Two types of patterned ∼25-amino-acid methionine-rich motifs are reported in reflectin sequence: N-terminal motif (RM_N) [MEPMSRM(T/S)- MDF(H/Q)GR(Y/L)(I/M)DS(M/Q)(G/D)R(I/M)VDP (R/G)] and a series of conserved reflectin motifs (RMs) [M/FD(X)₅MD(X)₅MDX_3/4] (Crookes et al., 2004) (see Figure 1A and Supplementary Figure S1). The N-terminal region is more evolutionarily conserved across species (Doryteuthis opalescens, Doryteuthis pealeii, Loligo forbesii, Sepia officinalis, Euprymna scolopes, and Octopus bimaculoides) and reflectin isoforms (23 in 27 kinds of the most known reflectins) than canonical RM_N (Izumi et al., 2010). While almost all “X” sites are populated largely by one specific residue with minor alternative residues usually represented in only one or a few reflectin motifs in the entire known library. Specifically, RfC is the shortest reflectin containing a GMXX motif and RM*. The GMXX motif is a unique region of increased overall hydrophobicity composed of a four amino acid repeat, where ‘X’ represents less conserved locations within the repeat. Asterisk-marked RM* of RfC contains substantial deviations in the sequence not observed in any other reflectin motifs (Levenson et al., 2017). Supplementary Figure S1 shows reflectins and their RM_N and RMs.

At present, studies are mostly focused on their self-assembly properties in vitro (DeMartini et al., 2015; Levenson et al., 2016; Levenson et al., 2019). The dynamic reflectin assembly properties have already inspired the development of various next-generation tunable photonic (Qin et al., 2013; Phan et al., 2015; Dennis et al., 2017) and electronic platforms and devices (Ordinario et al., 2014; Yu et al., 2014; Phan et al., 2016).

Four reflectin proteins were constructed into pEGFP-C1 vectors and transfected into HEK 293T cells to further investigate their intracellular characteristics. Compared with cells transfected by no-load pEGFP-C1, all these four kinds of reflectins tend to form protein condensates or spherical droplets in cells (see Supplementary Figure S2 for large area immunofluorescence images and Supplementary Figure S3 for cell viability tested by CCK-8 kits). The formation of proteinaceous condensates is consistent with speculation that reflectins are the potential intrinsically disordered protein to execute phase separation (Levenson et al., 2019).

More significantly, RfA1 condensates were exclusively distributed in the cytoplasm, while RfA2, RfB1, and RfC droplets were highly enriched in nuclei (see Figures 1B, E). Since amino acid composition lays the foundation of a protein structure and function, the similarities and differences among reflectins are also determined by their primary structure. The reason why they are sorted into one protein family is reflectin motifs (RMs) that they share. Meanwhile, the most significant difference among reflectins is also the number of RMs repetitions. As suggested by Morse and coworkers, these canonical reflectin motifs (RMs) could be structural or functional elements of reflectins (Levenson et al., 2016). Seeing from this point, cytoplasm-retained RfA1 contains the largest numbers of reflectin motifs, while nucleoplasm-enriched reflectins contain less. Hence, the work cloned the six RMs of RfA1 (RM_N, and RM_1,2,3,4,5, primers in Supplementary Table S1) and introduced them into cells to explore the role of conserved RMs during the protein condensation and selective localization. Results showed that all individual RMs distributed freely in both cytoplasm and nucleoplasm, with no difference compared to GFP alone expressed in cells (see Figure 1C). Therefore, the cytoplasmic enrichment of reflectins other than conserved amino acid composition should be driven by its segmented sequence structure. Transfection efficiencies were calculated and shown in Figures 1D/F.

Five pairs of primers were designed to clone the DNA sequences from RfA1 genes (see Supplementary Table S2 for primers) and gradually extend peptide sequences and restore their segmented structure. PCR products responsible for coding RL_Nto1, RL_Nto2, RL_Nto3, RL_Nto4, and RL_Nto5 are subsequently ligated to vector pEGFP-C1. The common characteristic for cells expressed with RL_Nto1, RL_Nto2, and RL_Nto3 is their enrichment in nuclei (see Figures 2A, E). It is different from the cytoplasmic-localization-preference of RfA1 or free-distribution of single RMs, but quite similar to simpler reflectins (RfA2, RfB1, and RfC). Besides, RL_Nto2 and RL_Nto3 are enriched in cytoplasm and phased out from the crowded cellular milieu (see Figures 2A, E). As sharp contrasts, longer RL_Nto4 and RL_Nto5 start to be excluded from nuclei and form condensates in the cytoplasm (see Figures 2A, E), which extremely resembles RfA1(see Figures 1B, E).

Guan et al. (2017) reported that reflectin motifs may be traced to a 24-bp transposon-like DNA fragment from symbiotic bioluminescent bacterium Vibrio fischeri in 2017. Afterward, million years of self-replication and translocation of that transposon leads to the formation of reflectin motifs and the prosperous reflectin family. Here, the repetition numbers of RMs (as basic units) accurately determine the different properties (e.g., intracellular localization) among RfA1, A2, B1, and C. Being a subordinate element and evolutionary origin, the 24-bp transposon-like DNA fragment is expected to be the root source of reflectins diversification, which makes our finding a strong clue to support Guan’s evolution hypothesis.

On the other hand, if GFP is taken as a molecular cargo, reflectin derivatives can be regarded as intelligent vehicles to transport cargoes to pre-selected destinations (cytoplasm or nucleoplasm). Based on this consideration, we tested the application potentials of RfA1 variants as synthetic biology tools. As the shortest peptide strictly targets nuclei, RL_Nto2 was used in the following studies and regarded as a guiding tag, with transfection conditions optimized by dose-dependent and time-scale preliminary assays (see Supplementary Figure S5).

RL_Nto2 is constructed into Tet-On plasmids that can be easily switched on or off by doxycycline (dox) to exhibit its application potential as a synthetic biology component (see Figure 3A). When cell confluence reaches ∼30%, transfection is conducted according to Lipofectamine 3,000 protocol. Transfection efficiency is checked by fluorescent images after 24 h. Cells are then treated with concentration-gradient dox and cultivated for another 24 h. Afterward, the nuclei-targeted expression of RL_Nto2 is observed (see Figures 3B, C and Supplementary Figure S6) by confocal microscope. The expression level of RL_Nto2 is enhanced synchronously with the dox concentration gradient (see Figure 3C). This implies the successful activation of this Tet-On system and its controllability based on the dox dose-dependent behavior.

Foreseeable, if reporter gene mCherry was replaced by other functional or therapeutical peptides, the Tet-On-Rfs system can precisely transport the molecular cargos into nuclei to amplify their biological effects. Besides, if certain components prefer to fulfill their functions in the cytoplasm, RL_Nto2 can be replaced by RL_Nto4 or RL_Nto5. At this point, the programmable RfA1 sequences provide an editable and selectable engineered toolkit to precisely transport proteins or peptide cargoes to the preselected subcellular area.

Based on Figure 2, RfA1-derived peptides occupying more than 3 RM motifs tend to stay in the cytoplasm. Otherwise, shorter peptides are inclined to be enriched in nucleoplasm. Perspectively, peptides RL_Nto2 can travel across nuclear membranes, while its dimer or analog may be resisted. Hence, CoH2/DocS domains were introduced for their stable and high-affinity interaction to verify this speculation (Li et al., 2020; Yu et al., 2020).

Recombined proteins mCherry-RL_Nto2-CoH2 and GFP-RL_Nto2 can pass through nuclear membranes and get into nuclei (see Figure 4A). Since these two proteins share the same transmembrane mechanism, they induce similar molecular responses in nuclei and co-localized to each other. Similarly, co-expression of mCherry-RL_Nto2 and GFP-RL_Nto2-DocS leads to the same phenomenon (See Figure 4B). However, when CoH2 and DocS domains are fused to mCherry-RL_Nto2 and GFP-RL_Nto2, respectively, most of the fluorescence signal is retained in the cytoplasm (see Figure 4C). The results suggest that the repetition-dependent nucleocytoplasmic localization preference of RfA1-derived peptides can be adjusted by molecular splicing. Inspired by this, the CRY2–CIB1 system is then employed to generate photoactivatable subcellular localization. mCherry-RL_Nto2-CIB1 and GFP-RL_Nto2-CRY2 without blue-light stimulus tend to be enriched in nuclei and co-localize with each other (Figure 4D). Contrastly, the green fluorescent signal of GFP-RL_Nto2-CRY2 in a fraction of cells was prevented from entering the nuclei upon blue-light irradiation (Figure 4E). In this case, blue-light induced interaction between CIB1 and CRY2 elevates the RMs repetition level because RL_Nto2 turned into dimer analog RL_Nto2-Pairs- RL_Nto2 and leads to their retention in cytoplasm (Figure 4F) .

Building blocks used for synthetic biology should better be standardized and modalized. Although the RMs in reflectins are highly canonical and conserved, there are still subtle composition differences among them (Crookes et al., 2004; Izumi et al., 2010; Levenson et al., 2017). We replaced all RM_1∼5 in RfA1 with a unified RM₁ to eliminate this subtle composition discrepancy (see Supplementary Table S3 for sequence information and Figure 5B for sketches of artificial peptides). Being similar to the original RfA1, RM_N + RM₁*5 is also observed to be highly enriched in cytoplasm (Figures 5A, E). This unification process (RM_1∼5 into RM₁) will not change proteins’ properties at least for intracellular localization preferences. Moreover, one recombinational peptide RM₁*3 + RL₂*2 is designed (see Supplementary Table S3 for sequence information; Figure 5D for the sketches of artificial peptides). It is transported and enriched in nuclei Figure 5C, which is similar to RfA1-derived analog RL_Nto2 (Supplementary Figure S6C, E). Hence, building blocks or functional components derived from reflectin amino acid sequences can be standardized without losing their intracellular localization preferences, which favors their application in synthetic biology.

The nucleotide sequences of D. (Loligo) pealeii reflectin A1 (RfA1) (Genbank: ACZ57764.1), D. (Loligo) pealeii reflectin A2 (RfA2) (Genbank: ACZ57765.1), D. (Loligo) pealeii reflectin B1 (RfB1) (Genbank: ACZ57766) and D. (Loligo) Opalescens reflectin C (Genbank: AIN36559.1) were optimized for human-cell expression. Then, they were synthesized and sequencing-identified by Sangon Biotech^® (Shanghai, China) Primers (F-GAATTCTATGAATAGATATTTGAATAGACA; R-GGATCCATACATATGATAATCATAATA ATTT) were designed to introduce EcoR I and BamH I cutting sites, so the modified RfA1 CDS could be constructed into pEGFP-C1 via a standard restriction enzyme cloning process. As for truncated RfA1 derivatives, six pairs of primers were coupled. If RMN-F and RM3-R primers were selected, then a nucleotide sequence responsible for coding of RMN-RL1-RM1-RL2-RM2-RL3-RM3 (simplified as RMNto3 in the work) was obtained after PCR. Meanwhile, 5′GCA​TGG​ACG​AGC​TGT​ACA​AG 3′ and 5′ TTATGATCAG- TTATCTAGAT 3’ were added to those F-primers and R-primers separately during primers synthesis, which enabled sequences to be ligated to pEGFP-C1 by Ready-to-Use Seamless Cloning Kit from Sangon Biotech^® (Shanghai, China).

HEK-293T cells (ATCC^®, CRL-3216TM) were cultured on plastic dishes in Dulbecco’s Modified Eagle Medium (DMEM, GibcoTM) supplemented with 10% fetal bovine serum (FBS, GibcoTM) at 37°C and under 5% CO₂. Cells were seeded at ∼33% of the confluent density for the glass bottom dishes from Cellvis (California, United States) 1 day before transfection, and grown for another 24 h. Then transfection mixtures containing Lipofectamine 3,000 (Ther-mo Scientific) and recombinant vectors were added to the medium and incubated for ∼16–∼24 h. 1 × 10⁴ cells were seeded into each hole of 96-well plates 1 day before transfection for CCK-8 tests. Then they were transfected with recombinant vectors and incubated for another 24 h. After that, 10-μL CCK-8 solutions were added into wells for a ∼2- to ∼4-h chromogenic reaction. OD450 was detected by Multiskan FC (Thermo Scientific).

Transfected HEK-293T cells grown in Cellvis plastic dishes were firstly fixed with 4% paraformaldehyde at room temperatures for 30 min and then stained with DiD (diluted in 0.5% Triton X-100 PBS) for ∼30 min after PBS rinses. Fixed cells were embedded in DAPI-Fluoromount (Beyotime, Shanghai, China) after washing off the fluorescent dye with PBS, and characterized with a Leica TCS SP8 imaging system in fluorescence imaging mode. 405 Diode laser was used for DAPI detection; 488 Argon was used for GFP detection; DPSS 561 was used for mcherry detection; HeNeB 633 was used for DiD detection. The resulting images were analyzed with ImageJ (Java 1.8.0_172/1.52b) (Schindelin et al., 2012).