Engineered HaloTag variants for fluorescence lifetime multiplexing

Self-labeling protein tags such as HaloTag are powerful tools that can label fusion proteins with synthetic fluorophores for use in fluorescence microscopy. Here we introduce HaloTag variants with either increased or decreased brightness and fluorescence lifetime compared with HaloTag7 when labeled with rhodamines. Combining these HaloTag variants enabled live-cell fluorescence lifetime multiplexing of three cellular targets in one spectral channel using a single fluorophore and the generation of a fluorescence lifetime-based biosensor. Additionally, the brightest HaloTag variant showed up to 40% higher brightness in live-cell imaging applications.

M ultiplexed fluorescence microscopy based on fluorescence lifetime is an attractive approach to image multiple targets simultaneously. It requires only one spectral channel and therefore frees up other channels for further multiplexing [1][2][3][4][5] .
To facilitate the use of fluorescence lifetime multiplexing in living cells, we aimed to establish a method in which different variants of a self-labeling protein (SLP) tag are labeled with a single fluorophore but become distinguishable based on their fluorescence lifetime. To achieve this, we focused on the combination of HaloTag7 with fluorogenic rhodamine-based fluorophores, which increase fluorescence upon target binding (Fig. 1a) [6][7][8][9] . The fluorogenicity of rhodamines is based on an environmentally sensitive equilibrium between a closed, nonfluorescent spirocyclic and an open, fluorescent quinoid form (Fig. 1b) 8 . While numerous strategies have been introduced to control the properties of rhodamines via chemical modifications 8,[10][11][12] , little attention has been paid to the influence of the protein [13][14][15][16] . However, SLP engineering offers an alternative approach to control not only the fluorogenicity but also the photophysical properties of fluorogenic rhodamines. Here, we describe the engineering of HaloTag7 to modulate the brightness and fluorescence lifetime of bound rhodamines for use in live-cell fluorescence lifetime multiplexing.

Results
HaloTag engineering and characterization. The rhodamine binding site of HaloTag7 is formed by three helices (6)(7)(8), as revealed by the crystal structure of tetramethyl-rhodamine (TMR) bound to HaloTag7 (Protein Data Bank (PDB) identification number (ID): 6Y7A, 1.4 Å) 17 . To engineer the binding site, ten amino acids on helices 6-8 in close proximity to TMR were chosen for randomization by site-saturation mutagenesis (Fig. 1c,d and Supplementary Fig. 1). The resulting libraries were expressed in Escherichia coli (cytoplasmic) and the cleared lysates screened for changes in brightness upon labeling with fluorogenic silicon rhodamine (SiR-CA). This led to the identification of both brighter (ΔI = (I var -I HT7 ) × I HT7 −1 > 0) and dimmer (ΔI < 0) HaloTag variants with similar labeling kinetics and thermostability as HaloTag7 (Extended Data Fig. 1, Supplementary Fig. 3 Tables 4-11). HaloTag9 increased the fluorescence intensity of 16 fluorophores, including rhodamines based on the popular SiR, carbopyronine (CPY) and TMR scaffolds. The most pronounced changes for each scaffold were found for the fluorogenic rhodamines: JF 614 -CA 15 (ΔI = 102 ± 11%), MaP618-CA 11 (ΔI = 31 ± 4%), and MaP555-CA 11 (ΔI = 12.2 ± 1.7%). The changes in intensity were found to stem from changes in quantum yield and/or extinction coefficient. HaloTag10 and HaloTag11 decreased the fluorescence intensity of almost all tested fluorophores compared with HaloTag7. The decreases were more pronounced for HaloTag10 than for HaloTag11 but in both cases caused by changes in quantum yield (Fig. 1e, Supplementary Figs.  . The changes in brightness were stable over the physiological pH range ( Supplementary Fig. 10).

3-4 and Supplementary
Solving the crystal structures of HaloTag9-TMR (PDB ID: 6ZVY, 1.4 Å), HaloTag10-TMR (PDB ID: 7PCX, 1.4 Å) and HaloTag11-TMR (PDB ID: 7PCW, 2.3 Å) allowed us to rationalize how the protein surface influences the spectroscopic properties of the fluorophore. HaloTag9 showed altered electrostatic surface potential, as well as changes in dihedral angles and rotational energy barriers of the bound TMR compared with HaloTag7, explaining changes in extinction coefficient and quantum yield. For HaloTag10 and HaloTag11, the proximity of TMR to the introduced tryptophans suggests that photoinduced electron transfer (PET) quenching of the fluorophore is responsible for the decreased quantum yields (Supplementary Figs. 5-9 and  Supplementary Tables 12-14) 18 .
Changes in fluorescence lifetimes. The measured changes in fluorescence lifetime of six fluorophores on the four HaloTags correlated with changes in quantum yield. Hence, HaloTag9 showed the longest and HaloTag10 the shortest fluorescence lifetime.
Labeling with MaP555-CA or MaP618-CA resulted in evenly spaced distributions of fluorescence lifetimes, ranging from 1.1 to 2.8 ns and from 1.6 to 3.7 ns, respectively (Fig. 1f,g and Supplementary Table 8). Either of the two MaP fluorophores is therefore particularly well suited for fluorescence lifetime multiplexing of up to three different HaloTag variants in one spectral channel (Fig. 2a) Fig. 4). To the best of our knowledge, this is the first example of multi-target FLIM using a single fluorophore on a subcellular level in living cells. Additionally, three-species images can also be accessed by multiplexing two HaloTag variants with the corresponding MaP-based F-actin, microtubule or DNA probes (Supplementary Figs. 33-34) 11 . Super-resolved images were acquired by STED-FLIM using only one depletion laser, while separating two species based on fluorescence lifetime information ( Supplementary Fig. 35). Additionally, fluorescence lifetime multiplexing can be combined with spectrally resolved detection, which allowed us to multiplex six species in living cells within two spectral channels each with three lifetime components ( Fig. 2e and Supplementary Fig. 36). For this, we expressed HaloTag7 as a fusion with the membrane-bound glycoprotein of the Golgi apparatus beta-1,4-galactosyltransferase (β4Gal-T1), HaloTag9 as a Tomm20 fusion, the lysosome-associated membrane glycoprotein 1 (LAMP1) as a HaloTag11 fusion, and the tyrosine protein kinase Lyn11, which localizes at the plasma membrane, as a SNAP-tag 19 fusion, and combined them with two noncovalent MaP555 probes targeting F-actin and DNA 11 .
Lifetime-based biosensors. Subsequently, we exploited the difference in fluorescence lifetime of HaloTag9 and HaloTag7 for the generation of a fluorescence lifetime-based biosensor to monitor cell cycle progression. Our design is based on Fucci biosensors, which rely on the cell cycle-dependent degradation of hCdt and Geminin (hGem) fragments fused to a green and red fluorescent protein (FP) 20  Considering that only a limited number of point mutants was screened in this work, we believe that, by targeting other residues for mutagenesis, further increases in fluorescence lifetime of labeled HaloTag9 should be feasible. The HaloTag variants developed enable live-cell fluorescence lifetime multiplexing of three species in a single spectral channel, freeing up other spectral channels for further multiplexing. Fluorescence lifetime multiplexing was demonstrated for combinations of various cellular targets, for different fluorophores and by combining two spectral channels to multiplex six species. Another possible advantage of using a single fluorophore for multiplexing is that it could eliminate challenges associated with differences in cell permeability, tissue penetration or metabolic clearance rates commonly encountered when using multiple fluorophores for multicolor experiments.
An additional application of the different HaloTag variants is their use in fluorescence lifetime-based Fucci biosensors, which were generated by fusing HaloTag variants with different fluorescence lifetimes to proteins with different cell-cycle-dependent degradation patterns. These biosensors occupy only one spectral channel that can be freely chosen and they can therefore be multiplexed with other biosensors in a straightforward manner. Finally, it might be possible to generate other lifetime-orthogonal tags from SNAP-tag-fluorophore combinations or fluorogen activating proteins 6 , further expanding the options for fluorescence lifetime multiplexing. Moreover, SLP engineering might prove beneficial to modulate other photophysical properties of fluorophores, such as excitation and emission spectra, giving access to tailor-made SLPs for specific applications. It is expected that the exploitation of the synergistic interaction between synthetic fluorophores and the proteins to which they are bound will enable the generation of additional tools for bioimaging.

Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/ s41592-021-01341-x.  Protein production and purification. Proteins were expressed in the E. coli strain BL21(DE3)-pLysS. LB cultures were grown at 37 °C to an optical density at 600 nm (OD 600nm ) of 0.8, induced by the addition of 0.5 mm isopropyl-β-d-thiogalactopyranoside (IPTG) and grown at 17 °C overnight in the presence of 1 mM MgCl 2 . The cells were harvested by centrifugation (4,500g, 10 min, 4 °C) and lysed by sonication (5 min, cycle 5, 70%, SonoPlus Bandelin). The cell lysate was cleared by centrifugation (70,000g, 20 min, 4 °C). Proteins were purified using affinity-tag Ni-NTA (Qiagen) leading to purity higher than 95% (verified by SDS-PAGE Coomassie staining). Proteins were finally concentrated using an Ultra-0.5 ml centrifugal filter device (Amicon) with a molecular weight cut-off according to the protein size, followed by buffer exchange into activity buffer (<0.1 mM Imidazole). The proteins were stored in a glycerol 45% solution at −20 °C or flash frozen and stored at −80 °C. Protein amino acid sequences are listed in the Supplementary Methods. For X-ray crystallography, proteins were produced as described above but purified using a HisTRAP FF affinity column (GE-Healthcare) on an ÄKTAPure M FPLC (GE-Healthcare). The proteins were concentrated using an Ultra-4mL centrifugal filter device (Amicon) and diluted to a final concentration of around 0.3 mg ml −1 (around 40 ml) in TEV-cleavage buffer (25 mM Na 2 HPO 4 , 200 mM NaCl). β-mercaptoethanol (10 μl) and TEV protease (mass ratio substrate:TEV 30:1, TEV protease produced and purified inhouse by A. Bergner) were added and incubated at 30 °C overnight. The solution was filtered (0.22 μm) and the cleaved protein was harvested by reverse purification on a HisTRAP FF affinity column (GE-Healthcare), collecting the flow through. Proteins were concentrated using an Ultra-4mL centrifugal filter device (Amicon) and further purified by size exclusion chromatography on a HiLoad 26/600 Superdex 75 pg column (GE-Healthcare) exchanging the buffer to activity buffer. Proteins were concentrated again and prepared to a final concentration of 5 μM in activity buffer; 3 mg of protein was incubated in presence of TMR-CA (10 μM) at room temperature overnight. The labeled protein was concentrated and an Illustra MicroSPin G-50 desalting column (GE-Healthcare) was employed to remove excess of unreacted fluorophore. The final protein concentration was adjusted to 13.0-15.0 mg ml −1 using the absorbance at 280 nm, correcting the extinction coefficient of the protein by ε 280, TMR-CA = 0.16. X-ray diffraction data collection and structure determination. Single crystal X-ray diffraction data were collected at 100 K on the X10SA beamline at the SLS (PSI). Data were processed with XDS 33 . The structures of HaloTag9-TMR, HaloTag10-TMR, or HaloTag11-TMR labeled with TMR were determined by molecular replacement (MR) using Phaser 34 and HaloTag7-TMR coordinates (6Y7A) as a search model. Geometrical restraints for TMR were generated using Grade server 35 . The final model was optimized in iterative cycles of manual rebuilding using Coot 36 and refinement using Refmac5 (ref. 37 ) and phenix.refine 38 . Data collection and refinement statistics are summarized in Supplementary Table 14 We performed structural analysis using PyMOL 40 , phenix 38  Polyacrylamide gel electrophoresis. HaloTag7 proteins (2 µM, 15 µl) were labeled using SiR-CA (10 µM) in activity buffer for 1 h at room temperature. After labeling, the proteins were separated by PAGE (4-20% ten-well Mini-Protean TGX, Bio-Rad) as recommended by the manufacturer and revealed by in gel fluorescence using a ChemiDoc MD Imaging System (Bio-Rad). SiR-CA labeled proteins were imaged using red epi illumination (695/55 nm), the proteins were revealed by Coommassie staining (Bio-Rad) and colorimetric imaging.
Library generation. The plasmid libraries consisting of site-saturation mutagenesis performed on specific sites were prepared using degenerated primers according to Kille et al. 43 . The degenerated primers (Eurofins) were mixed in a ratio of NDT:VHG:TGG = 12:9:1 (N = any base, D = A, G, or T, V = A, C, or G, and H = A, C, or T) and used for PCR amplification of two DNA fragments of pET5 1b(+)-His-tev-HaloTag7-EGFP. The saturation site belonged to an overlapping sequence between two DNA fragments. Plasmid libraries were prepared via Gibson assembly 24 . After electroporation in the E. coli strain E. cloni 10 G (Lucigen), library diversity was evaluated by serial dilution, plating on selective LB agar plates (100 µg ml −1 ampicillin) at 37 °C overnight and verification that more than 1,000 transformants were obtained by colony counting. Concomitantly, the library was isolated by plasmid extraction from a selective liquid LB culture (100 µg ml −1 ampicillin) performed at 37 °C overnight (Qiagen kit). The plasmid libraries were sequenced by the Sanger method (Eurofins) to verify proper incorporation of degenerate codons. Libraries were employed to transform E. coli strain BL21(DE3)-pLysS that were plated on selective LB agar plates (100 µg ml −1 ampicillin) at 37 °C overnight. Single colonies were used to inoculate 400 µl selective LB medium (100 µg ml −1 ampicillin) in a 96-deep well plate. Five wells were reserved for parental HaloTag7, five wells for CLIP-tag as a negative control and eight wells for sterility controls. The bacterial cultures were incubated at 37 °C overnight and 500 rpm. Then, 50 µl of the stationary phase cultures were employed to inoculate 950 µl selective LB medium (50 µg ml −1 ampicillin) in a 96-deep-well plate and incubated at 37 °C for 4 h at 500 rpm. The remaining culture was centrifuged (5,000g, 15 min, 4 °C) and stored at 4 °C. Protein expression was induced by addition of 0.5 mm IPTG and grown at 17 °C overnight at 500 rpm. The cells were harvested by centrifugation (5,000g, 15 min, 4 °C). The bacterial pellets were submitted to two cycles of freeze/thawing before resuspension in 300 μl lysis buffer (50 mM K 2 HPO 4 pH 8, 1 mg ml −1 lysozyme, 2 mM MgCl 2 and 2.5 units ml −1 benzonase (Turbo Nuclease, Jena Bioscience)) at 37 °C for 1 h. The cell lysate was cleared by centrifugation (5,000g, 20 min, 4 °C). The cleared supernatant was transferred into nonbinding black bottom 96-well plates for the screening assays.
Screening assay. Cell lysates (20 μl) were diluted in a nonbinding black bottom 96-well plate into activity buffer (100 μl final, 0.5 mg ml −1 bovine serum albumin (BSA; Sigma)). Background SiR (620/20 excitation, 680/30 emission) and GFP (485/20 excitation, 535/25 emission) fluorescence intensity were measured before spiking SiR-CA (5 μl) in each well (5 nM final SiR-CA concentration). After incubation at room temperature for 1 h, fluorescence intensities were measured again. We performed an additional second labeling step as previously described, reaching 10 nM final SiR-CA concentration and intensities were again measured. The five control wells allowed to access the mean and s.d. GFP and SiR fluorescence intensities of the parental protein. Wells with GFP intensities lower than 10% of the control were discarded for the screening (expression too low). Wells with SiR fluorescence intensities three or two times s.d. smaller/ larger than the control (first round: mean par ± 3 s.d. par , second and third round: mean par ± 2 s.d. par ) were selected for further characterization. Plasmids of selected wells were obtained from stored bacterial cultures and sequenced. Selected variants for characterization were produced and purified from 50 ml selective LB cultures (as described above).
Each protein (1 μM) was labeled with SiR-CA in 100 μl activity buffer (containing 0.5 mg ml −1 BSA (Sigma)) in a nonbinding black bottom 96-well plate and incubated for 2 h at room temperature. The SiR and GFP fluorescence intensities were measured as previously described. Measurements were performed in triplicate, performing the independent labeling reactions in three separate wells. Mean and 90% confidence intervals were calculated for every variant and compared with those of the parental protein. Variants with significant changes in SiR fluorescence intensity compared with the parental protein were picked for further characterization (one-sided t-test, α = 5%, d.f. = 4).
Fluorescence intensity characterization. The most promising variants were subcloned into a pET51b(+) vector without the C-terminal EGFP fusion. Variant proteins were produced and purified from selective LB cultures (500 ml) as described above.
Fluorophores (Supplementary Table 4) were distributed into a nonbinding black bottom 96-well plate (100 μl, 100 nM) and incubated at room temperature overnight. The next day, 100 μl protein (2 μM, activity buffer containing 0.5 mg ml −1 BSA (Sigma)) was added to the fluorophore and incubated for 4 h at room temperature. The respective fluorescence intensities were measured with a plate reader (TECAN Spark 20M). The labeling and measurements were performed in quadruplicates. Mean and 95% confidence intervals were calculated for every variant and compared with those of the parental protein (one-sided t-test, α = 5%, d.f. = 6). Fluorescence excitation and emission spectra were measured using a plate reader (TECAN Spark 20M). D 50 measurements. The free acid of the fluorophore (5 μM) was diluted in 200 μl water-dioxane mixtures of 0%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, and 100% in clear-bottom polypropylene 96-well plates (Greiner Bio-One). Absorbance spectra were recorded on a plate reader (TECAN Spark 20M) from 400 nm to 700 nm. Measurements were performed in triplicate. Data were baseline corrected and the absorbance values at the maximal absorbance wavelength (λ max ) were plotted against the dielectric constants of water-dioxane mixtures 44 . The data were fitted with a sigmoidal curve (1) and the D 50 was determined as the point of inflection (x c ).
Kinetics by plate reader. Labeling kinetics of HaloTag variants (80 nM) were measured by time course fluorescence anisotropy measurements using TMR-CA (20 nM) in 200 µl activity buffer supplemented with 0.5 mg ml −1 BSA at room temperature and on a plate reader (TECAN Spark 20M) using the above stated filters for TMR and an injector system. G-factor and gain were calculated from three control measurements (buffer only, fluorophore in buffer, fully labeled protein in buffer). The data were fitted with a monoexponential function (2), where y 0 corresponds to the y offset, x 0 to the x offset, A to the amplitude and τ to the time constant. Using equation (3), the apparent second-order rate constant (k app ) was calculated using the initial protein concentration [P] 0 . Fitted parameters are reported as means from at least two measurements.
Kinetics by stopped-flow. Labeling kinetics of HaloTag7 and HaloTag9 with TMR-CA were measured by recording fluorescence anisotropy changes over time using a BioLogic SFM-400 stopped-flow instrument (BioLogic Science Instruments) in single mixing configuration at 37 °C. Monochromator wavelengths for excitation was set to 555 nm and a 570-nm long pass filter was used for detection. Protein and substrates were mixed in a 1:1 stoichiometry in activity buffer supplemented with 0.5 mg ml −1 BSA. Concentrations were varied from 0.125 µM to 0.5 µM. The anisotropy of the free substrate was measured to obtain a baseline. The dead time of the instrument was measured according to the manufacturer's protocol (BioLogic Technical note no. 53) by recording the fluorescence decay during the pseudo first-order reaction of N-acetyl-l-tryptophanamide with a large excess of N-bromosuccinimide and fitting the data to the first-order reaction rate law. Recorded data were processed removing pretrigger time points and averaging replicates. The data were fit to a two-stage kinetic model (equations (4) and (5)) using the DynaFit software 45 . Baseline anisotropy of the free fluorophore, substrate concentrations and dead time of the instrument were taken into account. The s.d. (normal distribution verified) and confidence intervals of fitted parameters were estimated with the Monte Carlo method 46 with standard settings (N = 1,000, 5% worst fits discarded). The derived parameters K D (dissociation constant) and k app (apparent second-order rate constant) were calculated according to equations (6) and (7).
Thermostability. Thermostability of His-tev-HaloTag7, His-tev-HaloTag9, His-tev-HaloTag10, or His-tev-HaloTag11 was measured at 0.5 mg ml −1 in activity buffer on a nanoscale differential scanning fluorimeter Prometheus NT 48 (NanoTemper) over a temperature range from 20 °C to 95 °C with a heating rate of 1 °C min −1 by monitoring changes in the ratio of the fluorescence intensities at 350 nm and 330 nm. The indicated melting temperature (mean ± s.d., N = 2 samples) corresponds to the point of inflection (maximum of the first derivative).
Quantum yield. Quantum yields were determined using a Hamamatsu Quantaurus QY. Fluorophores (0.5 μM) were added directly to the target protein (2.5 μM) in activity buffer. After incubation for 4 h at room temperature, quantum yields were measured. Except for Cy-3, where proteins (5 μM) were labeled with fluorophores (1 μM) in activity buffer for 12 h at room temperature and an Illustra MicroSPin G-50 desalting column (GE-Healthcare) was employed to remove excess of unreacted fluorophore.
Extinction coefficient measurements. Proteins (12 μM) were labeled with fluorophores (6 μM) in activity buffer for 3 h at room temperature. The absorbance spectra of labeled proteins (0.5 μM, 1.0 μM, 1.5 μM, 2.0 μM, and 3.0 μM) were recorded in clear-bottom nonbinding 96-well plates (200 μl) on a plate reader (TECAN Spark 20M) from 400 nm to 700 nm. Data were baseline corrected and the maximum absorbance values plotted against concentration. The data were fitted to a linear function (equation (8)) and the extinction coefficients were calculated from the slope b.
Computational chemistry. We optimized the quinoid and the spirolactone form of TMR at the B3LYP/6-31G(d,p) level of theory using the software package Gaussian v.16 (ref. 47 ). Solvent effects were modeled using the polarizable continuum model SMD. Molecules were visualized using the Avogadro software 48 . We performed molecular modeling of TMR on the surface of HaloTag7 or HaloTag9 using MacroModel 49 and the Protein Preparation Wizard 50 , both part of Maestro 51 (Schrödinger Software). Relative energies were calculated by molecular modeling using the force field OPLS3e (ref. 52 ) in water constraining the protein as well as the remainder of the ligand apart from NMe 2 .
Cell culture and transfection. U-2 OS (ATCC) and U-2 OS Flp-In T-REx Cep41-HaloTag7 (ref. 25 ) cells were cultured in high-glucose phenol-red free DMEM (Life Technologies) medium supplemented with GlutaMAX (Life Technologies), sodium pyruvate (Life Technologies) and 10% FBS (Life Technologies) in a humidified 5% CO 2 incubator at 37 °C. Cells were split every 3-4 days or at confluency. Cell lines were tested regularly for mycoplasma contamination. Cells were seeded on eight-well glass bottom dishes (Ibidi) at three to one days before imaging. Transient transfections were performed using Lipofectamine 2000 reagent (Life Technologies) according to the manufacturer's recommendations: the DNA (0.3 μg) was mixed with OptiMEM I (10 μl, Life Technologies) and Lipofectamine 2000 (0.75 μl) was mixed with OptiMEM I (10 μl). The solutions were incubated for 5 min at room temperature, then mixed and incubated for an additional 20 min at room temperature. The prepared DNA-Lipofectamine complex was added to one of the wells in an eight-well glass bottom dish with cells at 50-70% confluency. After 12 h incubation in a humidified 5% CO 2 incubator at 37 °C, the medium was changed to fresh medium. The cells were incubated under the same conditions for 24-48 h before imaging. Confocal microscopy. Confocal fluorescence microscopy was performed on a Leica SP8 FALCON microscope (Leica Microsystems) equipped with a Leica TCS SP8 X scanhead; a SuperK white light laser, Leica HyD SMD detectors, a HC PL APO CS2 ×20/0.75 dry objective, a HC PL APO CS2 ×40/1.10 water objective and a water immersion microdispenser. Emission was collected as indicated in Supplementary Table 19. The microscope was equipped with a CO 2 and temperature controllable incubator (Life Imaging Services, 37 °C).
For brightness comparison, stable cell lines expressing HaloTag variants or HaloTag7 in the cytosol were seeded and labeled as described above. Cells were focused in the GFP channel and z-stacks were recorded with 1 μm step size over 22 μm. The summed stacks were analyzed as follows: the mean intensity of a rectangular region of interest (ROI) within the cell was normalized by the GFP intensity using a custom written Fiji macro 54,55 .
Images of Cep41 were acquired transiently by transfecting U-2 OS cells with CEP41-HaloTag7-T2A-EGFP or CEP41-HaloTag9-T2A-EGFP and labeling them as described above. Cells were focused in the GFP channel and z-stacks were recorded. The intensity of the MaP618-CA signal was normalized by the EGFP signal, and the resulting values compared for the two HaloTags. Cep41-Halo images were rescaled to the expression levels using the EGFP intensity values from within a ROI over the entire cell area and depicted using the same brightness and contrast settings.
Photostability measurements were performed using a PMT detector to collect emission. Stable cell lines expressing HaloTag7, HaloTag9, HaloTag10, or HaloTag11 as H2B fusions were seeded and labeled as described above. Cells were focused and a z-stack was recorded with 2 μm step size over 22 μm, using a pinhole of 5 Airy Unit (AU) and 2% (630 nm, SiR), 1%, 1.5% or 5% (615 nm, CPY, MaP618, or JF 614 ) and 2% (555 nm, TMR, MaP555) laser intensity. This was followed by the acquisition of eight consecutive photobleaching frames in the focal plane at 100% laser intensity. Z-stack and photobleaching was repeated 60 times. The summed stacks were analyzed as follows: the mean intensity of ROIs around the nuclei were normalized to the mean intensity found at t 0 .
Fluorescence excitation and emission spectra of MaP555-CA, MaP555-Actin, MaP555-Tubulin, MaP555-DNA, MaP555-SNAP, MaP618-CA, MaP618-Actin, and MaP618-DNA were measured in live U-2 OS cells expressing HaloTag variants in the cytosol or blank U-2 OS cells. MaP555 excitation: exciting at 475-575 nm in 2-nm steps collecting at 595-700 nm. MaP555 emission: exciting at 520 nm and collecting at 530-627 nm in 3-nm steps with a bandwidth of 10 nm. MaP618 excitation: exciting at 550-650 nm in 2-nm steps collecting at 670-780 nm. MaP618 emission: exciting at 600 nm and collecting at 610-707 nm in 3-nm steps with a bandwidth of 10 nm. FCS measurements. FCS was performed on a Leica SP8 FALCON microscope (as described above) using a HC PL APO CS2 ×40/1.10 water objective with a motorized correction collar. FCS traces (30 s) were measured in U-2 OS cells expressing either HaloTag7 or HaloTag9 in the cytosol (no induction). The cells were labeled with MaP618-CA (150 nM, 2 h) and washed twice for 1 min. Excitation and emission collection was performed as indicated in Supplementary  Table 19. Five traces per cell and a total of 36 cells from three biological replicates were measured. Data analysis (correlation and fitting) was performed using the LAS X software (Leica Microsystems) fitting a free three-dimensional diffusion model including a triplet component 56 . The diffusion amplitude (G(0)) as well as the mean photon counts (PC) over the 30 s trace were used to calculate the molecular brightness (mB, equation (9)).
STED microscopy. Imaging was performed on an Abberior easy3D STED/ RESOLFT QUAD scanning microscope (Abberior Instruments) built on a motorized inverted microscope IX83 (Olympus). The microscope was equipped with a pulsed STED lasers at 775 nm, and with a 640-nm excitation laser. Spectral detection was performed with avalanche photodiodes (APD) in the following spectral window: 650-725 nm. Images were acquired with a ×100/1.40 UPlanSApo Oil immersion objective lens (Olympus). Pixel size was 25 nm for all images. Laser powers and dwell times were kept constant so HaloTag9 and HaloTag7 images were comparable. For quantification of microtubule diameter, the perpendicular line profile of a 200-nm-wide microtubule section was measured. The profile was fitted with a Gaussian function (equation (10)) and the full-width at half-maximum (FWHM) derived thereof (equation (11)).
For STED laser only images, cells were focused using excitation only, followed by the sequential acquisition of a STED laser only image (excitation laser: off, STED laser: on) and a normal STED image (excitation laser: on, STED laser: on).
For photobleaching measurements, 31 consecutive STED frames were acquired and the intensity within a rectangular ROI was compared over time. Movement of mitochondria out of, or into, the ROI was neglected.
Fluorescence lifetime imaging microscopy. FLIM was performed on a Leica SP8 FALCON microscope (as described above) at a pulse frequency of 80 MHz unless otherwise stated. Emission was collected as indicated in Supplementary Table 19.
For determination of fluorescence lifetime, cells stably expressing HaloTag variants in the cytosol were imaged, collecting 1,000 photons per pixel. The acquired images of cells were thresholded to remove background signal from empty coverslip space. Mean fluorescence lifetimes were calculated in the LAS X software (Leica Microsystems) by fitting mono, bi, or triexponential decay models (n-exponential reconvolution) to the decay (χ 2 < 1.2).
For determination of average fluorescence lifetimes on different subcellular targets, cells were transiently transfected with the HaloTag constructs and imaged, collecting 500 photons per pixel. The acquired images were processed as described above.
Structural images (species separation) were acquired as indicated in Supplementary Table 19 and species separation was performed via phasor analysis, positioning the cluster-circles on the phasor plot at the position of the pure species (Leica Microsystems) 2,57,58 . Images in Supplementary Fig. 24 were analyzed using Pattern Matching in SymPhoTime64 (PicoQuant).
Long-term cell cycle measurements were performed using the Navigator function of the SP8. Cells were labeled with MaP618-CA (1 μM, no wash) and imaged after 1 h. FLIM images were acquired every 12 min and an autofocus z-stack measurement was performed before every image. Water immersion was controlled using the water immersion microdispenser. Biosensor multiplexing was performed on the same setup, acquiring FLIM images in both channels every 5 min for 2-4 h. Images were analyzed in LAS X (Leica Microsystems) applying pixel binning (equation 2). FastFLIM images (average photon arrival times per pixel) are shown. Additionally, we used phasor analysis to evaluate the different cell populations.
TauContrast microscopy. TauContrast microscopy was performed on a STELLARIS 8 FALCON microscope (Leica Microsystems, FALCON system only used for comparison). The microscope was equipped with a White Light Laser with tunable excitation wavelengths 440-790 nm operating at 80 MHz. Spectral detection was performed with Power HyD X photon-counting detectors in the following spectral window: 630-700 nm. Images were acquired with a ×86/1.20 STED WHITE water immersion objective lens (Leica Microsystems). Pixel size was 176 nm for all images. TauContrast images were analyzed using the LAS X software. For comparison FastFLIM images were acquired simultaneously using the FALCON system. STED-FLIM microscopy. Line sequential, confocal/STED-FLIM imaging was performed on a STELLARIS 8 STED FALCON microscope (Leica Microsystems). The microscope was equipped with a White Light Laser with tunable excitation wavelengths 440-790 nm operating at 80 MHz, and a pulsed STED laser at 775 nm operating at 80 MHz. Spectral detection was performed with Power HyD X photon-counting detectors in the following spectral window: 630-760 nm. Images were acquired with a ×86/1.20 STED WHITE water immersion objective lens (Leica Microsystems). Pixel size was 20 nm for all images. Images were analyzed using species separation via phasor analysis available in FALCON through the LAS X software.

Statistics and reproducibility.
Representative microscopy images are shown and unless otherwise stated experiments were performed on three different days.
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

data availability
Plasmids encoding HaloTag variants and fusions thereof have been deposited with Addgene. Accession codes can be found in Supplementary Table 18. The X-ray crystal structures of HaloTag9-TMR, HaloTag10-TMR, and HaloTag11-TMR have been deposited with PDB under deposition codes 6ZVY, 7PCX, and 7PCW. HaloTag7-TMR is available at PDB with deposition code 6Y7A. Correspondence and requests for materials should be addressed to K.J. Source data are provided with this paper. The data supporting the findings of this study are available within the paper and its Supplementary Information and are available from the corresponding author upon reasonable request.

Code availability
Custom written Fiji macros are available upon request to K.J. Fig. 1 | HaloTag7 engineering strategy and results. a Schematic representation of the in vitro engineering of HaloTag7. Site-saturation mutagenesis was performed onto the plasmid containing His x10 -HaloTag7-EGFP using degenerate primers and Gibson cloning. The plasmid library (for example P174X) was transformed in E. coli for protein production and extraction. The cell lysate (protein concentration estimated via EGFP signal: 50-150 nM) was labeled with a limiting amount of SiR-Ca (5 nM) and screened for increases in fluorescence intensity compared to HaloTag7. Brighter or dimmer hits (increased/decreased fluorescence intensity of SiR on variants in comparison to SiR on HaloTag7) were sequenced and validated in a separate fluorescence assay using purified protein.

Extended Data Fig. 5 | Multiplexed FLIM of the LT-Fucci(Ca) and Raichu-Rhoa-CR biosensors. a-e
The activity of the GTPase Rhoa as well as the cell cycle stage were followed over time for dividing cells. FLIM images are given for both Raichu-Rhoa-CR and LT-Fucci(Ca) labeled with MaP618-Ca (1 μM) along with a transmission light image and the phasor plot 1 h 20 min before cytokinesis (a), 30 min before cytokinesis (b), shortly after cytokinesis (c), and 1 h 20 min after cytokinesis (d). The FLIM images are color-coded according to the scales in (e). The two biosensors could be easily multiplexed measuring fluorescence lifetime in the GFP channel (Clover-Raichu-Rhoa-CR) and the CPy channel (LT-Fucci(Ca)-MaP618). as expected mitotic cells initially showed low Rhoa activity indicated by long donor fluorescence lifetimes (2.7 ns). However, closer to cytokinesis Rhoa activity started to increase (2.6 ns) until a donor fluorescence lifetime of 2.5 ns was reached directly after cytokinesis. after division the cells reached even higher Rhoa activity (2.2 ns) 69 . Scale bar, 10 μm.