HaloTag9: an engineered protein tag to improve fluorophore performance

HaloTag9 is an engineered variant of HaloTag7 with up to 40% higher brightness and increased fluorescence lifetime when labeled with fluorogenic rhodamines. Moreover, combining HaloTag9 with HaloTag7 and other fluorescent probes enabled live-cell multiplexing using a single fluorophore and the generation of a fluorescence lifetime-based biosensor. The increased brightness of HaloTag9 and its use in fluorescence lifetime multiplexing makes it a powerful tool for live-cell imaging.


Introduction
The need for brighter, more photostable, and spectrally diverse fluorophores for fluorescence microscopy has drawn the attention to small-molecule fluorophores, which exhibit superior photophysical properties than fluorescent proteins (FPs). Particularly, the combination of fluorogenic rhodamine-based fluorophores, which increase fluorescence upon target binding, with self-labeling protein (SLP) tags such as HaloTag7 has become a powerful tool for these purposes (Fig. 1A) [1][2][3] . The fluorogenicity of rhodamines is based on an environmentally sensitive equilibrium between a closed, non-fluorescent, spirocyclic and an open, fluorescent, quinoid form ( Fig. 1B) 4 . While numerous strategies have been introduced to control the open-closed equilibrium of rhodamines via chemical modifications [4][5][6][7] , little attention has been paid to the influence of the protein, even though the fluorogenic turn-on is mostly determined by the change in environment [8][9][10] . Consequently, SLP engineering offers an alternative approach to control the fluorogenicity and the photophysical properties of fluorogenic rhodamines. Here we describe the engineering of HaloTag7 to increase the brightness and fluorescence lifetime of bound rhodamines for live-cell microscopy.

Results
To engineer the rhodamine-binding site of HaloTag7, ten amino acids in close proximity to tetramethyl-rhodamine (TMR) on the HaloTag7-TMR crystal structure (PDB ID: 6Y7A, 1.4 Å) were chosen for randomization by site-saturation mutagenesis. The resulting libraries were screened for increases in brightness upon labeling with the fluorogenic fluorophore silicon rhodamine (SiR-CA) (Fig. 1B, Supplementary Fig. S1). This led to the identification of the double mutant HaloTag9 (HaloTag7-Q165H-P174R), which increased the brightness of SiR by 19.9±0.5% compared to HaloTag7 ( Supplementary Fig. S2, Supplementary Table S1-2). Characterization with a panel of 46 different fluorophores revealed that HaloTag9 increases the brightness of 14 rhodamines, including fluorophores based on the popular SiR, carbopyronine (CPY) and TMR scaffolds, which cover the spectral range between 550 and 650 nm (Fig. 1C, Supplementary Fig. S3, Supplementary  Table S3 -4). The most pronounced changes in fluorescence intensity (ΔI) for each scaffold were found for the fluorogenic rhodamines: JF614-CA 10 (ΔI = 102±11%), MaP618-CA 6 (ΔI = 31±4%), and MaP555-CA 6 (ΔI = 12.2±1.7%), rendering these fluorophores even more fluorogenic. These increases are comparable to those achieved through chemical modifications of rhodamines, for example through the introduction of azetidines (e.g. 5-42%) 11 . In addition to rhodamines, increases in brightness were also observed for a fluorescein derivative and one cyanine (Cy3-CA; Supplementary  Both are favorable for brightness as non-radiative decay pathways via rotational motion or photoinduced electron transfer are suppressed 12,13 . This mechanism helps to rationalize the relatively small effects on quantum yield observed for azetidine-bearing rhodamines as non-radiative decay pathways via rotational motion are already reduced for them 11 . Additionally, the two mutations (Q165H-P174R) influence the local electrostatic potential around the fluorophore, going from predominantly negative on HaloTag7 to slightly positive on HaloTag9 ( Supplementary Fig. S9). Such a change in the rhodamine's solvation shell might stabilize its carboxylate group, favoring the open form and therefore increasing the extinction coefficient.  We then investigated the brightness of labeled HaloTag9 relative to HaloTag7 in living U-2 OS cells by confocal (ΔIEnsemble) and fluorescence correlation spectroscopy (FCS) measurements (ΔIFCS). Nine rhodamines, which were brighter in vitro, were tested in cellulo and all except one showed significantly higher brightness when bound to HaloTag9 than to HaloTag7 ( Fig. 2A-B Fig. S12-14). Thus, HaloTag9 significantly increases the performance of fluorogenic fluorophores such as MaP618-CA in both live-cell confocal and STED microscopy and it is also expected to do so in other contexts such as biosensing (vide infra) or in vivo microscopy 7,14 .
Beyond improving fluorophore brightness, the change in fluorescence lifetime going from HaloTag7 (τMaP618 = 3.1 ns) to HaloTag9 (τMaP618 = 3.7 ns) creates the opportunity to use both tags simultaneously for fluorescence lifetime multiplexing using a single fluorophore (Fig. 2E) 15 . HaloTag7 and HaloTag9 were thus co-expressed as fusion proteins of the histone protein H2B and the outer mitochondrial membrane protein Tomm20 in U-2 OS cells. They were then labeled with a single fluorophore and the two tags distinguished by fluorescence lifetime imaging microscopy (FLIM) using phasor analysis 16 . We performed fluorescence lifetime multiplexing in two different spectral windows using either MaP555-CA or MaP618-CA and with different combinations of cellular targets either in fixed or living cells ( Supplementary Fig. S15-18). Furthermore, pairing the two HaloTags with probes such as MaP555-or MaP618-Actin, -Tubulin or -DNA 6 , allowed to multiplex even three species in living cells within the same spectral channel (Fig. 2F, Supplementary Fig. S19-20). Additionally, super-resolved images were acquired by STED-FLIM using only one depletion laser, while separating two species based on fluorescence lifetime information (Supplementary Fig. S21). To the best of our knowledge, this is the first time multitarget FLIM was achieved using a single fluorophore on a subcellular level in living cells, highlighting the potential of the combination of HaloTag9 and HaloTag7 for live-cell fluorescence lifetime multiplexing.  , 3 h). The brightness of the images was scaled to the expression level of the two proteins in the two cells. Scale bars, 10 μm. E Schematic view of fluorescence lifetime multiplexing using only one rhodamine for three targets (nucleus, mitochondria, and F-actin). Their spectral identity does not allow them to be distinguished but they can be separated using fluorescence lifetime information via phasor analysis. F Fluorescence lifetime multiplexing of U-2 OS cells expressing histone H2B as a HaloTag7 fusion and the outer-mitochondrial membrane protein Tomm20 as a HaloTag9 fusion. Both proteins were labeled with MaP555-CA (1 μM, 3 h). In addition, MaP555-Actin was used to label F-Actin (2 μM, 3 h). The composite, the total intensity and the three individual images with the separated structures are given. Scale bars, 10 μm. G Schematic overview of the LT-Fucci(CA) biosensor. During the G1 phase mainly HaloTag9-hCdt is present and the nuclei will therefore present long average photon arrival times (3.7 ns/orange). During S phase HaloTag7-hGem is predominant, resulting in shorter average photon arrival times ( As LT-Fucci biosensors only occupy one variable spectral channel, they are ideally suited for combination with other biosensors or probes. LT-Fucci(CA)-MaP618, for instance, can be combined with the green spectral region and, due to its narrow emission spectrum, also with the far-red region ( Supplementary Fig. S23) 6,19 . We thus simultaneously performed FLIM measurements of LT-Fucci(CA)-MaP618 and the RhoA GTPase activity biosensor Raichu-RhoA-CR (Clover-mRuby2) during cell division ( Supplementary Fig.  S24) 20 . The flexibility to choose the biosensor's color at the labeling step and the improved multiplexing capabilities sets LT-Fucci biosensors apart from the recently published FUCCI-Red (mKate2-mCherry) 19 .
In summary, HaloTag9 increases the brightness and fluorescence lifetime of fluorogenic rhodamines relative to HaloTag7. This improves its performance in fluorescence microscopy and opens up new possibilities for fluorescence lifetime multiplexing and the generation of fluorescence lifetime-based biosensors.

Data availability
Plasmids encoding for HaloTag9 and fusions thereof have been deposited on Addgene. Accession codes can be found in Supplementary Table S13. The X-ray crystal structure of HaloTag9-TMR has been deposited to the PDB with deposition code 6ZVY. Correspondence and requests for materials should be addressed to K.J.