Photoconvertible fluorophores are a valuable tool for studying dynamic cellular processes, such as protein turnover and mobility, and for tracking cell lineages (Nemet et al., 2015). Unlike conventional fluorophores such as GFP, photoconvertible fluorophores can be irreversibly converted from one colour state to another upon illumination with a specific wavelength (Wiedenmann et al., 2004; Lešková et al., 2019). Multiple photoconvertible fluorophores have been developed over the past decades, including EosFP, Dendra, Kaede, and KikGR (Wiedenmann et al., 2004; Tsutsui et al., 2005; Gurskaya et al., 2006; Habuchi et al., 2008). As photoconvertible fluorophores can vary in conversion rate, convertible fraction, photostability, and brightness, comparing different variants can help us identify the most suitable fluorophore for a given application.

In the nematode Caenorhabditis elegans , Dendra2 and mKikGR have been used successfully (Hamer et al., 2010; So et al., 2024). Both are green-to-red photoconvertible fluorophores that convert upon illumination with UV (100-400 nm) or violet light (405 nm) (Tsutsui et al., 2005; Gurskaya et al., 2006; Habuchi et al., 2008). Previous data indicate that mKikGR has a high photoconversion rate and improved photostability over other photoconvertible proteins, which helps to reduce negative effects from phototoxicity (Habuchi et al., 2008; Turkowyd et al., 2017). However, in C. elegans a direct comparison between the mKikGR and Dendra2 photoconvertible fluorophores is lacking. Here, we aimed to directly compare the properties of these two fluorophores in vivo to determine their suitability for measuring protein turnover.

To determine the in vivo behaviour of Dendra2 and mKikGR in C. elegans , we generated endogenous CRISPR/Cas9-mediated fusions to NRFL-1 , an abundant apical adapter protein that links the Ezrin/Radixin/Moesin ortholog ERM-1 to various other proteins ( Figure 1A, Sepers et al., 2022). High expression levels and low turnover rates make NRFL-1 an ideal candidate to measure and compare photoconversion rates, convertible fraction, photostability, and brightness.

We first compared the kinetics of green-to-red photoconversion by exposing L2 larvae to repeated 1 second pulses of a 405 nm laser, interspersed with imaging of green and red fluorescence for 100 ms each. mKikGR converted from green to red faster than Dendra2 ( Figure 1B, C). After two 1 second pulses of 405 nm laser exposure, green mKikGR levels decreased to less than 5% of initial levels, whereas Dendra2 required at least 8 consecutive 1 second pulses of 405 nm laser exposure to reach a comparable reduction ( Figure 1B, C). The maximum fraction converted was significantly higher for mKikGR (96.3% for Dendra2 vs. 97.5% for mKikGR, P = 0.0012), although the absolute difference was modest ( Figure 1D ).

Next, we compared the fluorescence intensity of NRFL-1 tagged with Dendra2 or mKikGR. We measured a higher fluorescence intensity for Dendra2, particularly in the green state, where its signal was approximately 3.5-fold higher than that of mKikGR ( Figure 1E ). This higher brightness may make Dendra2 advantageous for visualising low-abundance proteins. However, we cannot exclude the possibility that the two fusion proteins are expressed at different levels.

To test the photostability of Dendra2 and mKikGR, we performed an initial photoconversion (2 and 5 seconds of 405 nm light for NRFL-1 ::mKikGR and NRFL-1 ::Dendra2 respectively) followed by imaging of the red fluorescent channel at 5 second intervals (561 nm, 100 ms exposures). Additionally, we tested the effect of exposing animals to either 100 ms pulses of 488 nm (blue) light (mimicking imaging of a green fluorophore) or 1 second pulses of 405 nm light in between the image acquisition ( Figure 1F ). In all conditions, we measured remaining red fluorescence intensity after 3 seconds total exposure time to 561 nm (green) light.

With post-conversion exposure to 561 nm light only, 88% of the red Dendra2 signal remained after 3 second total exposure, compared with only 53% for mKikGR. Interestingly, the additional exposures to 405 nm or 488 nm strongly affected the results. Exposure to 405 nm light caused a decrease in average red Dendra2 intensity from 0.88 to 0.07 a.u. ( Figure 1G ). Thus, Dendra2 photostability is dependent on the illumination wavelength, and reduced by exposure to 405 nm light. mKikGR was much less affected by exposure to 405 nm light. Surprisingly, however, mKikGR was more photostable when exposed to 488 nm pulses in addition to the 561 nm pulses ( Figure 1G ). One possibility is that red mKikGR enters a dark state upon exposure to 561 nm light, which is reversed by exposure to 488 nm light. This hypothesis is based on the observation that red EosFP switches to a dark state upon 561 nm light illumination, which can be reversed with either 405 or 488 nm light exposure (Nienhaus & Nienhaus, 2021; Wulffelé et al., 2025).

The presence of dark states may also explain a second observation we made during our analyses. We found that illumination with a 488 nm laser to focus the microscope resulted in a rapid decline in green signal intensity for both Dendra2 and mKikGR. This decrease might be attributable to conversion to reversible dark green states as opposed to bleaching, based on in-depth analyses of conformational states in the mechanistically related EosFP family and Dendra2 (De Zitter et al., 2019; De Zitter et al., 2020; Tao et al., 2020; Nienhaus & Nienhaus, 2021; Wulffelé et al., 2025). To test this, we exposed NRFL-1 ::Dendra2 and NRFL-1 ::mKikGR animals to a brief (100 ms) pulse of 405 nm light following an initial period of exposure to 488 nm light. If the loss of green intensity were due to bleaching, the 405 nm pulse would be expected to further decrease green signal due to green-to-red conversion. If, however, the loss of green intensity were due to conversion to a dark state, the 405 nm pulse would be expected to shift a portion of the protein back to the green state, in addition to the green-to-red conversion, potentially resulting in a net increase in green signal. We observed a partial recovery of green signal, consistent with the hypothesis that the decrease in green signal upon 488 nm light exposure is due to conversion to a dark conformational state ( Figure 1H ).

Finally, we examined whether mKikGR can be used for long-term tracking of NRFL-1 protein turnover in the intestine. We exposed L2 larvae to the 405 nm LED light source of a fluorescent binocular microscope for 10 seconds and imaged at specific timepoints thereafter. Red fluorescence corresponds to the pool of NRFL-1 present at photoconversion, whereas green fluorescence indicates newly synthesised protein. Red signal decreased by ~85% after 24 h, suggesting that protein turnover of NRFL-1 takes around 24 hours ( Figure 1I ). This observation is consistent with the slow turnover rate of NRFL-1 at the apical domain previously measured by fluorescence recovery after photobleaching (FRAP) (Sepers et al., 2022) and demonstrates that mKikGR is suitable for the long-term tracking of protein turnover.

In summary, we compared the in vivo properties of Dendra2 and mKikGR in C. elegans using endogenous NRFL-1 fusions, and showed that mKikGR enables tracking of protein turnover over long periods of time. mKikGR has a higher photoconversion rate than Dendra2, but is less bright, particularly in the green state. Under imaging conditions combining a brief 405 nm exposure for photoconversion with subsequent red and green state imaging, we observed no difference in photobleaching between Dendra2 and mKikGR. Finally, we observed behaviour consistent with reversible dark green conformational states for mKikGR and Dendra2 under 488 nm illumination. This effect was stronger for mKikGR, making it more difficult to use when green fluorescence is essential to focus on the sample, or where obtaining pre-conversion images is essential. Ultimately, neither fluorophore consistently outperformed the other, and both are suitable for photoconversion studies in C. elegans .

C. elegans maintenance

C. elegans strains were cultured and maintained under standard conditions (Brenner, 1974). The animals were grown on nematode growth medium (NGM) plates seeded with the Escherichia coli strain OP50 . All experiments were performed at 20 °C.

CRISPR/Cas9 genome editing

C. elegans codon-optimised Dendra2 and mKikGR were designed with the use of Wormbuilder (https://wormbuilder.org/transgenebuilder/) (Frøkjær-Jensen et al., 2016), and ordered as gBlock Gene Fragments through IDT. gBlocks were cloned into pBSK by blunt ligation to yield plasmids pSRG103 and pSRG104. A 4x Gly linker was placed between the fluorescent protein and NRFL-1 . Silent mutations were introduced to prevent re-cutting after homologous recombination. The knock-in alleles were made using CRISPR/Cas9-mediated homology-directed repair in a wild-type animal ( CGC1 ). The genomic nrfl-1 sequence was obtained from WormBase (Sternberg et al., 2024). Young adult animals were injected with the repair template for either Dendra2 or mKikGR and the Cas9 (Alt-R S.p. Cas9 Nuclease V3)-guide RNA complex (Ghanta et al., 2021). Progeny were selected by fluorescence, and the knock-in sequences were PCR amplified and sequenced, using Sanger sequencing.

Microscopy

The animals were handled in daylight settings, as this did not produce any measurable level of conversion. Animals were mounted on a 5% agarose pad and immobilised with 20 mM tetramisole solution in M9 (22 mM KH ₂ PO ₄ , 42 mM Na ₂ HPO ₄ , 85 mM NaCl, 1 mM MgSO ₄ ). Spinning disc confocal imaging was performed on a Nikon Ti-U microscope, equipped with a Yokogawa CSU-X1 spinning disc and a Prime BSI Express Scientific CMOS camera (Teledyne Photometrics). We used a Nikon CFI Plan Apo λ 100x/1.45 0.13mm WD OIL objective for the Dendra2 and mKikGR comparisons, and a Nikon CFI Plan Apo λ 60x/1.40 0.13mm WD OIL objective for the 24 h protein turnover analysis. Prior to imaging, the objective was manually focused on the middle of the intestine using only backlight, without the use of the lasers. Identical settings were used for both green and red channels.

Photoconversion

Green-to-red conversion for the Dendra2 and mKikGR comparisons was done on the Nikon spinning disc confocal microscope, using the 405 nm BFP laser at a measured power of 0.605 mW. For the 24 h protein turnover experiment, animals were converted on an Axio Zoom.V16 using blue 405 nm light for 10 seconds to ensure optimal green-to-red conversion at a measured power of 86.0 mW.

Image analysis

Image analysis was conducted in FIJI. A 5 pixel-wide line was drawn along sections of the brush border of the int2-4 cells, and the average intensity was measured. All images were corrected for background fluorescence by measuring an area outside of the animal and subtracting this value from the absolute measured fluorescence value.

Statistics

All statistical analyses were performed using GraphPad Prism (10.4.1). The sample sizes, statistical test used, and p values are indicated in the Figure and legend. No statistical method was used to pre-determine sample sizes. No samples or animals were excluded from analysis. The experiments were not randomised, and the investigators were not blinded to allocation during experiments and outcome assessment. All presented graphs were made using GraphPad Prism and Adobe Illustrator. All tests were two-tailed. All data were treated as non-normally distributed, and a Mann-Whitney test was used for all comparisons.

Data availability

All data supporting the findings of this study are available within the paper. Raw microscopy image files are available from the corresponding author upon reasonable request. C. elegans strains are available upon reasonable request.