Near-infrared imaging in fission yeast using a genetically encoded phycocyanobilin biosynthesis system

Fluorescent proteins have become indispensable for visualization of biological processes in living cells and tissues (Lambert, 2019). Green fluorescent protein (GFP), the most widely used fluorescent protein, has been intensively modified to improve the brightness and the photo-, thermo- and pH-stabilities, as well as to change the excitation and emission spectra. The use of a variety of fluorescent proteins with different excitation and emission spectra enables multiplexed fluorescence imaging to monitor multiple biological events simultaneously at high spatial and temporal resolution.

Near-infrared fluorescent proteins (iRFPs) have been developed through the engineering of phytochromes, which are photosensory proteins of plants, bacteria and fungi (Chernov et al., 2017), or allophycocyanin, which is a light-harvesting phycobiliprotein of cyanobacteria (Rodriguez et al., 2016). RpBphP2 (also known as PhyB1) from photosynthetic bacteria has been engineered to function as an iRFP (later renamed iRFP713) by truncation and saturation mutagenesis (Filonov et al., 2011). Since the initial report of iRFP, tremendous efforts have been devoted to developing other iRFPs with higher brightness, monomer formation and longer wavelength (Filonov et al., 2011; Fushimi et al., 2019; Kamper et al., 2018; Matlashov et al., 2020; Oliinyk et al., 2019; Rodriguez et al., 2016; Rogers et al., 2019; Shcherbakova and Verkhusha, 2013; Shcherbakova et al., 2016, 2018; Stepanenko et al., 2016; Yu et al., 2014, 2015). Unlike the canonical fluorescent proteins derived from jellyfish or coral, phytochromes and allophycocyanin require a linear tetrapyrrole as a chromophore, such as biliverdin IXα (BV), phycocyanobilin (PCB) or phytochromobilin (PΦB); bacteriophytochromes bind to BV, allophycocyanin and cyanobacterial phytochromes bind to PCB, and plant phytochromes bind to PΦB. These photosensory proteins autocatalytically form a covalent bond with the chromophore (Fushimi and Narikawa, 2021). Linear tetrapyrrole chromophores are produced from heme (Beale, 1993; Terry and Lagarias, 1991). Heme oxygenase (HO) catalyzes oxidative cleavage of heme to generate BV with the help of ferredoxin (Fd), an electron donor, and ferredoxin–NADP⁺ reductase (Fnr) (Cornejo et al., 1998). In cyanobacteria, PCB is produced from BV through PcyA, Fd and Fnr, whereas in plants, PΦB is synthesized from BV using HY2, Fd and Fnr (Frankenberg et al., 2001; Kohchi et al., 2001; Muramoto et al., 1999). To exploit phytochromes that are required for PCB or PΦB in other organisms, our group and others have demonstrated reconstitution of BV, PCB and PΦB synthesis in bacteria, mammalian cells, frog eggs, budding yeast, methylotrophic yeast and fission yeast (Gambetta and Lagarias, 2001; Hochrein et al., 2017; Kyriakakis et al., 2018; Landgraf et al., 2001; Mukougawa et al., 2006; Müller et al., 2013; Shin et al., 2014; Tooley et al., 2001; Uda et al., 2017, 2020).

As the fluorescence of iRFP depends on chromophore formation, the BV concentration is of critical importance for imaging iRFP (Fig. 1A). Indeed, it has been reported that the addition of purified BV increases the fluorescence of iRFPs (Piatkevich et al., 2017; Shemetov et al., 2017). Alternatively, genetic modifications such as the overexpression of heme oxygenase-1 (HO1), which catalyzes degradation of heme to generate BV, and the knockout of biliverdin reductase A (BLVRA), which degrades BV to generate bilirubin, improve the brightness of iRFP through the additional accumulation of BV (Kobachi et al., 2020; Shemetov et al., 2017). Conversely, because Caenorhabditis elegans produces little or no BV (Ding et al., 2017), it is not possible to image biological processes in this nematode simply by introducing the gene encoding iRFP. In the case of multicellular organisms that cannot produce BV, including C. elegans, the introduction of genes required for BV production is more effective than the external addition of BV because of the low tissue penetration properties of BV. However, at present, only the introduction of the HO1 gene has been reported as a genetically encoded method for inducing the iRFP chromophore, and it has not been improved or optimized yet.

Here, we report that PCB acts as a better chromophore for iRFP than BV, and that genetically encoded PCB synthesis outperforms HO1-mediated BV production in terms of iRFP brightness in the fission yeast Schizosaccharomyces pombe. We accidentally found that iRFP did not fluoresce in fission yeast because of the lack of the gene encoding HO1, and therefore the lack of BV. Both external BV addition and heterologous HO1 expression rendered iRFP fluorescent in fission yeast. To our surprise, PCB biosynthesis with the SynPCB2.1 system, which we have previously reported (Uda et al., 2017, 2020), and treatment with purified PCB both yielded brighter iRFP fluorescence than that induced by either BV biosynthesis or BV treatment. We confirmed that PCB-bound iRFP showed higher fluorescence quantum yield than BV-bound iRFP. To facilitate the simple use of iRFP in fission yeast, we developed a plasmid for iRFP tagging of endogenous proteins at the C terminus, novel genome integration vectors and all-in-one plasmids carrying genes required for both the SynPCB2.1 system and iRFP-fused marker proteins. Finally, we proved the generality of PCB for iRFP imaging in budding yeast Saccharomyces cerevisiae and mammalian cells.

During the process of experiments, we accidentally found that iRFP did not fluoresce at all in fission yeast. We first tested whether iRFP could be used for near-infrared imaging in fission yeast. We established a cell strain stably expressing iRFP fused to two nuclear localization signals (NLSs), NLS–iRFP–NLS (Miura et al., 2018), under the constitutive adh1 promoter (Padh1). Two NLSs were fused with iRFP because the addition of a single NLS did not sufficiently localize the protein to the nucleus. No iRFP fluorescence was observed at an excitation wavelength of 640 nm (Fig. 1B). Because iRFP requires BV as a chromophore for emitting fluorescence (Fig. 1A), we hypothesized that fission yeast could not metabolize BV intracellularly. Upon the addition of external BV, the nuclear iRFP fluorescence signal was recovered (Fig. 1B). The titration of BV concentration yielded a dose-dependent increase in iRFP fluorescence up to 125 µM (Fig. 1C). We next examined the kinetics of BV uptake into fission yeast cells. Treatment with a high dose of BV (500 µM) gradually increased iRFP fluorescence until 24 h, suggesting slow uptake of BV in fission yeast cells (Fig. 1D). Since BV is produced from heme through the activity of HO, we searched for HO-like proteins encoded in the genomes of fission yeast and representative fungal species. As expected, we could not find any HO or HO-like protein in fission yeast (Fig. 1E). Interestingly, genes encoding HO and/or HO-like proteins, which have been found from bacteria to higher eukaryotes, are frequently and sporadically lost in the representative fungal species (Fig. 1E). Indeed, while iRFP has been widely used in the budding yeast S. cerevisiae, which retains gene encoding HO (Geller et al., 2019; Li et al., 2017; Tojima et al., 2019; Wosika et al., 2016), there have been no studies using iRFP in the fission yeast S. pombe. Taken together, these facts led us to conclude that iRFP does not fluoresce in fission yeast due to the lack of HO.

The above results showed that the external supply of BV required a high dose and long-term incubation to realize iRFP fluorescence in fission yeast, which prompted us to seek an alternative route to iRFP fluorescence by introducing genes for the biosynthesis of BV. Before starting to develop the reconstitution system, we developed novel stable integration vectors that met our specific requirements – stable one-copy integration into the genome, no effect on the auxotrophy of integrated cells and distant integration loci for crossing strains – rather than using one of the previously developed integration systems (Fennessy et al., 2014; Kakui et al., 2015; Keeney and Boeke, 1994; Matsuyama et al., 2004; Maundrell, 1993; Siam et al., 2004; Vještica et al., 2020). At first, we chose three gene-free loci on each chromosome at chromosome I positions 1,508,522 to 1,508,641 (near mug165, 1L), chromosome II positions 447,732 to 447,827 (near pho4, 2L), and chromosome III positions 1,822,244 to 1,822,343 (near nup60, 3R) (Fig. 2A). Next, we designed and developed plasmids that contain genes required for replication and amplification in Escherichia coli (Amp, ori), the constitutive promoter Padh1 or inducible nmt1 promoter (Pnmt1), a multiple cloning site (MCS), an adh1 terminator, a selection marker cassette encoding an antibiotic-resistance gene for fission yeast, and homology arms connected with the one-cut restriction enzyme recognition site for plasmid linearization (Fig. 2B). Expected genomic integration with these vectors was confirmed by genomic PCR using primers designed to span the integration boundary (Fig. 2C). None of these integrations affected the bulk growth of fission yeast (Fig. 2D), and the protein expression levels from these three loci were comparable or moderately higher than that from the Z locus, which is the locus adjacent to the zfs1+ gene of chromosome 2 (Sakuno et al., 2009) (Fig. 2E). We named this series of plasmids using the prefix pSKI (plasmid for stable knock-in; see Table S1) and used them for the following experiments.

HO is the crucial enzyme in the BV biosynthesis pathway, catalyzing the linearization of tetrapyrrole (Fig. 3A). Therefore, we established fission yeast cells stably expressing HO1 and NLS–iRFP–NLS using a pSKI plasmid and quantified the resulting iRFP fluorescence. As expected, the expression of HO1 derived from Thermosynechococcus elongatus strain BP-1 in mitochondria, where heme is abundant, resulted in iRFP fluorescence, and the iRFP fluorescence was brighter than that achieved by the external addition of BV (Fig. 3B, second and third columns). Because HO1 is known to catalyze heme in the presence of reduced Fd (Rhie and Beale, 1992), we next examined whether co-expression of HO1 and tFnr–Fd, a chimeric protein consisting of truncated T. elongatus Fnr (also known as PetH) and Fd (also known as PetF1) (Uda et al., 2020), would improve HO1-mediated iRFP fluorescence. However, the co-expression of HO1 and tFnr–Fd in mitochondria did not further enhance iRFP fluorescence as compared to the expression of HO1 alone (Fig. 3B, sixth column), suggesting that endogenous Fd in fission yeast sufficiently supports the catalytic reaction through HO1.

Unexpectedly, in a series of experiments, we found a further enhancement of iRFP fluorescence by PCB. When T. elongatus PcyA, the enzyme responsible for the production of PCB from BV, was co-expressed with HO1 and tFnr–Fd, the level of iRFP fluorescence was higher than other conditions (Fig. 3B, ninth column). To validate these results, we treated the cells expressing NLS–iRFP–NLS with purified PCB instead of BV. The addition of external PCB substantially outperformed the addition of BV with respect to iRFP fluorescence intensity (Fig. 3C,D). While the fluorescence intensities were quite different between PCB-bound iRFP (iRFP–PCB) and BV-bound iRFP (iRFP–BV), the effective concentration of the dose-response curve (Figs 1C and 3C) and the kinetics of chromophore incorporation (Figs 1D and 3D) were comparable between them.

The above data indicated the possibility that PCB might be a more suitable chromophore for iRFP than BV. To prove this hypothesis, we first examined whether the efficiency of holo-iRFP formation accounted for the difference in iRFP fluorescence between BV- and PCB-treated cells. PCB was added to the cells with HO1 expression, which exhibited constant intracellular production of BV. Therefore, iRFP already formed a holo-complex with BV before attaching to PCB (Fig. 4A). Given that iRFP–PCB is brighter than iRFP–BV, we reasoned that HO1 expression attenuated the increase in iRFP fluorescence when the cells were further treated with purified PCB due to competition between the PCB and already existing BV for binding to iRFP. As expected, the addition of purified PCB hardly increased iRFP fluorescence in cells that had been expressing HO1, despite a dose-dependent increase in iRFP fluorescence being observed upon PCB treatment of cells not expressing HO1 (Fig. 4B,C). These observations reveal that almost all iRFP forms a holo-complex with BV when HO1 is expressed.

To understand why iRFP–PCB was brighter than iRFP–BV, we prepared recombinant iRFP expressed in E. coli and purified apo-iRFP (Filonov et al., 2011) (Fig. S1A). Apo-iRFP was mixed with PCB and BV to form holo-iRFP (i.e. iRFP–PCB and iRFP–BV, respectively; Fig. S1B). The binding of PCB to iRFP resulted in a change in the absorption spectrum from the free PCB (Fig. 4D). The absorbance maximum of iRFP–PCB was blueshifted by 10 nm from that of iRFP–BV (Fig. 4E). Fluorescence excitation and emission spectra were also 10 nm blueshifted in iRFP–PCB compared to those of iRFP–BV (Fig. S1C). Notably, the fluorescence quantum yield of iRFP–PCB was nearly twice as high as that of iRFP–BV (0.094 versus 0.054), whereas their molecular extinction coefficient values were comparable (Fig. 4F). These results are consistent with previously published work (Stepanenko et al., 2019). Based on these results, we concluded that iRFP forms a complex with PCB as a holo-form and that iRFP–PCB is brighter than iRFP–BV at the molecular level.

For easy iRFP imaging using PCB as a chromophore, we introduced a system for efficient PCB biosynthesis, SynPCB2.1, in which tFnr–Fd, PcyA and HO1 genes are tandemly fused with the cDNAs of the mitochondrial targeting sequences (MTS) at their N termini and flanked by self-cleaving P2A peptide cDNAs for multi-cistronic gene expression (Uda et al., 2020) (Fig. 4G). The single cassette of SynPCB2.1 genes was knocked-in into cells expressing NLS–iRFP–NLS using the pSKI vector system and was expressed under the adh1 promoter. The cells expressing SynPCB2.1 showed higher iRFP fluorescence than both cells treated with PCB and cells expressing the three genes individually (Fig. 4G). The protein expression levels of iRFP were comparable between the cells treated with BV or PCB, and cells expressing HO1 or SynPCB2.1 (Fig. S2). These results indicate that the chromophore formation of iRFP has little impact on the protein stability of iRFP in fission yeast.

To determine to what extent iRFP formed a complex with PCB or BV in fission yeast cells, we quantified the number of fluorescent iRFP molecules using fluorescence correlation spectroscopy (FCS). FCS is a technique that exploits temporal fluctuations of fluorescent molecules in a confocal volume (∼1 fl) to enable estimation of the number of fluorescent molecules in the confocal volume and the diffusion coefficient (Kinjo et al., 2011; Shi et al., 2009; Sudhaharan et al., 2009). For this purpose, fission yeast cells expressing iRFP fused with mNeonGreen (iRFP–mNeonGreen) were treated with BV or PCB, or were transformed to co-express either HO1 or SynPCB2.1, and were then subjected to FCS measurement to quantify the numbers of fluorescent iRFP and mNeonGreen molecules (Fig. S3A). The more that iRFP forms a complex with the chromophore and fluoresces in cells, the more the ratio of the number of fluorescent iRFP molecules to the number of fluorescent mNeonGreen molecules measured by FCS approaches a value of 1.0. The cells expressing iRFP–mNeonGreen and SynPCB2.1, as well as the cells treated with PCB, exhibited ratio values of ∼0.8–1.0 (Fig. S3B,C), showing that 80–100% of iRFP formed a complex with PCB under these conditions. Importantly, HO1 expression resulted in the formation of a holo-iRFP complex with almost the same efficiency as SynPCB2.1 expression (Fig. S3C). Given the fact that the brightness of iRFP–BV was much weaker than that of iRFP–PCB (Fig. 4G), these results indicate that iRFP–PCB is a substantially brighter fluorescent protein than iRFP–BV in fission yeast. The external addition of BV resulted in lower values of iRFP to mNeonGreen ratio (Fig. S3C, first column), suggesting that the uptake of BV from outside of cells is the rate-limiting step in fission yeast. Recombinant iRFP also displayed comparable binding stoichiometry between BV and PCB (Fig. S3D,E).

Next, we measured the emission spectrum of iRFP in a living cell to compare the fluorescence properties of iRFP–BV and iRFP–PCB. As was observed for the emission spectra in vitro, the cells showed distinct emission spectra for iRFP–PCB and iRFP–BV, namely, a blueshifted emission spectrum of iRFP–PCB (Fig. S3F). A similar shift was observed when the emission spectrum of cells expressing SynPCB2.1 was compared to that of cells expressing HO1 (Fig. S3G and summarized in Fig. S3J). Importantly, cells separately expressing HO1, tFnr–Fd, and PcyA exhibited an intermediate emission spectrum, suggesting a mixture of iRFP–BV and iRFP–PCB in this cell line. The presence of iRFP–BV would explain why iRFP fluorescence upon SynPCB2.1 expression was brighter than that generated by separate expression of the three enzymes in fission yeast (Fig. 4G, fourth and sixth columns). Moreover, the emission spectra obtained from living fission yeast cells demonstrated that iRFP–PCB was much brighter than iRFP–BV (Fig. S3H,I).

During iRFP imaging experiments, we found that PCB synthesized in fission yeast cells expressing SynPCB2.1 leaks out of the cells and is incorporated into the surrounding cells. To clearly show the PCB leakage, we co-cultured cells expressing only SynPCB2.1 and cells expressing only NLS–iRFP–NLS (Fig. S4A). While neither strain exhibited any fluorescence when cultured singly, NLS–iRFP–NLS emanated fluorescence when cells were co-cultured with the cells expressing SynPCB2.1 (Fig. S4B,C). The data indicate that in fission yeast, PCB leaks into the extracellular space.

To further exploit the advantages of iRFP imaging in fission yeast, we first established C-terminal tagging plasmids based on a commonly used PCR-based tagging system (Longtine et al., 1998). The plasmids included an iRFP cassette followed by one of four different selection markers (Fig. 5A). Using these plasmids, we verified endogenous iRFP tagging and resulting fusion protein localization for several genes, including cdc2 (CDK, nucleus), rpb9 (RNA polymerase II subunit, chromatin), rpa49 (RNA polymerase I subunit, nucleolus), swi6 (heterochromatin), pds5 (cohesin), cut11 (nuclear envelope), mal3 (microtubule plus-end), sfi1 (spindle pole body, SPB), cox4 (mitochondria) and cnx1 (also known as cal1; endoplasmic reticulum, ER), with the expression of SynPCB2.1. All tested proteins showed the expected subcellular localization in fission yeast (Fig. 5B), although the signal-to-noise ratios depended on the expression level of the endogenously tagged proteins.

Second, we developed all-in-one plasmids carrying SynPCB2.1 and iRFP fusion protein genes to avoid a situation in which these two genes occupy two of the limited selection markers and integration loci. As a proof of concept, we introduced cDNA encoding Lifeact–iRFP (F-actin marker) or NLS–iRFP–NLS (nuclear marker) into the pSKI plasmid with the SynPCB2.1 gene cassette (Fig. 6A,B). Fission yeast transformed with these plasmids displayed bright fluorescent labeling of either F-actin, including actin patches, actin cables and contractile ring (Fig. 6A), or the nucleus (Fig. 6B), respectively. Taking full advantage of iRFP imaging using the SynPCB2.1 system in fission yeast, we established cells expressing five different fluorescent fusion proteins: the nucleus, plasma membrane, kinetochore, tubulin and F-actin were labeled with NLS–mTagBFP2–NLS, Turquoise2-GL–ras1ΔN200, endogenous Mis12–mNeonGreen, mCherry–Atb2 and Lifeact–iRFP, respectively (Fig. 6C).

To demonstrate the generality of our approach, we measured fluorescence intensities of BV- or PCB-bound miRFP670 and miRFP703, which are derived from a different branch of bacteriophytochrome RpBphP1 (Shcherbakova et al., 2016). Fission yeast expressing miRFP670, miRFP703 or iRFP (i.e. iRFP713) were treated with BV or PCB or were transformed to co-express SynPCB2.1 (Fig. S5). As with iRFP, PCB addition or SynPCB2.1 expression enhanced the fluorescence intensities of miRFP670 and miRFP703 compared to the effects of adding BV (Fig. S5). These data show that PCB biosynthesis by SynPCB2.1 is suitable for imaging not only with iRFP but also with miRFP670 and miRFP703 in fission yeast.

Next, we tested whether near-infrared fluorescent imaging with PCB enhances iRFP fluorescence in budding yeast. iRFP has been already applied to near-infrared fluorescent imaging in budding yeast (Geller et al., 2019; Li et al., 2017; Tojima et al., 2019; Wosika et al., 2016). This is reasonable because budding yeast harbors the HMX1 gene, a heme oxygenase homolog (Fig. 1E). At first, we confirmed that iRFP fluorescence depends on endogenous HMX1 in budding yeast expressing NLS–iRFP–NLS. Cells lacking the HMX1 gene (hmx1Δ) showed reduced iRFP fluorescence to almost a background level, and the reduction was rescued by treatment with exogenous BV (Fig. 7A and B, first and third columns). The addition of BV to wild-type budding yeast cells further enhanced iRFP fluorescence (Fig. 7A,B), indicating that BV produced by endogenous Hmx1 did not suffice to reach maximal fluorescence intensity of iRFP. Next, we treated cells with PCB or BV to compare the iRFP fluorescence intensity. As in fission yeast (Fig. 3C,D), treatment with PCB resulted in a higher intensity of iRFP fluorescence than that resulting from treatment with BV in both dose- (Fig. 7C) and time-dependent (Fig. 7D) manners. Moreover, PCB treatment resulted in a tenfold increase in the fluorescence of iRFP labeling of endogenous Hta2, a histone H2A, indicating the advantage of PCB for iRFP imaging in budding yeast (Fig. 7E,F).

Finally, we tested whether PCB could be used as an iRFP chromophore in mammalian cells. HeLa cells expressing iRFP along with EGFP, an internal control for iRFP expression, were treated with external BV or PCB. PCB treatment increased the brightness of iRFP in HeLa cells to the same degree as BV treatment (Fig. S6A,B). BLVRA knockout (KO) HeLa cells displayed higher iRFP fluorescence than parental HeLa cells, as reported previously (Kobachi et al., 2020), but did not show any change in iRFP fluorescence upon BV or PCB treatment (Fig. S6B). We hypothesized that pre-existing BV either generated by endogenous HO1 or derived from fetal bovine serum in the culture medium could occupy the iRFP before the treatment with PCB. To address this concern, we established HO1 KO and HO1/BLVRA double KO HeLa cell lines and cultured these cell lines with or without serum. Overall, the iRFP fluorescence in cells in the presence of serum (Fig. S6C) followed a similar trend to that in cells without serum (Fig. S6D) under our experimental conditions. HO1 KO partially, but not completely, diminished iRFP fluorescence (Fig. S6C and D, first versus fourth columns, and seventh versus tenth columns). The residual iRFP fluorescence can probably be attributed to other enzymes generating BV and/or residual BV in the cells and the culture medium. We observed that PCB treatment slightly increased iRFP fluorescence in BLVRA KO and HO1/BLVRA double KO HeLa cells cultured without serum (Fig. S6D). Taken together, these results led us to conclude that PCB is applicable to iRFP imaging in mammalian cells and that it is possible to increase iRFP fluorescence using PCB, although it does not offer a significant advantage over BV.

In this study, we demonstrated that iRFP does not fluoresce in fission yeast because of the lack of the BV-producing enzyme HO. Moreover, we found that PCB acts as a brighter chromophore for iRFP than BV both in vitro and in fission yeast expressing SynPCB2.1. Although PCB is not an authentic chromophore for iRFP or the original RpBphP2, our data strongly suggest that PCB forms a fluorescent chromophore in iRFP. For easy use of iRFP in fission yeast, we developed endogenous iRFP tagging plasmids and all-in-one plasmids carrying SynPCB2.1 and iRFP marker proteins. Finally, we demonstrated that PCB also enhances iRFP fluorescence in budding yeast. As an alternative to external chromophore addition, the SynPCB2.1 system has potential advantages for iRFP imaging, including being fully genetically encoded and capable of providing even brighter iRFP fluorescence in fission yeast.

Our data indicate that PCB is more suitable as an iRFP chromophore than BV in fission yeast for several reasons. The first reason is that iRFP–PCB has a 2-fold higher fluorescence quantum yield than iRFP–BV in vitro. Considering the molar extinction coefficient, the iRFP-PCB molecule is 1.61-fold brighter than iRFP-BV (Fig. 4F). The second reason is that the excitation and emission spectra of iRFP–PCB are blueshifted in comparison to those of iRFP–BV. This result is consistent with previous work describing the blueshifted spectra of PCB (Loughlin et al., 2016; Rumyantsev et al., 2015). The blueshifted spectra of iRFP–PCB possess favorable properties for most conventional confocal microscopes. Based on the emission and excitation spectra (Fig. S1C), iRFP–PCB is ∼1.3-fold more effectively excited by a 640 nm excitation laser and is detected ∼2.0-fold more efficiently with our emission filter (665–705 nm emission filter) in comparison to iRFP–BV. Based on these data, a rough estimation yields an overall 4.2-fold increase (1.61×1.3×2=4.2), which is comparable with our experimental results showing a roughly 4-fold increase in iRFP–PCB fluorescence compared to iRFP–BV (Fig. 4G, third and sixth columns). In contrast to fission and budding yeast, HeLa cells showed only a slight difference in iRFP fluorescence between PCB and BV chromophores (Fig. S6). This could be partly due to the metabolism and culture conditions in mammalian cells, including synthesis of BV by endogenous HO1 or other enzymes, degradation of BV and PCB by BLVRA (Kobachi et al., 2020; Terry et al., 1993; Uda et al., 2017), and the presence of BV and bilirubin in the serum of the culture medium. Based on the results obtained using fission yeast (Fig. 4A–C), we presume that the existence of BV within a HeLa cell and in the culture medium attenuates the increase in PCB-induced iRFP fluorescence. Moreover, other tetrapyrroles, such as protoporphyrin IX, could compete with BV or PCB for iRFP (Lehtivuori et al., 2013; Wagner et al., 2008).

The SynPCB2.1 system allows bright iRFP imaging without adding external chromophores. This fact led us to consider that PCB might be applicable to other BV-based fluorescent proteins and optogenetic tools, as with miRFP670 and miRFP703, which exhibited increased fluorescence with the SynPCB2.1 system (Fig. S5). Indeed, near-infrared fluorescent proteins that originate from allophycocyanin and cyanobacteriochrome, such as smURFP and iRFP670nano, respectively, exhibit high affinity to PCB because the original allophycocyanin and cyanobacteriochrome bind specifically to PCB (Oliinyk et al., 2019; Rodriguez et al., 2016). Bacteriophytochrome-based optogenetic tools using BV (Kaberniuk et al., 2016; Monakhov et al., 2020; Qian et al., 2020; Redchuk et al., 2017) would be a potential target for the application of the SynPCB2.1 system. We should note that it is not clear whether PCB, instead of BV, increases the fluorescence brightness of these near-infrared fluorescent proteins and maintains the photoresponsive properties of these optogenetic tools. Fission yeast is an ideal model to assess phytochrome-based tools in a cell – for example, testing the difference between BV and PCB as chromophores or establishing the efficacy of genetically encoded chromophore reconstruction – because there is neither a synthetic nor degradation pathway for BV in fission yeast.

We found that HO homologs are frequently lost in fungal species, including fission yeast, during evolution (Fig. 1E). In addition to fungi, C. elegans, one of the most popular model organisms, has very low, but not zero, BV-producing activity (Ding et al., 2017). Consistent with this fact, we could not find an HO homolog in the worm genome. The SynPCB2.1 system paves the way to utilizing iRFP in a broader range of organisms, including those that have lost an HO homolog during evolution. In addition, we observed that PCB produced by SynPCB2.1 leaks from the cells and is taken up by surrounding cells, as evidenced by iRFP fluorescence (Fig. S4). It is possible that the same events take place under ecological conditions; some organisms might exploit tetrapyrroles produced by other organisms in order to render their own phytochromes functional. In fact, Aspergillus nidulans and Neurospora crassa, both of which have lost an HO homolog from their genomes (Fig. 1E), harbor phytochrome genes that are required for chromophores (Blumenstein et al., 2005; Froehlich et al., 2005). The exchange of tetrapyrroles between living organisms might explain why the HO gene is sporadically lost in many organisms.

In this study, we have reported an iRFP imaging platform for fission yeast and a novel chromosome integration plasmid series, pSKI. The endogenous iRFP tagging system is based on a commonly used one, allowing anyone to introduce it quickly. The all-in-one plasmids carrying NLS–iRFP–NLS enable nuclear tracking without occupying green or red color fluorescence channels and automatic analysis of large-scale time-lapse images with nuclear translocation-type sensors (Regot et al., 2014). Further characterization and engineering will result in the wide use of iRFP and phytochrome-based optogenetic tools in living organisms.

The cDNAs of PcyA, HO1, Fd, and Fnr were originally derived from T. elongatus strain BP-1 as previously described (Uda et al., 2020). The nucleotide sequence of these genes and SynPCB2.1 were optimized for human codon usage (see Benchling links in Table S1). The mitochondrial targeting sequence (MTS; MSVLTPLLLRGLTGSARRLP) was derived from human cytochrome C oxidase subunit VIII. The cDNAs were subcloned into vectors through conventional ligation with Ligation high Ver.2 (TOYOBO, Osaka, Japan) or NEBuilder HiFi DNA Assembly (New England Biolabs, Ipswich, MA) according to the manufacturers' instructions. The nucleotide sequence of mNeonGreen and Turquoise2-GL were optimized for fission yeast codon usage (see Benchling links in Table S1). The pSKI vectors include Amp, colEI ori (derived from pUC119), selection marker cassettes [derived from pFA6a-3FLAG-bsd, pFA6a-kan, pAV0587 (pHis5Stul-bleMX), pMNATZA1, and pHBCN1], Padh1, Tadh1 (derived from pNATZA1), Pnmt1, Tnmt1 (derived from pREP1), and MCSs (synthesized as oligo DNA; Fasmac, Kanagawa, Japan). To construct pSKI-SynPCB2.1-Lifeact-iRFP, Pact1 (822 bp upstream of the start codon) was cloned from the fission yeast genome, and the cDNA of Lifeact was introduced by ligating annealed oligo DNAs. The cDNAs of miRFP670 and miRFP703 were obtained from pmiRFP670-N1 and pmiRFP703-N1 (Addgene plasmids #79987 and #79988, deposited by Vladislav Verkusha), and subcloned to obtain pMNATZA1-miRFP670 and pMNATZA1-miRFP703, respectively. pMNATZA1-iRFP-mNeonGreen was generated by inserting the cDNA of iRFP into the upstream of the mNeonGreen gene. To construct pCold-TEV-linker-iRFP and pCold-TEV-linker-iRFP-mNeonGreen, the cDNA of iRFP and iRFP–mNeonGreen were subcloned into pCold-TEV through conventional ligation, respectively. pCold-TEV plasmid was gifted from Dr Koichi Kato (Exploratory Research Center on Life and Living Systems, Aichi, Japan). All plasmids used in this study are listed in Table S1 with Benchling links, which include the sequences and plasmid maps.

Biliverdin hydrochloride was purchased from Sigma-Aldrich (30891-50MG), dissolved in DMSO (25 mM stock solution), and stored at −30°C. PCB was purchased from Santa Cruz Biotechnology (sc-396921), dissolved in DMSO (5 mM stock solution), and stored at −30°C. Of note, PCB and BV are insoluble at 625 µM in phosphate-buffered saline (PBS) solution and fission yeast culture medium, and are insoluble at 25 µM in mammalian cell culture medium, because insoluble PCB or BV debris is observed.

All fission yeast strains established and used in this study are listed in Table S2 with their origins. The growth medium, sporulation medium, and other techniques for fission yeast were based on the protocol described previously (Moreno et al., 1991), unless otherwise noted. The transformation protocol was modified from that of Suga and Hatakeyama (2005). Genome integration by pSKI was confirmed by colony PCR using KOD One (TOYOBO) and the primers listed in Table S3. For fluorescence microscope imaging, the fission yeast cells were concentrated by centrifugation at 860 g mounted on a slide glass and sealed by a cover glass (Matsunami).

All budding yeast strains established and used in this study are listed in Table S4 with their origins. Wild-type budding yeast strain BY4741 was a gift from Dr Yoshiaki Kamada (National Institute for Basic Biology, Aichi, Japan). Budding yeast strains were grown in yeast peptone dextrose adenine (YPDA) medium according to standard Cold Spring Harbor Protocols (Kaiser et al., 1994). Budding yeast strains were transformed with DNA fragments using a LiOAc-based approach. For fluorescence microscope imaging, budding yeast cells were concentrated by centrifugation at 860 g, mounted on a slide glass, and sealed by a cover glass (Matsunami).

HeLa cells were the kind gift of Michiyuki Matsuda (Kyoto University, Japan). BLVRA KO HeLa cells have been established previously (Uda et al., 2017). HeLa cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM) high glucose (Nacalai Tesque) supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich) at 37°C in 5% CO₂. For live-cell imaging, HeLa cells were plated on CELLview cell culture dishes (glass bottom, 35 mm diameter, four compartments; Greiner Bio-One). One day after seeding, transfection was performed with 293fectin transfection reagent (Thermo Fisher Scientific). Two days after transfection, cells were imaged using fluorescence microscopes. BV or PCB was added into the DMEM medium containing 10% FBS and cultured for 3 h at 37°C in 5% CO₂. For the experiment with serum-free medium, the medium was replaced with FluoroBrite DMEM (Thermo Fisher Scientific) supplemented with GlutaMax (Thermo Fisher Scientific) and 0.1% bovine serum albumin (BSA, Nacalai Tesque) just after the transfection.

To establish HO1 KO and HO1/BLVRA double KO cell lines, the parental HeLa cells and BLVRA KO cells were transfected with pX459-HO1, which targets the third exon of HO1, using 293fectin. Two days after transfection, the cells were selected by treatment with 1.0 µg ml⁻¹ puromycin for 48 h, followed by single-cell cloning. The genomic DNA of each single cell clone was extracted using Quick Extract DNA Extraction Solution (AR Brown Life Science). The exon 3-containing sequence was amplified using KOD One (TOYOBO; see Table S3 for the primer information), followed by direct sequencing (Fasmac) or Zero Blunt TOPO cloning (Thermo Fisher Scientific) for sequencing each allele. Cell lines having frame-shift mutations in both alleles were selected and used for experiments.

Fission yeast cells were precultured at 32°C up to the optical density at 600 nm (OD600) of 1.0, followed by 1:20 dilution. A Compact Rocking Incubator Biophotorecorder TVS062CA (Advantec, Japan) was used for culture growth (32°C, 70 rpm) and OD660 measurement. Growth curves were fitted using the logistic function [x=K/(1+(K/x₀−1)e^−rt)], and doubling time (ln2/r) was calculated using Python 3 (https://www.python.org/) and Scipy (https://scipy.org/).

For the purification of His-tag-fused iRFP, pCold-TEV-linker-iRFP was transformed into BL21(DE3) pLysS E. coli (Promega, L1195) and selected on LB plates containing 0.1 mg/ml ampicillin at 37°C overnight. A single colony was picked and used to inoculate 2.5 ml liquid LB medium supplemented with 0.1 mg/ml ampicillin and 30 µg/ml chloramphenicol at 37°C overnight. The preculture was further inoculated into 250 ml liquid LB medium (1:100) containing ampicillin and chloramphenicol. The culture was shaken at 37°C for 2–4 h until the OD600 reached 0.6–1.0. The culture was then cooled to 18°C, and 0.25 mM isopropyl β-D-1-thiogalactopyranoside (IPTG; Wako, 094-05144) was added to induce the expression of His-fused protein. After overnight incubation at 18°C, cells were collected and suspended in PBS (Takara, T900) containing 20 mM imidazole (Nacalai Tesque, 19004-22). Suspended cells were lysed by sonication (VP-300N; TAITEC), followed by centrifugation at 215,000 g for 15 min to collect the supernatant. The supernatant was mixed with 250 µl Ni-NTA sepharose (Qiagen, 1018244), and incubated at 4°C for 2 h. Protein-bound beads were washed with PBS containing 20 mM imidazole, and proteins were eluted by the addition of 300 mM imidazole in PBS. Eluted fractions were checked by SDS–PAGE with a protein molecular weight marker, Precision Plus Protein™ All Blue Standards (Bio-Rad, #1610373), followed by Coomassie Brilliant Blue (CBB) staining (BIOCRAFT, CBB-250) and detection using an Odyssey CLx system (LI-COR). Protein-containing fractions were dialyzed using a Slide-A-Lyzer Dialysis Cassette 3500 MWCO (Thermo Scientific, 66110) to remove the imidazole. To concentrate the recombinant protein, an Amicon ultra 3 K 500 µl (Millipore, UFC500308) was used. To measure the protein concentration, Pierce BCA Protein Assay Kit (Thermo Scientific, 23227) was used. Purified His–iRFP in PBS solution was mixed with an excess amount (1:5 molar ratio) of BV or PCB dissolved in DMSO, followed by size exclusion chromatography with NAP-5 columns (Cytiva, 17085301) to remove free BV or PCB.

For the purification of His–iRFP–mNeonGreen used in FCS analysis, pCold-TEV-linker-iRFP-mNeonGreen was transformed into BL21(DE3) pLysS E. coli. The cells were selected and precultured as described above. The preculture was further inoculated into 1 l liquid LB medium (1:100) containing ampicillin and chloramphenicol. The culture was shaken at 37°C for 2–4 h until the OD600 reached 0.6. The culture was rapidly cooled to 15°C and incubated for 30 min. Then, a final concentration of 1 mM IPTG was added to induce the expression of the protein. After overnight incubation at 15°C, cells were collected and lysed as described above. The supernatant was mixed with 500 µl Ni-NTA sepharose and incubated at 4°C overnight. Protein-bound beads were washed with PBS containing 20 mM imidazole, and proteins were eluted by the addition of 50 mM imidazole in PBS. Eluted fractions were checked for mNeonGreen absorption by eye. The protein was concentrated using an Amicon ultra 3 K 500 µl, followed by dialysis. Protein concentration was determined by CBB staining with a reference albumin standard (Thermo Fisher Scientific, #23209).

The absorption of BV (100 µM), PCB (100 µM), and His–iRFP (12 µM) bound to chromophore was measured using a P330 nanophotometer (IMPLEN) with a 10 mm quartz glass cuvette (TOSOH, T-29M UV10). The absorption spectrum was measured in a wavelength range of 200–950 nm. For measurements of absolute fluorescence quantum yield, BV- or PCB-bound His–iRFP (1 µM) in PBS was subjected to analysis with a Quantaurus-QY C11347-01 system (Hamamatsu Photonics). The excitation wavelength was 640 nm. For the measurements of excitation and emission spectra, BV- or PCB-bound His–iRFP (12 µM) was subjected to analysis with an F-4500 fluorescence spectrophotometer (Hitachi). The protein solution was excited in a wavelength range of 500–720 nm, and fluorescence at 730 nm was detected to measure the excitation spectrum. To measure the emission spectrum, the protein solution was excited at 640 nm, and fluorescence was detected in a wavelength range of 660–800 nm.

Protein extraction was based on previously reported methods (Koch et al., 2012). In brief, fission yeast cells were cultured at 32°C, and ∼2.5–5.0×10⁷ cells were collected by centrifugation. If necessary, the culture medium was incubated with a final concentration of 125 µM BV or PCB for 1.5 h at room temperature before collection. For cell lysis, the collected cells were suspended in 1 ml of ice-cold 20% trichloroacetic acid (TCA), and then the cells were washed with 1 ml of 1 M Tris-HCl (pH 7.5). Afterward, the cells were suspended in 200 μl of 2× SDS sample buffer (25 mM Tris-HCl, pH 6.8, 24% glycerol, 4% SDS, 0.008% Bromophenol Blue and 10% 2-mercaptoethanol) and incubated at 95°C for 10 min. Subsequently, the samples were transferred to a tube containing 7.0 mm zirconia beads (bio medical science), disrupted using a Cell Destroyer PS2000 (bio medical science) and then incubated at 95°C for 10 min, followed by centrifugation (16,000 g, 10 min, 4°C). The supernatants were loaded on 5–20% polyacrylamide gels (Nacalai Tesque, 13063-74). Proteins were separated by SDS–PAGE and transferred to polyvinylidene difluoride membranes (Merck Millipore, IPFL00010). After blocking with Intercept TBS Blocking Buffer (LI-COR, 927-60001), the membranes were probed with primary antibodies against T7-tag (1:1000, rabbit; Cell Signaling, 13246) and α-tubulin (1:1000, mouse; Sigma-Aldrich, T5168), followed by IRDye800CW donkey anti-mouse secondary antibody (1:5000; LI-COR, 926-32212) and IRDye680RD goat anti-rabbit secondary antibody (dilution 1:5000, LI-COR, 926-68071). Proteins were detected using an Odyssey infrared scanner (LI-COR).

Zinc blot analysis was based on previously reported methods (Uda et al., 2017). His–iRFP–mNeonGreen protein solution (5.1 µM) was mixed with various concentrations of BV or PCB and incubated for 30 min at room temperature. Then, the equivalent amount of 2× SDS sample buffer was added to the solution and incubated at 95°C for 5 min, followed by SDS–PAGE. The gel was incubated in the buffer containing 150 mM zinc acetate (Wako, 268-01882) and 150 mM Tris-HCl (pH 6.8) for 3 h at room temperature. BV or PCB fluorescence was detected using an Odyssey infrared scanner (LI-COR) with 680 nm excitation.

The lambda-scan function of the Leica SP8 Falcon confocal microscope system was used for measurement of the fluorescence emission spectrum. The excitation wavelength was fixed at 633 nm, and the 20 nm emission window was slid in 3 nm increments from 650 nm to 768 nm. Each emission spectrum was normalized by the peak emission value.

Cells were imaged with an IX83 inverted microscope (Olympus) equipped with an sCMOS camera (ORCA-Fusion BT; Hamamatsu Photonics), an oil objective lens (UPLXAPO 100×, NA=1.45, WD=0.13 mm; or UPLXAPO 60×, NA=1.42, WD=0.15 mm; Olympus) and a spinning disk confocal unit (CSU-W1; Yokogawa Electric Corporation). The excitation laser and fluorescence filter settings were as follows: excitation laser, 488 nm and 640 nm for mNeonGreen (or EGFP) and iRFP (or miRFP670, miRFP703), respectively; excitation dichroic mirror, DM405/488/561/640; emission filters, 525/50 for mNeonGreen or EGFP, and 685/40 for iRFP, miRFP670 or miRFP703 (Yokogawa Electric). For the multiplexed five color imaging, cells were imaged with Leica SP8 Falcon (Leica) equipped with an oil objective lens (HCPL APO CS2 100×/1.40 OIL). The excitation laser and fluorescence detector settings were as follows: excitation laser, 405 nm, 470 nm, 488 nm, 560 nm and 633 nm for mTagBFP2, Turquoise2-GL, mNeonGreen, mCherry and iRFP, respectively; detector bandwidth, 420–450 nm, 480–500 nm, 500–550 nm, 580–650 nm and 680–780 nm for mTagBFP2, Turquoise2-GL, mNeonGreen, mCherry and iRFP, respectively. Images were obtained with 10 z-slices of 0.5 µm intervals. Images were subjected to deconvolution using Lightning (Leica).

All fluorescence imaging data were analyzed and quantified using Fiji (Image J; https://fiji.sc/). The background was subtracted using the rolling-ball method. Some images were obtained with 10–30 z-slices of 0.2 µm intervals and shown as 2D images by the maximal intensity projection, as noted in each figure legend. For the quantification of signal intensity in fission yeast and in a part of budding yeast, appropriate regions of interest (ROIs) were manually selected, and mean intensities in ROIs were measured. For the quantification of signal intensities in budding yeast or HeLa cells, reference mNeonGreen or EGFP signal was used for the segmentation by Stardist (Schmidt et al., 2018; Weigert et al., 2020), followed by the measurement of iRFP signal. Segmented ROIs were further cleaned-up by size filters in the case of HeLa cells. To measure the Hta2–iRFP signals, iRFP signals were directly segmented using Stardist. Data visualization and graph creation were performed using Python 3.10 with Numpy 1.21.3, Pandas 1.3.4, Matplotlib 3.4.3 and Seaborn 0.11.2 modules.

Time-series data of fluorescence fluctuation were obtained using a Leica SP8 Falcon confocal microscope equipped with an objective lens, HC PL APO 63×/1.20 W motCORR CS2, and analyzed on Leica software essentially as described previously (Komatsubara et al., 2019; Sadaie et al., 2014).

For FCS in vivo, fission yeast cells expressing iRFP–mNeonGreen fusion protein, whose molecular ratio of iRFP to mNeonGreen was 1:1, were measured as follows: excitation wavelength, 488 nm (mNeonGreen) and 640 nm (iRFP); emission window, 500–620 nm (mNeonGreen) and 680–768 nm (iRFP). The structural parameter and effective confocal volume were calibrated using 500 nM Rhodamine 6G (TCI, R0039) in double-distilled water (DDW) based on the result that the diffusion constant of Rhodamine 6G in DDW is 414 µm² s⁻¹ at room temperature (Müller et al., 2008). The Rhodamine 6G solution was measured with 561 nm excitation and the emission from 580 nm to 700 nm. Structural parameter and the effective confocal volume were estimated as 3.70 and 0.616 fl, respectively. Note that iRFP excitation laser power was increased when iRFP–BV was measured, due to its dim fluorescence compared to iRFP–PCB.

For FCS in vitro, 5.1 µM His–iRFP–mNeonGreen protein solution, or Rhodamine 6G for FCS calibration, were mounted on 35 mm glass-base dishes (IWAKI) and mixed with fluorescent beads (Invitrogen, I14785). Structural parameter and the effective confocal volume were estimated as 8.57 and 1.63 fl, respectively. His–iRFP–mNeonGreen was measured with 488 nm excitation and the emission from 500 nm to 620 nm, and then was measured with 640 nm excitation and the emission from 680 to 795 nm with BV or PCB added to a final concentration of 160 µM, which is the saturated stoichiometry in the zinc blot analysis. Since the Förster resonance energy transfer (FRET) from mNeonGreen to iRFP was considered, firstly only the mNeonGreen fluorescence fluctuation was measured before chromophore addition to the protein solution. Then, BV or PCB was added. For each chromophore, two independent measurements were performed, three times for each, for a total of six measurements. Note that iRFP excitation laser power was again increased when iRFP–BV was measured, due to its dim fluorescence compared to iRFP–PCB.

The time-series data of fluorescence fluctuations were obtained for 30 s, corrected by the photobleach correction algorithm of the Leica FCS analysis software, and subjected to the calculation of the auto-correlation function G(τ) of the Leica FCS analysis software. The calculated auto-correlation functions were fitted with the equation concerning a single-component normal diffusion and triplet model on Leica FCS analysis software. The reciprocal of G (τ=0), which is the amplitude of the correlation function, is inversely correlated with the number of fluorescent molecules (N) in the confocal volume as N=1/G (τ=0). To estimate the fraction of holo-iRFP in all iRFPs, the number of fluorescent iRFP molecules was divided by that of fluorescent mNeonGreen molecules, assuming that all mNeonGreen formed chromophores.

We searched for HO-like sequences in representative fungal species using BLASTp (for details, see Table S5). We adopted human HO1 (Uniprot P09601) and S. cerevisiae HMX1 (Uniprot P32339) as the queries (e-value<1×10⁻⁵). The phylogenetic relationship is based on recent studies using multiple genes (Li et al., 2021; Nguyen et al., 2017). We have manually drawn the evolutionary relationship among representative species (Nguyen et al., 2017) based on a recent genome-scale phylogeny (Li et al., 2021), which is consistent with the current consensus view of the fungal tree of life (James et al., 2020). Since the results suggested sequence divergence among HO1 homologs, we also used HO-like proteins of Laccaria bicolor and Saitoella complicata obtained from the BLASTp hits, although no additional sequence was found. Note that the absence in A. nidulans and the existence in C. albicans are consistent with previous studies (Blumenstein et al., 2005; Pendrak et al., 2004). Concerning C. elegans, we searched for an HO-like sequence by the BLASTp interface provided on the WormBase website (http://www.wormbase.org; release WS280, date 20-Dec-2020, database version WS279). We used the same protein queries – human HO1 and S. cerevisiae HMX1 – although we obtained no hits (e-value<1×10⁻²).

We thank all members of the Aoki Laboratory for their helpful discussions and assistance. The pCold-TEV plasmid was a kind gift of Dr Koichi Kato (ExCELLS). We thank Dr Takuya Norizuki for his critical reading and comments. Some fission yeast strains were provided by the National Bio-Resource Project (NBRP), Japan. Budding yeast wild-type strains and plasmids were a kind gift from Dr Yoshiaki Kamada. We thank the Functional Genomics Facility of the NIBB Core Research Facilities for their technical support with fluorescence spectrometry.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.259315.