Serendipitous crystal structure of mJuniper, a rationally designed cyan fluorescent protein

We report the crystal structure of mJuniper, a rapidly maturing cyan fluorescent protein.

Abstract

mJuniper is a recently developed monomeric cyan fluorescent protein (CFP) optimized for bacterial imaging, distinguished by the fastest maturation kinetics recorded for a CFP. mJuniper shows significant promise for time-lapse microscopy, but was used here to improve the solubility of a target protein. Here, we report the high-resolution X-ray structure of mJuniper, obtained serendipitously during an attempted structural characterization of CagG, an essential component of the Helicobacter pylori Cag type IV secretion system. The structure reveals a classical β-barrel scaffold, as well as a chromophore with two rotameric states, resulting from specific radiation damage occurring in its vicinity.

1. Introduction

Fluorescent proteins (FPs) are ubiquitous tools that have revolutionized molecular biology by serving as reporters for gene expression and protein localization (Chalfie et al., 1994 ▸). The majority of engineered FPs are derivatives of the Aequorea victoria green fluorescent protein (GFP; Shimomura et al., 1962 ▸), with the demand for multi-colour imaging driving the development of spectral variants such as cyan fluorescent proteins (CFPs). CFPs are generated by replacing the chromophore-forming tyrosine at position 66 of GFP with tryptophan (Heim et al., 1994 ▸), and have been engineered into high-performance variants such as mTurquoise2 or mCerulean3 (Goedhart et al., 2012 ▸; Markwardt et al., 2011 ▸), which are suitable for FRET and multi-colour microscopy (Mastop et al., 2018 ▸). Recently, a set of Escherichia coli-optimized FPs was developed, including mJuniper, which exhibits remarkable maturation kinetics (Fraikin et al., 2025 ▸).

In this study, we employed mJuniper to assist in the structural characterization of a protein from Helicobacter pylori, a Gram-negative bacterium that colonizes the human stomach and is a major risk factor for gastric cancer. The virulence of H. pylori is partially dependent on a type IV secretion system (T4SS) encoded by the cag (cytotoxin-associated genes) pathogenicity island, which is responsible for translocating the CagA oncoprotein into host cells (Cover et al., 2020 ▸). Among the Cag components, CagG is essential for CagA translocation, yet its structural and functional role remain obscure (Merino et al., 2017 ▸). Structural characterization of CagG is hindered by the lack of reliable structural predictions; AlphaFold3 models of CagG yield low confidence scores (ipTM/pTM ≃ 0.21) with a poor predicted local distance difference test (pLDDT).

To address the solubility and expression issues encountered during CagG production, we designed a construct encoding a fusion of the protein with mJuniper, since fluorescent proteins are known to enhance the solubility and stability of challenging targets (Esposito & Chatterjee, 2006 ▸). mJuniper was derived from mChartreuse, which itself was engineered from superfolderGFP (sfGFP). As a result, the amino-acid sequence of mJuniper differs from that of sfGFP by ten residues: N39I, Y66W, S72A, I128S, D129G, F145Y, N146F, H148D, N149K and V206K. Compared with other CFPs such as mTurquoise2 or SCFP3A, mJuniper displays an enhanced photostability and maturation rate (Fraikin et al., 2025 ▸). While our primary objective was crystallization of the full-length CagG-mJuniper fusion, almost complete proteolysis of CagG occurred during crystal growth, leading to the serendipitous production of mJuniper crystals. Here, we present the crystal structure of mJuniper at a resolution of 1.6 Å and compare it with its parent sfGFP, and with Cerulean, one of its closest cyan relatives (Malo et al., 2007 ▸; Lelimousin et al., 2009 ▸).

2. Materials and methods

2.1. Macromolecule production

The gene encoding H. pylori strain 26695 CagG from GenBank (AAD07608) was synthesized with optimized codons for E. coli, amplified by PCR and inserted at the 5′-end of the gene encoding mJuniper in a pQE30 vector via Gibson Assembly cloning (Gibson et al., 2009 ▸), enabling the production of a 6×His-CagG-mJuniper fusion protein. The primers used were: Insert-F (5′-ACCATCACCATCACCATCACGACCAAAATACGGACATCAAAGA-3′), Insert-R (5′-CCTTATGACATACTTCCACCGTAGCCCAGTATCGGCGGTA-3′), Vector-R (5′-TTGATGTCCGTATTTTGGTCGTGATGGTGATGGTGATGGT-3′) and Vector-F (5′-TACCGCCGATACTGGGCTACGGTGGAAGTATGTCTAAAGGT-3′).

The construct was introduced into the E. coli T7 Express strain for protein production. Cultures were grown in lysogeny broth (LB) supplemented with ampicillin (100 µg ml⁻¹) until the OD₆₀₀ reached 0.6, at which point protein production was induced with 0.5 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) at 20°C for 16 h. The cells were harvested and resuspended in buffer A [20 mM Tris pH 7.4, 300 mM NaCl, 5%(v/v) glycerol] supplemented with 1%(v/v) antiprotease cocktail (0.1 mg l⁻¹ chymostatin, 0.1 g l⁻¹ pepstatin, 0.1 g l⁻¹ leupeptin, 0.1 g l⁻¹ antipain, 0.4 g l⁻¹ aprotinin). The cells were lysed by a cycle of freeze–thaw followed by sonication, and the lysate was clarified by centrifugation at 16 000g for 20 min at 4°C. The soluble fraction was loaded onto Protino Ni-TED resin (Macherey-Nagel). The resin was washed with 20 column volumes of buffer A and elution was performed using six column volumes of buffer A supplemented with 500 mM imidazole. The elution fraction was concentrated to 5 mg ml⁻¹ using an Amicon concentrator (10 kDa cutoff) and was then injected onto a Superdex S200 10/300 GL column (Cytiva) equilibrated with buffer A. Elution fractions corresponding to CagG-mJuniper were pooled and concentrated to 20 mg ml⁻¹. The yield for the purification of CagG-mJuniper is around 1.9 mg per litre of culture. Fluorescence spectra were recorded on a Spark plate reader (Tecan) using a 5 nm bandpass monochromator.

2.2. Crystallization

Crystallization screening was performed using the sitting-drop vapour-diffusion method with a Mosquito nanodispenser robot (SPT Labtech) following our previously described procedure (Dugelay et al., 2024 ▸). Drops consisting of 150 nl protein solution (20 mg ml⁻¹) and 150 nl reservoir solution were equilibrated against reservoir solution at 19°C. Crystals appeared within 48 h in Natrix (Hampton Research) condition C3: 0.2 M ammonium acetate, 0.01 M magnesium acetate, 0.05 M sodium cacodylate pH 6.5, 30%(w/v) PEG 8000.

2.3. Data collection and processing

For data acquisition, crystals were cryoprotected by supplementing the crystallization solution with 10%(v/v) glycerol before flash-cooling in liquid nitrogen. Diffraction data were collected on beamline ID30A-3 at the European Synchotron Research Facility, Grenoble, France (Von Stetten et al., 2020 ▸). Data reduction was performed using XDS and the grenades_parallelproc pipeline (Kabsch, 2010 ▸; Monaco et al., 2013 ▸). The crystals diffracted to a resolution of 1.6 Å and belonged to space group P2₁2₁2₁. The structure was solved by molecular replacement using an AlphaFold3 model (Abramson et al., 2024 ▸) of mJuniper as a template in Phaser (McCoy, 2007 ▸). Refinement was carried out using REFMAC5 in the CCP4 program suite (Agirre et al., 2023 ▸) and phenix.refine in Phenix (Afonine et al., 2012 ▸). Model quality was assessed using MolProbity (Williams et al., 2018 ▸) and the Phenix validation tools (Liebschner et al., 2019 ▸). Figures were generated using PyMOL (https://www.pymol.org). Data-collection and refinement statistics are given in Table 1 ▸. The model and structure factors were deposited in the Protein Data Bank (PDB entry 29zn).

Table 1. X-ray diffraction data-collection and refinement statistics for mJuniper.

Values in parentheses are for the highest resolution shell.

Wavelength	0.9677
Resolution range	41.54–1.623 (1.68–1.62)
Space group	P212121
a, b, c (Å)	51.588, 61.034, 70.049
α, β, γ (°)	90, 90, 90
Total reflections	258729 (26051)
Unique reflections	28600 (1359)
Multiplicity	9.0 (9.4)
Completeness (%)	99.66 (97.1)
Mean I/σ(I)	7.7 (1.0)
Wilson B factor (Å2)	22.83
R merge	0.157 (1.763)
R meas	0.168 (1.935)
R p.i.m.	0.079 (0.889)
CC1/2	0.991 (0.587)
CC*	0.996 (0.802)
No. of reflections, working set	27199 (2673)
No. of reflections, test set	1341 (111)
R work	0.165 (0.253)
R free	0.192 (0.284)
No. of non-H atoms
Total	2055
Macromolecules	1792
Ligands	48
Solvent	215
Protein residues	225
R.m.s.d., bond lengths (Å)	0.008
R.m.s.d., angles (°)	0.95
Ramachandran favoured (%)	98.19
Ramachandran allowed (%)	1.81
Ramachandran outliers (%)	0.00
Rotamer outliers (%)	0.51
Clashscore	1.37
Average B factor (Å2)
Overall	28.75
Macromolecules	28.04
Ligands	18.87
Solvent	36.92

3. Results and discussion

3.1. Structure determination of mJuniper

While no soluble protein was obtained when expressing cagG alone, we were able to purify the CagG-mJuniper fusion in large amounts. Crystallization trials were thus performed with pure CagG-mJuniper protein at a concentration of 20 mg ml⁻¹. The protein crystallized within 48 h and diffraction data were collected from the yellow-coloured crystals at the European Synchrotron Radiation Facility (Fig. 1 ▸a; Table 1 ▸). The crystals diffracted to a resolution of 1.6 Å and belonged to space group P2₁2₁2₁. With a unit-cell volume of 220 557 ų, it was unlikely that the crystal contained the entire CagG-mJuniper fusion protein (40 kDa), since this fusion construct would result in a Matthews coefficient of 1.34 ų Da⁻¹ and a solvent content of 8% for one molecule per asymmetric unit. We thus ran SDS–PAGE with the resolubilized crystals as the sample and observed that these contained only a protein fragment of approximately 27 kDa, suggesting that the fusion protein had undergone extensive proteolysis (Fig. 1 ▸b). This degradation is consistent with the predicted biochemical properties of CagG, with sequence-based analysis yielding an instability index of 46.58, classifying it as an unstable protein, likely rendering the fusion susceptible to in situ proteolysis (Guruprasad et al., 1990 ▸). Because the crystals displayed a bright yellow colour (Fig. 1 ▸a), we hypothesized that they contained only the mJuniper moiety (26.7 kDa), in agreement with the apparent molecular weight of the protein fragment observed by SDS–PAGE. Hence, considering the sole presence of mJuniper in the crystal, the obtained Matthews coefficient was 2.07 ų Da⁻¹ with 40% solvent content and one molecule per asymmetric unit. Of note, the absorbance and emission spectra of mJuniper alone as purified in Fraikin et al. (2025 ▸) and CagG-mJuniper were identical (Fig. 1 ▸c).

Figure 1.

Structure of mJuniper. (a) Representative photograph of yellow-coloured mJuniper crystals. (b) SDS–PAGE analysis of the purified CagG-mJuniper protein sample utilized for crystallogenesis (lane 1) and solubilized mJuniper crystals (lane 2). (c) Spectral properties of mJuniper and of the fusion protein CagG-mJuniper. (d) Cartoon representation of the mJuniper structure exhibiting the characteristic 11-stranded β-barrel fold with the central chromophore-containing α-helix. Rainbow colouring from blue to red indicates the N- to C-terminal position of the residues in the model. (e) Superimposition of mJuniper (grey), Cerulean (orange) and sfGFP (green; PDB entry 2b3p) structures displayed in cartoon representation.

Structure determination using molecular replacement confirmed that the crystal asymmetric unit indeed contained a monomer of mJuniper with no extra density for CagG. The final model corresponds to residues 3–230 of the mJuniper sequence (Fraikin et al., 2025 ▸). mJuniper adopts the classical β-barrel fold characteristic of the GFP family, consisting of an 11-stranded β-barrel with a central α-helix containing the chromophore (Fig. 1 ▸c). Superposition of mJuniper with the crystal structure of its parent scaffold, sfGFP (PDB entry 2b3p; Pédelacq et al., 2006 ▸), results in a root-mean-square deviation (r.m.s.d.) of 0.424 Å (218 pruned atom pairs), while superimpositions with Cerulean (PDB entry 2wso; Lelimousin et al., 2009 ▸) gives an r.m.s.d. of 0.228 (222 pruned atoms) (Fig. 1 ▸d). While the Cerulean and mJuniper structures are almost indistinguishable, differences between mJuniper and sfGFP reside predominantly in the conformation of the β9–β10 loop and the elongation of β7 in mJuniper and Cerulean (Fig. 1 ▸d).

3.2. Structural analysis of the mJuniper chromophore

mJuniper differs from its parent scaffold sfGFP by ten amino-acid substitutions (Fraikin et al., 2025 ▸). One of these mutations, V206K, was originally introduced to prevent oligomerization in mChartreuse. Four mutations were introduced to enhance brightness (N39I, I128S, D129G and N149K) and one to improve photostability (F145Y). The Y66W mutation was introduced to shift the fluorescence emission from the green region (510 nm) to the cyan region (475 nm). The remaining three mutations, S72A, N146F and H148D, were introduced since they were found in mTurquoise2 (Goedhart et al., 2012 ▸). Together, these spectral mutations result in a chromophore pocket nearly identical to that of Cerulean except for position 146, which is an isoleucine in Cerulean and a phenylalanine in mJuniper (Lelimousin et al., 2009 ▸; Malo et al., 2007 ▸; Fig. 2 ▸).

Figure 2.

Specific radiation damage in the vicinity of the mJuniper chromophore. Detailed comparison of the local chromophore environments in the mJuniper and Cerulean (PDB entry 2wso) structures. Residues involved are displayed as sticks and water molecules as red spheres, and hydrogen bonds are represented by yellow and green dashed lines. (a) Two rotated views are shown for the mJuniper chromophore Thr65-Trp66-Gly67 [coloured in teal in (a) and (c)] and for the equivalent chromophore of Cerulean [coloured in orange in (b)]. Two rotamers of Thr65 are observed in mJuniper. In the ‘water facing’ conformation (a), the Thr65 side-chain hydroxyl points towards a water molecule, as observed in the Cerulean structure (b). (c) In the ‘Val facing’ orientation, the Thr65 side chain is rotated and makes two additional hydrogen bonds to Glu222 that are displayed as green dashed lines. (d) Close-up view of the mJuniper chromophore with the 2mF_obs(i) − DF_calc(i), α_calc(i) electron-density map displayed as a mesh at 1.2σ [or 1.5σ in (e)] and coloured blue and the F_obs(i) − F_obs(1), α_calc(1) difference map as a mesh and coloured red (−3σ) or green (+3σ).

Structural comparison of mJuniper with Cerulean (Lelimousin et al., 2009 ▸; Malo et al., 2007 ▸) reveals that while the global scaffold is highly conserved, the chromophore pocket of mJuniper exhibits conformational heterogeneity (Fig. 2 ▸). The chromophore is composed of Thr65-Trp66-Gly67 and adopts two rotameric states corresponding to two orientations of the Thr65 side chain. In one rotameric state (occupancy of 0.52), the hydroxyl group of Thr65 forms a hydrogen-bonding network with internal water molecules and with Glu222 (named the ‘water facing’ conformation; Fig. 2 ▸a). Furthermore, the chromophore is directly coordinated by Val61, Thr62, Gln94, Arg96, Ser205, Glu222 and two ordered water molecules, while being structurally constrained by the surrounding residues Gln69, Phe146, Val150, Phe165 and Thr203. This conformation of the chromophore is identical to that found in Cerulean (Fig. 2 ▸b). In the second rotamer (occupancy of 0.48), the Thr65 hydroxyl group is oriented towards Val61 (named the ‘Val facing’ conformation; Fig. 2 ▸c). In this Val facing conformation, the hydroxyl group of the chromophore is pulled away from Glu222 and the water network to instead interact with the backbone carbonyl of Val61.

The presence of these rotameric states could be the result of the acidic crystallization conditions (pH 6.5), which would lead to an equilibrium between a nonprotonated, fluorescent form and a protonated, nonfluorescent form. However, the relatively low pK_a of mJuniper (4.7) excludes this hypothesis. Previous studies have shown that residue Glu222 is particularly sensitive to radiation damage and can undergo decarboxylation in Cerulean structural studies (Gotthard et al., 2019 ▸). No significant global radiation damage was apparent, with the AIMLESS cumulative pairwise residual (R_cp) remaining approximately constant (∼0.22) throughout the dataset, with no dose-dependent increase. However, examination of the 2mF_obs(i) − DF_calc(i), α_calc(i) electron-density map and F_obs(i) − F_obs(1), α_calc(1) difference map indicates that specific radiation damage has occurred on the chromophore environment and partial decarboxylation of Glu222 has taken place (Fig. 2 ▸d; Garman & Weik, 2017 ▸). In the event of decarboxylation, the water network would be disrupted, potentially altering the orientation of Thr65. This phenomenon has been documented previously and explains the two conformations observed for Thr65 in mJuniper (Gotthard et al., 2019 ▸).

4. Conclusion

The serendipitous crystal structure of mJuniper presented here provides a first step toward understanding its characteristics. This CFP exhibits a quantum yield of 43%, a brightness of 13.76 and a maturation time t_1/2 = 1.7 min, representing a powerful tool for real-time bacterial imaging (Fraikin et al., 2025 ▸). While the structure presented here could help to decipher the properties of mJuniper, future work should attempt to determine its structure without any radiation damage and/or at a more physiological pH. Although the crystallization of the mJuniper-CagG fusion was not successful, this study shows that this CFP can be efficient at improving the expression and solubility of difficult protein targets.

Supplementary Material

PDB reference: mJuniper, 29zn

Funding Statement

This project was supported by a La Ligue Contre le Cancer regional grant (grant No. 2023 to Laurent Terradot).

Conflict of interest

The authors declare that there are no conflicts of interest.

Data availability

The crystal structure and structure factors have been deposited in the Protein Data Bank with the code 29zn.