time of simulations and analytical expression 2 is shifted totake into account the initial inactive regime in experiments.and vesicle arrival, and chose a random target region on the membrane for each event. Considering the normal size of clathrin-coated vesicles to be in the range of 50 to 100 nm, we allowed the extent of events to vary within 5 × ×
10 lattice sites.An endocytosis event leads to elimination of all receptors within the affected zone. The number of receptors carried by anincoming vesicle was chosen randomly from 0 to the maximum capacity of that vesicle and distributed randomly within thetargeted zone on the PM upon availability of empty sites. One may also switch on the receptor surface dynamics, includinglateral membrane diffusion and receptor-receptor interactions. Starting with an initial random configuration of receptors onthe lattice, the surface density evolves and finally reaches a nonequilibrium steady state by balancing the receptor gain andloss. There are density fluctuations in the steady state due to the stochastic origin of endocytosis and vesicle arrival events.As shown in Fig. 2B, the experimental data could be qualitatively reproduced in simulations by tunning the initial density atthe PM and the gain and loss rates. Notably, the amplitude of steady-state oscillations in simulations obtained for a singlerealization is comparable to the experimental data.
Models for receptor cycle
We first developed a minimal theoretical model for loss and gain of receptors during endocytosis and recycling back to thesurface. Assuming that the total number of receptors in the cell is conserved within our experimental time window, thefractions of receptors at the cell surface n surf and inside the cell n bulk are related as n surf + n bulk =
1. Denoting the endocytosis andvesicle arrival rates, respectively, with α loss and α gain , the evolution of the average fraction of receptors at the plasma membrane n surf ( t ) in a simple form can be described asd n surf dt = α gain n bulk − α loss n surf . (1)Denoting the initial and steady-state fractions of surface receptors with n o surf and n ∞ surf , one obtains n ∞ surf = α gain α loss + α gain , and the timeevolution of the average fraction of surface receptors follows n surf ( t ) = n ∞ surf + ( n o surf − n ∞ surf ) e − t / τ o , (2)with the characteristic time τ o = α gain + α loss . Thus, n ∞ surf and τ o are controlled by the rates α loss and α gain . Indeed, the cycleof receptors in the cell is more complicated; there are more influential parameters involved and the receptor conservationassumption does not hold in general. Nevertheless, as shown in Fig. 2B, Eq. 2 qualitatively reproduces the trends observed inexperiments, though the curvature change is not captured. he previous approach would reflect a situation where the transport of receptors to and from the plasma membrane isneither influenced by exclusion nor by self-amplification. However, if one assumes that binary excluded-volume interactionsbetween receptors have to be considered and/or self-amplification of the receptor transport plays a role, the evolution of thefraction of surface receptors can be roughly described asd n surf dt = A + Bn surf − Cn surf . (3)By fitting the free parameters, the above equation captures quantitatively well the in vivo observed dynamics over the wholetime window (solid line in Fig. 2B). Effect of temperature and ligand concentration on KDELR cluster formation
Prior to experimentally investigating if temperature changes affect KDELR/cargo clustering at the cell surface and follow thevan-‘t-Hoff’sche rules, we assumed that KDELR clustering at 25 ◦ C should slow down by a factor of 2-4 without changingthe overall kinetics and shape of the curve. Based on this assumption, we reduced the endocytosis and vesicle arrival ratesin simulations by a factor of 3. The results predicted a timely retardation of KDELR membrane cluster formation at 25 ◦ Cwhile the overall saturation level was similar to the one reached at 37 ◦ C (Fig. 3A, left panel). Since the experimental datanicely confirmed the numerical predictions (Fig. 3A, right panel), it can be concluded that KDELR/cargo cluster formation isa temperature-dependent process. It can be also deduced from the minimal analytical model presented in the previous section,that the final saturation level n ∞ surf remains invariant under a symmetric scaling of the rates (i.e. α loss → κα loss and α gain → κα gain ),while the characteristic relaxation time is rescaled as τ o → τ o / κ .To determine any effect of cargo concentration on cluster formation at the PM, HeLa cells were treated with differentdoses of eGFP-RTA HDEL ; the corresponding results revealed a strict dose-dependency of KDELR/cargo cluster formation at thecell surface (Fig. 3B). In contrast to the impact of temperature, the variation of the saturation level with changing the cargoconcentration indicated that the endocytosis and vesicle arrival rates are differently affected, i.e. they scale as α loss → κα loss and α gain → κ ′ α gain with κ = κ ′ . This is indeed necessary to reproduce the experimental data in simulations (Fig. 3B, left panel).Denoting the steady-state fraction of surface receptors at low and high concentrations with n ∞ low and n ∞ high ( n ∞ high > n ∞ low ), fromthe minimal theoretical model one finds that α high gain α low gain > α high loss α low loss , thus, the vesicle arrival rate is more sensitive to the concentrationchanges than the endocytosis rate. Interestingly, reduction of the saturation level at lower cargo concentrations suggests thatmammalian cells can somehow modulate the response depending on extracellular ligand concentration. They are capable tosense the actual concentration of cargo binding and subsequently regulate the total amount of KDELRs at the cell surface.Hence, the combination of experimental data and numerical results provides a first mechanistic insight into KDELR/cargoclustering at the mammalian cell surface. Anterograde KDELR transport to preferential plasma membrane arrival sites
The cellular plasma membrane resembles a thoroughly regulated and highly dynamic compartment that contains cell surfacemicro-domains like lipid rafts or caveolea. It is well documented that plasma membrane receptors such as AchR, EGFRor TGF- β are associated with lipid rafts and that preferential receptor cluster formation in distinct micro-domains ofthe PM provides an important means to regulate downstream signaling as shown for EGFR. Moreover, it has been proposedthat T cell receptor pre-clustering at the cell surface contributes to a significant increase in ligand sensitivity and acceleratessignaling pathway activation. To understand how KDELR/cargo clustering evolves and whether or not KDELRs are likewise arranged in receptor pre-clusters or micro-domains, cluster size distribution P ( s ) of the model cargo eGFP-RTA HDEL was determined at different timepoints during the clustering process. As shown in Figs. 4A and 4B, relatively small clusters were visible at early time points,while larger clusters appeared at longer times. A detailed analysis indicated that the growth of the largest cluster eventuallystopped after reaching the stationary state. Notably, the cluster size distribution nearly followed a power-law decay P ( s ) ∼ s − β with a rather time-invariant exponent β ≈ t <
20 min).The functional form of the cluster size distribution P ( s ) indeed implies which of the underlying mechanisms of receptordynamics is dominant. In Monte Carlo simulations, we first assumed that the target zones for endocytosis and vesicle arrivalevents are randomly chosen, without allowing the distributed receptors to move on the surface. Next, we examined the mainpossible scenarios for receptor surface dynamics including lateral diffusion on the membrane and receptor-receptor attrac-tion (both with the short range of nearest-neighbor sites). Figure 4C shows that none of the resulting aggregation patternswas capable of producing a power-law size distribution; the tails of the resulting distributions rather follow an exponential-likedecay. The algebraic form can be recovered under the assumptions that KDELRs have distinct and preferred arrival sites atthe plasma membrane and the transport is self-amplified. This was achieved in the following way. The spatial distribution of B t (min) f r a c t i o n o f s u r f a c e r e c e p t o r s t (min) f r a c t i o n o f s u r f a c e r e c e p t o r s -4 -4 -4 -5 Figure 3.
KDELR/cargo clustering is dose- and temperature-dependent. (A) Changes in cargo accumulation of eGFP-RTA
HDEL at the surface of HeLa cells cultivated at 25 ◦ C or 37 ◦ C. The 3-fold reduced activity in simulations (left) represents the knowneffect of temperature on intracellular transport processes (e.g. endocytosis and exocytosis). (right) The experimental resultsat 25 ◦ C and 37 ◦ C. (B) Effect of changing the concentration of the model cargo eGFP-RTA
HDEL on KDELR/cargo clustering atthe plasma membrane. The indicated rates in simulations (left) were adopted to obtain the best fits to the experimental data(right).targeting probability was changed from a uniform to a multiple-peaked Gaussian one. The peaks represent the places whereMTs approach the cell cortex and distribute their vesicles, which are supposed to diffuse on the actin filament network untilthey finally reach the membrane. Additionally, the surface was divided into subdomains obtained by Voronoi tessellationof area around each peak, and a newly generated vesicle choses a target subdomain with a probability proportional to thetransport-activity history of the corresponding MT.Let us consider a rather simple process of receptor aggregation, in which the time evolution of the probability P ( s ) ofhaving receptor clusters of size s is expressed via the master equation dP ( s ) dt = ∑ i + j = s P ( i ) P ( j ) − ∑ i P ( i ) P ( s ) . The gain andloss terms on the right-hand side account for creation of s -size clusters from the coalescence of two smaller ones of sizes i and j , and merging of s -size clusters with other ones, respectively. Starting from an initial configuration with randomlydistributed single receptors, the master equation can be recursively solved to get P ( s ) ∼ exp [ − ln ( + t ) s ] , i.e. P ( s ) decaysexponentially with a time-dependent exponent. At long times, the slope decreases and P ( s ) evolves towards a flat distribution.Our attempts to consider more complicated aggregation scenarios, such as introducing diffusion or aggregation with input,failed to simultaneously reproduce the power-law and time-invariant features of P ( s ) . In contrast, it can be verified thatpreferential attachment of receptors to the existing clusters reproduces the experimentally observed distribution. We considera simple clustering process in which a new receptor is added to the surface at each time step, and it attaches to the cluster of size s i with a probability p s i which is proportional to the cluster size, i.e. p s i = s i ∑ j s j . The sum runs over all clusters, thus, reflectsthe total number of receptors and grows linearly with time. The rate at which s i changes can be assumed to be proportional to p s i ∂ s i ∂ t = p s i = s i ∑ j s j = s i t . (4)Denoting the initiation time of cluster i with t i , one obtains s i ( t ) = t / t i . The probability that a cluster is smaller than s is givenas P [ s i ( t ) < s ] = P ( t i > ts ) . (5)The probability P ( t i ) of initiation at time t i has a constant probability density with respect to time, i.e. P ( t i ) = t . Substituting =30 mint=60 mint=120 min best fit t=160 min -3 -5 -1 -2 s( m ) -1 -2 -1 -3 -5 -1 A CE Figure 4.
Preferential arrival sites of KDELRs at the plasma membrane. (A) The log-log plots of cluster-size distribution P ( s ) of eGFP-RTA HDEL (160 µ g/ml) treated HeLa cells at the indicated time points. The dashed line corresponds to the bestpower-law fit P ( s ) ∼ s − β with β ≃ . ± .
1. (B) Evolution of the receptor clusters at the plasma membrane. A randomlychosen region of the cell surface is shown at different time points (see
Suppl. Info. for the detailed description of themethodology of distinguishing the clusters and obtaining the cluster-size distribution). (C) A comparison of the resulting P ( s ) from different receptor dynamic scenarios in simulations. The solid, dashed, and dotted lines denote the shape of P ( s ) at t =
120 min for randomly distributed immobile receptors, aggregation process including lateral diffusion of receptors andnearest-neighbor attraction between them, and preferential attachment process, respectively. (D) The frequency of vesiclearrival at a sample cell periphery over a time window of 500 s. (E) In vivo dynamics of mCherry-ERD2.1. The transfectedHeLa cells with mCherry-ERD2.1 were analyzed by CLSM (720 frames/h). The illustrated heat map represents theaccumulated fluorescent signals of successive frames. The regions with high traffic load, e.g. around Golgi, are eliminated toprovide a more clear color distinction near the cell surface. (F) The frequency of vesicle transport near the plasma membraneof untreated or eGFP-RTA
HDEL treated cells. The data is averaged over bins of size 10 µ m ( ∗∗ , P ≤ .
01, t test).this into Eq. 5 we get P ( t i > ts ) = − ts · t = − s . (6)Finally, the cluster-size distribution can be obtained using P ( s ) = ∂ P [ s i ( t ) < s ] ∂ s = ∂∂ s ( − s ) = s , (7)which shows a power-law decay with a time-independent exponent 2, in a remarkable agreement with the experimental resultsshown in Fig. 4A. Since the receptor trafficking mainly occurs along MTs, vesicle exchange near the PM happens mostly inthe vicinity of the regions where MTs approach actin filaments near the cell cortex. Our data are consistent with a feedbackmechanism that amplifies receptor transport towards the plasma membrane in the presence of receptor clusters.Although the in vivo observed receptor dynamics is indeed more complicated and also depends on other factors (e.g.membrane thickness, lipid composition etc.) and involves both surface dynamic processes and the membrane-cytoplasmreceptor cycle, our numerical and analytical findings suggest that the intracellular transport of vesicles along the microtubulenetwork, which induces preferential zones for vesicle exchange at the PM, crucially controls the clustering at the mammaliancell surface. To experimentally prove this hypothesis, HeLa cells were transfected with mCherry-labeled KDELR1 (Erd2.1)and receptor dynamics was analyzed by live cell imaging (Supplementary Movie S3). By monitoring the frequency of vesiclearrival at the plasma membrane over a time window of 500 s, it is shown in Fig. 4D that anterograde KDELR transport isnon-uniformly distributed along the plasma membrane, i.e. there are hot spots on the cell periphery which are targeted morefrequently by the arrival of vesicles. The heat map of intracellular KDELR transport in Fig. 4E illustrates the spatio-temporaldistribution of KDEL receptors. Notably, a comparison between untreated and eGFP-RTA HDEL treated cells in Fig. 4F indicates anincrease in the transport rate in treated cells. Thus, the experimental findings support the numerical predictions and underlinethe importance of preferential absorption in regulating KDELR cluster formation. igure 5.
Microtubule-assisted KDELR transport is required for cargo clustering at the cell surface. (A) Tracking of singleKDELR clusters (red) moving along the microtubule network (green). A sequence of five successive live cell imagingpictures (720 frames/h) of HeLa cells expressing mCherry-tagged Erd2.1 and GFP-tagged β -tubulin is shown. The arrowsindicate an example of tubulin/KDELR signal co-localization. (B) Co-localization of GFP-tubulin and mCherry-Erd2.1. Theratio of correlated tubulin and KDELR pixels of the live cell imaging experiment is shown during 150 s. (C) KDELR/cargocluster formation in colchicine (red) and phalloidin (inset) pre-treated cells (2 . µ M colchicine, 60 min or 10 µ M phalloidin,90 min) after incubation with 160 µ g/ml eGFP-RTA HDEL . Temporal evolution of the accumulated KDELR/cargo is comparedwith the untreated control cells. The inset shows a comparison between untreated (solid line) and phalloidin-treated HeLacells (dashed line).
Microtubule and actin assisted receptor transport and membrane clustering
Active protein transport along the cytoskeleton is mediated by actin filaments and microtubules (MTs). While filamentousF-actin is mainly localized in the cell cortex and involved in cell migration, endocytosis and vesicle-mediated cargo transport,MTs are responsible for dynein/kinesin driven active transport of vesicles and organelles.
To verify that the MT-network is involved in intracellular KDELR trafficking, cells were co-transfected with GFP-tagged β -tubulin and mCherry-ERD2.1 and KDELR transport along MTs was visualized by CLSM. Despite the limited resolution oflive cell imaging, directed transport of KDELR signals along MTs could be clearly observed (see e.g. Fig. 5A), indicating thatanterograde receptor transport is indeed MT based. A more quantitative analysis also revealed a relatively high probability oftubulin/KDELR signal co-localization (Fig. 5B). Moreover, we observed that colchicine-mediated inhibition of MT assemblyhighly affects KDELR dynamics in vivo (Supplementary Movie S4A and S4B) and considerably reduces receptor clustering atthe cell surface, see Fig. 5C. We conclude that active receptor transport along MTs is a prerequisite for KDELR/cargo clusterformation at the plasma membrane. It has been shown that disruption of MTs causes the majority of EGFR or cAMPRclusters to be immobile, or affects the endocytosis of EGFRs; thus, MTs play a crucial role in organizing receptor clusters atthe plasma membrane. Interestingly, KDELR mobility at the cell periphery was not completely blocked in colchicine treatedcells indicating that cortical actin filaments which are unaffected by the drug are involved in and responsible for the observedKDELR endocytosis from the PM. Consistently, phalloidin-mediated actin inhibition caused a severe impairment of KDELRcluster development at the cell surface (Fig. 5C, inset). Discussion
Until recently, cargo recognition by KDEL-receptors has been assumed to mainly occur within the Golgi complex duringretrograde transport of soluble ER residents back to the ER.
More recent studies, however, indicated that KDELRs arealso responsible for cell surface binding of extracellular ligands such as the neurotrophic factor MANF or the microbial A/Btoxin K28.
Based on these findings we postulated that a model cargo containing a C-terminal H/KDEL amino acid motifand receptor binding site should likewise bind to cells and, thus, be suitable to track and analyze cargo binding and subsequent ellular responses. Using this approach we now demonstrate that treatment of cells with a fluorescent variant of RTA extendedby a C-terminal H/KDEL motif is required and sufficient to promote specific cargo binding and clustering at the mammaliancell surface. Based on the experimental data presented here it can be deduced that the initial binding of eGFP-RTA
H/KDEL to thecell periphery occurs within seconds and is immediately followed by the formation of plasma membrane-associated clusterswithin 20 minutes. Thereafter, cluster development follows an exponential growth and eventually saturates at time points >
80 min. During this process, KDELR/cargo clusters are also internalized, however the precise temporal resolution ofsuch endocytosis events has not yet been analyzed and will be subject of future studies. Since all cell binding studies wereperformed under conditions of natural KDELR in vivo expression, and H/KDEL motifs on cellular proteins have solely beenattributed to be exclusively recognized by KDELRs, our present data strongly point towards a function of KDELRs at thecell surface. Furthermore, our observation that eGFP-RTA
H/KDEL shows a strong increase in toxicity and in vivo uptake (data notshown), likewise supports a role of KDELR-mediated cargo/toxin transport from the plasma membrane to the cytosol.In an approach to characterize the observed clustering at the cell surface after cargo addition, we combined experimentsand numerical methods and thereby demonstrate that cargo/KDELR cluster formation over time is a process that equallydepends on temperature and cargo concentration. In particular, the later observation strongly points towards a regulated cellularmechanism to respond to an extracellular receptor ligand. Extensive simulations of cluster formation and size distributionindeed indicate that cells are capable to sense external cargo concentration and appropriately adapt the number of receptorsat the plasma membrane. External KDEL-cargo addition likewise resulted in a response leading to an increase in anterogradereceptor traffic to the plasma membrane. In addition, the lower cargo cluster numbers in the absence of active endocytosisand exocytosis (phalloidin-treatment or 4 ◦ C) indicate that under these conditions significantly less receptor molecules reachthe plasma membrane, indirectly supporting our assumption that regulated KDELR transport is important for receptor clusterformation at the cell surface. One of the most striking features of the observed clustering is the power-law decay of cluster-sizedistribution P(s) with an approximately time-independent slope at longer times. To elucidate the origin of this behavior, weisolated and examined the role of surface dynamic processes such as receptor diffusion and receptor-receptor attraction insimulations, which mainly led to a fast (exponential-like) decay of the tail of P(s). We showed that preferential adsorptionof receptors is a natural way to obtain adsorption kinetics and cluster-size distribution. Experimental data indicated thatintracellular KDELR transport to the PM occurs along microtubules, and hot spots form in the vicinity of the regions whereMTs approach the cell cortex and distribute anterograde arrival of KDELR-containing vesicles at the PM as well as collectingnewly formed KDELR/cargo complexes from the cell surface by actin-mediated endocytosis. Both responses represent thedominant mechanisms that control receptor distribution at the cell surface, even if affected by additional factors such as e.g.receptor surface dynamics. We showed that both microtubule network and cortical actin are important key players in theclustering process, and inhibitation of MTs or filamentous actin strongly impaired the dynamic clustering at the cell surface.In future studies we will try to further dissect the underlying molecular processes and to identify the cellular componentsinvolved in KDELR/cargo cluster formation at the mammalian plasma membrane. In future studies we intend to use single-molecule cargo tracking in conjunction with high-resolution imaging to further dissect the underlying molecular processesand to identify the cellular components involved in KDELR/cargo cluster formation at the plasma membrane.
Methods
A detailed description of the experimental procedures and methods can be found in
Supplementary Information , includingcultivation and transfection of mammalian cells, genetic techniques, affinity purification and immunochemical analysis ofPM-localized KDELR1.
Acknowledgments
We thank K. Salo and L. Ruddock for the supply of cDNA of human KDELR1, L. Roberts for cDNA of RTA and anti-RTA antibody, and H. Rieger, K. Kruse and Z. Sadjadi for helpful discussions. This study was supported by the DeutscheForschungsgemeinschaft (DFG) within the collaborative research center SFB 1027 (projects A6 and A7).
Author contributions statement
B.B., M.J.S., M.R.S. and L.S. designed the research. D.R., T.B. and B.B. performed the experiments, M.R.S. analyzed theexperimental data and performed simulations. M.R.S. and L.S. developed the theoretical framework and B.B., M.R.S. andM.J.S. wrote the manuscript.
Competing financial interests
The authors declare no competing financial interests.
References Park I & Lee HS. EphB/ephrinB signaling in cell adhesion and migration.
Mol Cell , 14-19 (2015). . Munoz P et al.
Antigen-induced clustering of surface CD38 and recruitment of intracellular CD38 to the immunologicsynapse.
Blood , 3653-3664 (2008). Wajant H, Pfizenmaier K & Scheurich P. Tumor necrosis factor signaling.
Cell Death Differ , 45-65 (2003). Renner M, Specht CG & Triller A. Molecular dynamics of postsynaptic receptors and scaffold proteins.
Curr Opin Neuro-biol , 532-540 (2008). Jaqaman K & Grinstein S. Regulation from within: the cytoskeleton in transmembrane signaling.
Trends Cell Biol ,515-26 (2012). Casaletto JB & McClatchey AI. Spatial regulation of receptor tyrosine kinases in development and cancer.
Nat Rev Cancer , 387-400 (2012). Hanlon CD & Andrew DJ. Outside-in signaling - a brief review of GPCR signaling with a focus on the Drosophila GPCRfamily.
J Cell Sci , 3533-3542 (2015). Abulrob A et al.
Nanoscale imaging of epidermal growth factor receptor clustering.
J Biol Chem , 3145-3156 (2010). Bray D, Levin MD & Morton-Firth CJ. Receptor clustering as a cellular mechanism to control sensitivity.
Nature ,85-88 (1998).
Abrami L, Liu S, Cosson P, Leppla SH & van der Goot FG. Anthrax toxin triggers endocytosis of its receptor via a lipidraft–mediated clathrin-dependent process.
J Cell Biol , 321-328 (2003).
Eierhoff T, Hrincius ER, Rescher U, Ludwig S & Ehrhardt C. The epidermal growth factor receptor (EGFR) promotesuptake of influenza A viruses (IAV) into host cells.
PLoS Pathog , e1001099 (2010). Gao H, Shi W & Freund LB. Mechanics of receptor-mediated endocytosis.
Proc Natl Acad Sci USA , 9469-9474(2005).
El-Rayes BF & LoRusso PM. Targeting the epidermal growth factor receptor.
Br J Cancer , 418-424 (2004). Riffer F, Eisfeld K, Breinig F & Schmitt MJ. Mutational analysis of K28 preprotoxin processing in the yeast Saccha-romyces cerevisiae.
Microbiology , 1317-1328 (2002).
Eisfeld K, Riffer F, Mentges J & Schmitt MJ. Endocytotic uptake and retrograde transport of a virally encoded killer toxinin yeast.
Mol Microbiol , 926-940 (2000). Schmitt MJ & Breinig F. Yeast viral killer toxins: lethality and self-protection.
Nat Rev Microbiol , 212-221 (2006). Henderson MJ, Richie CT, Airavaara M, Wang Y & Harvey BK. Mesencephalic astrocyte-derived neurotrophic factor(MANF) secretion and cell surface binding are modulated by KDEL receptors.
J Biol Chem , 4209-4225 (2013).
Siggs OM et al.
Mutation of the ER retention receptor KDELR1 leads to cell-intrinsic lymphopenia and a failure to controlchronic viral infection.
Proc Natl Acad Sci USA , 5706-5714 (2015).
Cancino J et al.
Control systems of membrane transport at the interface between the endoplasmic reticulum and the golgi.
Dev Cell , 280-294 (2014). Semenza JC, Hardwick KG, Dean N & Pelham HR. ERD2, a yeast gene required for the receptor-mediated retrieval ofluminal ER proteins from the secretory pathway.
Cell , 1349-1357 (1990). Pelham HR. Evidence that luminal ER proteins are sorted from secreted proteins in a post-ER compartment.
Embo J ,913-918 (1988). Hezel M, de Groat WC & Galbiati F. Caveolin-3 promotes nicotinic acetylcholine receptor clustering and regulates neu-romuscular junction activity.
Mol Biol Cell , 302-310 (2010). Balass M, Katchalski-Katzir E & Fuchs S. The alpha-bungarotoxin binding site on the nicotinic acetylcholine receptor:analysis using a phage-epitope library.
Proc Natl Acad Sci USA , 6054-6058 (1997). Sekine-Aizawa Y & Huganir RL. Imaging of receptor trafficking by using alpha-bungarotoxin-binding-site-tagged recep-tors.
Proc Natl Acad Sci USA , 17114-17119 (2004).
Wales R, Roberts LM & Lord JM. Addition of an endoplasmic reticulum retrieval sequence to ricin A chain significantlyincreases its cytotoxicity to mammalian cells.
J Biol Chem , 23986-23990 (1993).
Becker B & Schmitt MJ. Adapting yeast as model to study ricin toxin a uptake and trafficking.
Toxins , 834-847 (2011). Reddy AS, Chilukuri S & Raychaudhuri S. The Network of Receptors Characterize B Cell Receptor Micro- and Macro-clustering in a Monte Carlo Model.
J Phys Chem B , 487-494 (2010). Guo C & Levine H. A thermodynamic model for receptor clustering.
Biophys J , 2358-2365 (1999). Agrawal NJ, Nukpezah J & Radhakrishnan R. Minimal mesoscale model for protein-mediated vesiculation in clathrin-dependent endocytosis.
PLoS Comput Biol , e1000926 (2010). Hancock JF. Lipid rafts: contentious only from simplistic standpoints.
Nat Rev Mol Cell Biol , 456-462 (2006). Chen CL et al.
Euphol from euphorbia tirucalli negatively modulates TGF- β responsiveness via TGF- β receptor segrega-tion inside membrane rafts. PLoS One , e0140249 (2015). Campagna JA & Fallon J. Lipid rafts are involved in C95 (4,8) agrin fragment-induced acetylcholine receptor clustering.
Neuroscience , 123-132 (2006).
Castro M et al.
Receptor pre-clustering and T cell responses: insights into molecular mechanisms.
Front Immunol , 132(2014). Boggara M, Athmakuri K, Srivastava S, Cole R & Kane RS. Characterization of the diffusion of epidermal growth factorreceptor clusters by single particle tracking.
Biochim Biophys Acta , 419-426 (2013).
Saffman PG & Delbr¨uck M. Brownian motion in biological membranes.
Proc Natl Acad Sci USA , 3111-3113 (1975). Huber F, Boire A, Lopez MP & Koenderink GH. Cytoskeletal crosstalk: when three different personalities team up.
CurrOpin Cell Biol , 39-47 (2015). Hancock WO. Bidirectional cargo transport: moving beyond tug of war.
Nat Rev Mol Cell Biol , 615-628 (2014). Soldati T & Schliwa M. Powering membrane traffic in endocytosis and recycling.
Nat Rev Mol Cell Biol , 897-908(2006). de Keijzer S, Galloway J, Harms GS, Devreotes PN & Iglesias PA. Disrupting microtubule network immobilizes amoeboidchemotactic receptor in the plasma membrane. Biochim Biophys Acta , 1701-1708 (2011).
Kharchenko MV, Aksyonov AA, Melikova MM & Kornilova ES. Epidermal growth factor (EGF) receptor endocytosis isaccompanied by reorganization of microtubule system in HeLa cells.
Cell Biol Int , 349-359 (2007). Sakai T, Yamashina S & Ohnishi S. Microtubule-disrupting drugs blocked delivery of endocytosed transferrin to thecytocenter, but did not affect return of transferrin to plasma membranes tatsuy.
J Biochem , 528-533 (1991).
Majoul I, Straub M, Hell SW, Duden R & Soling HD. KDEL-Cargo regulates interactions between proteins involved inCOPI vesicle traffic: measurements in living cells using FRET.
Dev Cell , 139-153 (2001)., 139-153 (2001).