StayGold variants for molecular fusion and membrane targeting applications

doi:10.21203/rs.3.rs-2941917/v1

Download PDF

Article

StayGold variants for molecular fusion and membrane targeting applications

https://doi.org/10.21203/rs.3.rs-2941917/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 30 Nov, 2023

Read the published version in Nature Methods →

Version 1

posted

You are reading this latest preprint version

Although StayGold is a bright and highly photostable fluorescent protein (FP), its obligate dimer formation may prevent its application in molecular fusion and membrane targeting. With the objective of attaining monovalent as well as bright and photostable labeling, we engineered tandem dimers of StayGold to be dispersible. On the basis of the crystal structure of this FP, we disrupted the dimer interface to develop monomeric variants of StayGold. We applied the new StayGold tools to live cell imaging experiments using spinning-disk laser scanning confocal microscopy or structured illumination microscopy. We achieved cell-wide, high–spatiotemporal-resolution, and sustained imaging of subcellular dynamic events, including the targeting of endogenous condensin I to mitotic chromosomes at the onset of mitosis, the movement of the Golgi apparatus and its membranous derivatives along microtubule networks, the distribution of cortical filamentous actin near the plasma membrane, and the remolding of cristae membranes within mobile mitochondria.

Biological sciences/Biological techniques/Imaging/Fluorescence imaging

Biological sciences/Biological techniques/Sensors and probes/Fluorescent proteins

Biological sciences/Cell biology/Cellular imaging/Super-resolution microscopy

Biological sciences/Biological techniques/Microscopy/Wide-field fluorescence microscopy

In live fluorescence imaging, an increase in illumination intensity (irradiance) results in an increase in brightness (photon budget), thereby improving the spatiotemporal resolution of the observation. However, elevated irradiance naturally leads to photobleaching of fluorescent dyes and/or photodamage of observed cells. If the photobleaching problem could be solved, how would the spatial and temporal scales of fluorescence imaging be extended? In a previous study, we attempted to answer this question by developing a bright and highly photostable fluorescent protein (FP) that we provocatively named StayGold and demonstrated its usefulness by imaging a cell’s endoplasmic reticulum (ER) and mitochondria with enhanced spatiotemporal resolution and an extended observation period¹. Being a dimer, this FP was expressed as luminal soluble markers of these organelles. In the present study, to further expand the applications of this FP to include attachment to membranes or naturally oligomeric proteins, we aimed to develop techniques that enable monovalent tagging with StayGold.

We first used a compromised but powerful approach to generate tandem dimer (td) constructs² of StayGold with various adaptors and flexible linkers. Although this approach can, in principle, achieve monovalent tagging, some td constructs were found to exhibit cohesive tendencies under physiological conditions by our fluorescence-based technology detecting protein-protein interaction (Fluoppi) assay³ as well as the organized smooth ER (OSER) assay^4–6. Therefore, we optimized the adaptors and linkers to create td variants that showed good dispersibility.

Our second approach was a directed evolution to drive the monomerization of StayGold. Although we have been refining the crystal structure of StayGold to elucidate the molecular mechanism responsible for its outstanding photostability, we here characterized the dimeric structure and, on the basis of the information, we introduced mutations into the dimer interface to disrupt the dimerization. Although breaking the dimer interface of StayGold without losing its high photostability and brightness was really challenging, we eventually produced practically useful monomeric variants that exhibit dispersibility in combination with high brightness and photostability.

In our previous study, to achieve high-speed super-resolution or volumetric imaging for extended times, we used structured illumination microscopy (SIM)⁷ and spinning-disk super-resolution microscopy (SDSRM)⁸, both of which can demonstrate the best features of StayGold¹. In the present study, continuous sustainable observation of molecules and membranes labeled with the new StayGold variants enabled us to visualize subcellular dynamics more comprehensively and quantitatively in space and time than before.

Assessment of monomericity or dispersibility of FPs. A previous OSER assay by Cranfill et al. investigated the monomeric quality of various FPs commonly used for molecular fusion applications⁶; monomeric FPs (mFPs) were basically characterized to show high OSER scores (percentages of whorl-free cells). We performed an OSER assay on 8 common FPs (Fig. 1). We first confirmed the effect of the A206K mutation for monomerization of Aequorea GFP variants⁹. For example, mEGFP scored 89.8%, which was indeed higher than the score of its parent EGFP (83.4%). A larger difference was observed between mVenus and Venus, which scored 92.8% and 55.9%, respectively. Among three mFPs that have been characterized to be very bright, mClover3¹⁰ gave a high score (87.0%), whereas mNeonGreen¹¹ and mGreenLantern¹² showed relatively modest scores (74.2% and 83.1%, respectively). Remarkably, tdTomato¹³ showed a low score (57.7%), consistent with that reported by Cranfill et al. for this tdFP (57.6%)⁶.

A substantial number of whorls tend to appear in very bright cells in OSER assays even with mFPs⁶; the ER membrane should have a capacity for exogenous protein accumulation, and high-level expression of CytERM-FP exceeding the capacity might produce whorl structures efficiently. In fact, the results suggested a positive correlation between the degree of whorl appearance and the expression level of CytERM-Venus or CytERM-tdTomato. Although such gross overexpression artifacts have been eliminated deliberately, a complicated issue of how the threshold should be set has emerged.

To evaluate FP monomericity or dispersibility on the basis of a different physico-chemical principle, we took advantage of Fluoppi, the genetically encoded protein–protein interaction visualization system that harnesses the dynamics of condensed liquid-phase transitions (Extended Data Fig. 1)³. We developed a new method in which an FP of interest is simply fused to the Phox and Bem1p (PB1) domain of p62/SQSTM1. After transfection into cultured cells, oligomerization or assembly of the FP and homo-oligomerization of PB1 should result in crosslinking to form liquid-phase droplets that emit green fluorescence in the cytoplasmic compartment (Supplementary Fig. 1). We fused eight common FPs to the C-terminus of PB1 for the assessment of their monomericity (Extended Data Fig. 2a) and obtained results similar to those of the OSER assay (Fig. 1) except that PB1-mNeonGreen produced a substantial number of puncta.

oxStayGold. We previously carried out combinatorial saturation mutagenesis on StayGold for cysteine residues. Simultaneous C174I and C208I mutations combined with an additional mutation H169Y led to the development of oxStayGold¹. This variant of StayGold was engineered to efficiently label the ER from the inside, where the oxidizing environment might cause the formation of unnecessary disulfide bonds during folding of the b-barrel⁵. Because oxStayGold was found to label all subcellular components including the cytoplasm, very brightly, it has replaced StayGold in many experiments. However, oxStayGold remains a dimer.

Tandem dimer constructs. The original td construct, tdStayGold, was composed of n1, StayGold, c4, a 116-amino-acid linker (EV linker), n1, and StayGold (Extended Data Fig. 3), where n1 and c4 are adaptors that facilitate fusions to the N- and C-termini of StayGold, respectively¹. In the OSER assay, the fusion of tdStayGold to CytERM labeled the ER network with moderate brightness and led to a very low propensity to form ER whorls, with an OSER score of 95.0% (Fig. 1)¹. Subsequently, oxStayGold was substituted in the td construct to create tdoxStayGold (Extended Data Fig. 3), which consistently achieved brighter molecular fusions than tdStayGold. However, we found that CytERM-tdoxStayGold showed an undesirable score (58.3%) (Fig. 1). In Fluoppi assays, PB1-StayGold, PB1-tdStayGold, and PB1-tdoxStayGold scored 57.7%, 85.8%, and 80.0%, respectively, roughly corroborating the OSER assay results (Extended Data Fig. 2a).

To improve the dispersibility of tdoxStayGold, we reexamined the polypeptide linker between the two copies of (ox)StayGold. First, we trimmed the EV linker (116 residues) into three polypeptides that spanned 32, 64, and 97 amino acids. However, none of the td constructs with these shortened EV linkers (td2oxStayGold, td3oxStayGold, and td4oxStayGold, respectively) showed improved OSER or Fluoppi scores. We subsequently attempted to harness the linker used for generating tdTomato². The use of a 21-residue linker resulted in the generation of td5StayGold and td5oxStayGold (Extended Data Fig. 3), both of which gave high OSER scores (91.5% and 88.5%, respectively) (Fig. 1) and high Fluoppi scores (80.0% and 97.0%, respectively) (Extended Data Fig. 2a).

Monomeric versions. We determined the crystal structure of StayGold to 1.56 Å resolution (Fig. 2, Extended Data Fig. 4, Supplementary Table 1). The solved structure also shows the dimer formation of this FP as well as an environment around the chromophore. In the present study, we focused on the dimeric interface to adopt the monomerization approach² using (n1)oxStayGold as a starting material (Extended Data Fig. 3). We introduced mutations into the interface to generate dozens of monomeric versions of (n1)oxStayGold. Because the dimerization might be responsible for the outstanding photostability of StayGold, we initially suspected that creating a highly photostable and bright variant that is truly monomeric would be difficult. In fact, we found that complete disruption of the interactions at the interface resulted in the generation of only light-sensitive products that were hardly recovered by subsequent random mutagenesis.

We subsequently adopted a moderate approach to monomerization. We comprehensively prepared several libraries of (n1)oxStayGold variants that carried different partial mutations at the interface. In one attempt, we focused on threonine substitution at Pro¹⁵¹ and Leu¹⁵⁵ (Fig. 2) and screened candidates in a library of (n1)oxStayGold P151T/L155T iteratively with multiple cycles of random mutagenesis to produce QC2-6 (Extended Data Fig. 3). This variant contained four mutations (P151T, L155T, N132D, and K162E) relative to (n1)oxStayGold. When observing the fluorescence in the cytosolic and nuclear compartments of transfected cells under intense illumination, we confirmed that QC2-6 was highly photostable and bright. When QC2-6 was expressed in bacteria and purified, the protein product showed monomer-like behavior in pseudo-native SDS-PAGE analysis (Supplementary Fig. 2). Transfection of CytERM-QC2-6 into HeLa cells generated only a small number of ER whorls in an OSER assay (score: 85.7%) (Fig. 1). We noticed, however, that PB1-QC2-6 yielded a substantial number of puncta in a Fluoppi assay (score: 59.6%) (Extended Data Fig. 2a). With this unacceptable result, we decided to drive the monomerization further on the basis of QC2-6. Performing combinatorial saturation mutagenesis at Tyr¹⁸⁷, Arg¹⁴⁴, and Thr¹⁵⁵ and screening for brightness, photostability and monomericity yielded QC2-6 FIQ, which contained Y187F, R144I, and T155Q mutations relative to QC2-6 (Extended Data Fig. 3). QC2-6 FIQ gave high scores for both OSER (92.3%) (Fig. 1) and Fluoppi (100%) (Extended Data Fig. 2a) assays.

A new appendage at the StayGold C-terminus. In one experiment to amplify the full length of QC2-6, we accidentally used a defective reverse primer to introduce a mutation at the termination codon. The translation readthrough led to the addition of a short C-terminal tail (PT) composed of YSRTKLE. Interestingly, this protein product QC2-6(PT) (Extended Data Fig. 3) showed high scores in OSER (82.6%) (Fig. 1) and Fluoppi (100%) (Extended Data Fig. 2a) assays. Because the StayGold dimeric interface appears to involve the carboxyl group of the end residue (Leu²¹⁷), such a simple extension might improve the monomericity of QC2-6. However, the reality is complicated. For example, c4 did not replace PT in this situation; c4 works well as an adaptor but not as a tail at the C-terminus, in principle. Assuming that QC2-6(PT) could be a potentially useful byproduct, we considered its further characterization worthwhile (see below).

We also found that PT can be used as another adaptor for fusion at the StayGold C-terminus. By replacing c4 with PT in td5oxStayGold, we generated td8oxStayGold (Extended Data Fig. 3). In addition, we performed saturation mutagenesis on td8oxStayGold for Tyr¹⁶⁹ and generated tdox2StayGold with a Y169F mutation (Extended Data Fig. 3). Both td8oxStayGold and td8ox2StayGold exhibited excellent dispersibility (Fig. 1, Extended Data Fig. 2a).

Photostability. To examine the performance of monovalent tagging with bright and photostable fluorescence, we characterized four tandem dimers (td5StayGold, td5oxStayGold, td8oxStayGold, and td8ox2StayGold) and two monomers (QC2-6 FIQ and QC2-6(PT)) of StayGold. After preparing purified protein products, we determined the extinction coefficients (es) at 488 nm (Supplementary Fig. 3) and the fluorescence quantum yields (QY_fs) of these six variants (Table 1). Their molecular brightness (i.e., the product of e and QY_f) was nearly the same as that of StayGold. To assess their photostability, we expressed them in cultured HeLa cells as fusions to histone 2B (H2B) to be immobilized on the chromatin structures inside the nucleus¹⁴. We exposed live cell samples to continuous illumination under an unattenuated light-emitting diode (LED) lamp and normalized the photobleaching curves using the standard method that considers the molecular brightness of each FP^13–15(Fig. 3a). We also comparatively examined the brightness-adjusted photostability of fluorescence from living cultured HeLa cells that expressed H2B-EGFP, H2B-mEGFP, H2B-mClover3, H2B-mNeonGreen, and H2B-mGreenLantern as well as H2B-StayGold and H2B-tdStayGold in this experimental system. The time required for photobleaching from an initial emission rate of 1,000 to 500 photons s^− 1 molecule^− 1 (t_1/2) of each FP is presented in Table 1. All of the new variants (four tandem dimers and two monomers) are clearly similar to the original StayGold in terms of photostability and molecular brightness and are thus more than one order of magnitude more photostable than any currently available FPs.

Cellular brightness. We next examined the actual brightness of the six aforementioned StayGold variants in living cells in comparison with the brightness of some reference green-emitting FPs (Fig. 3b, Table 1). We used an expression system (cotranslation via the T2A peptide) to correct the cellular brightness of each green-emitting FP by that of mCherry. EGFP and mGreenLantern showed 37% and 66% brightness, respectively, compared with StayGold; our previous study revealed that mClover3 and mNeonGreen were 84% and 99% as bright as StayGold, respectively¹. td5StayGold was 2.09 times brighter than StayGold, indicating the general merit of such td constructs for doubling the fluorescence brightness per unit of host protein compared with the brightness achieved with the regular monomer. Interestingly, td5oxStayGold, td8oxStayGold, and td8ox2StayGold were all further brighter, suggesting that oxStayGold folds better than StayGold. It is noted that QC2-6 FIQ and QC2-6(PT) showed cellular brightness similar to that of StayGold, demonstrating the practical usefulness of these two StayGold monomers. Accordingly, we designated QC2-6 FIQ and QC2-6(PT) as mStayGold and mStayGold2, respectively.

N- or C-terminal tagging with StayGold variants. All of the StayGold variants have an n1 adaptor at their N-terminus and are useful for C-terminal tagging, as observed for CytERM-FP (Fig. 1) and PB1-FP (Extended Data Fig. 2a). In this case, the original C-terminus of StayGold can be safely kept; its extension with polypeptides appears to affect expression of the fusions. An exception is the PT adaptor, which can function also as a C-terminal tail.

By contrast, tagging a protein at the N-terminus with a StayGold variant requires an adaptor (c4 or PT) at the FP C-terminus. Whereas CytERM permits only C-terminal tagging, PB1 can be tagged at both termini. Thus, we fused StayGold and its variants to the N-terminus of PB1 via the c4 adaptor for a Fluoppi assay (Extended Data Fig. 2b). StayGold(c4)-PB1 exhibited a low score (22.2%), whereas high Fluoppi scores were obtained with td5StayGold (96.9%), td8ox2StayGold (100%), mStayGold (100%), and mStayGold2 (99.4%).

Condensin I as a migrant chromosome stabilizer. During mitosis, condensins play a central role in chromosome assembly and segregation¹⁶. Two types of condensin complexes are present in most eukaryotic cells: condensin I and condensin II. Whereas condensin II binds to chromosomes throughout the cell cycle, condensin I does so in a specific time window; it is excluded from the nucleus in interphase and prophase but associates with mitotic chromosomes after nuclear envelope breakdown (NEBD), which causes a mixing of nuclear and cytoplasmic macromolecules. It is thus interesting to visualize at high spatiotemporal resolution how this cytoplasmic chromosome-stabilizing factor gets to interact with mitotic chromosomes. In a previous seminal study, HeLa cell clones stably expressing functional EGFP-tagged subunits of condensin I were generated to time-lapse image volumes every 1 min using single-beam LSCM¹⁷. After a lapse of many years, we took a more advanced approach in the present study. Specifically, we used the CRISPR-Cas9 genome-editing technique^18,19 to produce HCT116 cells in which a subunit of condensin I (CAP-H) was endogenously tagged at the C-terminus with td5oxStayGold (CAPH-td5oxStayGold) (Supplementary Fig. 4). Prior to observation, nuclei were stained with SiR-DNA, which is a far-red, fluorogenic, cell permeable, and highly specific live cell DNA probe. We used SpinSR10, a type of spinning-disk LSCM, to image a physiological concentration of CAP-H-td5oxStayGold in parallel with chromosomal DNA in multiple mitosing cells. First, we carried out a volumetric imaging experiment in which z-stack images were acquired every 1 min for 30 min (Fig. 4, Supplementary Video 1). In a different experiment, we carried out a high-speed continuous imaging at one z-position at a temporal resolution of 1 frame per second for 30 min (Extended Data Fig. 5a, Supplementary Video 2). These two imaging modes complementarily followed the targeting behavior of condensin I to chromosomes. During prophase, the far-red fluorescence of SiR-DNA was partially condensed inside the nucleus and the green fluorescence of CAP-H-td5oxStayGold was homogeneously distributed in the cytoplasmic compartment. After NEBD began in late prophase, all the condensing chromosomes became green fluorescent in a converted manner within the nucleus. The green fluorescence was highly concentrated in the centromeric region; such uneven distribution of the green/far-red ratio in each chromatid was invariable subsequently. Thus, our fast continuous sustainable imaging revealed that CAP-H molecules associate in unison with mitotic chromosomes. This process being limited by reaction rather than diffusion may corroborate the previous finding that the association occurs in a single kinetic binding step¹⁷.

In prometaphase, the CAP-H-td5oxStayGold signals clearly delineated two closely associated sister chromatids. The bright and photostable labeling enabled continuous volumetric imaging (Extended Data Fig. 5b, Supplementary Video 3), which revealed that fully condensed chromosomes swayed constantly in a unified body, likely because of their attachment to spindle microtubules at this stage. We used the CRISPR-Cas9 technique to generate HCT116 cells expressing CAP-H-mClover3 (Supplementary Fig. 4) and found the labeling vulnerable to the same illumination (Extended Data Fig. 5b, Supplementary Video 3).

Golgi membrane dynamics. As discussed earlier regarding the OSER assay mechanism, fluorescent membrane labeling, in principle, requires monovalent fusion of an FP to a membrane-resident protein. To visualize the Golgi apparatus, we previously fused StayGold through c4 to the N-terminus of a region (amino acids 3,131–3,259) of human Giantin^1,20, the gigantic Golgi matrix protein. This region, which we refer to as GianCreg (Giantin C-terminal region), consists of a short cytoplasmic domain, a membrane-spanning domain and a short luminal C-terminal domain and is suggested to form a dimer via a disulfide bond in the lumen²¹. Accordingly, we do not consider obligate dimer formation of StayGold to be a serious problem in the labeling approach. However, as it is not clear how the monomer/dimer equilibrium of Giantin is regulated, we substituted td5StayGold to create td5StayGold(c4) = GianCreg in this study.

We transfected cultured HeLa cells transiently with the constructed cDNA. The Golgi stack comprising several flattened cisternae and many vesicular structures were highlighted against the 4’,6-diamidino-2-phenylindole (DAPI)-stained nucleus in each transfected cell (Fig. 5a). Continuous, cell-wide observation using SDSRM (SpinSR10) enabled us to track movements of the Golgi apparatus and its membranous derivatives, including long tubular structures emerging from the cisternae (Fig. 5b, Supplementary Video 4). The Golgi-derived tubules exhibited irregular dynamics^22,23 but mostly detached and moved away from the Golgi stack, similar to the dynamics of tubular post-Golgi carriers^24,25. We subsequently attempted to image the microtubule network that should guide Golgi-derived vesicles throughout the cell. We noticed substantial photobleaching of far-red chemical dyes for microtubules, such as SiR-Tubulin, under our imaging conditions and therefore transfected td8ox2StayGold(c4) = b-tubulin into cells in addition to td5StayGold(c4) = GianCreg to achieve eccentric dual-target imaging in a single channel. We tracked all the labeled vesicles, including long tubular ones, moving along microtubules over extended periods (Supplementary Video 5, Supplementary Fig. 5).

We observed similar fluorescence patterns of the Golgi apparatus in HeLa cells that stably expressed td5StayGold(c4) = GianCreg. These cell samples were fixed and subjected to immunocytochemical studies using various Golgi markers for identification of the fluorescently labeled components. We observed the most significant overlap with Giantin (medial)²¹ and partial overlap with GM130 (cis)²⁶ and TGN46 (trans-Golgi network)²⁷ signals (Extended Data Fig. 6). These results suggested that td5StayGold was most concentrated in medial-Golgi but distributed throughout the Golgi membrane, possibly due to the N-terminal truncation and/or over-expression.

mStayGold targeting filamentous actin and inner mitochondrial membranes. Size-wise, FP monomers should be more suitable than FP td constructs for labeling protein networks and complexes that are susceptible to steric hindrance. To study filamentous actin (F-actin) dynamics in live cells, we tagged actin-binding domains²⁸ with mStayGold. First, we simply cotransfected Lifeact-mStayGold into Vero cells with F-tractin-mScarlet-I and confirmed their colocalization on stress fibers, actin bundles in filopodia, actin networks in lamellipodia and cortical actin networks underlining the plasma membrane (Supplementary Fig. 6). We then used SDSRM (SpinSR10) to continuously image COS7 cells expressing F-tractin = mStayGold or mStayGld(c4) = UtrCH in single confocal slices sectioned nearest the cell bottom (Supplementary Fig. 7). Whereas stress fibers tended to be static, filopodia and lamellipodia were constantly moving. Cortical networks spread like cobwebs and some loose wires showed active fluctuations (Supplementary Video 6). Considerably dynamic were tiny ordered architectures that came out to “surf” on the cortical networks. They were mostly asterisk-like structures that could be categorized as actin stars and asters^29,30. Occasionally they swirled about to form rings that have been speculated to reflect endocytic machineries²⁵ or actin vortices²⁹. These punctate structures disappeared quickly after the application of latrunculin A, a drug that sequesters monomeric actin; they were stabilized on the plasma membrane by the application of cytochalasin D, a drug that caps filament plus ends (Extended Data Fig. 7, Supplementary Video 7).

Imaging the inner mitochondrial membrane (IMM) live is important for understanding the development of many diseases but remains challenging^31,32. To date, IMM-selective organic dyes or SNAP-tagged IMM-resident proteins have been used mainly in combination with stimulated emission depletion (STED) microscopy^33–35; however, the number of images acquired has been limited, likely because of substantial photobleaching of the dyes. The development of photostable chemical fluorophores that stain IMMs, such as MitoPB Yellow (ref. 36) and MitoESq-635 (ref. 37), greatly increased the spatial resolution of cristae imaging by STED microscopy. However, the extremely strong laser light brought a situation where stained mitochondria were time-lapse imaged with long time intervals (> 50 s)³⁷ or were otherwise photodegraded rapidly (< 1 min)³⁶. On the other hand, state-of-the-art deconvolution^38,39 or generative artificial intelligence methods⁴⁰ greatly reduced the excitation light of SIM, enabling long-duration high–spatiotemporal-resolution imaging of mitochondrial cristae. However, the computational approaches assume a priori knowledge and may miss rare but important changes in IMM structures.

To aim for sustained but simple live imaging of IMMs, we fused mStayGold to the C-terminus of subunit CoxVIIIa of cytochrome c oxidase. After the construct (COX8a = mStayGold) was transfected into HeLa cells, substantial cell-to-cell variation in the expression level was observed. Among transfectants, we chose cells with modestly labeled mitochondria for continuous observations. We first used the conventional lattice SIM technique to perform volumetric imaging and observed cristae membrane dynamics in a cell-wide manner (Fig. 5c, Supplementary Video 8). We next used SDSRM (SpinSR10) to perform a high-speed imaging at one z-position at a temporal resolution of 2–4 frame/s for 5–6 min. We noticed a variety of behaviors of mitochondria and their interior structures. Some were stable (Supplementary Video 9), whereas others were extremely inconstant and complicated (Supplementary Video 10). The observation period was sufficiently long for successive administration of three drugs; histamine to initiate physiological Ca²⁺ mobilization, an antihistamine to shut it down, and ionomycin to induce Ca²⁺ mobilization again (Supplementary Video 11). We found that movement of individual mitochondria was attenuated upon Ca²⁺ mobilization but that remodeling of the IMMs was constantly active (Extended Data Fig. 8). Further studies that adopt genetical or pharmacological approaches will be needed to elucidate the power source of the cristae movements.

Although the crystal structure of StayGold has been determined to have 1.56 Å resolution, we could not fully understand the structural basis of the high photostability of this FP and hope that future photophysical studies will provide straightforward clues. This paper reports a directed evolution of the dimeric StayGold toward useful monomers. Similarly to previous FP monomerization studies, we used the crystal structure of this FP to identify the mutational hotspots in the dimer interface. However, it is not easy to disrupt the dimerization while retaining both the brightness and photostability of the original FP fluorescence. Because high-throughput screening for photostable FPs in living cells is an ambitious task^41,42, we adopted a low-throughput approach in the present study; thus, room likely remains for further refinements of QC2-6 FIQ and QC2-6(PT). Nevertheless, the possibility of using StayGold monomers as valuable alternatives to popular bright monomeric green-emitting FPs, such as mEGFP and mNeonGreen, for molecular fusion and membrane-targeting applications warrants investigation.

For the generation of tdStayGold variants, on the other hand, we preserved the dimer interface and optimized both the linker and the adaptor for tandemization to maximize the dispersibility, which was also quantitated by both OSER and Fluoppi assays. As shown in Fig. 3b, a td construct doubles the fluorescence brightness per unit of host protein compared with the regular monomer and is indeed effective for observing low–copy-number targets such as the endogenous cohesion I subunit (Fig. 4, Extended Data Fig. 5). However, the doubled size of a td construct might cause steric hindrance. In our experience, for example, td8ox2StayGold did not fuse to COX8a for successful labeling of the inner mitochondrial membrane.

Tagging a protein at the N-terminus with a StayGold variant requires an adaptor (c4 or PT) at the C-terminus of the FP variant. Likewise, for C-terminal tagging with a StayGold variant, it is recommended to put an adaptor (n1 or n2) at the N-terminus of the FP variant. In this case, it is safe to keep the original C-terminus of StayGold; extension with c4 appears to affect the expression of the entire fusion. It is also advised that a flexible and durable linker be used for fusion to minimize both the folding interference of the two domains and proteolysis. For this purpose, we basically used a ‘Coupler linker’, a triple repeat of the amino acid linker Gly-Gly-Gly-Gly-Ser [(GGGGS)3] (ref. 43), which is denoted as ‘=’ in this paper.

To advance toward a real understanding of the spatiotemporal regulation of a biological event through live cell imaging, in general, careful consideration of the expression levels of FP-tagged proteins is required. Whereas the high photostability of StayGold variants is expected to allow for visualization with low-level expression, we discuss this point as follows. First, we believe that the HCT116 cell line expressing endogenous condensin I tagged with td5oxStayGold should provide a nearly perfect situation. Second, we reason that the C-terminal region of Giantin (GianCreg) tagged with td5StayGold can be adequately overexpressed for the purpose of labeling the whole Golgi apparatus. Third, we verified that microtubules and actin filaments labeled by StayGold variants behaved in the same manner as those labeled with common FPs. Fourth, because mitochondrial complex IV containing subunit COX8a was found to exist in free form as well as to be complexed with complexes I and III, we reason that COX8a tagged with mStayGold can be overexpressed to label the inner mitochondrial membrane. In all the cases with overexpression, we selected dim or moderately bright cells for observation. We also confirmed that all the stable transformants used for sustained imaging in this study—specifically, HCT116 cells expressing CAH-H-td5oxStayGold (Fig. 4), HeLa cells expressing td5StayGold(c4) = GianCreg (Extended Data Fig. 6), and HeLa cells expressing COX8a = mStayGold (Supplementary Fig. 8)—proliferated normally. It is thus assumed that the StayGold variants entered the biological systems with minimal perturbation.

Data availability

The accession numbers in the DDBJ/GenBank databases are LC756333 for mStayGold (QC2-6 FIQ), LC756334 for mStayGold2 (QC2-6(PT)), LC756335 for td5StayGold, LC756336 for td5oxStayGold, and LC756337 for td8ox2StayGold. The entry IDs in the Protein Data Bank are 8ILK and 8ILL for atomic structures of StayGold crystallized at pH 8.5 and pH 5.6, respectively. All data generated in this study are available through the R2DMS (Riken Research Data and copyrighted-work Management System) (https://dmsgrdm.riken.jp/). Plasmid DNAs containing mStayGold or tdStayGold variants are available from the RIKEN Bio-Resource Center (BRC) (http://en.brc.riken.jp) under a material transfer agreement with RIKEN.

Acknowledgments

The authors thank K. Higuchi, Y. Ue, and Y. Tabushi at the RIKEN CBS-Evident Collaboration Center (BOCC) for technical assistance with LSCM; Common Use Equipment in the Support Unit for Bio-Material Analysis, Research Resource Division, RIKEN CBS for technical assistance with multi-beam LSCM; H. Hijikata at Kyoto University and Dr. Anna Löschberger at Carl Zeiss for technical assistance in Lattice SIM at the iCeMS Analysis Center, Kyoto University; Dr. T. Tojima and Dr. A. Nakano at RIKEN Center for Advanced Photonics, Dr. T. Watanabe at Japan Tabacco Inc., Dr. Z. Yang at RIKEN CBS, Dr. Y. Yonemaru at BOCC for valuable advice; Dr. I. Imayoshi and Dr. M. Kengaku at Kyoto University, Dr. N. Imamoto at RIKEN CPR and Dr. R. Kageyama at RIKEN CBS for continuous support. This work was supported in part by Grant-in-Aid for Scientific Research (S) (21H05041 to A.M.), Grant-in-Aid for Innovative Areas: Resonance Bio (15H05948 to A.M.) and Information Physics of Living Matters (19H05794, 19H05795 to Y.O.), Japan Science and Technology Agency Core Research for Evolutionary Science and Technology program, "Spatiotemporal dynamics of intracellular components" (JPM JCR20E2 to Y.O.) and Moonshot R&D program "Realization of ultra-early disease prediction and intervention by 2050" (JPMJMS2025-14 to Y.O.), Marine Biomass Innovation Project (NFRFT-2020-00452 to A.M.), the Brain Mapping by Integrated Neurotechnologies for Disease Studies from AMED (Brain/MINDS, JP15dm0207001 to A.M.), RIKEN Collaboration Seed Fund (to M.T. and R.A.) and Platform Project for Supporting Drug Discovery and Life Science Research (Basis for Supporting Innovative Drug Discovery and Life Science Research (BINDS)) from AMED (JP21am0101070 to M.Y.).

Author Contributions

A.M. conceived the whole study. R.A. developed mStayGold and tdStayGold variants. S.S. developed tdStayGold variants. R.A. and S.S. analyzed dispersibility of the variants and performed subcellular targeting of selected constructs. R.A. analyzed photobleaching of FPs and performed live cell imaging experiments using SpinSR10. S.S. analyzed brightness of FPs and performed (immuno)cytological experiments. H.A. and M.H. crystallized StayGold. H.A., G.U. and M.Y. collected the crystallographic data and performed the atomic structure determination and refinement. M.T. performed condensin I imaging experiments. M.H. analyzed spectroscopic properties of FPs. F.I. and T.F. performed imaging experiments using lattice SIM. Y.O. performed cytoskeleton imaging experiments. M.S. and H.K. analyzed IMM structural dynamics. H.K. devised the algorithms for analyzing large-volume data. A.M. and M.S. prepared figures. A.M. designed and wrote the manuscript and supervised the project.

Competing Interests Statement

R.A., M.H. and A.M. are inventors on Japanese patent application No. 2021-065373 that covers the creation and use of StayGold. The remaining authors declare no competing interests.

Hirano, M. et al. A highly photostable and bright green fluorscent protein. Nature Biotechnology 40, 1132–1142 (2022).
Campbell, R.E. et al. A monomeric red fluorescent protein. Proc. Natl. Acad. Sci. USA 99, 7877–7782 (2002).
Watanabe, T. et al. Genetic visualization of protein interactions harnessing liquid phase transitions. Sci. Rep. 7, 46380 (2017).
Costantini, L.M., Fossati, M., Francolini, M. & Snapp, E.K. Assessing the tendency of fluorescent proteins to oligomerize under physiological conditions. Traffic 13, 643–649 (2012).
Costantini, L.M. et al. A palette of fluorescent proteins optimized for diverse cellular environments. Nat. Commun. 6, 7670 (2015).
Cranfill, P.J. et al. Quantitative assessment of fluorescent proteins. Nature Methods 13, 557–562 (2016).
Gustafsson, M.G. Surpassing the lateral resolution limit by a factor of two using structured illumination microscopy. J. Microsc. 198, 82–87 (2000).
Hayashi, S. & Okada, Y. Ultrafast superresolution fluorescence imaging with spinning disk confocal microscope optics. Mol. Biol. Cell 26, 1743–1751 (2015).
Zacharias, D.A., Violin, J.D., Newton, A.C. & Tsien, R.Y. Partitioning of lipid-modified monomeric GFPs into membrane microdomains of live cells. Science 296, 913–916 (2002).
Bajar, B.T. et al. Improving brightness and photostability of green and red fluorescent proteins for live cell imaging and FRET reporting. Sci. Rep. 6, 20889 | DOI: 10.1038/srep20889 (2016).
Shaner, N.C. et al. A bright monomeric green fluorescent protein derived from Branchiostoma lanceolatum. Nat. Methods 10, 407–409 (2013).
Campbell, B.C. et al. mGreenLantern: a bright monomeric fluorescent protein with rapid expression and cell filling properties for neuronal imaging. Proc. Natl. Acad. Sci. USA 117, 30710–30721 (2020).
Shaner, N.C. et al. Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein. Nat. Biotechnol. 22, 1567–1572 (2004).
Shaner, N.C. Fluorescent proteins for quantitative microscopy: important properties and practical evaluation. Methods Cell Biol. 123, 95–111 (2014).
Shaner, N.C. et al. Improving the photostability of bright monomeric orange and red fluorescent proteins. Nat. Methods 5, 545–551 (2008).
Batty, P & Gerlich, D.W. Mitotic chromosome mechanics: how cells segregate their genome. Trends Cell Biol. 29, 717–726 (2019).
Gerlich, D., Hirota, T., Koch, B., Peters, J.-M. & Ellenberg, J. Condensin I stabilizes chromosomes mechanically through a dynamic interaction in live cells. Curr. Biol. 16, 333–344 (2006).
Takagi, M. et al. Ki-67 and condensins support the integrity of mitotic chromosomes through distinct mechanisms. J. Cell Sci. doi: 10.1242/jcs.212092 (2018).
Koch, B., Nijmeijer, B., Kueblbeck, M., Cai, Y., Walther, N. & Ellenberg, J. Generation and validation of homozygous fluorescent knock-in cells using CRISPR-Cas9 genome editing. Nature Protocols 13, 1465–1487 (2018).
Bindels, D.S. et al. mScarlet: a bright monomeric red fluorescent protein for cellular imaging. Nat. Methods 14, 53–56 (2017).
Frisbie, C.P., Lushnikov, A.Y., Krasnoslobodtsev, A.V., Riethoven, J.-J.M., Clarke, J.L, Stepchenkova, E.I. & Petrosyan, A. Post-ER stress biogenesis of Golgi is governed by Giantin. Cells 8, 1631; doi:10.3390/cells8121631 (2019).
Bottanelli, F. et al. A novel physiological role for ARF1 in the formation of bidirectional tubules from the Golgi. Mol. Biol. Cell 28, 1676–1687 (2017).
Weigel, A.V. et al. ER-to-Golgi protein delivery through an interwoven, tubular network extending from ER. Cell 184, 2412–2429 (2021).
Hirschberg, K., Miller, C.M., Ellenberg, J., Presley, J.F., Siggia, E.D., Phair, R.D. & Lippincott-Schwartz, J. Kinetic analysis of secretory protein traffic and characterization of Golgi to plasma membrane transport intermediates in living cells. J. Cell Biol. 143, 1485–1503 (1998).
Li, D. et al. Extended-resolution structured illumination imaging of endocytic and cytoskeletal dynamics. Science 349, aab3500 (2015).
Tie, H.C., Ludwig, A., Sandin, S. & Lu, L. The spatial separation of processing and transport functions to the interior and periphery of the Golgi stack. eLIFE 7, e41301 (2018).
Luzio, J.P., Brake, B., Ganting, G., Howell, K.E., Braghetta, P. & Stanley, K.K. Identification, sequencing and expression of an integral membrane protein of the trans-Golgi network (TGN38). Biochem. J. 270, 97–102 (1990).
Malek, M., Plessner, M., & Grosse, R. Actin visualization at a glance. J. Cell Sci. 130, 525–530 (2017).
Fritzsche, M. et al. Self-organizing actin patterns shape membrane architecture but not cell mechanics. Nat. Comm. 8, 14347 (2017).
Xia, S., Lim, Y.B., Zhang, Z., Wang, Y., Zhang, S., Lim, C.T., Yim, E.K.F. & Kanchanawong, P. Nanoscale architecture of the cortical actin cytoskeleton in embryonic stem cells. Cell Reports 28, 1251–1267 (2019).
Kondadi, A.K., Anand, R. & Reichert, A.S. Cristae membrane dynamics-A paradigm change. Trends Cell Biol. 30, 923–936 (2020).
Ngo, J., Osto, C., Villalobos, F. & Shirihai, O.S. Mitochondrial heterogeneity in metabolic diseases. Biology 10, 927 (2021).
Stephan, T., Roesch, A., Riedel, D. & Jakobs, S. Live-cell STED nanoscopy of mitochondrial cristae. Sci. Rep. 9, 12419 (2019).
Kondadi, A.K. et al. Cristae undergo continous cycles of membrane remodelling in a MICOS-dependent manner. EMBO rep. 21, 49776 (2020).
Stephan, T. et al. MICOS assembly controls mitochondrial inner membrane remodeling and crista junction redistribution to mediate cristae formation. EMBO J. 39, 104105 (2020).
Wang, C., Taki, M., Sato, Y., Yaginuma, H., Okada, Y. & Yamaguchi, S. A photostable fluorescent marker for the super-resolution live imaging of the dynamic structure of the mitochondrial cristae. Proc. Natl. Acad. Sci. USA 116, 15817–15822 (2019).
Yang, X. et al. Mitochondrial dynamics quantitatively revealed by STED nanoscopy with an enhanced squaraine variant probe. Nat. Comm. 11, 3699 (2020).
Huang, X. et al. Fast, long-term, super-resolution imaging with Hessian structured illumination microscopy. Nat. Biotech. 36, 451–459 (2018).
Zhao, W. et al. Sparse decopnvolution improves the resolution of live-cell super-resolution fluorescence microscopy. Nat. Biotech. 40, 606–617 (2022).
Qiao, C. et al. Rationalized deep learning super-resolution microscopy for sustained live imaging of rapid cellular processes. Nat. Biotech. 41, 367–377 (2023).
Dean, K.M., Lubbeck, J.L., Davis, L.M., Regmi, C.K., Chapagain, P.P., Gerstman, B.S., Jimenez, R. & Palmer, A.E. Microfluidics-based selection of red-fluorescent proteins with decreased rates of photobleaching. Integr. Biol. 7, 263–273 (2015).
Dean, K.M., Davis, L.M., Lubbeck, J.L., Manna, P., Friis, P., Palmer, A.E. & Jimenez, R. High-speed multiparameter photophysical analyses of fluorophore libraries. Anal. Chem. 87, 5026–5030 (2015).
Shimozono, S. & Miyawaki, A. Engineering FRET constructs using CFP and YFP. Methods Cell Biol. 85, 381–393 (2008).
Ward, W.W. Biochemical and physical properties of green fluorescent protein. In : Gren Fluorescent Protein, Ch. 3 (Wiley-Liss, 1998).
Tsien, R.Y. The green fluorescent protein. Ann. Rev. Biochem. 67, 509–544 (1998).

Protein purification. Recombinant proteins with a polyhistidine tag at the N-terminus were expressed in Escherichia coli [JM109 (DE3)]. Transformed E. coli were incubated in a Luria–Bertani (LB) medium containing 0.1 mg mL⁻¹ ampicillin at room temperature (RT) with gentle shaking for several days. Protein purification by Ni²⁺ affinity chromatography was performed as described previously⁴⁶.

In vitro spectroscopy. Absorption spectra were acquired using a spectrophotometer (U-2910, Hitachi). Fluorescence excitation and emission spectra were acquired using a fluorescence spectrophotometer (F-2500, Hitachi). Absolute fluorescence quantum yields were measured using an absolute photoluminescence quantum yield spectrometer (C9920-02, Hamamatsu Photonics). Protein concentrations were measured using a Protein Assay Dye Reagent Concentrate kit (#5000006, Bio-Rad) with bovine serum albumin (BSA) as the standard.

Pseudonative SDS/PAGE analysis. Non-heated protein samples were separated on 10% polyacrylamide gels as described previously⁴⁷.

Gene construction for bicistronic expression in mammalian cells. The T2A⁴⁸ gene was synthesized with 5'-HindIII and 3'-EcoRI sites, and the restricted product was cloned into the HindIII/EcoRI sites of pBlueScript (pBS) to generate pBS/T2A. The mCherry gene was amplified using primers containing 5'-XhoI and 3'-HindIII sites, and the restricted product was cloned in frame into the XhoI/HindIII sites of pBS/T2A to generate pBS/mCherry-T2A. The green-emitting FP (EGFP, mGreenLantern, StayGold, td5StayGold, td5oxStayGold, td8oxStayGold, td8ox2StayGold, QC2-6 FIQ, or QC2-6(PT)) gene was amplified using primers containing 5'-BamHI and 3'-XbaI sites, and the restricted product was cloned in frame into the BamHI/XbaI sites of pBS/mCherry-T2A to generate pBS/mCherry-T2A-green-emitting FP. Lastly, XhoI/XbaI fragments encoding mCherry-T2A-green-emitting FP were subcloned into pCSII-EF to generate pCSII-EF/mCherry-T2A-green-emitting FP plasmids.

Cellular brightness assay. HeLa cells were seeded into 24-well glass-bottom plates (5826-024, IWAKI) and maintained in growth medium (Dulbecco’s Modified Eagle Medium low glucose, supplemented with 10% fetal bovine serum). On the following day, cells were transfected with 0.5 mg of pCSII-EF/mCherry-T2A green-emitting FP per well. Forty-eight hours after transfection, the cells were imaged on an inverted microscope (IX-83, Evident) equipped with an LED light bulb (X-Cite XYLIS, Excelitas Technologies), an objective lens (UPlanXApo 4×/0.16 NA, Evident), and a scientific CMOS camera (ORCA-Fusion, Hamamatsu Photonics). Green-emitting FPs were observed using a filter cube (U-FBNA, Evident). mCherry was observed using a filter cube (U-FMCHE, Evident). Green-emitting FP fluorescence was divided by mCherry fluorescence and normalized to the ratio of StayGold/mCherry, as described previously¹². See Fig. 3b.

OSER assay. The cDNA fragment encoding CytERM (ref. 4) was synthesized according to the sequence information of Emerald-CytERM-N-17 (addgene #56290) with 5'-HindIII and 3'-BamHI sites. As the CytERM gene has BamHI, EcoRI, and HindIII sites internally, all these sites were eliminated in the synthesis. The FP gene was amplified using primers containing 5'-BamHI and 3'-XhoI sites. The restricted products were cloned into the HindIII/XhoI sites of pcDNA3 to generate pcDNA3/CytERM-FP. Twenty hours after transfection, HeLa cells on a standard 35-mm glass-bottom dish were incubated in Hanks’ Balanced Salt Solution (HBSS, 14025, Thermo Fischer Scientific) containing 15 mM HEPES-NaOH (pH 7.4) and imaged on an inverted microscope (IX-83, Evident) equipped with a 20× objective lens (UPlanXApo 20×/0.8 NA, Evident) and a camera (ORCA-Fusion, Hamamatsu Photonics). At an xy position, nine images were serially acquired with a z-step size of 0.59 µm, from which an in-focus image was mathematically generated by the extended focus imaging (EFI) function of the cellSens Dimension (Evident) software. A logarithmic transformation was applied to all image data that had a wide range of fluorescence intensity distributions. The number of transfected cells showing whorl structures was counted. In addition, the number of transfected cells avoiding whorl formation was counted. Three independent experiments were carried out for each construct.

Fluoppi assay. The FP gene was amplified using primers containing 5'-BamHI and 3'-EcoRI sites, and the restricted product was cloned into the BamHI/EcoRI sites of pAsh-MCL (Medical Biological Laboratory, Japan) to generate a plasmid DNA for expression of PB1-FP. Also, the FP gene was amplified using primers containing 5'-BamHI and 3'-XhoI sites, and the restricted product was cloned into the BamHI/XhoI sites of pAsh-MNL (Medical Biological Laboratory) to generate a plasmid DNA for expression of FP-PB1. Twenty-four hours after transfection, HeLa cells on a standard 35-mm glass-bottom dish or a 24-well glass-bottom plate (5826-024, IWAKI) were incubated in HBSS (14025, Thermo Fischer Scientific) containing 10 mM HEPES-NaOH (pH 7.4) and imaged on an inverted microscope (IX-83, Evident) equipped with a 20× objective lens (UPlanXApo 20×/0.8 NA, Evident) and a camera (ORCA-Fusion, Hamamatsu Photonics). The mirror units used for imaging green-, yellow-, and red-emitting FPs were U-FBNA, U-FYFP, and U-FGNA (Evident), respectively. A logarithmic transformation was applied to all image data that had a wide range of fluorescence intensity distributions.

Expression and purification for crystallography. StayGold, in a pET-47b(+) vector (Novagen) carrying ampicillin resistance and an HRV 3C-cleavable N-terminal polyhistidine tag, was expressed in Escherichia coli (BL21[DE3]). Transformed E. coli were incubated at 25 °C in an LB medium containing 20 mg mL⁻¹ kanamycin with gentle shaking (63 rpm) for 5 days. Protein purification by Co²⁺ affinitychromatography was performed using TALON resins (Clontech). Cleavage of the polyhistidine tag was performed during dialysis into 50 mM Tris-HCl (pH7.5), 0.3 M NaCl, and 1 mM dithiothreitol (DTT) using HRV 3C protease at 4 ℃ overnight. The sample was loaded onto TALON resins, and the unbound fraction was applied to a HiPrep 16/60 Sephacryl S300 HR column (cytiva) equilibrated with 20 mM HEPES-NaOH (pH7.5), and 0.15 M NaCl for preparative separation of StayGold. Finally, the untagged product was concentrated to 8.3 mg ml⁻¹ using Amicon Ultra (3,000 MW cutoff) (Merck).

Crystallization and X-ray data collection. Crystals of StayGold were grown at 20 °C using the sitting-drop vapor diffusion method by mixing 0.1 mL of protein solution (8.3 mg mL⁻¹ in 20 mM HEPES-NaOH (pH 7.5) and 0.15 M NaCl) with 0.1 mL of reservoir solution I (25%(w/v) PEG3350, 0.2 M MgCl₂, and 0.1 M Tris-HCl (pH 8.5)) or reservoir solution II (20%(w/v) PEG4000, 20%(v/v) 2-propanol, and 0.1 M sodium citrate (pH5.6)). The mixture was sealed over a well containing 50 mL of reservoir I or reservoir II solution, respectively. Individual crystals were soaked in 1 mL of reservoir I or reservoir II solution containing 250 mg trehalose, scooped using a nylon loop, and flush-cooled in liquid nitrogen. The diffraction data were collected at 100 K using the BL26B2 beam line at the SPring-8 (Hyogo, Japan), and were processed using the DIALS program⁴⁹.

Structure determination and refinement. The structure of StayGold was determined by the molecular replacement technique with a model of GFP (PDB: 2Q57) as a search model using phenix.phaser⁵⁰. The model was refined using phenix.refine⁵¹ and repeatedly corrected using Coot⁵². Refinement statistics of structures are summarized in Supplementary Table 1. Structural figures were prepared using PyMOL.

Gene construction (nuclear targeting). The mouse histone 2B (H2B) gene (Fantom3) was amplified using primers containing 5'-XhoI and 3'-HindIII sites, and the restricted product was cloned into the XhoI/HindIII sites of pBS Coupler 1 (ref. 43) to generate pBS Coupler 1/H2B. In addition, the green-emitting FP (EGFP, mEGFP, mClover3, mNeonGreen, mGreenLantern, StayGold, tdStayGold, td5StayGold, td5oxStayGold, td8oxStayGold, td8ox2StayGold, QC2-6 FIQ, or QC2-6(PT)) gene was amplified using primers containing 5'-BamHI and 3'-XbaI sites, and the restricted product was cloned in frame into the BamHI/XbaI sites of pBS Coupler 1/H2B=FP. Lastly, XhoI/XbaI fragments encoding H2B=FPs were subcloned into pCSII-EF for transfection.

WF photobleaching (living cells in HBSS). Living cells on 35-mm glass-bottom dishes were incubated in Hanks’ Balanced Salt Solution (HBSS) containing 15 mM HEPES-NaOH (pH 7.4) and imaged on an inverted microscope (IX-83, Evident) equipped with an LED light bulb (X-Cite XYLIS, Excelitas Technologies), a 60× objective lens (UPlanSApo 60×/1.35 NA), and a scientific CMOS camera (ORCA-Fusion, Hamamatsu Photonics). The data were analyzed using Excel (2019). The fluorescence intensity at t = 0 was normalized to 1,000 photons s⁻¹ molecule⁻¹, and the time axis was adjusted according to the standard method. See Fig. 3a.

Gene construction for genome editing. pMT690-2 is a plasmid that contains a 1,231-bp genomic fragment around the termination codon of NCAPH (ref. 53). A series of “cassette constructs” encoding FP tags plus selection markers⁵⁴ were provided by Dr. Masato T. Kanemaki at the National Institute of Genetics in Mishima, Japan. They can be used for generating FP knock-in cells at the C-terminal end of any protein of interest via homology-directed repair. Among the constructs, pMK281 (mCherry2-Hygro) and pMK278 (mClover3-Hygro) were selected in the present study. Also, the mCherry2 gene in pMK281 was replaced with the td5oxStayGold gene to generate a new cassette construct, pMT892 (td5oxStayGold-Hygro). The td5oxStayGold-Hygro and mClover3-Hygro cassette genes were amplified using pMK892 and pMT278 as templates, and the PCR products were inserted in frame via Gibson assembly (NEB, Ipswich, MA) at a site immediately upstream of the termination codon of NCAPH in pMT690-2 to generate pMT897 and pMT899, respectively. In both constructs, the C-terminus of CAP-H was linked to FP via a linker amino-acid tract GSGAAS.

Genome-edited cell lines. pMT691, a derivative of pX330 (Addgene plasmid #42230), can be used for cleaving the genome with Cas9 near the termination codon of NCAPH (ref. 53). HCT116 cells were cotransfected with pMT691 and pMT897 using FuGene HD (Promega, Fitchburg, WI) and were then cultured in the presence of 100 mg mL⁻¹ hygromycin B (Nacalai Tesque, Kyoto, Japan) for selection of cell clones in which CAP-H was endogenously tagged at the C-terminus with td5oxStayGold. Single cell colonies growing normally and exhibiting green fluorescence in the cytoplasm in interphase cells and on chromosomes in mitotic phase cells were picked up. Among them was a cell clone designated as #897, which was further characterized for the tagging (CAP-H-td5oxStayGold). Likewise, cotransfection of HCT116 cells with pMT691 and pMT899 resulted in the generation of a cell line #899 carrying CAP-H-mClover3. Cells were cultured at 37 ºC with 5% CO₂ in DMEM supplemented with 10% fetal bovine serum.

Validation of FP integration in knock-in cell lines. First, junction PCR was performed using a forward primer 5' outside of the left homology arm (P758: 5'-GTTAATCTCTTACTGTGCCT-3’) and a reverse primer 3' outside of the right homology arm (P759: 5'-TCTCTTCCATTCTCCTCCGA-3'). Second, western blotting analysis was performed using a rabbit polyclonal anti-CAP-H antibody (Proteintech, #11515-1-AP).

Gene construction (Golgi targeting). The td5StayGold(c4) gene was amplified using primers containing 5'-KpnI and 3'-EcoRI sites. The restricted product was substituted for the StayGold(c4) gene in pcDNA3/StayGold(c4)-20aa-Giantin¹, which is identical to pcDNA3/StayGold(c4)=GianCreg. GianCreg is a C-terminal domain that contains amino acids 3,131–3,259 of human Giantin. The resultant plasmid was pcDNA3/td5StayGold(c4)=GianCreg.

Gene construction (microtubule targeting). The td8ox2StayGold(c4) gene was amplified using primers containing 5'-NotI and 3'-EcoRI sites. Also, the b-tubulin gene was amplified using primers containing 5'-HindIII and 3'-XhoI sites. The two restricted products were sequentially cloned into the BamHI/EcoRI and HindIII/XhoI sites of pBS Coupler 4. Finally, the BamHI/XhoI fragment was cloned into pcDNA3 to generate pcDNA3/td8ox2StayGold(c4)=b-tubulin.

Stable transformants. Replication-defective, self-inactivating lentiviral vectors were used¹. The pCSII-EF-MCS vector encoding td5StayGold(c4)=GianCreg or COX8a=mStayGold was co-transfected with the packaging plasmid (pCAG-HIVgp) and the VSV-G-/Rev-expressing plasmid (pCMV-VSV-G-RSV-Rev) into 293T cells. High-titer viral solutions were prepared and used for transduction into HeLa cells (MOI = 1–10). Most (> 95%) of the resultant cells uniformly exhibited green fluorescence and were used as stable transformants.

Immunocytochemistry of the Golgi apparatus. After being washed in phosphate-buffered saline (PBS) three times, cells stably expressing td5StayGold(c4)=GianCreg were chemically fixed (see below), and then incubated in blocking solution (PBS containing 3% BSA and 0.1% Triton X-100) for 60 min at RT. The cells were then reacted with primary antibodies (Abs) in blocking solution at RT for 60 min. After being washed in PBS-T [PBS containing 0.1% Triton X-100] three times, the cells were reacted with secondary Abs in blocking solution at RT for 60 min. After the cells were washed in PBS-T three times, nuclear staining was performed using DAPI (Fuji Film, 340-07971, 1:1000 dilution) at RT in PBS for 5 min. The fixation conditions and used Abs are as follows.

<GM130>

Fixation: 4% paraformaldehyde (PFA)/PBS at RT for 5 min.

Primary Ab: anti-GM130 Ab (MBL, PM061), 1:250 dilution.

Secondary Ab: Donkey anti–rabbit IgG (H+L) highly cross-adsorbed secondary Ab, Alexa Fluor^TM 647–conjugated (Thermo Fisher, A-31573), 1:500 dilution.

Fixation: 4% PFA/PBS at RT for 5 min.

Primary Ab: anti-Giantin Ab (POTEINTECH, 22270-1-AP), 1:250 dilution.

Secondary Ab: Donkey anti–rabbit IgG (H+L) highly cross-adsorbed secondary Ab, Alexa Fluor^TM 647–conjugated (Thermo Fisher, A-31573), 1:500 dilution.

<TGN46>

Fixation: 4% PFA + 0.05% glutaraldehyde/PBS at RT for 5 min.

Primary Ab: anti-TGN46 Ab (Sigma-Aldrich, SAB4200355), 1:100 dilution.

Secondary Ab: Donkey anti–mouse IgG (H+L) highly cross-adsorbed secondary Ab, Alexa Fluor^TM 647–conjugated (Thermo Fisher, A-31571), 1:500 dilution.

Cell samples were imaged using an inverted laser scanning confocal microscopy system (FV3000, Evident, Tokyo, Japan) equipped with a 60× water objective lens (UPlanSApo 60×/1.2 NA). The size of the confocal aperture was 1 Airy disk. For a zoom factor of 4× and a pixel array size of 512 × 512, the size of each pixel was calculated to be 0.104 mm. Confocal images were acquired every 0.52 μm along the z-axis to create z stacks (25–27 slices) that covered the entire Golgi apparatus.

td5StayGold, Alexa 647, and DAPI were excited at 488 nm, 640 nm, and 405 nm, respectively, using a dichroic mirror (DM405/488/561/640). Their fluorescence signals were acquired sequentially in each scanning line. A scatter plot was generated between td5StayGold(c4)=GianCreg and GM130, Giantin, or TGN46 in each z-slice, and the plots across the z-range were merged. On the basis of the Otsu method, threshold values were automatically optimized for the fluorescence of td5StayGold and Alexa Fluor 647 using Fiji (https://fiji.sc). After exclusion of data points below the thresholds in both colors and data points showing signal saturation, colocalization was quantified by correlation analysis. The Pearson correlation coefficient (r) was determined using R (https://www.r-project.org). Three independent experiments (different immunostaining experiments) were carried out for each combination. See Extended Data Fig. 6.

Cytochemistry of the Golgi apparatus. HeLa cells were fixed 2 days after transfection with the cDNA of td5StayGold(c4)=GianCreg. Cell samples were imaged using an inverted laser scanning confocal microscopy system (FV3000, Evident, Tokyo, Japan) equipped with a 60× water objective lens (UPlanSApo 60×/1.2 NA). The size of the confocal aperture was 1 Airy disk. For a zoom factor of 1× and a pixel array size of 2048 × 2048, the size of each pixel was calculated to be 0.104 mm. Confocal images were acquired every 1.0 μm along the z-axis to create z-stacks (20 slices) that covered the entire Golgi apparatus. td5StayGold and DAPI were excited at 488 and 405 nm, respectively, using a dichroic mirror (DM405/488/561/640). See Fig. 5a.

Colocalization of Lifeact-mStayGold and F-tractin-mScarlet-I. Vero cells were cotransfected with the two plasmids. Cells were imaged live using an inverted laser scanning confocal microscopy system (TCS SP8, Leica) equipped with a 93× objective lens (HC PL APO 93X/1.30 GLYC motCORR objective lens). The pinhole size was 132 nm (back-projected size), and the size of each pixel was calculated to be 30 nm. See Supplementary Fig. 6.

Gene construction (inner mitochondrial membrane targeting). The mouse COX8a cDNA was amplified using primers containing 5'-BamHI and 3'-EcoRI sites, and the restricted product was cloned into the BamHI/EcoRI sites of pBS Coupler 4 (ref. 43) to generate pBS Coupler 4/COX8a. The mStayGold gene was amplified using primers containing 5'-HindIII and 3v-XhoI sites, and the restricted product was cloned in frame into the HindIII/XhoI sites of pBS Coupler 4/COX8a to generate pBS Coupler 4/COX8a=mStayGold. Lastly, the BamHI/XhoI fragment encoding COX8a=mStayGold was subcloned into pcDNA3 for transfection.

Gene construction (F-actin targeting). The rat F-tractin cDNA that corresponds to an N-terminal domain consisting of 41 amino acids was amplified using primers containing 5'-KpnI and 3'-XhoI sites, and the restricted product was cloned into the KpnI/XhoI sites of pBS Coupler 1 (ref. 43) to generate pBS Coupler 1/F-tractin. The mStayGold gene was amplified using primers containing 5'-BamHI and 3'-NotI sites, and the restricted product was cloned in frame into the BamHI/NotI sites of pBS Coupler 1/F-tractin to generate pBS Coupler 1/F-tractin=mStayGold. Lastly, the KpnI/NotI fragment encoding F-tractin=mStayGold was subcloned into pcDNA3 for transfection. The human Utrophin cDNA that corresponds to an N-terminal domain consisting of 261 amino acids was amplified using primers containing 5'-KpnI and 3'-XhoI sites, and the restricted product was cloned into the BamHI/NotI sites of pBS Coupler 1 (ref. 43) to generate pBS Coupler 1/UtrCH. The mStayGold gene was amplified using primers containing 5'-KpnI and 3'-XhoI sites, and the restricted product was cloned in frame into the KpnI/XhoI sites of pBS Coupler 1/UtrCH to generate pBS Coupler 1/mStayGold=UtrCH. Lastly, the KpnI/NotI fragment encoding mStayGold=UtrCH was subcloned into pcDNA3 for transfection.

Lattice SIM for live imaging. Super-resolution 3D-SIM images were acquired continuously on a ZEISS Elyra 7 equipped with a PlanApo 63×/1.46 NA oil immersion objective at 37 ºC. The Leap mode for lattice SIM was used to increase the temporal resolution of volumetric imaging. Image analysis was carried out with ZEN 2014 (ver. 9.1). See Fig. 5c, Supplementary Fig. 5, and Supplementary Video 8.

SpinSR10. Living cells on 35-mm glass-bottom dishes in HBSS containing 15 mM HEPES-NaOH (pH 7.4) were imaged using a SpinSR10 imaging system (Evident) built on an Evident inverted microscope (IX83P2ZF) equipped with an ORCA-Flash 4.0 V3 camera (Hamamatsu Photonics), a motorized stage (IX3-SSU) and a 100× oil objective lens (UPLAPO100XOHR, NA 1.50). With the SoRa spinning disk, the optical resolution in xy-plane at 488 nm excitation is approximately 160 nm. The total magnification of the system was considered to determine the best sampling interval of the camera (pixel binning).

SDSRM (spinning-disk super-resolution microscopy)

A 3.2× magnification changer was used for observing the Golgi apparatus (Fig. 5b, Supplementary Videos 4 and 5), cytoskeletons (Extended Data Fig. 7, Supplementary Fig. 7, Supplementary Videos 6 and 7), and IMM (Extended Data Fig. 8, Supplementary Fig. 8, Supplementary Videos 9–11). Among these, the following figures and videos were processed with deconvolution using a commercial algorithm “Olympus Sper Resolution (OSR)” to achieve super-resolution imaging: Extended Data Figs. 7 and 8; Supplementary Fig. 7; Supplementary Videos 6, 7, 9, 10 and 11.

spinning-disk LSCM (laser-scanning confocal microscopy)

The 3.2× magnification changer was not used for confocal imaging of condensin I (Fig. 4, Extended Data Fig. 5, Supplementary Videos 1–3).

When single-plane images were acquired rapidly, the autofocus function of a z-drift compensator (IX3-ZDC2, Evident) was set continuously active. Image acquisition and analysis were carried out using the Evident cellSens software (ver. 3.1.1).

Measurement of irradiance (W cm⁻²). The power of excitation light (W) above the objective at the focal plane was measured using a microscope slide power meter sensor (S170C; Thorlabs, Newton, NJ) and an optical power and energy meter (PM100D; Thorlabs, Newton, NJ). The power was divided by the area of the illumination field (cm²) to obtain the irradiance. In all cases of WF microscopy, the illuminator (collimator lens) was adjusted to achieve Köhler illumination. A color acrylic plate (Tokyu Hands, Japan) was placed at the focal plane to evaluate illumination uniformity on a CCD (CMOS) image.

References

46. Nagai, T. et al. A variant of yellow fluorescent protein with fast and efficient maturation for cell-biological applications. Nat. Biotechnol. 20, 87–90 (2002).

47. Karasawa, S., Araki, T., Yamamoto-Hino, M. & Miyawaki, A. A green-emitting fluorescent protein from Galaxeidae coral and its monomeric version for use in fluorescence labeling. J. Biol. Chem. 278, 34167–34171 (2003).

48. Szymczak, A.L. et al. Correction of multi-gene deficiency in vivo using a single ‘self-cleaving’ 2A peptide-based retroviral vector. Nat. Biotechnol. 22, 589–594 (2004).

49. Winter, G., Waterman, D. G., Parkhurst, J. M., Brewster, A. S., Gildea, R. J., Gerstel, M., Fuentes-Montero, L., Vollmar, M., Michels-Clark, T., Young, I. D., Sauter, N. K. & Evans, G. DIALS: implementation and evaluation of a new integration package. Acta. Cryst. D74, 85–97 (2018). doi: 10.1107/S2059798317017235

50. McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. & Read, R. J. Phaser crystallographic software (2007). J. Appl. Cryst. 40, 658–674 (2007). https://doi.org/10.1107/S0021889807021206

51. Afonine, P. V., Grosse-Kunstleve, R. W., Echols, N., Headd, J. J., Moriarty, N. W., Mustyakimov, M., Terwilliger, T. C., Urzhumtsev, A., Zwart, P. H. & Adams, P. D. Towards automated crystallographic structure refinement with phenix.refine. Acta. Cryst. D68, 352–367 (2012). https://doi.org/10.1107/S0907444912001308

52. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta. Cryst. D66, 486–501 (2010). https://doi.org/10.1107/S0907444910007493

53. Takagi, M., Ono, T., Natsume, T., Sakamoto, C., Nakao, M., Saitoh, N., Kanemaki, MT., Hirano, T. & Imamoto, N. Ki-67 and condensins support the integrity of mitotic chromosomes through distinct mechanisms. J. Cell Sci. 131, jcs202192 (2018).

54. Yesbolatova, A., Natsume, T., Hayashi, K. & Kanemaki, MT. Generation of conditional auxin-inducible degron (AID) cells and tight T control of degron-fused proteins using the degradation inhibitor auxinole. Methods 164–165, 73–80 (2019).

Table 1 is available in the Supplementary Files section.

Yes there is potential Competing Interest. R.A., M.H. and A.M. are inventors on Japanese patent application No. 2021-065373 that covers the creation and use of StayGold. The remaining authors declare no competing interests.

SupplementaryInformation.pdf
Supplementary Information
SupplementaryVideo1.mpg
Supplementary Video 1
SupplementaryVideo2.mp4
Supplementary Video 2
SupplementaryVideo3.mpg
Supplementary Video 3
SupplementaryVideo4.mpg
Supplementary Video 4
SupplementaryVideo5.mpg
Supplementary Video 5
SupplementaryVideo6.mpg
Supplementary Video 6
SupplementaryVideo7.mp4
Supplementary Video 7
SupplementaryVideo9.mpg
Supplementary Video 9
SupplementaryVideo10.mpg
Supplementary Video 10
SupplementaryVideo11.mpg
Supplementary Video 11
ExtendedDataFigure1.pdf
ExtendedDataFigure2.pdf
ExtendedDataFigure3.pdf
ExtendedDataFigure4.pdf
ExtendedDataFigure5.pdf
ExtendedDataFigure6.pdf
ExtendedDataFigure7.pdf
ExtendedDataFigure8.pdf
Table1.docx
AndoetalrevisedSupplementaryInformation.pdf
Supplementary Information updated

Download PDF

Journal Publication

published 30 Nov, 2023

Read the published version in Nature Methods →

Version 1

posted

You are reading this latest preprint version

StayGold variants for molecular fusion and membrane targeting applications

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results

Discussion

Declarations

References

Methods

Table

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1