Red-Emitting Fluorophores with Tailored Properties for Microscopy and Nanoscopy

Abstract

Extended series of red-emitting rhodamine dyes were synthesized and tested. This revealed the real factors determining the quality of STED (stimulated emission depletion) imaging, dye photostability in general, and stability of the corresponding reactive labels. Previously overlooked and underestimated observations are now being highlighted and discussed. So are some very important syntheses-related aspects. The studies were also expanded on far-red emitting oxazine and silicon rhodamine dyes. An intriguing discovery emerged when certain fluorophores, which were originally designed for STED, demonstrated an ability to discern between viable and dead bacterial cells. Recent publications by Lucidi and co-workers showed that certain positively charged dyes allow imaging of bacterial membranes using both conventional techniques, for example, confocal laser scanning microscopy (CLSM), and STED. This opens avenues for investigating membrane biogenesis and diagnostics.

1. Introduction

Far-field optical microscopy in its modern applications such as STED (stimulated emission depletion) [1, 2, 3, 4], PALM (photoactivation localization microscopy) [5, 6, 7], STORM (stochastic reconstruction microscopy) [8], and GSDIM (ground state depletion with individual molecular return) [9] poses very strict and often contradictory requirements on fluorescent markers involved [10, 11]. Among those, there is a large quantum yield of fluorescence, a high extinction coefficient, high photostability, and good solubility in water. No less important is a proper reactive group for conjugation to biological substrates [4, 11]. The labels (usually dye active esters) need to be stable upon storage in a freezer in the neat state and as stock solutions. Indeed, all these features seem to be hardly met by a single substance. However, in the course of our targeted research work, we showed that it is possible, at least by a dye scaffold with multiple variable positions. One very important advantage of rhodamine dyes as a class is the unique combinatorial amenability, as we call it [11]. It means an opportunity to combine and vary diverse functional groups and linkers. As a result, one can obtain fluorophores with tailored properties, staying within the same dye family. Here, we are trying to summarize our studies on this subject. We will tell how one can attach the required moieties to different sites of a rhodamine dye scaffold and take advantage of this.

2. Dye design and applications

2.1 Design of photostable and hydrophilic red-emitting STED fluorophores available as reactive labels with improved stability

Red-emitting rhodamines with absorbance/emission maxima of around 635/655 nm are now quite numerous, as they have been known for more than a decade [10, 11, 12]. The first dye of this family suitable for STED microscopy was synthesized and explored by Kolmakov and co-workers. It is known as KK 114 (see Figure 1 and Ref. [11]) and was supposed to be a better and cheaper alternative for the dye ATTO 647 N. The latter, in turn, was the first dye purposefully designed for STED microscopy by ATTO-TEC GmbH [13]. For this application, photostability and, especially, stability toward photo-induced reactive oxygen species were crucial. Its photophysical properties were indeed greatly improved compared to those of previously known labels, for example, Alexa Fluor 633. However, despite its merits, ATTO 647 N is too lipophilic (see Figure 4, right) and, as it is, leaves no opportunity for further chemical modification. In aqueous media, the dye sticks to glass and, with antibodies, often gives unspecific labeling, strong background, and aggregation. So, a better candidate needed to have polar functional groups. As the most promising alternative dye scaffold, a fluorinated red-emitting rhodamine was chosen, which had been already known from patent literature (see Figure 1 and Ref. [14]).

Figure 1.

Dye scaffold and a family of red-emitting rhodamines, which were screened as fluorophores for STED microscopy. For details and syntheses see Refs. [9, 11]. * Compound KK 1115 is also known as Abberior STAR 635 [11, 12]. Rh denotes the dye core (see also Figure 2), while Y and Z are variable functional groups.

Figure 2.

Red-emitting rhodamine KK 114 modified with an intramolecular stabilizer (NPA) and optional charged groups. *Rh denotes the core of the red-emitting rhodamine scaffold according to Figure 1.

The new rhodamine dye KK 114 (later known as Abberior STAR RED [15]) soon gained recognition due to its excellent STED imaging performance. It showed a very high brightness, high signal-to-noise ratio, and high resistance to bleaching, accompanied by low background and low unspecific labeling [9, 11, 12]. The image quality with certain substrates, particularly, mitochondria organelles [16, 17], is unparalleled. Importantly, with this fluorophore, one can even operate at “higher” STED beams (i.e., 800 nm instead of 775) [18], which are, by definition, less harmful for samples given very high beam powers. Later, phosphorylated, hydroxylated and photostabilized fluorophores were prepared (see Figures 1–5). Apart from other benefits, the analogs of KK114 demonstrate the so-called “good application tolerance”: they are far less sensitive to the changes in illumination/imaging conditions and other setup settings, developing fewer re-excitation artifacts. Here, the advantages of KK 114 compensate for the fact that it is not as suited to the far-red spectral area as, for example, Si-rhodamines [11, 18]. Last but not the least, the synthesis of KK 114, its derivatives, and building blocks is not very difficult. In fact, one needs no metalorganic chemistry, no dangerous chemicals, and no tedious separations either. All the chromatographic separations can be done using conventional hand-made columns. On the contrary, the synthesis of ATTO 647 N takes more than 11 steps and utilizes much more sophisticated precursors and recipes [13]. Finally, the dye KK 114, even taken as it is, can be further functionalized via aromatic nucleophilic substitution in the meso-phenyl ring [11, 19]. Particularly, as we shown in Ref. [11], it is possible to directly decorate this dye (or its precursors) with extra polar ionizable groups (e.g., HO₃SCH₂CH₂S-) or intramolecular stabilizers (e.g., 4-NO₂PhCH₂S-). Regarding the stability of the active esters, usually NHS (N-hydroxysuccinimidyl) or TFP (tetrafluorophenyl) esters, we made three important observations: (1) the linker size improves the stability, especially after certain critical length is exceeded. This can be attributed to the influence of the bulky chromophore residue. Particularly, antibody labeling proved unfeasible with the non-sulfonated analog of KK 114 (see Ref. [9]), so no images were recorded for comparative experiments. This instability was mistakenly attributed to the positive net charge of its molecule. Also, it proved impossible to prepare active esters from the 4-nitro-L-phenylalanine-modified dye KK 114 (see Sections 2.3 and 2.4) unless the alanine terminus was modified. We found that a simple elongation of the linker group even by one methylene unit makes the active esters in all dyes of the family, including KK 114, much more stable, easier to handle and to ship. This applies to mixed NHS carbonates of the rhodamines, as it was shown, for example, in Ref. [20] (see also compound 4 in Figure 5) and for fluorophores of other families (see Figures 6 and 7), as well; (2) a large negative net charge of a label decreases the stability of its active esters, as we observed in a comparative study, which involved labels with extra SO₃H groups [11]; (3) chromatographic isolation of the NHS esters should be avoided where possible. All derivatives of dyes with multiple polar groups require aqueous or alcoholic media as the mobile phase, where their stability is low, especially upon concentrating. Particularly, rhodamine dyes of this family can react with methanol to give either Me-esters or, in basic conditions, compounds where one aromatic fluorine atom is exchanged for MeO or OH [9, 11, 19]. In view of all these, it is crucial to find such a recipe that easily brings the NHS esterification to completion. Having achieved that, one removes the excess of the reagents (small molecules and salts) by simple extraction or sonication of the solids in proper inert solvents [11, 21]. Among the NHS ester labels with the best stability, we should, first of all, mention STAR 635 (also called KK 1115 in the references above, with a net charge (z) of −1). It was purposefully developed as a label with improved stability upon handling and storage.

Figure 3.

Exemplary STED images of tubulin (left) and actin (right) labeled with Abberior STAR 635. For imaging and labeling conditions see Refs. [11, 12]. Scale bars: 1 μm; resolution ~40 nm.

Figure 4.

Left: Red-emitting rhodamine dye STAR635 with a blue color and intense red fluorescence in aqueous solution under visible light. Right: Distribution of Abberior STAR635 (1) and ATTO 647 N (2) between dichloromethane (lower phase) and water at pH 8.5 (labeling buffer; upper phase). The dye ATTO 647 N is so lipophilic that it remains exclusively in the lower organic phase.

Figure 5.

Synthesis of phosphorylated red-emitting rhodamines and their NHS esters. Compound 1 is the precursor for phosphorylation, and compound 2 is the desired NHS ester, which presents selectivity challenges during its synthesis. Compounds 3 and 4 represent approaches to solve these issues.

Figure 6.

Exemplary post-synthetic modification that affects the net charge and makes the fluorophores hydrophilic. A general formula of the dyes modified with custom-made branched linkers L² and auxiliary (solubilizing) groups.

Figure 7.

Fluorophores from other families for multi-color imaging shifted to the far-red (KK 1441, and 2-H, see Ref. [18]) and rhodamine shifted to the orange spectral region (14-H). * – KK 1905-NHS is a custom-made non-phosphorylated oxazine label with a positive net charge for bacterial imaging [21]. ** – for synthesis, properties, and STED imaging see Ref. [20] and Figure 13.

Particularly, STAR 635-NHS withstands aqueous workup in neutral or slightly acidic media and, therefore, can be extracted with proper organic solvents that are easy to evaporate. This is followed by drying and simple filtration. These procedures quite well remove salts and silica gel dust, which is, at times, a big problem [11]. Preparative chromatography (hand-made columns or HPLC) at pH ≤ 7 followed by concentrating at room temperature under reduced pressure makes no harm either. The NHS esters of KK 114 L-NHS, KK 1553-NHS, KK 1555-NHS (all three have z = −1), and especially KK 1517-NHS (z = +1, with no negatively charged groups) are sufficiently stable too [11]. Regarding our far-red emitting labels, the stability of KK 1905-NHS (z = +1) proved to be surprisingly good (see Figure 7 for structure and [21] for preparation and properties). Straightforward and reliable recipes for the preparation of the NHS esters can be found in the references above (see also Figure 8). Finally, imaging experiments with various labels showed that a dye linker does not need to be as short as possible, as was mistakenly assumed earlier [4]. Particularly, in the sections below, one will see the dyes with bulky extra moieties and with longer and/or branched linkers. They all performed very well.

Figure 8.

Left: dye KK 114 as a fluffy solid after freeze-drying. Note the red glittering under daylight. Right: evaporation of the KK 114-NHS ester aliquots on small scale (1 mg of the solid) using a vacuum-inert gas manifold [11].

2.2 Phosphorylated dyes as an alternative to the conventional sulfonated compounds

The introduction of sulfonic acid groups (sulfonate) to aromatic compounds, known for more than 150 years, drastically improves the solubility in water, which is important for many industrial dyestuffs, fluorophores for microscopy, and laser dyes [4, 11]. Having obtained the diol precursor 1 (Figure 5) by oxidation of the corresponding rhodamine fluorophore at the allylic sites of the parent rhodamine, we sought to modify the hydroxy (OH) groups so that the dye could get ionizable functional groups with acidic protons. Free OH groups offer a variety of opportunities for derivatization and, at the same time, remain unreacted while the carboxyl-containing linker is subjected to a peptide-type coupling. So, an alternative rhodamine dye candidate, which is modified at the OH sites, must be an ester of a strong mineral acid. Esters of mineral acids raise hydrolytic stability concerns in multistep syntheses and while labeling. Therefore, one should exclude nitric and sulfonic acid esters as candidates. Phosphate was thus chosen as an alternative to the conventional sulfonate. Indeed, this acid has a medium strength, an increased number of acidic protons relative to sulfuric acid, and most importantly, its primary and secondary alkyl esters are hydrolytically stable in basic media. So, one can increase the net charge of a dye as a whole, which is very important for electrokinetic analyses. Also, one can call the phosphate group “biologically friendly”, as it is very abundant in living nature. However, before our first studies on dye phosphorylation were published in 2012, no attempts had been known to attach phosphoric acid (phosphate) residues to fluorescent dyes [22]. Phosphate groups (-OP(O)(OH)₂), despite these advantages, pose selectivity concerns while preparing reactive labels [20]. These groups do react with most coupling reagents to a noticeable extent to form P-NHS esters (see compound 4 in Figure 5). Unexpectedly, the latter proved inert toward amines. Despite the praiseworthy attempts, an efficient and reasonable way how to selectively activate a carboxyl group (-CO₂H) in the presence of a phosphate was not found. Remarkable is the approach that involved a dye with free phosphate groups, an azide, and a functionalized alkyne as a counterpart. The dye azide was subjected to a metal-free click reaction with an alkyne that had a pre-synthesized amino reactive (NHS) site. It was a linker-containing dibenzocyclooctyne-NHS ester (DBCO-NHS). Obviously, such a bulky and lipophilic building block, which comprises more than 30% of the molecular mass of the entire conjugate (compound 3), does not seem to be a reasonable solution. On the other hand, metal-catalyzed click reactions on smaller synthons proved unsuitable due to strong complexation [20]. Remarkably, the real (remeasured) fluorescence quantum yields Φ_f of the phosphorylated dye (compound 2 in Figure 5 as acid; see also the dye KK 1550 in Figure 1) did not exceed those of “the benchmark dye” KK 114 with conventional SO₃H groups. Certain increases in the quantum yield were, however, observed [11], but only at pH above 9, which is unacceptable for most biology-related applications. So, summarizing, phosphate groups as auxiliary groups for microscopy-related fluorophores proved to be not worthwhile, yet they did involve certain interesting chemistry by their preparation. The phosphorylated fluorophores, however, proved to be very useful for electrokinetic-related applications, particularly for high-throughput carbohydrate analyses, as was reasonably assumed in [22] and shown in applications [23, 24].

2.3 Post-synthetic (last minute) modifications of the fluorophores

In the course of our research work, we very often made use of the combinatorial possibilities of the dye scaffold depicted in Figure 1. Particularly, we performed the so-called “post-synthetic” or “last minute” modifications. This approach was first realized by Romieu and co-workers in 2008 [25, 26, 27, 28, 29, 30]. It is usually the linking group of a dye (as shown in Figure 6) that allows such transformations. High-yielding peptide-type coupling reactions were utilized. One can either modify the linker step by step or attach a pre-synthesized building block containing the required auxiliary (i.e., ionizable and/or solubilizing) groups. We had shown already how one could attach a required moiety to different sites of our specific dye scaffold. It can be done even in the final steps of the synthesis. Moreover, intramolecular photostabilizers (see Section 2.4 and Figure 2) can be installed following this straightforward methodology (“last minute” modifications). Luckily, sulfonic acid groups (sulfonate, -SO₃H), the classical dyestuff solubilizing moieties, are unaffected neither in amidation reactions nor in the preparation of active esters [11, 27, 28, 30]. Phosphate groups, on the other hand, pose selectivity problems, as it was mentioned above and in Ref. [20].

2.4 Fluorophores with intramolecular photostabilization

Self-healing dyes with photostabilizing moieties represent an efficient solution for the fatigue/bleaching problem [11, 31, 32, 33, 34, 35, 36] (see Figure 2). However, too many questions remained unanswered. In our recent study, we tried to address them [11]. We wanted to establish to what extent the photostability can be further improved, and what it would be traded for. Particularly, the effect of intramolecular stabilizers on rhodamines, which are alone much more stable than cyanines, still remained unclear. The basic photophysical properties of the individual rhodamine conjugates containing photostabilizing groups (PSG), which were originally called triplet state quenchers (TSQ). Fluorescence quantum yields Φ_f (fl. QY), fluorescence lifetimes τ (FLT), and the photostability of the individual conjugates were thoroughly measured. These data were compared to their STED performance in the immobilized state and in bulk aqueous solutions, as seen in Figures 9 and 10 (which contain parts reproduced from Ref. [11]). The authors took a closer look at the photostability and STED performance of the conjugate “NPA-KK 114” (dye KK 114 decorated with a 4-nitro-L-phenylalanine moiety – NPA (the conjugate was also mentioned as KK 1553, see Figure 2 for structure and Figure 10 for STED image series). Also, 4-nitrobenzyl thiol (NBT) was shown to be a good alternative for NPA (4-nitrophenylalanine) or NPAA (4-nitrophenylacetic acid) as a protective moiety [11, 36, 37]. As examples, one can see compounds KK 1555 and KK 1556 in Figure 1.

Figure 9.

Photolysis in solution upon stirring in air: Reference dye KK 114 and an “everlasting dye” - its conjugate with an intramolecular stabilizer (NPA). The blue color fades, as seen in daylight, and so does the initial red color of the fluorescence, as seen under UV illumination (365 nm) in the dark. Note the drastic spectral shift of fluorescence (bluing). The unmodified dye is thus almost completely “burned out” under illumination with a powerful 400-wt. halogen lamp [11].

Figure 10.

The reference dye KK 114 L (top) and its NPA-modified derivative KK 1553 (bottom) in STED imaging. Presented are a series of ten STED frames of a mitochondrial inner membrane Mic60 with secondary antibodies custom-labeled with the fluorophores. Scale bar: 1 μm. The unmodified label fades completely after taking ten STED frames [11].

Experiments on aerobic photolysis in bulk dye solutions showed the following: the unmodified rhodamines always lose their extinction consistently, while the dyes are being split into smaller colorless molecules. At the same time, the PSG-modified dyes paradoxically increase their fluorescence signal in the beginning of the photolysis [11] and survive much longer than the parent dyes, as seen in Figures 9 and 10. However, the extinction of the modified dyes does constantly decrease, yet not as fast as it does in the unmodified compounds. The paradoxical rise in the fluorescence signal in the PSG-modified dyes under illumination can be explained by assuming a dye-mediated photo-cleavage of the photostabilizing group (4-nitrobenzyl). In the course of the cleavage, the “free dye” with a much higher Φ_f is released (see Ref. [11] and Table 1 therein), so the resulting fluorescence signal becomes higher than it was at the beginning. We assume that 4-nitrobenzyl, as a part of NPA and NBT moieties, forms radicals and/or quinoidal structures, analogously to 2-nitrobenzyl, a conventional photolabile protecting group (PPG) [38]. This allows the 4-nitrobenzyl group to relieve the excitation energy from the fluorophore and even to play a sacrificial role. In a FRET pair, the FLT of the donor becomes shorter, as a favorable channel of energy transfer from the excited (S1) state appears [39]. Here the transfer is, however, non-radiative. Also, neither the NPA nor NBT groups absorb visible light. This is completely different from the way how the PPGs work. Summarizing, one can say that the NPA- or NBT-modified dyes can fulfil the demand for stable fluorophores for prolonged microscopical observations. It is noteworthy that the attempts to obtain an NHS ester from the conjugate “NPA-KK 114” failed until an extra linker was installed (see Figure 2). Finally, a praiseworthy idea to improve the photostability by means of a partial deuteration did not work for KK 114 [9]. One could assume that, back then, a wrong site of the fluorophore was modified. However, a decade later, a noticeable increase [10] in photostability was observed when fluorophores were deuterated at their real weak points – terminal N-alkyl sites.

2.5 Do not count the photons, make the photons count!

In addition to the dye design, we tried to discover the real determining factors in the STED microscopy performance of a dye family. In view of our recently presented data, one had to reconsider some well-established views on this topic [11]. Our image series showed that, contrary to the well-established opinion [9, 12], the photostability, not the fluorescence quantum yield of the fluorophore, is the real determining factor in STED imaging. Also, the presence of polar groups and the net charge are not as important as they seemed to be. So, the “race” for very high fluorescence quantum yields is not always justified. Immunolabeling and imaging with the “control dye,” which has no extra polar groups at all, demonstrated no increase in unspecific labeling and no increase in the background either, as seen from the control STED images in [11] and in related supporting information. Note that unspecific labeling and background were previously attributed exclusively to the absence of polar functional groups [9, 12, 22]. Their significance was clearly over-exaggerated. But, as a matter of fact, the decoration of fluorophores with polar functional groups, hydroxy group (-OH) including, was not in vain. As we established [11], these groups do increase photostability, which proved to be the real deciding factor in STED imaging. It was shown that, even with a fluorophore whose Φ_f is as small as 14–17%, good-quality STED images can be taken, as seen in Figure 11 (partly reproduced from Ref. [11]). They were recorded with a sufficient photon budget and with only a small loss in resolution and signal-to-noise ratio, compared to the reference dyes. Very importantly, one should keep in mind that, in immunolabeling one operates with antibodies whose molecular weight (MW) is usually around 150 kDa, whereas the MW of usual small-molecule fluorophores does not exceed 1 kDa. So, a label is obviously not big enough to influence the properties of an antibody conjugate as a whole, particularly its hydrophilicity and polarity. In other words, the antibodies that are usually used for immunolabeling are very hydrophilic and bulky. On top of that, the objects are of micrometer size and, in biology-related applications, are by definition never identical. Neither is the actual amount of a fluorophore in the site under the focus. Substance diffusion and heat dissipation in a viscous medium might be hindered. Finally, the objects can drift away from the focus. All these make it problematic to compare and quantitatively estimate the performance of dyes, especially without statistics. One should definitely hold himself from making far-going conclusions on STED performance of dyes without having at least some statistics. Immunolabeling involves two steps, so the image quality and an unspecific background are very often just a matter of good fortune and/or tidiness. There is, however, one obvious thing that one could state for sure: in the ideal case, one needs a good batch of antibodies, a good label, a good experimentalist, and favorable imaging settings.

Figure 11.

Exemplary STED images with a photostabilized model dye KK 1556 having a fluorescence quantum yield (Φ_f) of only 14%. Presented are the first frame and an add-up of 15 frames of an immunolabeled nuclear pore complex. See Ref. [11] for imaging details and conditions. Scale bar: 1 μm.

2.6 Additional spectrally shifted fluorophores for multicolor imaging

For multi-color imaging, a spectral separation of different labels in two or more excitation or acquisition channels is essential. Alternatively, time-resolved imaging can be performed, where fluorophores with different excited state lifetimes (FLT) are employed [20, 40]. Neither spectral nor FLT separation of two labels is usually not possible within a single dye scaffold where simple functional group transformations are performed. To that end, one needed to switch to another chromophore family. As such, custom-made far-red oxazine and Si-rhodamine dyes (see Figure 7 for structures and Figure 12 for images) were studied [18], which allow both options (a FLT difference and a spectral shift). For imaging in a two-color STED with spectrally separated labels, we used compounds STAR RED and 14-H (another red-emitting rhodamine [20] with solubilizing OH groups, see Figure 7 for structure), whose absorbance and fluorescence maxima are 636/659 and 579/609 nm, respectively. Direct color separation in two observation channels with a negligible background signal was achieved with different cellular substrates, as seen in Figure 13.

Figure 12.

Exemplary STED images with a Si-rhodamine- (KK 1441) and a phosphorylated oxazine-containing (2-H) far-red emitting fluorophores [18]. Scale bar: 1 μm. See Figure 7 for structures.

Figure 13.

Two-color STED image with a pair of spectrally separated red-emitting fluorophores STAR RED (spectrally identical to KK 114, see Figure 1 and [15, 20]) on PMP (in red) and 14-H on NUP 153 (in green, see Figure 7 for structure). Spectral difference is clearly seen in solutions, as well. For details see Ref. [20]. Scale bar: 1 μm.

A later work by Winter and co-authors [40] was a milestone for multi-color imaging: it illustrates the approach showing four-color nanoscopy of fixed and living cellular samples. In that work, a far-red emitting silicon rhodamine KK1441, one of the four dyes, (see Figure 7 for structure) played an important role. The authors came up with “hyperSTED” – a simple and versatile STED instrument design for multicolor nanoscopy. This utilized only two laser sources. Particularly, only one excitation laser and a single STED beam are needed. By hyperspectral detection of four channels, separation of four different markers was accomplished. Discrimination of the markers by the differences in emission spectra allowed fast and simultaneous data acquisition [40]. To realize such an approach in practice, a four-color fixed cell sample was stained with a suitable combination of four of the tested dyes with the following abs./fl. maxima (nm): Atto594 (601/627) for peroxisomes, Abberior STAR 635P (633/659) for vimentin, KK1441 (661/679) for giantin and CF680R (680/701) for nuclear pores.

3. New bacteriology-related applications of red and far-red emitting dyes

3.1 The microbiologist palette: synthetic dyes and fluorescent proteins

The increasing use of fluorescence microscopes and CLSM caused a high demand for new dyes. Max Haitinger (1868–1946) introduced for the first time the term ‘fluorochroming’ or fluorescent staining of specimens, and he developed some staining techniques for fluorescence microscopy [41]. Numerous advancements have emerged in the realm of fluorophore development after the groundbreaking discovery of the green fluorescent protein (GFP) in 1962 by Shimomura and colleagues [42, 43]. Despite the invaluable utility of GFP and their variants in tracking protein expression and localization within cells, they often exhibit a lower quantum yield compared to organic dyes [42]. Additionally, challenges such as fluorescent protein oligomerization, fluorescence attenuation during fixation, and moderate cytotoxicity upon overexpression persist [42]. Consequently, there is a growing demand for novel organic dyes, particularly, characterized by reduced steric bulk and accelerated labeling kinetics [44, 45]. These dyes are based on various molecular scaffolds including fluorescein, cyanine, oxazine, squaraine, BODIPY, xanthene, benzobisthiadiazole, and rhodamine [44]. For a more in-depth discussion on the use of these fluorophores in bacteriology, the reader is referred to the reviews [44, 45, 46]. The impact of fluorescent dyes in bacteriology further increased when some fluorophores were ad hoc generated to differently stain viable and dead bacteria, thus giving additional information on bacterial metabolic activity and/or membrane integrity under the exposition to environmental stressors (e.g., antibiotic treatment). The predominantly employed combination of synthetic dyes in bacteriology involves SYTO 9 and propidium iodide (PI), commercially available in the LIVE/DEAD BacLight Bacterial Viability and Counting Kit (Thermo Fisher). The green fluorescing dye SYTO 9 penetrates inside a bacterial cell, and it is used to assess the total cell count, whereas the red fluorescing dye PI enters only cells with damaged membranes [47]. The emission properties of the stain mixture bound to DNA change due to the displacement of SYTO 9 by PI and quenching by fluorescence resonance energy transfer [48]. Increasing the presence of membrane discontinuities in cells facilitates the entry of PI, resulting in higher levels of SYTO 9 fluorescence quenching and emission of red fluorescence. Other techniques involve LIVE/DEAD staining, spanning from flow-cytometry [49], fluorimeter and other automatic fluorescence readers, confocal microscopy imaging [50], and fiber-based spectroscopic systems known as optrodes [51, 52].

3.2 The dark side of the microbiologist toolbox: the red and near-infrared emitting dyes

Red, far-red (FR), and near-infrared (NIR) emitting dyes offer significant benefits in bio-imaging within living systems. Their fluorescence in the longer wavelength range minimizes phototoxicity to biological components, enables deep tissue penetration, and reduces background interference from auto-fluorescent biomolecules [53]. In the last years, many efforts have been focused on the repurposing of fluorophores for bacterial imaging [44]. Luckily, among other candidates, the dye KK 114 (commercially available under the name Abberrior STAR RED [15, 21]) was tested as a contrast agent. It proved to perform very well, which was nicely demonstrated in the recent works on eukaryotic cells [21, 54, 55, 56]. This fluorophore is water-soluble, highly photostable, and negatively charged (z = −1 in the bound state). This makes its conjugates tentatively not permeable for eukaryotic cells. Surprisingly, despite its negative charge, this molecule demonstrated an ability to adhere to the membranes of various bacterial species, encompassing both Gram-positive and Gram-negative bacteria [54, 55]. Notably, the outer membrane of Gram-negative bacteria and the cytoplasmic membrane/cell wall of Gram-positive bacteria are decorated with lipopolysaccharide and teichoic acids, respectively, both making their outer shells negatively charged. This suggests that charge repulsion alone is insufficient to hinder the dye from binding to the membrane, implying the presence of other yet-to-be-identified mechanisms governing the interaction between the dye and bacterial macromolecules. Despite the unclear mechanism of action, KK 114 proved to be suitable for staining different members of the ESKAPE group. ESKAPE is the acronym for the group of Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter species, which cause life-threatening hospital-acquired infections, mainly among critically ill and immunocompromised patients. These species are considered critical-priority pathogens, representing a global menace because of their high transmissibility, mortality, limited treatment options for the rapid diffusion of multidrug-resistant strains, and healthcare and community burden [57].

The capability to label bacterial membranes opens the vault to track the arrangements that occur during cell division. The inhibition of the bacterial division machinery (namely the divisome) gives a background for the generation of new antimicrobial compounds capable of interfering with the natural dynamics and functions of bacterial replication [58] with implications in ESKAPE containment, treatment, and eradication. Nevertheless, the imaging of bacterial membranes faces challenges due to the <10 nm dimensions of this structure, significantly below the maximum limit of spatial resolution achievable by modern confocal microscopes, which is approximately 200–250 nm [59]. To overcome this limitation, super-resolution approaches are required to achieve a suitable resolution for membrane imaging. STED microscopy is in fact an example of a CLSM-derived super-resolution technique. It is capable of pushing the resolution limits beyond the diffraction barrier [1, 2, 3, 59]; a thorough examination of super-resolution approaches is provided in the reviews by Refs. [46, 60]. Utilizing the remarkable photostability of KK 114, STED imaging was performed on different bacterial species (Figure 14), thus enabling the observation of bacterial membranes with a resolution one order of magnitude greater than that achievable with CLSM [21].

Figure 14.

STED imaging of different bacterial species stained with the dye KK 114: E. faecium (A); S. aureus (B); Bacillus subtilis (C); Streptococcus pyogenes (D); Escherichia coli (E); Achromobacter xylosoxidans (F); A. baumannii ATCC 17978 (G); A. baumannii ATCC 19606^T (H); and K. pneumoniae (I). Cyan arrows indicate KK 114 penetration inside the bacterial cytoplasm, resulting in higher fluorescence emission. A different magnification was used for each bacterial species. See Ref. [21] for labeling and imaging conditions.

In STED imaging, a secondary KK 114 mechanism of action emerged: this molecule demonstrated a capability to penetrate bacterial cells with compromised membranes with a higher fluorescence emission (as indicated by the cyan arrows in Figure 14), suggesting a potential for conducting LIVE/DEAD staining. To that end, one should combine KK 114 with other compatible fluorophores [21, 54, 55]. Apart from its potential application in labeling division septa for exploring molecules to enhance our therapeutic options against ESKAPE pathogens, another interesting feature of KK114 was unveiled. It is also a powerful tool to perform rapid detection of colistin resistance in A. baumannii [55]. Colistin is a positively charged membrane-disrupting peptide used as a “last-resort” drug to treat infections caused by A. baumannii multidrug-resistant strains. This peptide interacts with the outer membrane, a distinct feature of Gram-negative bacteria, leading to the displacement of divalent cations that stabilize the outer membrane, thus causing derangement of the cell membrane, leakage of intracellular contents, and ultimately cell death [56]. A multicolor labeling strategy based on the combination of SYTO 9 and KK 114 allowed discrimination of colistin-resistant from colistin-sensitive A. baumannii strains [55]. In colistin-sensitive strains, exposure to colistin causes loss of membrane integrity that results in KK 114 intracellular penetration accompanied by an enhancement in FR fluorescence emission. Conversely, the membranes of colistin-resistant strains are impermeable to KK 114, with no detectable increase in FR emissions even at high colistin concentrations. Therefore, in colistin-sensitive strains, the ratio between SYTO 9 and KK 114 emissions swiftly diminishes in response to colistin concentrations and treatment duration, whereas it remains constant in colistin-resistant strains [55]. This alternative method for detecting colistin resistance in A. baumannii offers a comparable duration (i.e., less than 2 hours) to other techniques that are not based on bacterial cultivation and have a direct application in the clinical routine (e.g., resazurin reduction-based assay) [61].

The dye KK 114 served as an example of an intriguing case study, highlighting the importance of expanding the palette of available dyes in the development of new diagnostic and therapeutic approaches for treating bacterial infections. Encouraged by these promising results, we tested an extended set of dyes [21]. To elucidate the labeling mechanism, the dyes were used both in their reactive form (NHS esters) and inactive. Also importantly, the net charges varied in the range from +2 to −4. It was found that all the rhodamine dyes of the set and an exemplary oxazine dye KK 1905 (also as NHS ester), as well, can be used to label dead bacterial cells, thus expanding the range of available dyes for LIVE/DEAD applications. As we established, these dyes do not interfere with bacterial replication and can be used in live-cell imaging experiments, which is essential in bacteriology for gauging the physiological state of bacteria, untangling the intricate mechanisms at the basis of the bacterial divisome, performing real-time transcriptional analysis at the single-cell level, investigating phenotypic heterogeneity in bacterial populations, and exploring the host-bacteria interactions [62]. Among the tested dyes, only a few could effectively stain bacterial membranes. Out of the whole dye set, only KK 1517, KK 1518, and KK 1905-NHS could stain viable bacteria. In particular, the zwitter-ionic rhodamine KK 1517 (z = o) was able to decorate the cell periphery of all the B. subtilis cells and of some E. coli cells, partially penetrating cytoplasm of both species. On the other hand, KK 1518 and KK 1905-NHS (both positively charged, see Ref. [21] for structures) selectively stained the cell periphery of B. subtilis and E. coli cells, with the latter generating brighter and more contrasted images. Note that the dye KK 1518 was a purposefully synthesized label with a net charge of +2 and a water-solubilizing moiety.

4. Conclusions

Very reasonable strategies were employed to develop user-friendly red and far-red emitting microscopy labels with tailored properties. Particularly, the authors summarize and discuss the syntheses of well-performing STED fluorophores with enhanced photostability. Some issues are highlighted that did not receive attention in the previous publications. Straightforward modifications at proper sites of a rhodamine dye scaffold enable synthesis and evaluation of diverse candidates. It was shown that the properties of the fluorophores alone, i.e., polar functional groups and the fluorescence quantum yields, are not as crucial for STED microscopy as they were considered to be. As one can see from the comparative studies [11], the “race” for high quantum yields of fluorescence is not always justified. The photostability, proper instrument settings, the quality of antibodies, and the quality of labeling are the real deciding factors. Also, in recent years, new techniques were discovered that allow to reduce the high intensity of a STED beam and further increase the resolution, on the other hand, as one sees, for example, in [63, 64] and in the references therein. Importantly, apart from the conventional immunolabeling with antibodies or with HaloTag® ligands, there exists a great potential for non-covalent binding to cell structures, particularly bacterial membranes. The rhodamine dyes that were originally designed for STED microscopy a decade ago, for example, KK 114, have thus gotten a “new life” [21, 54]. This offers fresh avenues and rich toolboxes for microbiological research and diagnostic applications. Especially important could be custom-made fluorophores with multiple positive net charges [21], which were previously underestimated as STED dyes. Also, possibilities may be broadened in single-molecule studies and in medical diagnostics, as well. By diversifying the palette of the dyes available, researchers can enhance their ability to observe and analyze biological samples with greater precision and sensitivity, not necessarily using STED. This advancement holds promise for deeper insights into microbiology and related fields, as well as improving diagnostic and analytical techniques in medical and scientific research.

Acknowledgments

The authors are grateful to Prof. S. W. Hell for the opportunity to work at the Department of NanoBiophotonics (MPI-BPC Göttingen) and perform STED imaging in 2008—2015. Acknowledged are the grants, particularly, FKZ 13 N11066 and FKZ 13 N14122 (Bundesministerium für Bildung und Forschung), which financed the research work at that period. The studies would not be possible without extra images and good advice kindly provided by Dr. Daniel C. Jans, Dr. Franziska R. Winter, Dr. Fabian Göttfert, Dr. Maksim V. Sednev (MPI-BPC Göttingen) and Dr. Sergey M. Borisov (Graz University of Technology). Highly acknowledged are the European Society for Photobiology, the European Photochemistry Association, and the Royal Society of Chemistry for the license agreement allowing the authors to reproduce their figures. Very helpful were the good advice, instrumental, informational, and moral support from glyXera GmbH (Germany) and Peptide Technologies Ltd. (Russia). Dr. M. Lucidi especially acknowledges the support of NBFC to the University of Roma Tre (CN00000033).

Conflict of interests

The authors have no conflicts of interest to declare. None of the authors is employed by the companies whose products or instruments are mentioned.