RefSOFI for Mapping Nanoscale Organization of Protein-Protein Interactions in Living Cells.

HHS Public Access Author manuscript Author Manuscript

Cell Rep. Author manuscript; available in PMC 2016 May 17. Published in final edited form as: Cell Rep. 2016 January 12; 14(2): 390–400. doi:10.1016/j.celrep.2015.12.036.

RefSOFI for Mapping Nanoscale Organization of Protein-protein Interactions in Living cells Fabian Hertel1, Gary C. H. Mo1, Sam Duwé3, Peter Dedecker3, and Jin Zhang1,2 1Department

of Pharmacology, University of California, San Diego, 9500 Gilman Drive, San Diego, CA 92093, USA

Author Manuscript

2Department

of Pharmacology and Molecular Sciences, The Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA

3Department

of Chemistry, University of Leuven, Celestijnenlaan 200F, 3001 Heverlee, Belgium

Summary

Author Manuscript

It has become increasingly clear that protein-protein interactions (PPIs) are compartmentalized in nanoscale domains that define the biochemical architecture of the cell. Despite tremendous advances in super-resolution imaging, strategies to observe PPIs at sufficient resolution to discern their organization are just emerging. Here we describe a strategy in which PPIs induce reconstitution of fluorescent proteins (FPs) that are capable of exhibiting single-molecule fluctuations suitable for Stochastic Optical Fluctuation Imaging (SOFI). Subsequently, spatial maps of these interactions can be resolved in super-resolution in living cells. Using this strategy, termed reconstituted fluorescence-based SOFI (refSOFI), we investigated the interaction between the endoplasmic reticulum Ca2+ sensor STIM1 and the pore-forming channel subunit ORAI1, a crucial process in store-operated Ca2+ entry (SOCE). Stimulating SOCE does not appear to change the size of existing STIM1/ORAI1 interaction puncta at the ER-plasma membrane junctions, but results in an apparent increase in the number of interaction puncta.

Introduction

Author Manuscript

Highly dynamic protein-protein interaction (PPI) networks form the basis of virtually all cellular processes (Bonetta, 2010; Koh et al., 2012). The subcellular locations of these interactions encode information critical for understanding the molecular logic behind cellular functions. For instance, within the small junctions between the endoplasmic reticulum (ER) membrane and plasma membrane (PM), the interaction between subunits of

Contact: Correspondence should be addressed to Jin Zhang ([email protected]). Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Author Contributions FH, PD and JZ designed the experiments. FH, GM, SD and PD performed research. FH, PD and JZ wrote the paper. Competing Financial Interests The authors declare no competing financial interests.

Hertel et al.

Page 2

Author Manuscript

the pore forming Ca2+ release-activated Ca2+ (CRAC) channel ORAI1 and stromal interaction molecule 1 (STIM1) is crucial for store-operated Ca2+ entry (SOCE) (Carrasco and Meyer, 2011). Protein complexes that result from PPIs are often further organized into microdomains or nanodomains. The often submicroscopic size of these functional domains makes it difficult to characterize them using current diffraction-limited methods of detecting PPIs such as Förster resonance energy transfer (FRET) (Fernández-Dueñas et al., 2012; Sun et al., 2013) and bimolecular fluorescence complementation (BiFC; Kodama and Hu, 2012). Elucidation of the functional organization of protein interaction networks in living cells therefore requires the development of methods capable of resolving PPIs at an enhanced spatial resolution.

Author Manuscript

However, currently available strategies that provide super-resolution information focus on the visualization of individual fluorescent molecules (Han et al., 2013). One class of superresolution imaging techniques relies on the use of special illumination schemes, represented by stimulated emission depletion (STED; Hell, 2007) and saturated structured illumination microscopy (SSIM; Gustafsson, 2005). Another class is based on the application of photoswitchable dyes for single-molecule localization or fluctuation imaging, such as photo activated localization microscopy (PALM; Betzig et al., 2006), stochastic optical reconstruction microscopy (STORM; Rust et al., 2006) or stochastic optical fluctuation imaging (SOFI; Dertinger et al., 2010). By contrast, the number of approaches for the visualization of PPIs at the super-resolution level is currently limited.

Author Manuscript Author Manuscript

To develop a general method for imaging PPIs in super-resolution in living cells, the formation of target protein complexes at specific loci needs to be translated into spatially constrained fluorescent signals compatible with super-resolution imaging. Our strategy utilizes the interaction of a pair of target proteins to bring complementary, non-fluorescent fragments into nanometer proximity and initiate BiFC (Kodama and Hu, 2012) of suitable FPs, which in turn generates detectable single molecule fluorescence fluctuations. The overall performance of this strategy depends on efficient fragment reconstitution, fast chromophore maturation, and robust single molecule fluctuations from the reconstituted FP. While photochromic properties depend on the coupling between the fluorophore and surrounding residues (Dedecker et al., 2013; Gayda et al., 2012), the reconstitution/ maturation of the chromophore depends on the efficient folding of the β-can and accurate assembly of the proton network (Remington, 2006). Fulfilling these distinct requirements in a single fluorescent protein complicates the task of identifying suitable fragments. Recently, super-resolution imaging utilizing BiFC has been demonstrated using PALM (Liu et al., 2014; Nickerson et al., 2014; Xia et al., 2014). Although development of BiFC-PALM represents a valuable advance, live-cell imaging requires adjustments that limit its actual performance. Furthermore, PPIs that are acutely induced through a signaling event (signalinduced PPIs) cannot be observed over time in the same cells using BiFC-PALM, since it utilizes probes that require long maturation times and often involves photobleaching the probes upon imaging which during the reiterative cycle depletes the pool of probes available. In this study, we develop a method based on photochromic Stochastic Optical Fluctuation Imaging (pcSOFI), which takes advantage of single-molecule fluctuations generated when reversibly photoswitchable fluorescent proteins (FPs) such as Dronpa (Ando et al., 2004) are illuminated at specific wavelengths. Statistical analysis of the fluorescence fluctuations Cell Rep. Author manuscript; available in PMC 2016 May 17.

Hertel et al.

Page 3

Author Manuscript

allows super-resolution images to be computed, currently providing an up to three-fold enhancement of the spatial resolution beyond the diffraction limit (Dedecker et al., 2012a; Duwé et al., 2015; Moeyaert et al., 2014). A pcSOFI-based method would be complementary to BiFC-PALM and provide the following advantages, especially for livecell experiments: possibility of imaging formed complexes in the same cell at multiple time points during a single experiment; improved temporal resolution for a given imaging condition; and lower susceptibility to noise, therefore being less demanding on exposure/ excitation resulting in reduced photo-damage to cells (Dertinger et al., 2013).

Author Manuscript

Here, we report the successful development of a set of molecular tools that can reveal both constitutive and signal-induced PPIs in super-resolution in living cells, based on an enhanced Dronpa variant that we developed and the yellow fluorescent protein Venus. In particular, we demonstrate that using this strategy, which we termed reconstituted fluorescence based Stochastic Optical Fluctuation Imaging (refSOFI), it is possible to visualize the signal-induced occurrence of STIM1/ORAI1 interactions at ER-plasma membrane (PM) junctions in super-resolution, quantify the STIM1/ORAI1 interaction puncta, which are clusters of labeled STIM1/ORAI1 complexes, and distinguish the changes in number versus size upon Ca2+ store depletion.

Results Identification of complementary Dronpa fragments suitable for BiFC


Given that pcSOFI was initially demonstrated using the reversibly photoswitching FP Dronpa (Dedecker et al., 2012a), we first set out to identify fragments that can successfully reconstitute Dronpa. Although BiFC using Dronpa has already been described (Lee et al., 2010), the reported long (24 h) incubation time effectively prohibits live-cell imaging of signal-induced PPIs. To help identify Dronpa fragments for efficient BiFC, we generated a series of circularly permutated (cp) Dronpa variants based on the notion that circular permutation sites are good predictors for generating BiFC pairs (Kodama and Hu, 2012). cpDronpa-181 was identified as the best candidate, and its maturation efficiency was then further improved via the known point mutation V60A (Moeyaert et al., 2014). Random mutagenesis identified two further mutations, M40I and F173V, which increased in vivo brightness. The final mutant, termed cpDMVF-181, exhibited a fluorescence off-switching rate approximately 10 times faster than that of wild-type Dronpa under similar conditions (Figure 1A). With a fluorescence quantum yield of 0.65 and an extinction coefficient of 72,000 M−1cm−1, the molecular brightness of cpDMVF is approximately 70% that of Dronpa (0.74 and 93,000 M−1cm−1). We measured the refolding kinetics of purified cpDMVF after denaturation and found that it renatured faster than Dronpa while retaining a comparable level of renaturation (Figure 1B). The recovery of cpDMVF fluorescence after combined denaturation and chromophore reduction was slightly slower than that of Dronpa (Figure 1C). These data show that the brightness and maturation of cpDMVF was on par with wild-type Dronpa, making it a good BiFC candidate. Next, we separated cpDMVF at the circular permutation site and fused DMVF-181-C (Cterminal) and DMVF-181-N (N-terminal) to FK506 binding protein (FKBP) and FKBPrapamycin binding domain (FRB), respectively, and used this inducible heterodimerization Cell Rep. Author manuscript; available in PMC 2016 May 17.

Hertel et al.

Page 4

Author Manuscript

system to characterize DMVF reconstitution (Figure 1D-F). The FRB fused N-terminal fragment was targeted to the plasma membrane using the N-terminal localization sequence of Lyn kinase (Dedecker et al., 2012a) or to actin via Life-Act (Riedl et al., 2008), and the FKBP-fused C-terminal fragment remained diffusible. The induction of hetero-dimerization by rapamycin (DeRose et al., 2013; Robida and Kerppola, 2009) resulted in reconstitution of DMVF, and responding cells exhibited an increase in fluorescence which reached a plateau after approximately 4 h (Figure 1E). Cells of the corresponding negative control experiments exhibited negligible fluorescence from reconstituted DMVF, independent of the targeting sequence (Figure 1G), indicating that the occurrence of fluorescence is clearly coupled to the PPI. Super-resolution imaging with reconstituted DMVF


We performed a side-by-side comparison of reconstituted DMVF with wild-type Dronpa using the fusions FKBP-Dronpa and Lyn-FRB. Average whole-cell fluorescence intensities (3062 ± 527 a.u., n=15 for Dronpa versus 5368 ± 815 a.u., n=12 for reconstituted DMVF) suggested that DMVF exhibited high reconstitution efficiency and was present at a level comparable to wild-type Dronpa. The fluctuation behaviors were quantified using normalized pcSOFI values (Supplemental Experimental Procedures), which are independent from the local concentration of the fluorophore and directly proportional to the molecular brightness and the probability of the fluorophore residing in the OFF-state at any given time (POFF). We found that the averaged normalized pcSOFI values across the cell membrane were 49.07 ± 4.73 a.u. (n=15) for Dronpa and 50.99 ± 5.08 a.u. (n=12) for reconstituted DMVF (Figure 1F), suggesting that both chromophores behaved similarly, and that reconstituted DMVF is suitable for pcSOFI imaging. Given the lower molecular brightness of cpDMVF, these results also indicate that reconstituted DMVF exhibits a higher POFF, which is consistent with the observed faster off-switching of cpDMVF.

Author Manuscript

To demonstrate super-resolution imaging using DMVF, we induced reconstitution of DMVF targeted to the membrane (Lyn) and to actin (LifeAct) in HeLa cells, acquired image sequences of at least 800 frames with an exposure time of 35 ms and performed pcSOFI analysis (Figure 2A, Movie S1). More specifically, cross-cumulant analysis is carried out on a pixel-by-pixel basis, obtaining a ’pcSOFI value’ for each pixel of the newly generated super-resolution image. This value is proportional to the number of labels, but also depends on the molecular brightness and the switching kinetics of the dye (Supplemental Experimental Procedures), thus bearing an analogy to an intensity value of a conventional fluorescence image. An important, open parameter for this analysis is the cumulant order. While increasing the order is more demanding regarding the source data, it results in further contrast enhancement, background reduction, and spatial resolution increase (Dertinger et al., 2010). As a simplistic estimate, the use of nth order calculations gives an n-fold improvement in resolution. We were able to obtain high quality images employing 2nd and 3rd order analysis, which provides a two- and three-fold enhancement in spatial resolution compared to conventional TIRF images (Figure 2B-K, Figure S1), corresponding to about 100 nm for the imaging equipment used. These properties allow for the identification of details that cannot be resolved in the conventional image (Figure 2C-F and 2H-K). Overall, the obtained data suggest that super-resolution imaging of PPIs is feasible using FP

Cell Rep. Author manuscript; available in PMC 2016 May 17.

Hertel et al.

Page 5

Author Manuscript

fragments derived from Dronpa. We term this strategy reconstituted fluorescence based Stochastic Optical Fluctuation Imaging (refSOFI).

Author Manuscript

To demonstrate the utility of DMVF fragments beyond the FRB/FKBP model interaction, we focused on the receptor tyrosine kinases (RTKs) HER2 and HER3. Their interaction in combination with overexpression is suggested to promote breast tumorigenesis in multiple ways (Aceto et al., 2012). Previous super-resolution microscopy has revealed that HER2 and HER3 individually form submicroscopic clusters (Kaufmann et al., 2011), suggesting that direct investigation of the interaction at super-resolution will be beneficial. HER2 and HER3 were C-terminally tagged with the DMVF fragments, expressed in HeLa cells and the formation of HER2/HER3 heterodimers was induced through a 4 h incubation with neuregulin 1 (NRG1), which is a potent HER3 ligand that activates HER2/HER3 signaling (Lee et al., 2014). While a no drug control exhibited no significant fluorescence, we found that NRG1 induced the formation of small fluorescent puncta, highlighting the locations where the interaction occurred, generally resembling the appearance of HER2 clusters shown previously (Figure S2). Super-resolution imaging of protein-protein interactions using reconstituted Venus


In order to further improve the versatility of our approach, we set out to expand the molecular tool set focusing on identifying FP fragments that exhibit a faster maturation upon association compared to DMVF, which should allow following signal-induced PPIs without much delay. The lower requirements of pcSOFI on switching kinetics allowed us to screen a wider range of FPs than those typically associated with super-resolution imaging. We discovered that Venus (Nagai et al., 2002) displays robust single-molecule fluorescence fluctuations under certain conditions, consistent with a recent study (David et al., 2012). This observation suggested that Venus may be suitable for pcSOFI. However, the high molecular brightness of Venus and the low POFF can result in a broad distribution of intensity values over the image, making it difficult to capture single-molecule fluorescence fluctuations due to the limited dynamic range intrinsic to all sensitive cameras. To identify suitable imaging conditions, we targeted Venus to the plasma membrane using the Cterminal CAAX sequence of KRas (Dedecker et al., 2012a) (Figure 3A) and examined its behavior under continuous excitation with 514 nm laser light, during which the population of fluorescent Venus decreased monotonically due to photobleaching. The initially very bright signal decreased sufficiently so that single-molecule fluctuations could be captured, which provided conditions where pcSOFI could be used to recover super-resolution information (Figure 3B-J; Figure S3). To test the scope of Venus-based pcSOFI, we additionally fused Venus to the LifeActtag and Histone H2B (Figure 3A) and were able to identify structures and complexes with enhanced resolution and contrast (Figure 3E-J), demonstrating the versatility of Venus-based pcSOFI. We next tested the suitability of Venus fragments for refSOFI. Using the well-characterized complementary fragments Venus-173-N (VN) and Venus-173-C (VC), (Shyu et al., 2006) we generated the fusions Lyn-FRB-VN, LifeAct-FRB-VN and FKBP-VC (Figure 4A). After stimulation with rapamycin, we obtained super-resolution images comparable to those obtained using DMVF fragments or wild-type Venus (Figure 4B-G; Figure S4; Movie S2).


Hertel et al.

Page 6

Author Manuscript

The observed cells exhibited an average normalized pcSOFI value of 28.62 ± 2.08 a.u. (n=16) for reconstituted Venus and 38.86 ± 11.14 a.u. (n=8) for wild-type Venus. This difference is consistent with the lower molecular brightness of the corresponding circularlypermuted Venus (about 80%) (Meng and Sachs, 2012) and implies that reconstituted Venus exhibits a similar fluctuation behavior to Venus.

Author Manuscript

The half-time of Venus complementation has been reported to be 1 h in living cells (Robida and Kerppola, 2009), during which biologically important dynamics could take place. We hypothesized that a sufficient number of fluorophores could be complemented to allow super-resolution images to be obtained prior to the half-time of complementation determined at the bulk fluorophore level. This would permit refSOFI to monitor signal-induced PPIs within this shorter time window. To test this hypothesis, we acquired image sequences of responding cells at different time points during the reconstitution process, over a period of 90 min. The average intensity was monitored to track reconstitution, while the normalized pcSOFI value was used to assess the fluctuations. A 15-fold increase in the average intensity was observed after 60 min, whereas the apparent normalized pcSOFI value rose to halfmaximum after just 20 min and started to reach a plateau phase after 30 min. The stabilization of the value is indicative of a pool of fluorophores exhibiting robust singlemolecule fluctuations sufficient for pcSOFI analysis. Accordingly, we found that refSOFI using Venus can map structures formed by PPI complexes approximately 30 minutes after the interaction occurs (Figure 4H).

Author Manuscript

The two fluorescent proteins that we have identified as suitable for refSOFI, DMVF and Venus, provide complementary characteristics for use in different contexts. Reconstituted DMVF is reversibly photoswitchable, making it generally useful for obtaining high-contrast images of complex features comprised of constitutive PPIs and robust signal-induced PPIs, and allows high-quality SOFI images to be calculated from fewer fluorescence images. On the other hand, Venus has a lower POFF and exhibits intensity fluctuations on a slower time scale, which may appear as “stable” fluorescence at the ensemble level at typical labeling densities. Thus Venus-based refSOFI may require reducing the fluorophore population for instance by photo-bleaching to permit super-resolution imaging. Nevertheless, we demonstrated that the higher rate of Venus reconstitution allows for the observation of a stochastic fraction of PPIs within 30 minutes of induction, which makes Venus highly suitable for investigating the spatial organization of acutely induced PPIs. Investigating the activation of CRAC channels using refSOFI

Author Manuscript

SOCE is a ubiquitous pathway for extracellular Ca2+-entry into cells, and has been the subject of intense research in recent years. The critical step of this process has been narrowed down to a single intermembrane interaction between two proteins. Upon depletion of intracellular Ca2+ stores, the ER Ca2+ sensor STIM1 oligomerizes and translocates to ERPM junctions, where it tethers plasma membrane-localized ORAI1, leading to the formation of functional CRAC channels within these assemblies (Deng et al., 2009; Gwozdz et al., 2008). Various approaches including FRET imaging, electron microscopy and single particle tracking have been used to analyze these assemblies (Navarro-Borelly et al., 2008; Wu et al., 2006, 2014). Recently, super-resolution imaging of ER-PM junctions suggests a


Hertel et al.

Page 7

Author Manuscript

submicroscopic size on average that does not change upon store depletion (Chang et al., 2013), therefore we reasoned that refSOFI provides a suitable strategy to resolve the locations where STIM1 and ORAI1 interact in living cells at super-resolution straightforwardly. Since the interaction starts to occur within minutes after stimulation (Muik et al., 2008), we chose to employ Venus fragments to map the STIM1/ORAI1 interactions induced by store depletion: VN was fused to the N-terminus of ORAI1 and VC to the C-terminus of STIM1 (Figure 5D), so that both fragments face the cytosol to facilitate complementation. An interaction will then trigger the reconstitution of Venus, and the complexes of STIM1, ORAI1 and reconstituted Venus then highlight membrane areas where the interaction occurred, which we will refer to as interaction puncta.

Author Manuscript

Several studies have shown that the employed labeling scheme does not affect the functionality of STIM1/ORAI1 proteins (Muik et al., 2008; Navarro-Borelly et al., 2008). Furthermore, heterologous expression of both proteins has shown to be a reliable tool in several studies to discover and characterize their interaction (Fahrner et al., 2009), since general characteristics of the interaction and subsequent puncta formation are not perturbed (Gwozdz et al., 2008). In order to generate a negative control for STIM1/ORAI1 interaction, a truncated STIM1 protein was fused to VC in place of full-length STIM1. This truncated mutant (STIM1ΔCAD) localizes properly in cells but lacks the CRAC channel activation domain (CAD) and therefore cannot bind and activate CRAC channels (Park et al., 2009).


In order to monitor the formation of STIM1/ORAI1 complexes, HeLa cells expressing the fusion constructs (along with an ER-marker comprised of rsTagRFP (Subach et al., 2010) fused to an N-terminal fragment of Cytochrome P450 (Sakaguchi et al., 1987), ERrsTagRFP, Figure 5D) were stimulated with the SERCA inhibitor thapsigargin (TG) (Treiman et al., 1998) to cause Ca2+ store depletion. The same cells were imaged before and 30 min after stimulation. Prior to stimulation, the cells exhibited a basal fluorescent signal. Using the negative-control construct, we found that the removal of the CAD domain decreased the fluorescence intensity, normalized to an expression marker (Supplemental Experimental Procedures), by about 80%, indicating that a large part of this basal signal can be attributed to specific interactions between STIM1 and ORAI1 and not to spontaneous association (Figure S5). In addition, multicolor pcSOFI (Dedecker et al., 2012a) using Dronpa and rsTagRFP for labeling showed that the vast majority of STIM1 and ORAI1 are not co-localized in the absence of stimulation (Pearson's coefficient of 0.177 ± 0.061 for STIM1 (n=13), 0.162 ± 0.093 for STIM1ΔCAD (n=9)), which is consistent with previous studies using HEK293 cells (Gwozdz et al., 2008). These findings suggest that the preformed complexes are caused by sporadic interactions between a small fraction of STIM1 and ORAI1 molecules in the basal state, thereby highlighting pre-existing junctions between the ER and plasma membrane where these interactions occur sufficiently for detection. Next, we focused on analyzing changes in the density and size of interaction puncta in individual cells upon TG treatment. To assess the benefit of super-resolution information in characterizing this PPI, we performed a direct side-by-side comparison of standard TIRF images and the corresponding refSOFI images of the same cells. In the TIRF images, TG stimulations (n=26) appeared to result in a moderate increase in puncta number (7.9 ± 0.9 to 11.5 ± 0.9 puncta per 100 μm2) as well as a large increase in the average puncta size per cell


Hertel et al.

Page 8

Author Manuscript

(0.57 ± 0.10 to 1.16 ± 0.27 μm2), where the puncta size doubled upon treatment. In contrast, super-resolution images obtained using refSOFI (Figure 5A-G) revealed a different picture. In 18 out of 26 cells, TG stimulation induced a two-fold increase in puncta density and greater-than-three-fold increase in the pcSOFI value. The non-responding cells (8 out of 26 cells) showed a significantly higher puncta density in the basal state and did not show a change in puncta density, average size or pcSOFI value upon TG stimulation (Gwozdz et al., 2008). However, the lack of response may not directly result from STIM1/ORAI1 overexpression since the parental HeLa cells show a similar variability of TG-induced calcium responses without overexpression (Figure S5). Overall, the average puncta density increased from 16.2 ± 2.2 to 21.9 ± 1.9 puncta per 100 μm2 in cells treated with TG (Figure 5H) whereas control cells receiving no drug treatment did not show any changes.

Author Manuscript

To allow a side-by-side comparison of the performance of refSOFI and BiFC-PALM for this particular experiment, we replaced the fragments of Venus with fragments of PAmCherry1, PAmCherry1-159-N and PAmCherry1-160-C (Nickerson et al., 2014), incubated the cells for several hours after addition of TG to account for the prolonged reconstitution kinetics, and performed PALM-imaging in live cells. Due to the small number of reconstituted PAmCherry molecules, we could only localize 0.163 ± 0.044 molecules per μm2 per s (n=8), which is much lower than typical results in live cells (> 25 molecules per μm2 per s; Shroff et al., 2008), and the majority of molecules were isolated and not located in punctate structures. As a consequence, the reconstructed images provided a low density (4.7 ± 0.6 puncta per 100 μm2) of very small puncta, which did not yield much information. This data illustrates that using BiFC-PALM, we cannot localize a sufficient number of reconstituted fluorophores to map the STIM1/ORAI1 interaction and perform in-depth analysis.

Author Manuscript

The enhanced resolution and contrast provided by refSOFI allowed us to better detect small puncta and resolve closely positioned puncta (Figure 5F-G). In contrast to what we observed using standard TIRF, we could not detect a significant change in average puncta size (0.192 ± 0.020 to 0.239 ± 0.035 μm2) before and after TG treatment in our experiments (Figure 5I). Additionally, the distribution of individual puncta sizes did not change substantially (Figure 5J). These data, in conjunction with similar observations that have been reported for ER-PM junctions (Chang et al., 2013; Wu et al., 2006), suggest that Ca2+ store depletion results in the formation of additional STIM1/ORAI1 interaction puncta rather than the expansion of existing interaction puncta.

Discussion Author Manuscript

In this study, we introduced refSOFI for visualizing PPIs in living cells in super-resolution. This method offers high sensitivity, high spatiotemporal resolution, and a broadly applicable platform for adapting many existing Venus-BiFC-based assays. We obtained a three-fold improvement in spatial resolution compared to the conventional resolution of our imaging setup, achieving a resolution of about 100 nm. Further optimization regarding the signal-tonoise ratio could allow for higher-order pcSOFI analysis and thus a better spatial resolution. More specifically, photochromic BiFC pairs can be engineered or further enhanced to provide a higher molecular brightness of the reconstituted FP and a more effective maturation while retaining photoswitching properties. In addition, other reported Venus-


Hertel et al.

Page 9

Author Manuscript

based BiFC assays could be tested with refSOFI, particularly BiFC pairs that have been optimized to further reduce self-association (Kodama and Hu, 2010; Ohashi et al., 2012), which could lead to a lower background depending on the experimental context.

Author Manuscript

We developed two sets of refSOFI probes that exhibit different strengths. The photoswitchable Dronpa was optimized for BiFC, and the resulting reconstituted mutant DMVF is suitable for imaging complex features comprised of constitutive or inducible longlasting PPIs using refSOFI, such as the interaction between HER2 and HER3. We also demonstrated pcSOFI and refSOFI with Venus, a highly engineered FP. We showed that PPIs can be detected within 30 min of complex formation, making Venus highly suitable for monitoring signal-induced PPIs without much delay. Moreover, assays that use Venus fragments can be easily adapted, which adds several prospective applications. Future development of the refSOFI tool set will involve expanding the color palette of suitable FP fragments. While the development of complementing fragments for photoactivatable FPs (paFPs) is challenging, our approach to identifying and enhancing Dronpa fragments for BiFC set an example.

Author Manuscript

One current limitation is that the fluorophore maturation in both refSOFI and BiFC is largely irreversible, which restricts the ability to monitor dynamic or reversible processes. It can also introduce a bias if the signal is integrated over a long period of time compared to the duration of the interaction. However, this problem should not introduce a significant bias in this study, since we observe the signal 30 min after inducing the interaction and the signals we detect after 30 minutes originate with high probability from interactions that occurred towards the beginning of the time window. Nonetheless, the development of approaches that exhibit reversibility would be beneficial and allow the investigation of dynamic PPIs or even enzyme activities in super-resolution. Another limitation is that the introduction of sensor proteins can perturb cellular signaling. Specifically for STIM1 and ORAI1, it is known that their overexpression, which has been used in several studies to investigate related signaling, can directly affect STIM1/ORAI1-signaling, resulting in the amplification of Ca2+ currents and a portion of resting cells that exhibit STIM1 puncta whose distribution did not change upon stimulation (Gwozdz et al., 2008). In our study, although expression of STIM1 and ORAI1 fusions does not result in interaction-mediated co-localization in unstimulated cells, we could detect a basal signal that is largely dependent on a functional STIM1/ORAI1 interaction interface. Future studies using a CRISPR/Cas9 based biosensor-knock-in approach (Krentz et al., 2014) could determine whether this small amount of basal interactions occur at the endogenous level.

Author Manuscript

A BiFC-PALM approach was recently developed using PAmCherry1, which involves a straightforward adaption of mCherry BiFC to PAmCherry. This method was demonstrated by imaging abundant, constitutive PPIs in fixed cells, as well as by single-molecule tracking (Nickerson et al., 2014). In addition, fragments of PAGFP have been used to localize and characterize constitutively formed protein dimers (Xia et al., 2014). Moreover, the photoactivatable FP mEos3.2 has also been used to demonstrate BiFC-PALM, predominantly in bacteria, to investigate constitutive PPIs (Liu et al., 2014). Although these PALM-based techniques currently provide greater spatial resolution enhancement than SOFI-based methods under optimal conditions, refSOFI can serve as a complementary


Hertel et al.

Page 10


strategy, since it offers several advantages that allow outperforming BiFC-PALM depending on the experimental context. First, the relaxed requirements on switching kinetics of dyes allow for use of labels that are not suitable for PALM, such as Venus, where we could take advantage of its fast maturation and follow signal-induced PPIs without much delay. Furthermore, refSOFI provides a better temporal resolution under comparable conditions (due to the shorter total exposure time required for generating a super-resolution image), a better live-cell imaging compatibility due to higher tolerance to noise and labeling density (Dertinger et al., 2013) and capability to repeatedly follow signal-induced PPIs. It has to be noted that suboptimal live-cell imaging conditions for BiFC-PALM can eventually be addressed by either higher excitation power or exposure, which results in more photodamage, and/or adjustments of parameters that effectively reduce the resolution and sensitivity. In our hands, BiFC-PALM with PAmCherry did not yield much information when applied to STIM1/ORAI1, showing only a low density of very small interaction puncta, whereas refSOFI provided an information-rich assessment of the size and density of STIM1/ORAI1 interaction puncta in living cells.


While the interaction between STIM1 and ORAI1 has been investigated by FRET, colocalization and BiFC (Navarro-Borelly et al., 2008; Wei-Lapierre et al., 2013; Yuan et al., 2009), super-resolution imaging techniques have only been applied to investigate related aspects: Single-particle tracking of STIM1 and ORAI1 has been used to study the effect of store-depletion on their diffusion behavior, showing that ORAI1 is effectively trapped in STIM1 puncta (Wu et al., 2014). Moreover, STED microscopy has been employed to directly investigate ER-PM junctions, demonstrating a submicroscopic size of about 0.05 μm2 on average and no significant change of size upon store depletion (Chang et al., 2013). Although ER-PM junctions serve as platforms for the interaction between STIM1 and ORAI1, the interaction may only occur in a subset of junctions and exhibit different dynamics (Carrasco and Meyer, 2011). For instance, it has been shown that mechanisms exist that render the translocation of STIM1 polarized or localized (Lioudyno et al., 2008; Lur et al., 2009). We used refSOFI to directly investigate STIM1/ORAI1 interactions in super-resolution, and our data indicated that Ca2+ store depletion resulted in the formation of new, closely localized interaction puncta with no significant change of the average size or substantial impact on the size distribution within the resolution of our imaging. The sizes of STIM1/ORAI1 interaction puncta are potentially limited by the number of ORAI molecules recruited to the STIM1 oligomer, by the size of the region to which they are recruited, or by the local concentration of STIM1. Overall, our observations regarding the interaction puncta follow the behavior that was reported for ER-PM junctions using super-resolution imaging (Chang et al., 2013). refSOFI is critical for these findings, since this assay provides greater precision in determining the number and size of closely positioned puncta compared to standard TIRF imaging. Thus, the ability to map PPIs in super-resolution in living cells is instrumental in providing insights into biological processes involving the formation of submicroscopic features by PPIs.


Hertel et al.

Page 11

Author Manuscript

Experimental procedures Construct generation

Author Manuscript

Standard cloning procedures were performed, including PCR with specific primers using Phusion polymerase (New England Biolabs, Ipswich, MA), enzymatic DNA restriction (Thermo Scientific, Waltham, MA), and ligation using T4 DNA ligase (New England Biolabs). DH5α E. Coli cells were employed for cloning and amplification of plasmids. For expression in mammalian cells, DNA fragments were cloned into a pcDNA3 vector (Life Technologies, Carlsbad, CA) with a modified multiple cloning site. Details are provided in the Supplemental Experimental Procedures. All constructs were verified by sequencing. Plasmids encoding ORAI1 and STIM1 were kindly provided by Dr. Paul Worley, FRB and FKBP were generous gifts of Dr. Takanari Inoue, HER2 and HER3 were generously provided by Dan Leahy (all Johns Hopkins University, Baltimore), GCaMP3 was kindly supplied by Dr. Loren Looger (Janelia Research Campus, Ashburn) and histone H2B was obtained from the Ultimate ORF collection (Life Technologies). Cloning, mutagenesis and screening of cpDronpa

Author Manuscript

Specific primers were used to generate cpDronpa inserts from a tandem construct of two Dronpa genes (Supplemental Experimental Procedures) for screening and mutagenesis. Rational and random mutagenesis were performed as previously described (Miyazaki and Takenouchi, 2002; Sawano and Miyawaki, 2000). After overnight growth of transformed JM109(DE3) cells on LB agar plates, the colonies were screened for brightness and photochromic behavior using a homebuilt colony screening system. This setup consisted of a high-power Xenon lamp (Max-302, Asahi Spectra, Torrance, CA), appropriate emission (530/40) and excitation (400/30 (on-switching) & 480/40 (fluorescence and off-switching)) filters, an EMCCD camera (Cascade 512B, Photometrics, Tucson, AZ) and a watercooling unit. All illumination and detection settings were controlled by custom software, making use of Igor Pro (WaveMetrics, Lake Oswego, OR). Bright switchable clones were picked and grown, and the plasmid DNA was extracted and sequenced (LGC Genomics, Beverly, MA). Characterization of purified Proteins Fluorescent proteins were purified following standard protocols (Supplemental Experimental Procedures) and characterized using a home-built laser setup, described previously (Moeyaert et al., 2014). Quantum yields were determined at 473 nm excitation with respect to EGFP in TN buffer (QY=0.60) and fluorescein in 0.1 M NaOH (QY=0.95). Extinction coefficients were determined according to Ward's method (Ward, 1981) using the value of EGFP (53,000 M−1cm−1) as a control.

Author Manuscript

Reversible photoswitching was measured by following both the absorbance and fluorescence emission of a fluorescent protein solution with an optical density of approximately 0.2 in TN buffer. The protein solution was illuminated for 30 steps of 60 s with 488 nm laser light, followed by 10 steps of 60 s 405 nm laser light. The laser powers used were 1.07 mW and 15.2 mW for the 405 nm and 488 nm illumination, respectively.


Hertel et al.

Page 12

Author Manuscript

Denaturation/renaturation experiments were carried out as described previously (Reid and Flynn, 1997; Supplemental Experimental Procedures). Cell culture and transfection HeLa and NIH3T3 cells were grown in DMEM growth media (Mediatech, Manassas, VA) containing 10% FBS and 1% penicillin/streptomycin and maintained at 37°C in an atmosphere of 5% CO2. Cells were seeded in sterile 35 mm glass-bottom imaging dishes and transfected at a confluency of about 70% using Lipofectamine 2000 (Life Technologies) following the manufacturer's protocol. After incubation for 24 h, the cells were washed twice with Hank's Balanced Salt Solution (HBSS, 14065, Life Technologies) buffer containing 1.26 mM Ca2+ and subsequently imaged in HBSS buffer at room temperature. Total Internal Reflection Fluorescence (TIRF) microscopy


The TIRF imaging setup consisted of an Eclipse Ti microscope (Nikon, Tokyo, Japan), a perfect focus system (Nikon), a 100x oil immersion objective (NA 1.49), an Evolve 512 EMCCD camera (Photometrics), and a laser TIRF-system (Nikon). For illumination, a 40mW 440 nm diode laser (JDSU, Milpitas, CA), a 25-mW 457-514 nm argon laser (JDSU) and a 100-mW 561 nm Sapphire diode laser (Coherent, Santa Clara, CA) were employed. Acquired images have a dimension of 512 × 512 pixels and an optical pixel size of 160 nm. The components were controlled with the software package NIS elements (Nikon). Laser power, electronic gain, and TIRF angle were held constant during all experiments. In general, time-lapse image sequences for pcSOFI analysis were captured for 30 s with an exposure time of 35 ms and no delay interval, resulting in the acquisition of about 800 frames per imaging experiment. Conventional TIRF images were acquired using exposure times of 100 - 500 ms which were kept consistent during experiments. Detailed descriptions of individual experiments are provided in the Supplemental Experimental Procedures. pcSOFI analysis

Author Manuscript

The Localizer software package (Dedecker et al., 2012b), implemented in Igor Pro (WaveMetrics), was used to perform 2nd and 3rd order SOFI analysis on the recorded image sequences as described previously (Dedecker et al., 2012a). Subsequently, the pcSOFI image was deconvolved using the built-in Richardson-Lucy algorithm (Igor Pro) with the following parameters: 3 iterations and a Gaussian PSF with a standard deviation of 1.5 pixels. Average intensity images resembling conventional TIRF images were generated by averaging the intensity values of each pixel over the whole image sequence. The pcSOFI analysis provides values for every pixel (pcSOFI values), and the information this value provides and ways it can be used to evaluate image data is described in the Supplemental Experimental Procedures. Epifluorescence microscopy The epifluorescence imaging setup consisted of an Axiovert 200M Microscope (Zeiss, Oberkochen, Germany), a 40x oil immersion objective (NA 1.3), a MicroMAX CCD Camera (Princeton Instruments, Trenton, NJ), and a XBO 75 xenon arc lamp. The


Hertel et al.

Page 13

Author Manuscript

MetaFluor Software (Molecular Devices, Sunnyvale, CA) was used to control the components. Photoactivated Localization Microscopy (PALM)

Author Manuscript

The PALM imaging setup consisted of an IX-71 inverted microscope (Olympus, Tokyo, Japan), a 60x TIRFM oil immersion objective (NA 1.45) and an IXON DU897E EMCCD camera (Andor Technology, Belfast, United Kingdom). To activate and excite PAmCherry1 simultaneously, a 50-mW CUBE™ 405-nm laser (Coherent), a 100-mW 561 nm Sapphire solid-state laser (Coherent) and a dual-band emission filter (510/19 and 620/20, Chroma Technology, Bellows Falls, VT) were used. Image sequences of 4,000 frames were recorded with an exposure time of 35 ms per frame and an imaging area of 150 × 150 pixels. For analysis and image reconstruction, we employed the localizer software package (Dedecker et al., 2012b), implemented in Igor Pro (WaveMetrics). For analysis, the default settings were used with a GLRT Insensitivity of 15 and a standard deviation of the PSF of 1.5 pixels. For reconstruction, the pixel size and the PSF model used for plotting were chosen to mimic the properties of 2nd order pcSOFI images (80 nm pixel size, Gaussian PSF model with a FHWM of 120 nm). Imaging data analysis Imaging data was analyzed using Fiji (Schindelin et al., 2012). Experiment-specific details are provided in the Supplemental Experimental Procedures. Statistical analysis

Author Manuscript

Statistically significant differences between datasets were identified using Student's t-test. Significance levels were indicated using * for P < 0.05, ** for P < 0.01 and *** for P < 0.001. Point and bar graphs illustrate the mean value with error bars showing the standard error of the mean (s.e.m.).

Supplementary Material Refer to Web version on PubMed Central for supplementary material.

Acknowledgements

Author Manuscript

We thank Brian Ross, Carla Coltharp, Jie Xiao, Scot Kuo and the staff of The Johns Hopkins University School of Medicine Institute for Basic Biomedical Sciences microscope facility for technical assistance. We are further grateful to Sohum Mehta for carefully reading the manuscript. This work was supported by a NIH Director's Pioneer Award DP1 CA174423 (to J.Z.). P.D. thanks the Research Foundation-Flanders (FWO Vlaanderen) for a postdoctoral fellowship and the KULeuven for a BOF-ZAP grant. S.D. thanks the Flemish Agency for Innovation by Science and Technology (IWT) for a doctoral fellowship.

References Aceto N, Duss S, Macdonald G, Meyer DS, Roloff T-C, Hynes NE, Bentires-Alj M. Co-expression of HER2 and HER3 receptor tyrosine kinases enhances invasion of breast cells via stimulation of interleukin-8 autocrine secretion. Breast Cancer Res. 2012; 14:R131. [PubMed: 23062209] Ando R, Mizuno H, Miyawaki A. Regulated fast nucleocytoplasmic shuttling observed by reversible protein highlighting. Science. 2004; 306:1370–1373. [PubMed: 15550670]


Hertel et al.

Page 14

Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Betzig E, Patterson GH, Sougrat R, Lindwasser OW, Olenych S, Bonifacino JS, Davidson MW, Lippincott-Schwartz J, Hess HF. Imaging intracellular fluorescent proteins at nanometer resolution. Science. 2006; 313:1642–1645. [PubMed: 16902090] Bonetta L. Protein-protein interactions: Interactome under construction. Nature. 2010; 468:851–854. [PubMed: 21150998] Carrasco S, Meyer T. STIM proteins and the endoplasmic reticulum-plasma membrane junctions. Annu. Rev. Biochem. 2011; 80:973–1000. [PubMed: 21548779] Chang C-L, Hsieh T-S, Yang TT, Rothberg KG, Azizoglu DB, Volk E, Liao J-C, Liou J. Feedback regulation of receptor-induced Ca2+ signaling mediated by E-Syt1 and Nir2 at endoplasmic reticulum-plasma membrane junctions. Cell Rep. 2013; 5:813–825. [PubMed: 24183667] David CC, Dedecker P, De Cremer G, Verstraeten N, Kint C, Michiels J, Hofkens J. Spectroscopic characterization of Venus at the single molecule level. Photochem. Photobiol. Sci. 2012; 11:358– 363. [PubMed: 22212707] Dedecker P, Mo GCH, Dertinger T, Zhang J. Widely accessible method for superresolution fluorescence imaging of living systems. Proc. Natl. Acad. Sci. U. S. A. 2012a; 109:10909–10914. [PubMed: 22711840] Dedecker P, Duwé S, Neely R, Zhang J. Localizer: fast, accurate, open-source, and modular software package for superresolution microscopy. J. Biomed. Opt. 2012b; 17 Dedecker P, De Schryver FC, Hofkens J. Fluorescent proteins: Shine on, you crazy diamond. J. Am. Chem. Soc. 2013; 135:2387–2402. [PubMed: 23317378] Deng X, Wang Y, Zhou Y, Soboloff J, Gill DL. STIM and Orai: dynamic intermembrane coupling to control cellular calcium signals. J. Biol. Chem. 2009; 284:22501–22505. [PubMed: 19473984] DeRose R, Miyamoto T, Inoue T. Manipulating signaling at will: chemically-inducible dimerization (CID) techniques resolve problems in cell biology. Pflugers Arch. 2013; 465:409–417. [PubMed: 23299847] Dertinger T, Colyer R, Vogel R, Enderlein J, Weiss S. Achieving increased resolution and more pixels with Superresolution Optical Fluctuation Imaging (SOFI). Opt. Express. 2010; 18:18875–18885. [PubMed: 20940780] Dertinger T, Pallaoro A, Braun G, Ly S, Laurence TA, Weiss S. Advances in superresolution optical fluctuation imaging (SOFI). Q. Rev. Biophys. 2013; 46:210–221. [PubMed: 23672771] Duwé S, De Zitter E, Gielen V, Moeyaert B, Vandenberg W, Grotjohann T, Clays K, Jakobs S, Van Meervelt L, Dedecker P. Expression-Enhanced Fluorescent Proteins Based on Enhanced Green Fluorescent Protein (EGFP) for Super-Resolution Microscopy. ACS Nano. 2015 Fahrner M, Muik M, Derler I, Schindl R, Fritsch R, Frischauf I, Romanin C. Mechanistic view on domains mediating STIM1-Orai coupling. Immunol. Rev. 2009; 231:99–112. [PubMed: 19754892] Fernández-Dueñas V, Llorente J, Gandía J, Borroto-Escuela DO, Agnati LF, Tasca CI, Fuxe K, Ciruela F. Fluorescence resonance energy transfer-based technologies in the study of protein-protein interactions at the cell surface. Methods. 2012; 57:467–472. [PubMed: 22683304] Gayda S, Nienhaus K, Nienhaus GU. Mechanistic insights into reversible photoactivation in proteins of the GFP family. Biophys. J. 2012; 103:2521–2531. [PubMed: 23260054] Gustafsson MGL. Nonlinear structured-illumination microscopy: wide-field fluorescence imaging with theoretically unlimited resolution. Proc. Natl. Acad. Sci. U. S. A. 2005; 102:13081–13086. [PubMed: 16141335] Gwozdz T, Dutko-Gwozdz J, Zarayskiy V, Peter K, Bolotina VM. How strict is the correlation between STIM1 and Orai1 expression, puncta formation, and ICRAC activation? Am. J. Physiol. Cell Physiol. 2008; 295:C1133–C1140. [PubMed: 18768920] Han R, Li Z, Fan Y, Jiang Y. Recent advances in super-resolution fluorescence imaging and its applications in biology. J. Genet. Genomics. 2013; 40:583–595. [PubMed: 24377865] Hell SW. Far-field optical nanoscopy. Science. 2007; 316:1153–1158. [PubMed: 17525330] Kaufmann R, Müller P, Hildenbrand G, Hausmann M, Cremer C. Analysis of Her2/neu membrane protein clusters in different types of breast cancer cells using localization microscopy. J. Microsc. 2011; 242:46–54. [PubMed: 21118230]


Hertel et al.

Page 15


Kodama Y, Hu C-D. An improved bimolecular fluorescence complementation assay with a high signalto-noise ratio. Biotechniques. 2010; 49:793–805. [PubMed: 21091444] Kodama Y, Hu C-D. Bimolecular fluorescence complementation (BiFC): a 5-year update and future perspectives. Biotechniques. 2012; 53:285–298. [PubMed: 23148879] Koh GCKW, Porras P, Aranda B, Hermjakob H, Orchard SE. Analyzing protein-protein interaction networks. J. Proteome Res. 2012; 11:2014–2031. [PubMed: 22385417] Krentz, N. a J.; Nian, C.; Lynn, FC. TALEN/CRISPR-mediated eGFP knock-in add on at the OCT4 locus does not impact differentiation of human embryonic stem cells towards endoderm. PLoS One. 2014; 9:e114275. [PubMed: 25474420] Lee CYF, Lin Y, Bratman SV, Feng W, Kuo AH, Scheeren FA, Engreitz JM, Varma S, West RB, Diehn M. Neuregulin autocrine signaling promotes self-renewal of breast tumor-initiating cells by triggering HER2/HER3 activation. Cancer Res. 2014; 74:341–352. [PubMed: 24177178] Lee YR, Park J-H, Hahm S-H, Kang L-W, Chung JH, Nam K-H, Hwang KY, Kwon IC, Han YS. Development of bimolecular fluorescence complementation using Dronpa for visualization of protein-protein interactions in cells. Mol. Imaging Biol. 2010; 12:468–478. [PubMed: 20373040] Lioudyno MI, Kozak JA, Penna A, Safrina O, Zhang SL, Sen D, Roos J, Stauderman KA, Cahalan MD. Orai1 and STIM1 move to the immunological synapse and are up-regulated during T cell activation. Proc. Natl. Acad. Sci. U. S. A. 2008; 105:2011–2016. [PubMed: 18250319] Liu Z, Xing D, Su QP, Zhu Y, Zhang J, Kong X, Xue B, Wang S, Sun H, Tao Y, et al. Super-resolution imaging and tracking of protein-protein interactions in sub-diffraction cellular space. Nat. Commun. 2014; 5:4443. [PubMed: 25030837] Lur G, Haynes LP, Prior IA, Gerasimenko OV, Feske S, Petersen OH, Burgoyne RD, Tepikin AV. Ribosome-free Terminals of Rough ER Allow Formation of STIM1 Puncta and Segregation of STIM1 from IP3 Receptors. Curr. Biol. 2009; 19:1648–1653. [PubMed: 19765991] Meng F, Sachs F. Orientation-based FRET sensor for real-time imaging of cellular forces. J. Cell Sci. 2012; 125:743–750. [PubMed: 22389408] Miyazaki K, Takenouchi M. Creating random mutagenesis libraries using megaprimer PCR of whole plasmid. Biotechniques. 2002; 33:1033–1038. [PubMed: 12449380] Moeyaert B, Nguyen Bich N, De Zitter E, Rocha S, Clays K, Mizuno H, van Meervelt L, Hofkens J, Dedecker P. Green-to-Red Photoconvertible Dronpa Mutant for Multimodal Super-resolution Fluorescence Microscopy. ACS Nano. 2014; 8:1664–1673. [PubMed: 24410188] Muik M, Frischauf I, Derler I, Fahrner M, Bergsmann J, Eder P, Schindl R, Hesch C, Polzinger B, Fritsch R, et al. Dynamic coupling of the putative coiled-coil domain of ORAI1 with STIM1 mediates ORAI1 channel activation. J. Biol. Chem. 2008; 283:8014–8022. [PubMed: 18187424] Nagai T, Ibata K, Park ES, Kubota M, Mikoshiba K, Miyawaki A. A variant of yellow fluorescent protein with fast and efficient maturation for cell-biological applications. Nat. Biotechnol. 2002; 20:87–90. [PubMed: 11753368] Navarro-Borelly L, Somasundaram A, Yamashita M, Ren D, Miller RJ, Prakriya M. STIM1-Orai1 interactions and Orai1 conformational changes revealed by live-cell FRET microscopy. J. Physiol. 2008; 586:5383–5401. [PubMed: 18832420] Nickerson A, Huang T, Lin L-J, Nan X. Photoactivated Localization Microscopy with Bimolecular Fluorescence Complementation (BiFC-PALM) for Nanoscale Imaging of Protein-Protein Interactions in Cells. PLoS One. 2014; 9:e100589. [PubMed: 24963703] Ohashi K, Kiuchi T, Shoji K, Sampei K, Mizuno K. Visualization of cofilin actin and Ras-Raf interactions by bimolecular fluorescence complementation assays using a new pair of split Venus fragments. Biotechniques. 2012; 52:45–50. [PubMed: 22229727] Park CY, Hoover PJ, Mullins FM, Bachhawat P, Covington ED, Raunser S, Walz T, Garcia KC, Dolmetsch RE, Lewis RS. STIM1 clusters and activates CRAC channels via direct binding of a cytosolic domain to Orai1. Cell. 2009; 136:876–890. [PubMed: 19249086] Reid BG, Flynn GC. Chromophore formation in green fluorescent protein. Biochemistry. 1997; 36:6786–6791. [PubMed: 9184161] Remington SJ. Fluorescent proteins: maturation, photochemistry and photophysics. Curr. Opin. Struct. Biol. 2006; 16:714–721. [PubMed: 17064887]


Hertel et al.

Page 16


Riedl J, Crevenna AH, Kessenbrock K, Yu JH, Neukirchen D, Bista M, Bradke F, Jenne D, Holak TA, Werb Z, et al. Lifeact: a versatile marker to visualize F-actin. Nat. Methods. 2008; 5:605–607. [PubMed: 18536722] Robida A, Kerppola T. Bimolecular fluorescence complementation analysis of inducible protein interactions: effects of factors affecting protein folding on fluorescent protein fragment. J. Mol. Biol. 2009; 394:391–409. [PubMed: 19733184] Rust MJ, Bates M, Zhuang X. Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM). Nat. Methods. 2006; 3:793–795. [PubMed: 16896339] Sakaguchi M, Mihara K, Sato R. A short amino-terminal segment of microsomal cytochrome P-450 functions both as an insertion signal and as a stop-transfer sequence. EMBO J. 1987; 6:2425– 2431. [PubMed: 2822391] Sawano A, Miyawaki A. Directed evolution of green fluorescent protein by a new versatile PCR strategy for site-directed and semi-random mutagenesis. Nucleic Acids Res. 2000; 28:E78. [PubMed: 10931937] Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, Preibisch S, Rueden C, Saalfeld S, Schmid B, et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods. 2012; 9:676–682. [PubMed: 22743772] Shroff H, Galbraith CG, Galbraith JA, Betzig E. Live-cell photoactivated localization microscopy of nanoscale adhesion dynamics. Nat. Methods. 2008; 5:417–423. [PubMed: 18408726] Shyu J, Liu H, Deng X, Hu C-D. Identification of new fluorescent protein fragments for bimolecular fluorescence complementation analysis under physiological conditions. Biotechniques. 2006; 40 Subach FV, Zhang L, Gadella TWJ, Gurskaya NG, Lukyanov KA, Verkhusha VV. Red fluorescent protein with reversibly photoswitchable absorbance for photochromic FRET. Chem. Biol. 2010; 17:745–755. [PubMed: 20659687] Sun Y, Rombola C, Jyothikumar V, Periasamy A. Förster resonance energy transfer microscopy and spectroscopy for localizing protein-protein interactions in living cells. Cytometry. A. 2013; 83:780–793. [PubMed: 23813736] Treiman M, Caspersen C, Christensen SB. A tool coming of age: Thapsigargin as an inhibitor of sarcoendoplasmic reticulum Ca2+-ATPases. Trends Pharmacol. Sci. 1998; 19:131–135. [PubMed: 9612087] Ward, WW. In Bioluminescence and Chemiluminescence: Basic Chemistry and Analytical Applications. DeLuca, MA.; McElroy, WD., editors. Academic Press; New York: 1981. p. 235-242. Wei-Lapierre L, Carrell EM, Boncompagni S, Protasi F, Dirksen RT. Orai1-dependent calcium entry promotes skeletal muscle growth and limits fatigue. Nat. Commun. 2013; 4:2805. [PubMed: 24241282] Wu MM, Buchanan J, Luik RM, Lewis RS. Ca2+ store depletion causes STIM1 to accumulate in ER regions closely associated with the plasma membrane. J. Cell Biol. 2006; 174:803–813. [PubMed: 16966422] Wu MM, Covington ED, Lewis RS. Single-molecule analysis of diffusion and trapping of STIM1 and Orai1 at ER-plasma membrane junctions. Mol. Biol. Cell. 2014 Xia P, Liu X, Wu B, Zhang S, Song X, Yao PY, Lippincott-Schwartz J, Yao X. Superresolution imaging reveals structural features of EB1 in microtubule plus-end tracking. Mol. Biol. Cell. 2014; 25:4166–4173. [PubMed: 25355949] Yuan JP, Zeng W, Dorwart MR, Choi Y-J, Worley PF, Muallem S. SOAR and the polybasic STIM1 domains gate and regulate Orai channels. Nat. Cell Biol. 2009; 11:337–343. [PubMed: 19182790]


Hertel et al.

Page 17

Author Manuscript Author Manuscript Figure 1. Characterization of an enhanced cpDronpa variant cpDMVF-181 and development of its BiFC


(A) Reversible photoswitching of Dronpa and cpDMVF upon illumination with 488 and 405 nm laser light. (B) Fluorescence time-course of the renaturation of denatured Dronpa and cpDMVF. The decrease of measured fluorescence from cpDMVF, observable after approximately 200 seconds, is caused by fluorescence off-switching. (C) Fluorescence time-course of the renaturation of denatured and reduced Dronpa and cpDMVF. (D) Schematic representation of the DMVF-based constructs used in this study. (E) Time-course of DMVF reconstitution in HeLa cells upon rapamycin (Rap, 100 nM) induced heterodimerization of Lyn-FRB-DMVF-181-N and FKBP-DMVF-181-C. The maturation process is reflected by the normalized fluorescence intensity of reconstituted DMVF (rcDMVF) in responding cells (n≥13; data shown as mean ± s.e.m.). (F) Comparison of the fluctuation behavior of Lyn-targeted Dronpa (Lyn-FRB*FKBPDronpa) and reconstituted DMVF (Lyn-FRB*FKBP-rcDMVF) 4h after induction with rapamycin by means of the normalized 2nd order SOFI value (n=15 for Dronpa and n=12 for rcDMVF; data shown as mean ± s.e.m.). (G) Scatter plot showing the comparison of the basal signal and the signal after rapamycininduced reconstitution (4 h) of DMVF in individual HeLa Cells expressing FKBPDMVF-181-C and Lyn-FRB-DMVF-181-N or LifeAct-FRB-DMVF-181-N, respectively, normalized to the fluorescence of TagRFP-T-CAAX serving as an expression marker (Lyn: n=133 for negative and n=130 for positive control; LifeAct: n=123 for negative and n=151 for positive control; black horizontal line indicates mean ± s.e.m.).


Hertel et al.

Page 18

Author Manuscript Author Manuscript Author Manuscript Figure 2. Reconstituted DMVF for refSOFI

Author Manuscript

(A) Schematic representation of the underlying principle of refSOFI. Two proteins of interest (POIs) are fused to non-fluorescent fragments of a FP suitable for refSOFI. Upon interaction, the FP is reconstituted, exhibits intensity fluctuations and can be imaged in super-resolution by acquiring an image sequence and applying pcSOFI analysis. (B) Representative pcSOFI image (3rd order analysis, triple pixel density) of Lyn-FRB/ FKBP/reconstituted DMVF complexes in HeLa cells. (C) to (F) are expansions of (B), displaying a side by side comparison of a conventional TIRF image (C), a 2nd order pcSOFI image (double pixel density, D), a 3rd order pcSOFI image (E) and the final deconvolved image (F), illustrating the increase in spatial resolution, enhanced contrast and a reduction of background signal.


Hertel et al.

Page 19

Author Manuscript

(G) Representative pcSOFI image (3rd order analysis, triple pixel density) of LifeAct-FRB/ FKBP/reconstituted DMVF complexes in HeLa cells. (H) to (K) are expansions of (G) analogous to (C) to (F). (Scale bar: 2 μm in all expansions) See also Figure S1, S2 and Movie S1.

Author Manuscript Author Manuscript Author Manuscript Cell Rep. Author manuscript; available in PMC 2016 May 17.

Hertel et al.

Page 20

Author Manuscript Author Manuscript Author Manuscript

Figure 3. Super-resolution imaging using Venus

Author Manuscript

(A) Schematic representation of the Venus-constructs used for super-resolution imaging. (B) Representative pcSOFI TIRF image (3rd order) of a HeLa cell expressing Venus-CAAX. (C) and (D) show expansions of an area of (B) allowing a side-by-side comparison of the average fluorescence image and the corresponding pcSOFI image, which features triple pixel density. (E) Representative pcSOFI TIRF image (3rd order) of a HeLa cell expressing LifeAct-Venus. (F) and (G) display expansions of (E) analogous to (C) and (D). (H) Representative pcSOFI image (near-TIRF illumination, 3rd order) of a nucleus of a NIH3T3 cell expressing histone H2B-Venus. (I) and (J) display expansions of (H) analogous to (C) and (D). (Scale bar: 1 μm in all expansions) See also Figure S3.


Hertel et al.

Page 21

Author Manuscript Author Manuscript Author Manuscript

Figure 4. Reconstituted Venus for refSOFI

Author Manuscript

(A) Schematic representation of the Venus-based constructs used in this study. (B) Representative pcSOFI TIRF image (3rd order) of Lyn-FRB*FKBP/reconstituted Venus in HeLa cells 2 h after rapamycin treatment. (C) and (D) show expansions of (B) allowing a side-by-side comparison of the average fluorescence image and the corresponding pcSOFI image. (E) Representative pcSOFI TIRF image (3rd order) of LifeAct-FRB*FKBP/reconstituted Venus in Hela cells 2 h after rapamycin treatment. (F) and (G) display expansions of (E) analogous to (C) and (D). (Scale bar: 1 μm in all expansions) (H) Time-course of average intensity and normalized pcSOFI value of cells expressing FKBP-VC and Lyn-FRB-VN after addition of rapamycin (100 nM at × = 0). The data were normalized to the respective average values at the time × = 0 (n≥9, data shown as mean ± s.e.m.). It has to be noted that the observed broad increase of the apparent normalized value mainly results from the initial occurrence of fluctuations and the high impact of noise due to the low population of intact Venus molecules at the start. See also Figure S4 and Movie S2.


Hertel et al.

Page 22

Author Manuscript Author Manuscript Author Manuscript Figure 5. Applying refSOFI to investigate the interactions between STIM1 and ORAI1

Author Manuscript

(A) Representative pcSOFI image (2nd order) of STIM1/ORAI1/reconstituted Venus complexes in an unstimulated HeLa cell expressing STIM1-VC, VN-ORAI1 and ERrsTagRFP (not shown). (B) Image shown in (A) scaled to the color scale of (C) for direct comparison. (C) Image of the cell shown in (B) 30 min after treatment with 1 μM Thapsigargin (TG). (D) Schematic representation of the constructs used in these experiments. (E) Multicolor pcSOFI image displaying the localization of STIM1/ORAI1 complexes (green) and the ER (red) in this cell 30 min after TG addition. (F) Expansion of an area of the average florescence image for comparison with the corresponding pcSOFI image in (G) (Scale bar: 1 μm). All images were acquired under TIRF conditions.


Hertel et al.

Page 23

Author Manuscript

(H) Bar graph displaying the interaction puncta density of individual cells before and 30 min after treatment with 1 μM TG (n=26), illustrating the difference between the analysis of standard TIRF images and the corresponding refSOFI images. (I) Analogous bar graph showing the average interaction puncta size of individual cells under the indicated conditions. Data shown as mean + s.e.m.. (J) Normalized frequency of the sizes of the individual puncta detected in HeLa cells using refSOFI presented in (I), before (n=3301) and 30 minutes after stimulation with TG (n=4173). A bin size of 0.025 μm2 is used and 90% of the dataset is displayed for clarity. See also Figure S5.

Author Manuscript Author Manuscript Author Manuscript Cell Rep. Author manuscript; available in PMC 2016 May 17.

Nanoscale optomechanical actuators for controlling mechanotransduction in living cells.

Spectroscopic techniques for monitoring protein interactions in living cells.

Active organization of membrane constituents in living cells.

Optimization of molecular organization and nanoscale morphology for high performance low bandgap polymer solar cells.

STED microscopy for nanoscale imaging in living brain slices.

Interactions between semiconductor nanowires and living cells.

Direct temperature mapping of nanoscale plasmonic devices.

Measurement of nanoscale three-dimensional diffusion in the interior of living cells by STED-FCS.

Concentration- and chromosome-organization-dependent regulator unbinding from DNA for transcription regulation in living cells.

Interactions between living and zinc-fixed cells in culture.

Simultaneous Nanoscale Surface Charge and Topographical Mapping.

Nanoscale organization of human GnRH-R on human bladder cancer cells.

Nanoscale label-free bioprobes to detect intracellular proteins in single living cells.

Optimizing fluorescent protein trios for 3-Way FRET imaging of protein interactions in living cells.

Nanoscale compositional mapping of cells, tissues, and polymers with ringing mode of atomic force microscopy.

The nanoscale organization of signaling domains at the plasma membrane.

Visualizing mechanical modulation of nanoscale organization of cell-matrix adhesions.

Recent progress in electromagnetic field mapping at the nanoscale.

Nanoscale carrier multiplication mapping in a Si diode.

Photoactivated localization microscopy with bimolecular fluorescence complementation (BiFC-PALM) for nanoscale imaging of protein-protein interactions in cells.

Nanoscale probing of image-dipole interactions in a metallic nanostructure.

Correlated 3D Nanoscale Mapping and Simulation of Coupled Plasmonic Nanoparticles.

Large-Scale Arrays of Bowtie Nanoaperture Antennas for Nanoscale Dynamics in Living Cell Membranes.

Visualization of microtubule organization and dynamics in living Arabidopsis embryonic cells.