A general method to optimize and functionalize red-shifted rhodamine dyes

Abstract

Expanding the palette of fluorescent dyes is vital to push the frontier of biological imaging. Although rhodamine dyes remain the premier type of small-molecule fluorophore due to their bioavailability and brightness, variants excited with far-red or near-infrared light suffer from poor performance due to their propensity to adopt a lipophilic, nonfluorescent form. We report a framework for rationalizing rhodamine behavior in biological environments and a general chemical modification for rhodamines that optimizes long-wavelength variants and enables facile functionalization with different chemical groups.

Introduction

The development of hybrid small-molecule:protein labeling strategies enable the use of chemical fluorophores in living cells and in vivo¹. Optimizing small-molecule dyes for these complex biological environments is important, as synthetic fluorophores are often brighter and more photostable than fluorescent proteins². We recently developed general methods to improve³ and fine-tune⁴ rhodamine fluorophores by incorporating four-membered azetidines into the structure, yielding the ‘Janelia Fluor’ dyes. Although our existing tuning strategies allow optimization of short-wavelength rhodamines, we discovered these methods cannot be applied to analogs excited with far-red and near-infrared (NIR) light due to their propensity to adopt a colorless form. Here, we report a new complementary tuning strategy that allows rational optimization of a broader palette of fluorophores. This general method also serves as a basis for facile functionalization, enabling the synthesis of novel cell- and tissue-permeable rhodamine labels for biological imaging experiments.

Results

A general rubric to predict rhodamine performance.

A key property of rhodamine dyes is an equilibrium between a lipophilic, colorless lactone and the polar, fluorescent zwitterion (Fig. 1a)⁵. Based on our previous work^{2–4, 6, 7}, we outlined a general rubric that directly correlates the lactone–zwitterion equilibrium constant (K_L–Z) to performance in biological environments (Fig. 1b). Dyes with high K_L–Z exist almost exclusively in the zwitterionic form, making them useful as environmentally insensitive biomolecule labels⁸. Rhodamines with intermediate K_L–Z values exhibit improved cell and tissue permeability due to the modestly higher propensity of the molecule to adopt the lipophilic lactone form and rapidly traverse biological membranes^{4, 7}. Dyes exhibiting even smaller K_L–Z values preferentially adopt the closed lactone form, which can be exploited to create ‘fluorogenic’ dyes^{7, 9–11} as binding of ligands and stains to their cognate biomolecular targets typically shifts the equilibrium to the fluorescent form. This property also decreases in vivo utility, however, due to problems with solubility and sequestration in membranes. Finally, dyes with extremely small K_L–Z values exist completely in the nonfluorescent lactone form, rendering them effectively unusable in biological experiments.

Figure 1. General rubric relating lactone–zwitterion equilibrium to rhodamine dye performance and optimizing short-wavelength dyes by decreasing K_L–Z.

(a) The lactone–zwitterion equilibrium of Janelia Fluor rhodamine dyes and the equilibrium constant K_L–Z. (b) Phenomenological plot categorizing the properties of different dyes based on K_L–Z. (c) Synthesis, structures, and λ_abs of dyes 1–8. (d) Plot of K_L–Z vs. λ_abs for dyes 1–8 and general tuning strategies for dyes with short or long λ_abs. (e) Structures of dyes 9–11. (f) Plot of K_L–Z vs. λ_abs showing decreased K_L–Z for dyes 9–11. (g) Structure of JF₄₇₉–HaloTag ligand (11_HTL). (h) Fluorescence excitation (ex) and emission (em) spectra of 11_HTL:HaloTag conjugate and GFP. (i) Two-color confocal imaging experiment of live U2OS cells expressing HaloTag–histone H2B labeled with either 11_HTL (500 nM, 3.5 h) or JF₅₀₃–HaloTag ligand (12_HTL; 500 nM, 3.5 h) excited with 488 nm (left panels) and TOMM20–SNAP-tag labeled with JF₅₂₅–cpSNAP-tag ligand (9_STL; 100 nM, 30 min, 3× wash) excited with 532 nm (right panels); scale bars: 10 μm; this experiment was duplicated with similar results. White arrows highlight the absence (11_HTL) or presence (12_HTL) of nuclear signal in the 532 nm-excited channel. (j) Structures of dyes 13–14. (k) Plot of K_L–Z vs. λ_abs showing decreased K_L–Z for dyes 13–14. (l) Structures of 2_STL and 13_STL. (m) Plot of fluorescently labeled mouse embryonic stem (ES) cells (%) vs. time determined by flow cytometry. Wild-type ES cells (WT) or ES cells expressing SNAP-tag–histone H2B (ST) were incubated with 2_STL or 13_STL (15 min); error bars show SE; experiments using 2_STL n = 7 independent cellular samples except for [ligand] = 3 nM where n = 5 independent cellular samples; experiments using 13_STL n = 3 independent cellular samples. (n) Plot showing fraction of chromatin-bound molecules per cell in single-particle tracking experiments using 2_STL and 13_STL in cells expressing SNAP-tag–histone H2B and 2_HTL in U2OS cells expressing HaloTag–histone H2B; center line indicates median; box limits indicate upper and lower quartiles; whiskers indicate min–max; one-way ANOVA gave adjusted P Value = 0.0013 (**) for 2_STL vs. 13_STL and adjusted P Value = 0.9963 (ns) for 13_STL vs. 2_HTL; n = 12 cells for experiments using 2_STL and n = 8 cells for experiments using 13_STL and 2_HTL.

We compared a series of Janelia Fluor rhodamine analogs with different fluorophoric systems (1–8, Fig. 1c). Compounds 2, 4, and 5 were described previously and include the azetidine-containing rhodamine (2), which we termed ‘Janelia Fluor 549’ (JF₅₄₉), and the carborhodamine^{12, 13} and Si-rhodamine^{9, 14} analogs 4 and 5 (JF₆₀₈ and JF₆₄₆, respectively; Fig. 1c)³. We expanded the wavelength range of the JF dyes by synthesizing new azetidine-containing analogs of known rhodamine structures using metalation of bis(2-bromoarenes) (Fig. 1c, Supplementary Note), which we previously established as a general method for rhodamine synthesis⁶. These included compounds based on classic dyes containing nitrogen^15–17 (X = NCH₃; 1) and sulfur¹⁸ (X = S; 3) atoms as well as recently described variants containing phosphinate^{19, 20} (X = PO₂H; 6), phosphine oxide^{21, 22} (X = P(O)Ph; 7), and sulfone²³ (X = SO₂; 8) moieties. We measured the absorption maximum (λ_abs), extinction coefficient at λ_abs (ε), fluorescence emission maximum λ_em, and fluorescence quantum yield (Φ) of these dyes in aqueous buffer and the K_L–Z in a dioxane:water mixture (Table 1). Comparing K_L–Z and λ_abs uncovered an inverse correlation (Fig. 1d), with the short wavelength NCH₃-containing JF₅₀₂ (1) exhibiting a high K_L–Z = 4.33 and the near-infrared (NIR) dyes containing P(O)Ph (7) and SO₂ (8) showing a low K_L–Z ≈ 10^–4. This correlation likely stems from the electron-withdrawing character of the X substituents²⁴ as recently demonstrated in a computational study of rhodamines containing O, C(CH₃)₂, and Si(CH₃)₂ moieties¹¹.

Table 1. Properties of Janelia Fluor dyes 1–11, 13–20, and 37–38.

All properties were measured in 10 mM HEPES, pH 7.3 except for K_L–Z, which was determined in 1:1 dioxane:H₂O. Properties for 2, 4, 5, 9, 10, 13, and 14were taken from previous work^{3, 4, 7}. ND: not determined due to low absorbance.

Compound	Name	X	Y	Z	NR2	λabs (nm)	ε (M–1cm–1)	λem (nm)	Φ	KL–Z
1	JF502		H	H		502	57,800	533	0.71	4.33
2	JF549		H	H		549	101,000	571	0.88	3.47
3	JF570		H	H		570	83,600	593	0.63	2.24
4	JF608		H	H		608	99,000	631	0.67	0.091
5	JF646		H	H		646	5,600	664	0.54	0.0014
6	JF668		H	H		668	26,700	687	0.34	0.0093
7	–		H	H		~704	<200	~723	ND	<0.0001
8	–		H	H		ND	<100	ND	ND	<0.0001
9	JF525		H	H		525	94,000	549	0.91	0.068
10	JF585		H	H		585	1,500	609	0.78	~0.001
11	JF479		H	H		479	47,900	517	0.62	2.88
13	JF552		F	H		552	95,000	575	0.83	0.70
14	JF526		F	H		526	19,000	550	0.87	0.005
15	JF669		H	F		669	112,000	682	0.37	0.262
16	JF690		H	F		690	150,000	707	0.24	2.90
17	JF722		H	F		722	87,200	743	0.11	0.026
18	JF724		H	F		724	6,600	748	0.05	~0.001
19	JF571		H	F		571	101,000	590	0.78	7.93
20	JF593		H	F		593	90,300	612	0.55	6.06
37	JF559		H	F		559	106,000	579	0.85	6.22
38	JF711		H	F		711	12,400	732	0.17	~0.001

Optimizing short-wavelength dyes.

In previous work, we focused on tuning the K_L–Z lower to improve tissue permeability and create fluorogenic ligands based on short-wavelength dyes. This allowed us to estimate thresholds between different categories of dyes based on measured K_L–Z values (Fig. 1b,d)^{4, 7, 10}. One general strategy to decrease K_L–Z involved incorporation of 3,3-difluoroazetidines⁴. This modification also elicits a concomitant hypsochromic shift of ~25 nm, transforming JF₅₄₉ (2; K_L–Z = 3.5) into JF₅₂₅ (9; K_L–Z = 0.68, Fig. 1e, Table 1). The HaloTag²⁵ ligand based on JF₅₂₅ shows improved cell permeability relative to the JF₅₄₉ derivative⁴, which is consistent with our K_L–Z rubric (Fig. 1b,f), and is blood–brain-barrier (BBB) permeable making it useful for in vivo voltage imaging using the Voltron indicator². This strategy also transformed JF₆₀₈ (4; K_L–Z = 0.091) into the fluorogenic JF₅₈₅ (10; K_L–Z = 0.001, Fig. 1e, Table 1), again supporting our K_L–Z-based framework (Fig. 1b,f). We applied this same tuning approach to JF₅₀₂ (1) to yield the fluorinated JF₄₇₉ (11; Extended Data Fig. 1a–c). Given the high K_L–Z = 4.33 for 1, however, this modification only moderately decreased K_L–Z to 2.88 (Fig. 1e–f, Table 1) and the JF₄₇₉–HaloTag ligand derivative (11_HTL, Fig. 1g) exhibited similar cell permeability to our previously described rhodol-based JF₅₀₃–HaloTag ligand (12_HTL; Extended Data Fig. 1d–i). Although the molecular brightness of the parent JF₄₇₉ (11; ε = 47,900 M^–1cm^–1, Φ = 0.62; Table 1) is lower than JF₅₀₃ (ε = 83,000 M^–1cm^–1, Φ = 0.87)⁴, the shorter λ_abs is advantageous for multicolor imaging experiments. JF₄₇₉–HaloTag ligand (11_HTL) exhibits similar spectral properties to enhanced green fluorescent protein (GFP) when attached to the HaloTag protein (Fig. 1h). This property allows efficient excitation of 11_HTL with 488 nm light but no appreciable signal when excited with 532 nm light, allowing two-color imaging with JF₅₂₅–cpSNAP-tag ligand (9_STL; Fig. 1i, Extended Data Fig. 2a–e)⁴ In contrast, the longer λ_abs of 12_HTL results in unwanted excitation by 532 nm light, which makes spectral separation difficult when paired with 9_STL (Fig. 1i, Extended Data Fig. 2b,f–h).

Another method to decrease K_L–Z involves direct fluorination on the xanthene system of rhodamine dyes. Using this strategy, we previously created JF₅₅₂ (13; K_L–Z = 0.70; Fig. 1j, Table 1); the JF₅₅₂–HaloTag ligand shows improved cell-permeability compared to JF₅₄₉ derivatives⁷. This approach is complementary to incorporation of 3,3-difluoroazetidines and combining these modifications yielded the fluorogenic JF₅₂₆ (14; K_L–Z = 0.005; Fig. 1j, Table 1)⁷. The performance of both of these dyes is consistent with our K_L–Z rubric (Fig. 1b,k) and we sought further validation by exploring a novel SNAP-tag ligand based on JF₅₅₂. The SNAP-tag is typically inferior to the HaloTag in live-cell imaging experiments due to slower labeling kinetics^{25, 26}, higher nonspecific interactions of the SNAP-tag ligands^{27, 28}, and other factors²⁹. We compared the performance of chloropyrimidine (cp)³⁰ derivatives of JF₅₄₉ and JF₅₅₂ (2_STL and 13_STL, respectively; Fig. 1l). The shift in K_L–Z resulted in the JF₅₅₂ compound (13_STL) showing low nonspecific staining across a wider range of concentrations (Fig. 1m) and faster live-cell labeling (Extended Data Fig. 3a–d) compared to the JF₅₄₉ compound (2_STL). 2_STL and 13_STL exhibited comparable brightness and photostability in single-particle tracking experiments using SNAP-tag–histone H2B fusions (Extended Data Fig. 3e–f), but the 13_STL compound showed significantly lower nonspecific cytosolic staining, matching the performance of the widely used JF₅₄₉–HaloTag ligand (2_HTL, Fig. 1n, Extended Data Fig. 3g–i)^{3, 31}.

Improving and derivatizing long-wavelength dyes.

Having conceived and validated our K_L–Z-based framework with the short wavelength dyes, we then turned to the far-red and NIR rhodamines (5–8, Fig. 1c), where the K_L–Z vs. λ_abs relationship reveals the need for alternative tuning strategy to increase K_L–Z (Fig. 1d). This would improve the in vivo performance of Si-rhodamines such as JF₆₄₆ and rescue the colorless P(O)Ph- and SO₂-containing dyes (7–8). We previously showed that halogenation of the pendant phenyl ring system can substantially increase the K_L–Z of Si-rhodamine dyes⁶, presumably by lowering the pK_a of the benzoic acid moiety; this substitution also elicits a bathochromic shift³². For example, JF₆₄₆ (5; λ_abs/λ_em = 646 nm/664 nm) exhibits a K_L–Z = 0.0014 but the fluorinated analog, JF₆₆₉ (15; λ_abs/λ_em = 669 nm/682 nm), is higher with K_L–Z = 0.262 (Fig. 2a–b, Extended Data Fig. 4a–b). This shift in K_L–Z manifests in a higher absorptivity in aqueous solution with 5 exhibiting ε = 5,600 M^–1cm^–1 but the fluorinated analog 15 showing ε = 112,000 M^–1cm^–1 (Fig. 2c, Table 1). We expected this strategy would be general and prepared the fluorinated PO₂H-, P(O)Ph-, and SO₂-containing rhodamines (16–18, Fig. 2a) by replacing phthalic anhydride with tetrafluorophthalic anhydride in our synthetic scheme (Extended Data Fig. 4a, Supplementary Note). This modification universally increased K_L–Z and ε, while eliciting a ~23 nm shift in λ_abs (Fig. 2b–c, Table 1, Extended Data Fig. 4b–e). In particular, the fluorinated phosphine oxide derivative 17 strongly absorbs visible light in aqueous solution (ε = 87,000 M^–1cm^–1; λ_abs = 722 nm) compared to the parent compound 7 (ε < 200 M^–1cm^–1; λ_abs = 704 nm, Fig. 1c, Table 1). This trend was generalizable to oxygen- and sulfur-containing rhodamines based on 2 and 3 where the fluorine substituents on the pendant phenyl ring also increased K_L–Z and λ_abs (19–20, Fig. 2d–e, Table 1, Extended Data Fig. 4f–g)⁶.

Figure 2. Optimizing long-wavelength rhodamine dyes by increasing K_L–Z and subsequent derivatization.

(a) Structures of dyes 15–18. (b) Plot of K_L–Z vs. λ_abs showing increased K_L–Z for dyes 15–18 relative to parent dyes 5–8. (c) Absorption spectra of 5–8 and fluorinated analogs 15–18. (d) Structures of dyes 19–20. (e) Plot of K_L–Z vs. λ_abs showing increased K_L–Z for dyes 19–20. (f) Nucleophilic aromatic substitution (S_NAr) of JF₆₆₉ (15) to form derivatives 21–24 and 35; subsequent synthesis of JF₆₆₉–HaloTag ligand (15_HTL). (g) Airyscan confocal fluorescence microscopy image of live U2OS cells expressing endoplasmic reticulum-localized HaloTag–Sec61β labeled with 15_HTL (100 nM, 30 min, no wash); this imaging experiment was duplicated with similar results. (h) Airyscan confocal fluorescence microscopy image of live U2OS cells expressing mitochondria-localized HaloTag–TOMM20 labeled with 15_HTL (100 nM, 30 min, no wash); this imaging experiment was duplicated with similar results; color scale in g and h indicates z-depth; scale bars: 5 μm. (i) Image of fixed coronal mouse brain slice from animal expressing GFP–HaloTag fusion protein in neurons after IV administration of 15_HTL (100 nmol); scale bar = 500 μm. (j) Plot of fluorescence from fixed cells expressing HaloTag–H2B labeled with 5_HTL or 15_HTL (200 nM, 30 min, 3× wash) over 30 bleach cycles; error bars indicate SE; n = 3 independent cellular samples. (k) Structure of JF₆₆₉–SNAP-tag ligand (15_STL). (l) Airyscan confocal fluorescence microscopy image of live U2OS cells expressing Sec61β–HaloTag labeled with 2_HTL (30 nM, 30 min, 3× wash) and nucleus-localized SNAP-tag–histone variant H2A.Z labeled with 15_STL (30 nM, 30 min, 3× wash); co-stained with Hoechst 33342 (1 μM, 30 min, 3× wash); scale bar: 5 μm; this imaging experiment was duplicated with similar results.

We then explored derivatives of these new far-red and NIR dyes. In addition to increasing λ_abs and K_L–Z, the halogenated phenyl ring motif can also serve as an electrophile in nucleophilic aromatic substitution (S_NAr) reactions. Thiol nucleophiles have been used for decades to prepare conjugatable derivatives of fluorinated xanthene fluorophore derivatives^33–35, including some Alexa Fluor dyes^{8, 36}. The reactivity of nucleophiles other than thiols was largely unexplored, however; we discovered that N₃^–, CN^–, NH₃, and NH₂OH could react with JF₆₆₉ (15) to provide derivatives 21–24 (Fig. 2f). This reaction type was generalizable to other fluorinated rhodamines and regioselective at the 6-position (Supplementary Note). Although beyond the scope of this report, we briefly investigated some of these derivatives, finding azide 21 was an excellent reactant in strain-promoted ‘click chemistry’ with cyclic alkynes³⁷ 25–26 to form triazole adducts 27–28 (Extended Data Fig. 5a), validating the regiochemistry of the amine addition to form 22 using intermediates 21 and 29 (Extended Data Fig. 5b), and testing the reactivity of amine-containing ion-chelating groups 30 and 31 which generated novel prototype far-red indicators for K⁺ and Zn²⁺ (32–33, Extended Data Fig. 5c–e).

Synthesis of fluorescent labels for cellular imaging.

We then sought derivatives optimized for labeling strategies such as the HaloTag and SNAP-tag. Ligands based on 6-carboxyrhodamines are particularly attractive since compounds with this regiochemistry show superior labeling efficiency and lower toxicity^{27, 30}. Although 6-carboxy-4,5,7-trifluororhodamines were unknown, we were encouraged by the selectivity of thiol and amine addition (Fig. 2f). We therefore explored malonates and related carbon nucleophiles, all of which gave a regioselective reaction at the 6-position (Supplementary Note). In particular, the addition of masked acyl cyanide³⁸ reagent 34, an umpolung-type acyl anion equivalent, to JF₆₆₉ resulted in intermediate 35, which could be deprotected to yield a reactive acyl cyanide suitable for direct conjugation with the HaloTag ligand amine (36) to form JF₆₆₉–HaloTag ligand (15_HTL; Fig. 2f, Extended Data Fig. 6a). As expected from the JF₆₆₉ K_L–Z value = 0.262 (Fig. 2b, Table 1), compound 15_HTL was useful in cell biological experiments (Fig. 2g–h) and was also BBB permeable, labeling HaloTag-expressing neurons throughout the mouse brain after intravenous (IV) administration (Fig. 2i, Extended Data Fig. 6b). The fluorination also improves photostability, with 15_HTL bleaching slower compared to the parent JF₆₄₆–HaloTag ligand (5_HTL, Fig. 2j, Extended Data Fig. 6c). The JF₆₆₉–SNAP-tag ligand (15_STL; Fig. 2k) was an excellent live-cell label with low nonspecific staining (Fig. 2l, Extended Data Fig. 6d–e). This chemistry was generalizable across fluorinated rhodamines, allowing facile synthesis of HaloTag ligands 15_HTL–20_HTL from dyes 15–20 with λ_abs ranging from the yellow to NIR (Fig. 3a, Extended Data Fig. 6a,f, Supplementary Note). These compounds selectively labeled HaloTag fusions in cells (Fig. 3a) and, like the JF₆₆₉–HaloTag ligand (15_HTL, Fig. 2j), the fluorinated dye ligands 19_HTL and 20_HTL showed higher photostability than their nonfluorinated congeners (2_HTL and 3_HTL; Extended Data Fig. 6c,g–h). We note this novel late-stage, regioselective introduction of a carboxy group has distinct advantages over classic rhodamine syntheses that generate isomeric mixtures^{39, 40} and will be useful for synthesizing derivatives beyond self-labeling tag ligands.

Figure 3. Fine-tuning of NIR rhodamines.

(a) Structures of HaloTag ligands 15 _HTL–20_HTL and high-magnification images of fixed U2OS cell nuclei expressing HaloTag–histone H2B and labeled with 15_HTL–20_HTL (100–200 nM, 30 min, 3× wash); scale bars: 10 μm; the imaging experiments were duplicated with similar results. (b) Absorption spectra of JF₇₂₄–HaloTag ligand in the absence (18_HTL) or presence (18_HTL:HT) of excess HaloTag protein. (c) Absorption spectra of JF₇₂₂–HaloTag ligand in the absence (17_HTL) or presence (17_HTL:HT) of excess HaloTag protein. (d) Structures of dyes 17 and 38. (e) Plot of K_L–Z vs. λ_abs showing decreased K_L–Z for dye 38. (f) Structure of JF₇₁₁–HaloTag ligand (38_HTL). (g) Absorption spectra of JF₇₁₁–HaloTag ligand in the absence (38_HTL) or presence (38_HTL:HT) of excess HaloTag protein. (h) Nuclear fluorescence vs. time upon addition of ligands 17_HTL or 38_HTL (100 nM) to fixed cells expressing HaloTag–histone H2B; error bars indicate SE; n = 100 nuclei from three fields of view except for t = 30 min with compound 17_HTL where n = 97 nuclei. (i) Confocal imaging experiment of fixed U2OS cells expressing HaloTag–TOMM20 labeled with 38_HTL (1 μM, 30 min, 3× wash); scale bar: 20 μm; this imaging experiment was duplicated with similar results.

Fine-tuning of NIR labels.

Finally, we sought to optimize NIR HaloTag ligands for biological imaging. The SO₂-containing rhodamine JF₇₂₄ (18) possessed a promising K_L–Z = 10^–3 for creating fluorogenic compounds; the JF₇₂₄–HaloTag ligand (18_HTL) showed a 15-fold increase upon reaction with HaloTag protein in vitro (Fig. 3b). Nevertheless, this dye was plagued with a low Φ = 0.05 (Table 1), making it suboptimal for imaging experiments. In contrast, the P(O)Ph-containing fluorophore JF₇₂₂ (17) exhibits a larger Φ = 0.11 but also a relatively high K_L–Z = 0.026; as expected the JF₇₂₂–HaloTag ligand (17_HTL) was not fluorogenic (Fig. 3c). We investigated whether our two tuning strategies could work synergistically, using JF₅₇₁ (19, Fig. 2d) as a proof-of-concept. We introduced a fluorine substituent on each azetidine ring to create JF₅₅₉ (37; Extended Data Fig. 7a) and found this dye exhibits K_L–Z = 6.22, intermediate between JF₅₄₉ (2; K_L–Z = 3.5) and JF₅₇₁ (20; K_L–Z = 7.93; Table 1, Extended Data Fig. 7b–d), demonstrating the compatibility of these strategies. The JF₅₅₉–HaloTag ligand (37_HTL) could be used in live cell labeling (Extended Data Fig. 7e–f). We then applied this modification to JF₇₂₂, synthesizing the 3-fluoroazetidinyl JF₇₁₁ (38, Fig. 3d, Extended Data Fig. 7g). The fluorination in JF₇₁₁ gave a further improvement in Φ = 0.17 (Table 1) and was predicted to yield fluorogenic ligands based on its K_L–Z = 10^–3 (Fig. 3e). In line with our K_L–Z rubric, the JF₇₁₁–HaloTag ligand 38_HTL (Fig. 3f) showed a 5-fold increase upon binding HaloTag (Fig. 3g). We compared JF₇₁₁ ligand 38_HTL with the parent JF₇₂₂–HaloTag ligand (17_HTL) in cells. Consistent with the Φ values of the parent dyes (Table 1) and our K_L–Z-based framework (Fig. 1b, Fig. 3e) we found that the JF₇₁₁ ligand 38_HTL shows higher brightness in fixed cells (Fig. 3h–i), but the JF₇₂₂–HaloTag ligand (17_HTL) shows better loading kinetics in live cells (Extended Data Fig. 7h) along with modestly higher photostability (Extended Data Fig. 7i). Thus, JF₇₁₁ derivatives could be most useful in experiments where high brightness and low background are crucial, and JF₇₂₂-based compounds could be better-suited for live-cell applications.

Discussion

In summary, we developed a rubric to relate the performance of simple rhodamine dyes to a single parameter, K_L–Z (Fig. 1b), and discovered an inverse correlation between K_L–Z and λ_abs (Fig. 1d). We validated this rubric by using our established tuning strategies⁴ to decrease K_L–Z and λ_abs, resulting in the GFP-like JF₄₇₉ (11; Fig. 1e–i) and an optimized SNAP-tag ligand based on JF₅₅₂ (13; Fig. 1j–n). The NIR-excited dyes 7 and 8 posed a new challenge, with low K_L–Z values that rendered the compounds unusable in biological environments (Fig. 1d). We therefore established a complementary general method to increase both λ_abs and K_L–Z by incorporating fluorines on the pendant phenyl ring of rhodamine dyes (Fig. 2a–b) followed by facile, generalizable S_NAr chemistry to install groups for bioconjugation (Fig. 2f). This strategy yielded the bioavailable JF₆₆₉ (15, Fig. 2g–l) along with other new fluorophores (16–20, Fig. 3a) and could be combined with our previous tuning method to rationally design the fluorogenic NIR-excited JF₇₁₁–HaloTag ligand (38_HTL, Fig 3f–i). Although we have focused here on HaloTag and SNAP-tag ligands and mammalian systems, we expect this general rubric relating K_L–Z to cellular performance (Fig. 1b) could be customized for other ligand types and biological systems⁷. More generally, we anticipate this expanded fluorophore palette and new derivatization chemistry will facilitate the synthesis of novel ligands, labels, stains, and indicators for biological imaging experiments in cells or animals.

Online Methods

Chemical synthesis.

Methods for chemical synthesis, full characterization of all novel compounds, and crystallographic confirmation of regioselective S_NAr can be found in the Supplementary Note.

General UV–vis and fluorescence spectroscopy (Fig. 2c; Table 1; Extended Data Fig. 1a–b; Extended Data Fig. 4b–g; Extended Data Fig. 5d–e; Extended Data Fig. 7d,g).

Fluorescent and fluorogenic molecules for spectroscopy were prepared as stock solutions in DMSO and diluted such that the DMSO concentration did not exceed 1% v/v. Spectroscopy was performed using 1-cm path length, 3.5-mL quartz cuvettes or 1-cm path length, 1.0-mL quartz microcuvettes from Starna Cells. All measurements were taken at ambient temperature (22 ± 2 °C). Absorption spectra were recorded on a Cary Model 100 spectrometer (Agilent). Fluorescence spectra were recorded on a Cary Eclipse fluorometer (Varian). Maximum absorption wavelength (λ_abs), extinction coefficient (ε), and maximum emission wavelength (λ_em) were measured in 10 mM HEPES, pH 7.3 buffer; reported values for ε are averages (n = 3). Normalized spectra are shown for clarity. For prototype ion indicators 32 and 33 (Extended Data Fig. 5d–e) the compounds were dissolved in 10 mM HEPES, pH 7.3 buffer alone or with either 100 mM KCl or 10 μM ZnCl₂; the fluorescence emission spectra of these solutions were recorded using λ_ex = 575 nm and λ_em = 625–825 nm.

Determination of K_L–Z (Fig. 1d,f,k; Fig. 2b,e; Fig. 3e; Table 1; Extended Data Fig. 7b–c).

We calculated K_L–Z using the following equation⁵: K_L–Z = (ε_dw/ε_max)/(1 – ε_dw/ε_max). ε_dw is the extinction coefficient of the dyes in a 1:1 v/v dioxane:water solvent mixture; this dioxane–water mixture was chosen to give the maximum spread of K_L–Z values⁴. ε_max refers to the maximal extinction coefficients measured in different solvent mixtures empirically determined depending on dye type: 0.1% v/v TFA in ethanol for the Si-rhodamines (5 and 15); 0.1% v/v trifluoroacetic acid (TFA) in 2,2,2-trifluoroethanol (TFE) for all the other rhodamine variants. We note that accurate determination of low K_L–Z values (≤ 10⁻³) is complicated by the relatively poor sensitivity of absorbance measurements. We estimated K_L–Z = 10⁻³ when we observed a small but significant absorbance signal in 1:1 v/v dioxane:water solvent mixture over the dye-free control, and K_L–Z ≈ 10⁻⁴ when we observed no significant absorbance of the dye solution.

Quantum yield determination (Table 1).

All reported absolute fluorescence quantum yield values (Φ) were measured in our laboratory under identical conditions using a Quantaurus-QY spectrometer (model C11374, Hamamatsu). This instrument uses an integrating sphere to determine photons absorbed and emitted by a sample. Measurements were carried out using dilute samples (A < 0.1) and self-absorption corrections⁴¹ were performed using the instrument software. Reported values are averages (n = 3).

Absorption increase of ligands upon binding HaloTag protein (Fig. 3b,c,g).

HaloTag protein was used as a 100 μM solution in 75 mM NaCl, 50 mM TRIS·HCl, pH 7.4 with 50% v/v glycerol (TBS–glycerol). Absorption measurements were performed in 1-mL quartz cuvettes. A solution of HaloTag ligands 17_HTL, 18_HTL, or 38 _HTL (5 μM) was prepared in 10 mM HEPES, pH 7.3 containing 0.1 mg·mL^–1 CHAPS. An aliquot of HaloTag protein (1.5 equiv, 7.5 μM final [HaloTag]) was added and the resulting mixture was incubated until consistent absorption signal was observed (60–120 min). An equivalent volume of TBS–glycerol blank was added in place of enzyme to record the ‘ligand-only’ absorption. Spectra are averages (n = 2).

Fluorescence spectroscopy of HaloTag conjugates (Fig. 1h; Extended Data Fig. 2b).

HaloTag protein was used as a 200 μM solution in PBS buffer pH 7.4. A solution of HaloTag ligands 11_HTL or 12_HTL (5 μM) was prepared in 10 mM HEPES, pH 7.3 containing 0.1 mg·mL^–1 CHAPS. An aliquot of HaloTag protein (2 equiv, 10 μM final) was added and the resulting mixture was incubated at 4 °C overnight. Fluorescence measurements were performed after the HaloTag conjugate solutions were diluted 5× (1 μM final [ligand]) into 10 mM HEPES, pH 7.3 buffer solution. Spectra are averages (n = 2). The spectra of GFP (Fig. 1h) was taken from FPbase (https://www.fpbase.org/protein/egfp/)⁴².

Multiphoton spectroscopy of dyes and HaloTag conjugates.

For compounds 1, 11, and the fluorescein control (Extended Data Fig. 1c) solutions of the free dyes (5 μM) were prepared in 10 mM HEPES buffer, pH 7.3. For other rhodamines (Extended Data Fig. 6f), spectra of the HaloTag conjugates were measured. As above, solutions of HaloTag ligands compounds 15_HTL–20_HTL and 37_HTL–38_HTL (5 μM) were prepared in 10 mM HEPES, pH 7.3 containing 0.1 mg·mL^–1 CHAPS. An aliquot of HaloTag protein (2 equiv, 10 μM final) was added and the resulting mixtures were incubated at 4 °C overnight. These HaloTag conjugate solutions were diluted 5× (1 μM final [ligand]) into 10 mM HEPES, pH 7.3 and the two-photon excitation spectra were measured as previously described^{43, 44}. Briefly, measurements were taken on an inverted microscope (IX81, Olympus) equipped with a 60×, 1.2NA water objective (Olympus). Dye–protein samples were excited with pulses from an 80 MHz Ti-Sapphire laser (Chameleon Ultra II, Coherent) for 710–1080 nm and with an OPO (Chameleon Compact OPO, Coherent) for 1000–1300 nm. Fluorescence collected by the objective was passed through a dichroic filter (675DCSXR, Omega) and a short pass filter (720SP, Semrock) and detected by a fiber-coupled Avalanche Photodiode (SPCM_AQRH-14, Perkin Elmer). All excitation spectra are corrected for the wavelength-dependent transmission of the dichroic and band-pass filters, and quantum efficiency of the detector. Spectra are averages (n = 2).

General cell culture and fluorescence microscopy.

All cell lines undergo regular mycoplasma testing by the Janelia Cell Culture Facility. Unless otherwise noted, U2OS cells (ATCC) were cultured in Dulbecco’s modified Eagle medium (DMEM, phenol red-free; Life Technologies) supplemented with 10% v/v fetal bovine serum (FBS, Life Technologies), 1 mM GlutaMAX (Life Technologies) and maintained at 37 °C in a humidified 5% (v/v) CO₂ environment. For confocal and widefield imaging of cell nuclei (Fig. 1i; Fig. 3a; Extended Data Fig. 1f,h; Extended Data Fig. 2c–h; Extended Data Fig. 7f), we used U2OS cells with an integrated HaloTag–histone H2B fusion protein expressing plasmid via the piggyBac transposon system unless otherwise noted. For confocal imaging of mitochondria (Fig. 3i), we used U2OS cells with an integrated TOMM20–HaloTag fusion protein expressing plasmid unless otherwise noted; TOMM20 is an outer mitochondrial membrane protein as part of a protein translocase complex. These cell lines were kept under the selection of 500 μg/mL Geneticin (Life Technologies). For confocal imaging of the cell surface (Extended Data Fig. 1g,i), we used U2OS cells transiently transfected by nucleofection (Lonza) with a plasmid expressing a C-terminal transmembrane anchoring domain from platelet-derived growth factor receptor (PDGFR) fused to the HaloTag protein (HaloTag–PDGFR). Unless otherwise noted, cells were imaged on the following microscopes: Nikon Eclipse Ti with a Plan APO λ 20×/0.75 air objective, Leica SP8 Falcon confocal microscope with an HC PL-APO 86×/1.20 water objective; Zeiss LSM 800 confocal microscope with a Plan APO 20×/0.8 air M27 objective or Plan APO 63×/1.4 oil DIC M27 objective; Zeiss LSM 880 with a C-APO 40×/1.2 W Corr FCS M27 objective. The Leica and Zeiss LSM 800 confocal images were processed using FIJI⁴⁵. Unless otherwise noted, live cells were washed in media and fixed cells were washed in phosphate buffered saline. We use the following shorthand in figures: HT = HaloTag; ST = SNAP-tag; H2B =histone H2B; PDGFR = C-terminal transmembrane anchoring domain from platelet-derived growth factor receptor; H2A.Z = histone variant H2A.Z

Multiplexed imaging comparison JF₅₀₃ and JF₄₇₉ (Fig. 1i; Extended Data Fig. 2c–h).

U2OS cells stably expressing HaloTag–histone H2B fusion protein were transiently transfected with plasmids encoding pSNAPf–TOMM20 fusion protein using nucleofection (Lonza). Live cells were incubated with JF₄₇₉–HaloTag ligand (11_HTL, 500 nM) or JF₅₀₃–HaloTag ligand (12_HTL; 500 nM; Extended Data Fig. 1d) for 3 h followed by addition of JF₅₂₅–cpSNAP-tag ligand (9_STL, 100 nM, Extended Data Fig. 2a) and incubated for an additional 30 min. These cells were then washed 3× in dye free media and imaged (Fig. 1i) using tunable white light laser (WLL) excitation at 488 nm or 532 nm on a Leica SP8 Falcon confocal microscope with an HC PL-APO 86×/1.20 water objective. The images are displayed as maximum intensity projections of confocal image stacks. These images were processed and associated line-scans were extracted using FIJI⁴⁵.

Flow cytometry loading experiments (Fig. 1m; Extended Data Fig. 3a–d).

This experiment utilized the mouse embryonic stem cell line JM8.N4, a gift from R. Tjian (Berkeley), derived from the C57BL/6N strain. The JM8.N4 cells were authenticated by short tandem repeat DNA profiling and approved by the NIH 4D Nucleome project as a Tier2 cell line. Wild-type mouse embryonic stem (ES) cells or ES cells stably expressing SNAP-tag–histone H2B fusion protein were plated into onto flat-bottom 96-well microplates precoated with 0.1% gelatin (Corning). Cells were washed 3× for 10 min, trypsinized, and loaded onto CytoFLEX S flow cytometer equipped with a plate loader (Beckman Coulter). The ES cell population was designated based on its forward light scatter (FSC) and side light scatter (SSC) characteristics (Extended Data Fig. 3a). The gating strategy to determine the nonfluorescent cell population used control ES cell samples not incubated with SNAP-tag ligands, plotting SSC vs. fluorescence from the Y585-PE channel (phycoerythrin; 561 nm laser excitation, 585 nm with a 42 nm bandpass emission, avalanche photodiode detector; Extended Data Fig. 3b). For the assays, cells at 70% confluency were stained with JF₅₄₉–cpSNAP-tag ligand (2_STL) or JF₅₅₂–cpSNAP-tag ligand (13_STL) at different concentrations (3 nM, 10 nM, 30 nM, 100 nM, 300 nM) for 15 min (Fig. 1m) or at different time points (15 min, 30 min, 60 min, 120 min, 210 min) using 10 nM ligand (Extended Data Fig. 3c–d). A typical experiment recorded 20,000 cells per condition and the percentage of fluorescently positive cells was determined. For the experimental replicates, the mouse embryonic stem cell line used for flow cytometry analysis had between 85–98% expression of SNAP-tag–histone H2B as determined against negative fluorescence gating. Sample dilution and flow rate were adjusted to optimize event recordings for the 96-well microplate format. The instrument settings were as follows: FSC avalanche photodiode (AP) detector gain setting = 12; SSC AP detector gain setting = 130; FSC threshold Automatic; PE channel AP gain setting = 1. The experimental data from the CytoFLEX was analyzed using FlowJo v.10.6.1. In some cases, not all concentrations or timepoints could be sampled during every run due to microplate and instrument constraints. For experiments varying loading concentration (Fig. 1m) replicates were as follows: experiments using 2_STL n = 7 except for [ligand] = 3 nM where n = 5; experiments using 13_STL n = 3. For experiments varying loading time (Extended Data Fig. 3d) replicates were as follows: experiments using 2_STL n = 3; experiments using 13_STL n = 4 except for t = 7.5 min where n = 2 and t = 210 min where n = 3.

Single-particle tracking (SPT) experiments (Fig. 1n; Extended Data Fig. 3e–i).

SPT experiments were performed in U2OS cells with an integrated SNAP-tag–histone H2B fusion protein expressing plasmid or an integrated HaloTag–histone H2B fusion protein expressing plasmid via the piggyBac transposon system. SNAP-tag–histone H2B expressing cells were labeled with 2 nM of either JF₅₄₉–cpSNAP-tag ligand (2_STL) or JF₅₅₂–cpSNAP-tag ligand (13_STL). HaloTag–histone H2B fusion protein expressing cells were labeled with 2 nM of JF₅₄₉–HaloTag ligand (2_HTL). Cells were incubated with dyes for 15 min at 37 °C and then washed 3× for 30 min each. SPT was performed at 100 Hz (10 ms frames) and 5000 frames were recorded for each cell. Single molecules were localized and tracked by a MATLAB implementation of multiple target tracing (MTT) and SLIMFast^{31, 46}. For SPT brightness (photons/s; Extended Data Fig. 3e) n = 19008 single-molecule events using 2_STL and n = 9511 single-molecule events using 13_STL. For SPT track length (s; Extended Data Fig. 3f) n = 10822 single-molecule events using 2_STL and n = 9387 single-molecule events using 13_STL. Trajectories were fitted into a two-state model: chromatin bound and diffusive (free) using diffusion coefficient (D) cut off Dbound: [0.0005,0.08] and Dfree: [0.15, 25]. The fraction of chromatin bound molecules per cell are plotted (Fig. 1n); n = 12 cells for experiments using 2_STL and n = 8 cells for experiments using 13_STL and 2_HTL.

Airyscan imaging experiments using 15_HTL (Fig. 2g–h).

U2OS cells were transiently transfected with HaloTag–Sec61β fusion protein expressing plasmid or HaloTag–TOMM20 fusion protein expressing plasmid using FuGENE HD (Promega) and maintained in DMEM containing 10% v/v FBS and penicillin–streptomycin–glutamine. Sec61β is an endoplasmic reticulum membrane protein translocator protein. Cells were incubated with 100 nM JF₆₆₉–HaloTag ligand (15_HTL) in full media at 37 °C for 30 min. Imaging was performed without intermediate washing steps using a Zeiss LSM 880 with Airyscan and a plan-apochromatic 63× oil objective (NA=1.4). The airyscan images were processed using the Zen software from Zeiss.

Mouse in vivo labeling experiments (Fig. 2i; Extended Data Fig. 6b).

A GFP–HaloTag fusion was expressed protein throughout the brain by systemic injection of adult C57/BL6 male mice, 2–4 months old, with the viral vector: PHP-eB-Syn-HaloTag-GFP (~5 × 10¹¹ infectious units per mL, 100 μL). The virus was injected using a 0.5 mL 27G syringe to the retro-orbital sinus. JF₆₆₉–HaloTag ligand (15_HTL) was administered to mice 3–4 weeks after the viral injection. Dye solution was prepared by first dissolving 100 nmol (76 μg) of 15_HTL in 20 μL DMSO. After vortexing, 20 μL of a Pluronic F-127 solution (20% w/w in DMSO) was added and this stock solution was diluted into 200 μL sterile saline for IV (retro-orbital) injection. For the reported images (Fig. 2i; Extended Data Fig. 6b) the mouse was injected with virus at 122 days old, injected with 15_HTL ligand at 148 days old, and perfused at 149 days old. Imaging was done on a TissueFAXS 200 confocal slide scanner (Tissuegnostics) using a SpectraX light engine (Lumencor) with the following peak powers and excitation filters: V = 395 nm (400 mW) with a 395 nm-centered extictaion filter and 25 nm band-pass; C = 475 nm (480 mW) with a 475 nm-centered excitation filter and 34 nm band-pass; G/Y = lightpipe with a 585 nm-centered excitation filter and 35 nm band-pass; R = 619 nm (629 mW) with a 635 nm-centered excitation filter and 22 nm band-pass. These excitation sources were fed by a lightguide to a Crest X-Light V2 confocal spinning disc (Crestoptics; 60 μm pinhole spinning disk) with the following dichroics: T425lpxr, T495lpxt, T600lpxr, and T660lpxr; and emission filters: ET460 nm/50 nm, ET525 nm/50 nm, ET625 nm/30 nm, ET700 nm/75 nm. The emission light was collected with the following Zeiss objectives: EC Plan-Neofluar 2.5×/0.085 M27 for tissue detection and Plan-Apochromat 20×/0.8 M27 imaging. Detection was done by a Zyla 5.5 sCMOS camera (Andor). Acquisition of the coronal sections was performed after semi-automated tissue detection and by using multiple autofocusing points per section (5×5 fields of view). Three z-planes with a 7 μum spacing were imaged and z-projected. All experimental protocols were conducted according to the National Institutes of Health guidelines for animal research and were approved by the Institutional Animal Care and Use Committee at the Janelia Research Campus, HHMI.

Photobleaching experiments (Fig. 2j; Extended Data Fig. 6g–h; Extended Data Fig. 7i).

U2OS cells expressing HaloTag–histone H2B fusion protein were co-fixed and labeled by incubation of 4% w/v PFA and 200 nM of ligands 5_HTL, 15_HTL, 2_HTL, 19_HTL, 3_HTL, 20_HTL, 17_HTL, or 38_HTL in 0.1 M phosphate buffer. Cells were washed 3× in phosphate-buffered saline and then imaged. To measure photobleaching for dyes 5_HTL, 15_HTL, 2_HTL, 19_HTL, 3_HTL, and 20_HTL, we used the tunable WLL excitation on the Leica SP8 Falcon confocal microscope to excite dyes at their λ_abs in constant power mode. For each dye wavelength we set the WLL laser percent power to equal the 50% power of the 549 nm excitation used to bleach 2_HTL. For dyes 17_HTL and 38_HTL we performed the photobleaching experiment with excitation at 660 nm with WLL power set to 40%; n = 3.

Dye loading kinetics (Fig. 3h; Extended Data Fig. 1e; Extended Data Fig. 7h).

For live-cell labeling, U2OS cells stably expressing HaloTag–histone H2B fusion protein were labeled over a time course of 0–4 h with 200 nM of JF₄₇₉–HaloTag ligand (11_HTL), JF₅₀₃–HaloTag ligand (12_HTL), JF₇₂₂–HaloTag ligand (17_HTL), or JF₇₁₁–HaloTag ligand (38_HTL). Cells were briefly washed 2× with dye-free media and immediately imaged live using widefield microscopy on a Nikon Eclipse Ti, Plan APO λ 20×/0.75 air objective. For fixed-cell labeling, U2OS cells stably expressing HaloTag–histone H2B were co-fixed in 4% w/v PFA in 0.1 M phosphate buffer with 200 nM of either JF₇₂₂–HaloTag ligand (17_HTL) or JF₇₁₁–HaloTag ligand (38_HTL) over a time course of 0–60 min. Fluorescence was quantified from the average of the summed intensity of nuclear signals in single-plane widefield images analyzed using Nikon NIS-Elements AR software. Fields of view were chosen to obtain approximately 30 nuclei per image and a maximum of 100 nuclear signals were plotted. For 38_HTL vs. 17_HTL in fixed cells (Fig. 3h) n = 100 except for t = 30 min with 17_HTL where n = 97. For 11_HTL vs. 13_HTL (Extended Data Fig. 1e) n = 100. For 38_HTL vs. 17_HTL in live cells (Extended Data Fig. 7h) n = 100 except for: t = 0.5 h with 38_HTL where n = 86; t = 1 h with 38_HTL where n = 94; t = 2 h with 38_HTL where n = 96; t = 0.5 h with 17_HTL where n = 94.

Airyscan imaging experiments using 15_STL (Fig. 2l; Extended Data Fig. 6d–e).

U2OS cells stably expressing SNAP-tag–histone H2AZ and HaloTag–Sec61b fusion proteins were labeled with JF₆₆₉-SNAP-tag ligand (15_STL; 30 nM) and JF₅₄₉–HaloTag ligand (2_HTL; 30 nM) for 30 min, co-staining with Hoechst 33342 (1 μM; ThermoFisher). Cells were washed (3× 10 min) and then imaged using the Zeiss LSM 880 platform under the Airyscan SR mode using a plan-apochromatic 63× oil objective (NA=1.4). The Airyscan images were processed using the Zen software (Zeiss) and the fluorescence intensity line-scan (Extended Data Fig. 6e) was extracted using FIJI⁴⁵.

Statistics and Reproducibility.

For spectroscopy measurements reported n values for absorption spectra, extinction coefficient (ε) and quantum yield (Φ) represent measurements of different samples prepared from the same dye DMSO stock solution or HaloTag conjugate stock solution. For flow cytometry experiments, reported n denotes separate cell samples taken from different microplate wells. For single-particle tracking brightness and track-length experiments, n indicates separate events extracted by the MTT algorithim. For fraction chromatin-bound from single-particle tracking experiments, n indicates the number of individual cells; one-way ANOVA gave adjusted P Value = 0.0013 (**) for 2_STL vs. 13_STL and adjusted P Value = 0.9963 (ns) for 13_STL vs. 2_HTL; F (2, 25) = 11.38. For photobleaching experiments, n indicates the number of separate cellular experiments where the intensity of the entire field of view was measured at the indicated time points. For cell loading experiments, n represents the number of intensity values from individual cells extracted from three fields of view at the indicated time points. For representative fluorescence microscopy and flow cytometry experiments, all procedures were duplicated at least once on a separate biological sample to ensure results were similar as indicated in the figure legends. Additional information can be found in the Life Sciences Reporting Summary.

Data Availability

The data that support the findings of this study are provided in the Source Data files or available from the corresponding author upon request.

Extended Data

Extended Data Fig. 1. Utility of JF₄₇₉–HaloTag (11_HTL) ligand in cellular imaging experiments.

(a–b) Fluorescence excitation (ex) and emission (em) spectra of 1 (a) and 11 (b). (c) Two-photon absorption spectra of 1, 11, and reference dye fluorescein. (d) Structure of JF₅₀₃–HaloTag ligand (12_HTL). (e) Nuclear fluorescence vs. time upon addition of ligands 11_HTL or 12_HTL (200 nM) to live cells expressing HaloTag–histone H2B; error bars indicate SE; n = 100 nuclei. (f–i) Confocal imaging experiments of fixed U2OS cells expressing either HaloTag histone–H2B fusion protein (f,h; nucleus) or HaloTag–PDGFR transmembrane domain (TMD) fusion protein (g,i; plasma membrane) labeled with JF₄₇₉–HaloTag ligand (11_HTL; 100 nM, 1 h, 2× wash; f,g) or JF₅₀₃–HaloTag ligand (12_HTL; 100 nM, 1 h, 2× wash; h,i); scale bars: 21 μm; these imaging experiments were duplicated with similar results.

Extended Data Fig. 2. Comparison of JF₄₇₉–HaloTag ligand (11_HTL) and JF₅₀₃–HaloTag ligand (12_HTL) in two-color experiments with JF₅₂₅–cpSNAP-tag ligand (9_STL).

(a) Structure of JF₅₂₅–cpSNAP-tag ligand (9_STL). (b) Fluorescence excitation spectra of JF₄₇₉–HaloTag ligand (11_HTL) or JF₅₀₃–HaloTag ligand (12_HTL) of bound to HaloTag protein. Dashed lines highlight 488 nm or 532 nm excitation. (c–e) Enlarged confocal images and line-scan from Figure 1i showing live U2OS cells expressing HaloTag–histone H2B labeled with 11_HTL (500 nM, 3.5 h, 3× wash) and TOMM20–SNAP-tag labeled with 9_STL (100 nM, 30 min, 3× wash) excited with 532 nm; this imaging experiment was duplicated with similar results. (c) Confocal image from Figure 1i with blue line indicating line-scan position; (d) Line-scan profile; (e) Over-exposed image showing low nuclear signal. (f–h) Enlarged confocal images and line-scan from Figure 1i showing U2OS cells expressing HaloTag–histone H2B labeled with 12_HTL (500 nM, 3.5 h, 3× wash) and TOMM20–SNAP-tag labeled with 9_STL (100 nM, 30 min, 3× wash) excited with 532 nm; this imaging experiment was duplicated with similar results. (f) Confocal image from Figure 1i with green line indicating line-scan position; (g) Line-scan profile; (h) Over-exposed image showing high nuclear signal. Scale bars for all images: 10 μm.

Extended Data Fig. 3. Utility of JF₅₅₂–cpSNAP-tag ligand (13_STL) in cellular imaging experiments.

(a) Representative flow cytometry plot showing forward light scatter (FSC) vs. side light scatter (SSC) demonstrating gating strategy to separate cells from debris; experiment was duplicated with similar results. (b) Representative flow cytometry plot showing SSC vs. fluorescence of 13_STL:SNAP-tag measured using the Y585-PE channel (561 nm laser excitation, 585 nm with a 42 nm bandpass emission) to demonstrate gating strategy to separate fluorescent and nonfluorescent cells; experiment was duplicated with similar results. (c) Representative flow cytometry plots showing the change in % fluorescent cells as a function of incubation time with 13_STL (10 nM); top row: wild-type (WT) embryonic stem (ES) cells; bottom row SNAP-tag–histone H2B (ST) expressing ES cells; experiment was duplicated with similar results. (d) Plot of fluorescent mouse ES cells (%) vs. time determined by the flow cytometry experiment exemplified in c. WT cells or ST cells were incubated with 2_STL or 13_STL (10 nM) for different times and % fluorescent cells were measured; error bars show SE; experiments using 2_STL n = 3; experiments using 13_STL n = 4 except for t = 7.5 min where n = 2 and t = 210 min where n = 3. (e) Violin plot of photon counts from a single-particle tracking (SPT) experiment using U2OS cells expressing SNAP-tag–histone H2B and labeled with 2_STL or 13_STL (2 nM, 30 min, 3× wash); lines indicate median and quartiles; n = 19008 single-molecule events for experiment using 2_STL and n = 9511 single-molecule events for experiment using 13_STL. (f) Histogram of track lengths from SPT experiment using cells expressing SNAP-tag–histone H2B and labeled with 2_STL or 13_STL (2 nM, 30 min, 3× wash); n = 10822 single-molecule events for experiment using 2_STL and n = 9387 single-molecule events for experiment using 13_STL. (g) Structure of JF₅₄₉–HaloTag ligand (2_HTL). (h) Image of individual SPT traces in live U2OS cells expressing SNAP-tag–histone H2B and labeled with 2_STL (2 nM, 30 min, 3× wash); dashed line indicates outline of nucleus; arrows highlight nonspecific staining in cytosol; scale bar: 2 μm. (i) 3D kymograph showing data from e detailing single-particle track position and length as a function of time. Diffusion coefficient values (D) are calculated from single-particle tracking data and color-coded; experiment was duplicated with similar results.

Extended Data Fig. 4. Synthesis and spectral properties of 15–20.

(a) Synthesis of Janelia Fluor dyes 15–20. (b–g) Fluorescence excitation (ex) and emission (em) spectra of JF₆₆₉ (15; b), JF₆₉₀ (16; c), JF₇₂₂ (17; d), JF₇₂₄ (18; e), JF₅₇₁ (19; f), and JF₅₉₃ (20; g).

Extended Data Fig. 5. Derivatization of JF₆₆₉.

(a) Reaction of azide 21 with strained alkynes 25 or 26 to form triazole adducts 27 or 28. (b) Synthesis of amine 22 via reaction of JF₆₆₉ (15) with NH₃, reduction of azide 21, or Curtius rearrangement starting from ester 29 showing consistent regioselectivity of S_NAr reactions. (c) Reaction of 15 with amine-containing chelator groups 30 and 31 to form far-red K⁺ indicator 32 and far-red Zn²⁺ indicator 33. (d) Fluorescence emission spectra of 32 in the absence or presence of 100 mM K⁺. (e) Fluorescence emission spectra of 30 in the absence or presence of 10 μM Zn²⁺.

Extended Data Fig. 6. Synthesis and properties of new HaloTag and SNAP-tag ligands based on 15–20 and 37–38.

(a) Expanded synthetic scheme of HaloTag ligands 15_HTL–20_HTL and 37_HTL–38_HTL starting with nucleophilic aromatic substitution (S_NAr) of 15–20 and 37–38 with masked acyl cyanide 34. (b) Two-color montage image of fixed coronal slices with zoom-in regions from a mouse expressing HaloTag–GFP in neurons transduced by IV administration of the viral vector PHP-eB-Syn-HaloTag–GFP followed by IV administration of 15_HTL (100 nmol), perfusion, and slicing; GFP signal in green and 15_HTL signal in magenta; scale bar = 3 mm; experiment was duplicated with similar results. (c) Structures of JF₆₆₉–HaloTag ligand (15_HTL), JF₆₄₆–HaloTag ligand (5_HTL), JF₅₉₃–HaloTag ligand (20_HTL), and JF₅₇₀–HaloTag ligand (3_HTL). (d–e) Confocal image and line-scan from Figure 2l showing live U2OS cells expressing Sec61β–HaloTag labeled with 2_HTL (30 nM, 30 min, 3× wash) and SNAP-tag–histone variant H2A.Z labeled with 15_STL (30 nM, 30 min, 3× wash); co-stained with Hoechst 33342 (1 μM, 30 min, 3× wash); (d) Confocal image with white line indicating line-scan position; (e) Line-scan profile; scale bar: 5 μm; experiment was duplicated with similar results. (f) Two-photon absorption spectra of the HaloTag conjugates of HaloTag ligands 15_HTL–20_HTL, and 37_HTL–38_HTL. (g) Plot of fluorescence from cells expressing HaloTag–H2B labeled with 2_HTL (200 nM) or 19_HTL (200 nM) over 30 bleach cycles; error bars indicate SE; n = 3 independent cellular samples. (h) Plot of fluorescence from fixed U2OS cells expressing HaloTag–H2B labeled with 3_HTL (200 nM) or 20_HTL (200 nM) over 30 bleach cycles; error bars indicate SE; n = 3 independent cellular samples.

Extended Data Fig. 7. Further fine-tuning of JF₅₇₁ (19) and JF₇₂₂ (17).

(a) Structure of JF₅₇₁ (19) and JF₅₅₉ (37). (b–c) Full plot of K_L–Z vs. λ_abs (b) and zoom-in (c) showing decreased K_L–Z for dye 37. (d) Fluorescence excitation (ex) and emission (em) spectra of JF₅₅₉ (37). (e) Structure of JF₅₅₉–HaloTag ligand (37_HTL). (f) Widefield imaging experiment of U2OS cells expressing HaloTag–histone H2B labeled with 37_HTL (100 nM, 30 min, 3× wash); scale bar: 51 μm; experiment was duplicated with similar results. (g) Fluorescence excitation (ex) and emission (em) spectra of JF₇₁₁ (38). (h) Nuclear fluorescence vs. time upon addition of ligands 17_HTL (200 nM) or 38_HTL (200 nM) to live cells expressing HaloTag–histone H2B; error bars indicate SE; n = 100 nuclei except for: t = 0.5 h with 38_HTL where n = 86 nuclei; t = 1 h with 38_HTL where n = 94 nuclei; t = 2 h with 38_HTL where n = 96 nuclei; t = 0.5 h with 17_HTL where n = 94 nuclei. (i) Plot of fluorescence from cells expressing HaloTag–H2B labeled with 17_HTL (200 nM) or 38_HTL (200 nM) over 30 bleach cycles; error bars indicate SE; ; n = 3 independent cellular samples.