Genetically Encodable Bacterial Flavin ... - ACS Publications

Dec 30, 2016 - untransfected cell lysate (Figure 1B, lane 6) or in the cells transfected ..... mg/mL using an Amicon Ultra-15 centrifugal filter (MW 1...
19 downloads 0 Views 5MB Size
Research Article pubs.acs.org/synthbio

Genetically Encodable Bacterial Flavin Transferase for Fluorogenic Protein Modification in Mammalian Cells Myeong-Gyun Kang,† Jumi Park,‡ Gianfranco Balboni,§ Mi Hee Lim,† Changwook Lee,‡ and Hyun-Woo Rhee*,† †

Department of Chemistry and ‡Department of Biological Sciences, Ulsan National Institute of Science and Technology (UNIST), Ulsan 44919, Korea § Department of Life and Environmental Sciences, Pharmaceutical, Pharmacological and Nutraceutical Sciences Unit, University of Cagliari, I-09124 Cagliari, Italy S Supporting Information *

ABSTRACT: A bacterial flavin transferase (ApbE) was recently employed for flavin mononucleotide (FMN) modification on the Na+-translocating NADH:quinone oxidoreductase C (NqrC) protein in the pathogenic Gram-negative bacterium Vibrio cholerae. We employed this unique post-translational modification in mammalian cells and found that the FMN transfer reaction robustly occurred when NqrC and ApbE were genetically targeted in the cytosol of live mammalian cells. Moreover, NqrC expression in the endoplasmic reticulum (NqrC-ER) induced the retrotranslocation of NqrC to the cytosol, leading to the proteasome-mediated ER-associated degradation of NqrC, which is considered to be an innate immunological response toward the bacterial protein. This unexpected cellular process of NqrC-ER could be exploited for the construction of an in cellulo proteasome inhibitor screening system, and our proposed approach yielded substantially improved results compared to a previous method. In addition, a truncated version of RnfG (half-RnfG) was found to be potentially useful as a genetically encoded tag for monitoring protein−protein interactions in a specific compartment, even in the ER, in a live cell according to its fluorogenic post-translational modification via ApbE. This new genetically encoded system in mammalian cells should serve as a valuable tool for anticancer drug screening and other applications in molecular and synthetic biology. KEYWORDS: flavin transferase, flavin mononucleotide modification, ER-associated degradation, proteasome inhibitor screening, protein−protein interaction, subcellular compartment

G

immunostaining method using expensive antibodies is required to visualize the modified product. The recently engineered bacterial LplA can accept a synthetic fluorophore such as coumarin-conjugated carboxylic acid15 or resorufin-conjugated carboxylic acid,16 resulting in a fluorescent-modified product, thereby eliminating the requirement of an antibody. Although this is a more convenient approach, all of these methods require expensive materials, including both the antibodies and synthetic fluorophores, making them not cost-effective for the construction of a high-throughput PPI monitoring system. In this article, we introduce a newly identified bacterial flavin transferase (ApbE), which can perform an orthogonal flavin transferase reaction onto its substrate protein, Na+-translocating NADH:quinone oxidoreductase C (NqrC) or recombinant Rhodobacter capsulatus electron transport protein (RnfG), using flavin nucleotide, an endogenous green fluorescent cofactor in mammalian cells. NqrC (27 kDa) is a bacterial protein subunit of the multiprotein complex of the Na+-pump machinery of Vibrio

enetically encoded reporter systems have become essential tools in biological research, which help to efficiently unveil and confirm biological events.1 In particular, reporter systems for visualizing protein−protein interactions (PPIs) have helped to gain a deeper understanding of the dynamic “interactome” in living cellular systems2−4 Moreover, construction of a high-throughput PPI monitoring system5 using a genetically encoded reporter has contributed to the development of an in vivo platform for efficient drug screening in diverse disease models.5,6 For PPI monitoring, genetically targeted bacterial enzymatic post-translational protein modification reactions have been introduced to mammalian systems. For example, bacterial biotin ligase,7 lipoic acid ligase (LplA),8 sortase,9 phosphopantetheinyltransferase,10,11 and farnesyltransferase12−14 have been successfully employed to monitor spatially restricted PPIs with their orthogonal substrate proteins or peptides in live mammalian cells. It is noteworthy that none of these bacterial enzymatic reactions are naturally observed in the human proteome; thus, detection of the modification of an acceptor protein could be regarded as a true PPI between proteins of interest in a specific compartment. The bacterial reactions that have been thus far employed for PPI monitoring do not emit fluorescence; thus, an © 2016 American Chemical Society

Received: October 7, 2016 Published: December 30, 2016 667

DOI: 10.1021/acssynbio.6b00284 ACS Synth. Biol. 2017, 6, 667−677

Research Article

ACS Synthetic Biology

PPI monitoring. For this assay, we attempted to detect this reaction within the cytosolic and endoplasmic reticulum (ER) proteome, which are two major subcellular compartments of mammalian cells with different physiological conditions (e.g., pH, redox potential). We therefore cloned nqrC and apbE of Vibrio cholerae with conjugation of the nuclear exclusion signal sequence (NES: LQLPPLERLTLD) at the C-terminus- or ERtargeting signal sequence (N-terminus: IgK chain signal sequence, C-terminus: KDEL), and an epitope tag for immunodetection (V5 for ApbE, and Flag-tag for NqrC) into a mammalian expression vector, pCDNA3 or pDisplay. These plasmids were transiently transfected into HEK293T cells, incubated at 37 °C for 24 h, and harvested. Cells were lysed in RIPA lysis buffer and loaded onto a denaturing sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel. The gel was analyzed by in-gel fluorescence using the GE Typhoon gel scanner, and the expression levels of proteins were determined by Western blotting using specific antibodies. As shown in Figure 1, both NqrC-NES and ApbE-NES were well expressed in HEK293T cells. Only cells cotransfected with both NqrC and ApbE showed a single strong green fluorescent protein band, corresponding with the size of NqrC (Figure 1B, lane 1). This labeled band was not observed in either the untransfected cell lysate (Figure 1B, lane 6) or in the cells transfected individually with either NqrC-NES or ApbE-NES (Figure 1B, lane 2 and 3), which indicated that there are no functional homologous proteins of NqrC and ApbE present in the mammalian cell. When extra FAD was added to the cell media, the fluorescence intensity of the labeled band did not increase, suggesting that there was a sufficient amount of FAD in the mammalian cellular environment (Figure 1B, lane 5). Furthermore, it is noteworthy that this fluorescent band was not observed when ApbE-NES was coexpressed with T229A NqrC-NES, in which the FMN labeling site (Thr-229) of NqrC was mutated to alanine (Figure 1B, lane 4). These results suggest that ApbE-mediated FMN labeling on NqrC is very robust and specific in mammalian cells. ER-Targeted NqrC Is Retro-translocated to the Cytosol and Degraded by the Proteasome. Next, we tested whether NqrC and ApbE also show a fluorogenic response in other mammalian compartments with different physiological conditions, such as the ER. We observed a fluorescent band of fluorescently labeled NqrC-ER when it was cotransfected with ApbE-ER (Figure 2A, lane 1). This FMN labeling reaction did not occur when ApbE-ER or NqrC-ER was solely expressed in the ER (Figure 2A, lanes 2 and 3). In addition, the fluorescence labeling intensity increased when the cells were treated with additional FAD (100 μM), indicating that, in contrast to the cytosol, the local concentration of FAD in the ER was not sufficient for this reaction under the physiological condition of ER (Figure 2A, lane 4). However, even this increased green fluorescence intensity of FMNmodified NqrC in the ER was quite low compared to that observed in the cytosol. Thus, we examined the total expression levels of NqrC-ER and NqrC-NES using the same amount of plasmid for transfection into the same amount of HEK293T cells, and found that the total expression level of NqrC-ER was much lower than that of NqrC-NES (anti-Flag Western blot, Figure S1), although the ApbE-ER expression level was similar to the ApbE-NES level (anti-V5 Western blot, Figure S1). This result implies that NqrC degradation occurs in the ER regardless of its flavinylation status.

Figure 1. FMN labeling of NqrC by ApbE in mammalian cells. (A) Scheme of the ApbE-mediated flavinylation of NqrC; the crystal structure of NqrC is from the protein database (PDB ID: 4U9S). The FMN-labeling site (Thr-229) of NqrC is shown in blue. (B) Gel analysis of FMN-modified NqrC-NES by ApbE-NES in HEK293T cells.

cholerae. NqrC possesses an unusual flavin mononucleotide (FMN) covalently attached via phosphoester bonds to a threonine residue (Thr-229),17 which is known to participate in a redox-driven Na+ translocation pathway.18 Its flavinylation mechanism remained undiscovered for a decade, until Bogachev and co-workers recently identified that ApbE (36 kDa) is an enzyme that transfers the FMN moiety from flavin adenine dinucleotide (FAD) to Thr-229 of the NqrC protein (Figure 1A).19 Owing to FMN’s strong green fluorescence around 520 nm (Q = 0.26, excitation at 470 nm, emission at 520 nm), labeled NqrC could be easily detected by fluorescence.19 Using this fluorogenic enzymatic reaction, we successfully constructed an in cellulo proteasome activity monitoring system that is expected to be beneficial for proteasome inhibitor drug screening.



RESULTS Site-Specific Flavin Transfer Reaction of Recombinant NqrC and ApbE in Mammalian Cells. We first checked whether the FMN labeling reaction between NqrC and ApbE occurs under the physiological condition of a mammalian system and orthogonality of this reaction with the mammalian proteome. If functional homologous proteins for NqrC or ApbE are present in the human proteome, these bacterial proteins would not be suitable as genetically encoded tags for 668

DOI: 10.1021/acssynbio.6b00284 ACS Synth. Biol. 2017, 6, 667−677

Research Article

ACS Synthetic Biology

fluorescent intensity was largely enhanced following MG-132 treatment, because the accumulated cytosolic NqrC-ER could react with ApbE-NES. Therefore, these experiments confirmed that NqrC-ER is indeed translocated to the cytosol and degraded by the cytosolic proteasome. By contrast, NqrC-NES was robustly expressed in the cytosol and its expression level was not changed by MG-132 treatment (Figure S2A), demonstrating that originally cytosol-targeted NqrC was not affected by ERAD pathway. This result indicates that the presence of NqrC in the ER is tightly controlled by an as-yetunknown mechanism. One possible mechanism to explain why NqrC in the ER follows the ERAD pathway is that NqrC might be unfolded in the acidic environment of the ER lumen and would therefore be recognized by the unfolded protein response for ERAD.23 To check the possibility of the structural variation of NqrC in different environments, we recombinantly expressed NqrC from Escherichia coli and analyzed its secondary structure and thermal stability based on comparison of the circular dichroism (CD) spectra of NqrC in low pH (pH 5.0) and neutral pH (pH 7.5) conditions, representing subcompartments of the ER and cytosol, respectively (see the Supporting Information). No significant difference was observed in the wavelength scans of the CD spectra between two pH conditions, indicating that pH variation did not affect the secondary structure of NqrC (Figure 3A). Furthermore, NqrC at a low pH showed even higher thermal stability than that in neutral pH (Figure 3B), indicating that NqrC remains folded in a low pH condition. We also performed size-exclusion chromatography of NqrC, which migrated as a small effective size on the column, indicating that NqrC formed a more compacted structure at a low pH (Figure 3C). On the basis of these results, we concluded that NqrC is not unfolded in the acidic conditions of the ER and therefore follows another mechanism for ERAD in mammalian cells. In addition, we tested whether the retro-translocation of NqrC-ER is mediated by p97/VCP.24 For this assay, we applied Eeyarestatin I (EerI), a well-known specific p97 inhibitor,25 to our NqrC-ER/ApbE-NES system. As shown in Figure S3A, the degradation of NqrC-ER was inhibited upon addition of EerI from 1 to 5 μM. We observed increased expression levels of both Flag-NqrC-ER (3.3 fold increase) and FMN-labeled NqrC-ER (3.6 fold increase) with anti-Flag Western blot analysis and in-gel fluorescence detection, respectively (Figure S3B). This result implies that the retro-translocation of NqrCER should be mediated by p97/VCP.24 NqrC-ER/ApbE-NES as an In Cellulo Proteasome Inhibitor Screening System. The observed fluorogenic response of NqrC-ER resulting from proteasome inhibition shows a potential application as an in cellulo proteasome inhibitor screening platform, particularly for developing anticancer drugs for multiple myeloma, which is strongly related to abnormal proteasome activity.22 To check whether our proposed system could be used to screen proteasome inhibitors in a dose-dependent manner, we treated NqrC-ER/ ApbE-NES-transfected HEK293T cells to a range of concentrations (0.1 to 5.0 μM) of MG-132 for 16 h. Cell lysates were obtained after 16 h of incubation and subjected to gel electrophoresis followed by analysis of the in-gel fluorescence band intensity of NqrC. As shown in Figure 4B, a more intense green fluorescent band was observed in samples treated with higher concentrations of MG-132. As a negative control, NqrCNES/ApbE-NES cotransfected cells showed no response to the MG-132 treatment (Figure S2B). Thus, we could propose that

Figure 2. Proteasome-mediated degradation of NqrC-ER. (A) Gel analysis of FMN-modified NqrC-ER by ApbE-ER in HEK293T cells. (B) Gel analysis of FMN-modified NqrC-ER by ApbE-NES in HEK293T cells. For the sample in lane 6, NqrC-ER/ApbE-NEStransfected HEK293T cells were incubated with MG-132 (5 μM) for 16 h.

In addition, we compared the expression level of NqrC-ER to ER-targeted genetically encoded peroxidase (e.g., APEX2-ER)20 in doxycycline-inducible Flp-In Hek293T-REx stable cell lines. We could observe that APEX2-ER was weakly expressed by doxycycline treatment (e.g., 10 ng/mL to 100 ng/mL) whereas NqrC-ER expression was not observed under the same condition (Figure S1B and C). We also checked that NqrCER expression can be rescued by MG-132 and reacted with ApbE-NES in stable cell line (Figure S1D). This result implies that NqrC-ER can be degraded by ER-associated degradation (ERAD) pathway.21 To check whether NqrC-ER is degraded by the cytosolic proteasome, we evaluated whether its expression could be rescued by the cytosolic proteasome inhibitor MG-132.22 As shown in Figure 2B, the MG-132-treated sample showed a higher expression level of NqrC-ER (anti-Flag Western blot, lane 2) than the untreated sample (anti-Flag Western blot, lane 1). This result indicates that NqrC-ER is specifically degraded by the cytosolic proteasome and that NqrC-ER should be retrotranslocated to the cytosol for degradation.21 Thus, we postulated that NqrC-ER would accumulate in the cytosol when the proteasome is inhibited by MG-132. To verify this hypothesis, we coexpressed ApbE-NES and NqrC-ER in the same HEK293T cells and observed that NqrCER was only weakly labeled by ApbE-NES due to its retrotranslocation to the cytosol (Figure 2B, lane 5). However, its 669

DOI: 10.1021/acssynbio.6b00284 ACS Synth. Biol. 2017, 6, 667−677

Research Article

ACS Synthetic Biology

Figure 3. In vitro characterization of NqrC structural stability in an acidic condition. (A) Circular dichroism (CD) spectra of NqrC in low pH (pH 5.0, blue line) and neutral pH (pH 7.5, orange line) conditions. (B) Thermal stability of NqrC in neutral condition (pH 7.5, orange line) and acidic condition (pH 5.0, blue line). (C) Gel filtration of NqrC in neutral condition (pH 7.5, orange line) and acidic condition (pH 5.0, blue line). Figure 4. In cellulo proteasome inhibitor screening assay using NqrCER and ApbE-NES in living HEK293T cells. (A) Scheme of the proposed mechanism. (B) In cellulo screening result of NqrC-ER- and ApbE-NES-transfected cells treated with MG-132 (0−5 μM) for 16 h, and gel analysis of FMN-modified NqrC in each treated sample. (C) In-gel fluorescence intensity and expression level of the expressed enzymes by Western blot (anti-Flag for NqrC-ER and anti-V5 for ApbE-NES) treated with different MG-132 concentrations. The maximum and minimum intensity values were obtained from triplicate experiments. (D) In cellulo screening result of NqrC-ER- and ApbENES-transfected cells treated with bortezomib (0−50 nM) for 24 h. (E) Chemical structures of the proteasome inhibitors MG-132 and bortezomib. (F) In cellulo screening result of NqrC-ER- and ApbE-

only the NqrC-ER/ApbE-NES system effectively exhibited a dramatic fluorescence response toward the proteasome inhibitor. The proposed mechanism of rescued NqrC labeling by ApbE-NES in the cytosol is shown in Figure 4A. We also tested another proteasome inhibitor, bortezomib (also known as Velcade), which is an FDA-approved drug for multiple myeloma treatment.26 Bortezomib treatment also increased the fluorescence level of NqrC-ER, and its level became saturated at over 20 nM of bortezomib (Figure 4C). This low nanomolar efficacious dose of bortezomib is 670

DOI: 10.1021/acssynbio.6b00284 ACS Synth. Biol. 2017, 6, 667−677

Research Article

ACS Synthetic Biology

NqrC by ApbE. Protocols for fixation and denaturation for the fluorescence microscope detection of FMN-labeled NqrC are described in the Supporting Information. We have explored other proteasome-specific degrons such as Ubi-G76 V34 with NqrC to check whether NqrC could be utilized as a general genetically encoded tag protein for proteasome inhibitor assays. As shown in Figure 5B, we found that degradation of Ubi-G76 V-NqrC was inhibited by MG-132 treatment, and the accumulated Ubi-G76 V-NqrC was FMNlabeled by ApbE-NES by gel electrophoresis analysis. This increased fluorescence of FMN-labeled Ubi-G76 V-NqrC was also observed by confocal microscopy (Figure 5C and Figure S4). The increased fluorescence intensity of NqrC-ER and UbiG76 V-NqrC were quantified by ImageJ program and this analysis result showed that the dynamic fluorescence response of Ubi-G76 V-NqrC to MG-132 is slightly less than that of NqrC-ER, but sensitivity at low concentrations (e.g., 2 μM) of MG-132 is superior to NqrC-ER (Figure 5D). From this experiment, we proposed that Ubi-G76 V-NqrC can be utilized as a proteasome-specific degradation reporter for proteasome inhibitor screening. Truncated RnfG Is Another Promising FMN Acceptor Protein of ApbE. In Vibrio cholerae, there are several bacterial proteins (NqrB, RnfD, and RnfG) that show high sequence homology with NqrC.19 Accordingly, these proteins are expected to be able to be FMN-modified by ApbE; however, this ability has not been fully characterized yet. Thus, we recombinantly expressed NqrB, RnfD, and RnfG in the ER with ApbE-ER to evaluate whether these bacterial homologous proteins are modified by ApbE in the ER. In this assay, we also introduced other bacterial species of ApbE, including ApbE of Salmonella enterica (Se.ApbE) and ApbE of Treponema pallidium (Tp.ApbE), because these enzymes also show considerable sequence homology with ApbE of Vibrio cholerae (Vc.ApbE); Se.ApbE and Tp.ApbE showed 43.8% and 28.4% sequence identity with Vc.ApbE, respectively. In this mix-match experiment, we attempted to find unexpected substrate tolerance of other species of ApbE for the proteins from Vibrio cholerae. As shown in Figure 6A, RnfG-ER showed considerably strong green fluorescence, which indicated that a substantial level of FMN modification had occurred by Se.ApbE-ER. Interestingly, Se.ApbE could recognize and modify both NqrC and RnfG of Vibrio cholerae (Figure 6A and B), whereas Vc.ApbE could not recognize its own RnfG (Figure 6A). It is also noteworthy that RnfG-ER was not degraded by ERAD, unlike NqrC-ER. Owing to its nondegradable property in the ER, RnfG can have an orthogonal labeling reaction with Se.ApbE in either compartment (i.e., the cytosol or ER) of mammalian cells (Figure 6A and B). We did not observe any cross-talk between RnfG and ApbE when they were expressed in the different compartments (Figure 6C). Moreover, the other NqrC homologues tested (NqrB and RnfD) did not show any labeling by any of the three ApbE species in the ER. Notably, NqrB and RnfD did not react with the ApbEs in the cytoplasm, whereas RnfG-NES showed reactivity with Se.ApbENES (Figure 6B). Because covalently FMN-modified NqrB and RnfD were previously characterized from the endogenous bacterial proteome,35,36 our result suggests that there should be another flavin transferase enzyme rather than NqrC that may conduct the FMN modification on NqrB and RnfD. It is noteworthy that flavinylated sites of NqrB and RnfD are exposed at the cytoplasmic side of the inner membrane (IM) of

Figure 4. continued NES-transfected cells treated with flavonoids (100 μM each). (G) Comparison of in cellulo screening results of mApple-degron-, NqrCER, and ApbE-NES-transfected cells treated with of MG-132 (0−5 μM). mApple fluorescence was measured in the RFP channel and FMN-modified NqrC fluorescence was measured in the GFP channel of the Typhoon gel scanner. Total expression levels of mApple-degron and NqrC-ER were analyzed by anti-Flag Western blot.

consistent with a previous report.27 We also tested various flavonoid molecules using our system, which have been reported to inhibit proteasome activity based on in vitro assays with purified proteasome.28−30 Specifically, we treated NqrCER/ApbE-NES-transfected cells with 100 μM of epigallocatechin gallate (EGCG), quercetin, and several aminoisoflavone molecules,31 including 1, 2, and 3 (see Figure S3 for the chemical structures); however, we did not observe any fluorescence increase of NqrC-ER in the flavonoid-treated samples, whereas the MG-132- and bortezomib-treated samples reproducibly showed more FMN-labeled NqrC-ER (Figure 4F). For comparison of other reported systems designed for proteasome inhibitor screening, we tested the mApple-degron system, which was designed to be degraded by the proteasome in live cells.6 This fluorescent protein-conjugated degron responds to proteasome inhibitor treatment by an increase in its total fluorescence level due to the accumulation of the nondegraded protein population. For comparison, we treated mApple-degron- and NqrC-ER/ApbE-NES-transfected HEK293T cells with the same amount of MG-132 (0, 2, 5 μM) respectively for 16 h. As shown in Figure 4G, mAppledegron showed negligible enhanced fluorescence intensity (1.3 fold increase) by MG-132 (5 μM), whereas NqrC-ER showed substantially increased fluorescence intensity (12.1 fold increase) by the same amount of MG-132 treatment. In addition, the basal level of mApple-degron was much higher than that of NqrC-ER (anti-Flag Western blot, Figure 4G), which is likely related to the higher resistance to proteolysis of beta-barrel fluorescent protein.32 Overall, our proposed proteasome inhibitor screening system using NqrC-ER and ApbE-NES showed significant improvement over the previous method and provided unbiased results. Cell-Based Detection of FMN-Labeled NqrC by Confocal Microscope. For high-throughput screening of proteasome inhibitor, cell-based detection of FMN-labeled NqrC by optical imaging33 might be preferable to gel electrophoresis experiments. Because FMN-labeled NqrC was detected in SDS-denaturing gel, we postulated that the addition of SDS after fixation and permeabilization steps might help to recover FMN fluorescence by denaturing the labeled proteins in the fixed cells. We tested this condition (e.g., 10% SDS incubation for 10 min after formaldehyde fixation) with NqrCER and ApbE-NES system, and successfully observed that fluorescent signal of FMN-labeled NqrC-ER was increased upon addition of MG-132 using confocal microscopy (Figure 5A). We also checked NqrC-ER and ApbE-NES expression by immunofluorescence imaging with anti-Flag (for Flag-NqrCER) and anti-V5 (for V5-ApbE-NES), respectively. As shown in Figure S2C, we confirmed that this green fluorescence signal appears in the cells expressing both NqrC-ER and ApbE-NES. This result supports that the green fluorescence signal observed by microscopy is not an artifact but a signal from FMN-labeled 671

DOI: 10.1021/acssynbio.6b00284 ACS Synth. Biol. 2017, 6, 667−677

Research Article

ACS Synthetic Biology

Figure 5. Cell-based detection of FMN-labeled NqrC by confocal microscopy. (A) Confocal microscope imaging results of Flag-NqrC-ER and V5ApbE-NES transfected HEK293T cells under MG-132 treatment (0−5 μM) overnight. (B) Proteasome inhibition assay using Ubi-G76 V-NqrC and V5-ApbE-NES by Western blot (anti-Flag for Ubi-G76 V-Flag-NqrC and anti-V5 for V5-ApbE-NES). (C) Confocal microscope imaging results of Ubi-G76 V-NqrC and ApbE-NES under MG-132 treatment (0−5 μM) overnight. (D) Quantified values of fluorescence intensities of NqrC-ER (A) and Ubi-G76 V-NqrC (C) using the ImageJ program. FMN-labeled NqrC was visualized after incubation with 10% SDS for 10 min after fixation. Please see the detailed protocol in the Supporting Information. Scale bar = 10 μm. 672

DOI: 10.1021/acssynbio.6b00284 ACS Synth. Biol. 2017, 6, 667−677

Research Article

ACS Synthetic Biology

Figure 6. Exploration of ApbE from other bacterial species and NqrC homologous proteins. FMN modification test of NqrB, RnfD, and RnfG from Vibrio cholerae by ApbE from Vibrio cholerae (Vc-ApbE), Salmonella enterica (Se.ApbE), and Treponema pallidium (Tp-ApbE) in the endoplasmic reticulum (A) and the cytoplasm (B) of the in the mammalian cell. (C) Spatial-restricted reaction of RnfG with Se.ApbE in the ER and cytosol of HEK293T cells. In-gel fluorescence was analyzed using the Typhoon gel scanner (GFP filter).

the bacteria while the flavinylated sites of NqrC and RnfG are exposed at the periplasmic side of the IM.18,37 Thus, a hypothetical flavin transferase which has substrate specificity for NqrB and RnfD might work in the cytoplasmic space of the Gram-negative bacteria. We further investigated whether a truncated version of RnfG could react with ApbE in mammalian cells. Recently, the RnfG topology was characterized at the bacterial cytosolic membrane, and the N-terminus domain of RnfG is proposed to have a transmembrane (TM) domain (Figure 7A).37 Thus, we cloned a 10-kDa fragment of the C-terminal domain of RnfG (amino acids 120−209; designated as Half-RnfG) that contained the Thr-175 residue required for FMN modification (Figure 7A). Interestingly, Half-RnfG-NES was strongly FMN-labeled by Vc.ApbE-NES but not by Se.ApbE-NES (Figure 7B). In addition, we did not observe any cross-talk between Vc.ApbE and Half-RnfG in the different compartments of the mammalian cell (Figure 7C). We also detected the green fluorescence signal of Half-RnfG when ApbE was expressed in the same subcellular space using confocal microscopy (Figure 7D). This result implies that a Half-RnfG and ApbE system can be employed for the high-throughput monitoring of spatially restricted protein−protein interactions in mammalian cells.

Mammalian expression of all of the constructs we tested was confirmed by imaging experiments (Figure S5).



DISCUSSION

In this study, we first introduced a bacterial FMN modification reaction in a mammalian system and discovered that two FMN acceptor proteins, NqrC and RnfG, behave differently in mammalian cells. Each unique property of NqrC and RnfG can be respectively employed for the construction of an in cellulo proteasome inhibitor screening platform (NqrC-ER/ApbENES). Furthermore, adoption of a spatially restricted protein labeling approach (e.g., Half-RnfG) might be useful to monitor specific PPIs in a certain cellular compartment. In particular, because Half-RnfG is a considerably small-sized protein, it might be beneficial for a genetic tagging approach, especially recently developed knock-in technology using the CRISPRCas9.38 In addition, FMN-binding proteins (e.g., LOV domain,39 minisog40) have been recognized for their unique photochemical properties, our covalent protein FMN labeling system can be further utilized for other potential applications in synthetic biology. 673

DOI: 10.1021/acssynbio.6b00284 ACS Synth. Biol. 2017, 6, 667−677

Research Article

ACS Synthetic Biology

modification is a covalent bond, only FMN-modified NqrC could be observed on the denaturing gel (Figure 1). Notably, in-gel fluorescence detection can avoid problems of autofluorescence background interference because all of the small greenish fluorescent biomolecules could be removed during the gel electrophoresis. In our experiment, the flavonoid molecules did not show their previously reported strong efficacy to inhibit proteasome activity in a test tube. This inconsistency in the results between in vitro and in cellulo assays may be due to alteration of assembled structure of the isolated proteasome during the purification step under the nonphysiological condition.43,44 Thus, a proteasome inhibition assay conducted in live cells can provide a more unbiased screening result to select active candidate molecules for development of next-generation antimyeloma drugs.26 Several previous studies have focused on bacterial protein degradation in mammalian cells through the ERAD pathway, which can be considered as an innate immunological response toward bacterial proteins.45−48 We postulated that NqrC expression in the ER might be recognized as the protein derived from a bacterial infection in the human cell. This might be reasonable because in most cases bacterial invasion occurs through endocytosis, which is connected to the secretory pathway of the ER. Thus, our results support that the recognition unit for bacterial proteins might be located in the ER,46−48 because NqrC expression in the cytosol was not affected by proteasome-mediated degradation. We postulated that NqrC-ER degradation by proteasome might not follow ubiquitinylation pathway because we did not observe significant molecular weight increase of NqrC-ER in the gel analysis. Because this innate immunological process for bacterial proteins is still largely unknown, further investigation might be required to understand and exploit the mechanisms.



CONCLUSION In summary, we genetically introduced a bacterial flavin transferase (ApbE) into a mammalian system, which performed a very specific fluorogenic modification on its substrate proteins, NqrC and RnfG, in live cells. We found that NqrC was degraded by the ERAD pathway, and we employed this unexpected behavior of NqrC in mammalian cells to construct an in cellulo proteasome inhibitor screening system. We also found that RnfG and its truncated version were robustly labeled by ApbE in both the ER and cytosol, which might be beneficial to monitor PPIs of interest in a specific compartment of mammalian cells.

Figure 7. Spatial-restricted reactivity of RnfG and its truncated version (Half-RnfG) with ApbE. (A) RnfG structure and its topology at the bacterial outer membrane. The RnfG structure was predicted using the Swiss-Model algorithm based on its structural homology with NqrC.49 (B) Species-specific test of truncated RnfG (Half-RnfG-NES) with VcApbE-NES and Se.ApbE-NES in the cytosol. (C) Confirmation of the spatial-restricted reaction of Half-RnfG with Vc.ApbE either in the cytosol or ER by Western blot (anti-Flag for Half-RnfG constructs and anti-V5 for ApbE constructs). (D) Confocal microscope imaging results of the spatial-restricted reaction of Half-RnfG with Vc.ApbE either in the cytosol or ER. Scale bar = 10 μm.



EXPERIMENTAL SECTION Plasmids and Cloning. Genes were cloned into the specified vectors using standard enzymatic restriction digest and ligation with T4 DNA ligase. To generate constructs where short tags (e.g., V5 or a Flag epitope tag) or signal sequences were appended to the protein, the tag was included in the primers used to PCR-amplify the gene. PCR products were digested with restriction enzymes and ligated into cut vectors (e.g., pcDNA3, and pDisplay). In all cases, the CMV promoter was used for expression in mammalian cells. Table S1 in the Supporting Information summarizes the genetic constructs cloned and used for this study. The genes of NqrC and ApbE of V. cholera were a generous gift of Professor Alexander Bogachev of Moscow State University.

It is noteworthy that FMN-NqrC exhibited weak fluorescence in living cells because conjugated FMN is partially buried in the binding pocket of NqrC.18,41 Since the isoalloxazine ring of FMN is an environment-sensitively quenched fluorophore via hydrophobic interactions,42 FMN fluorescence should be quenched in the native conformation. Indeed, the FMN-NqrC fluorescence could be recovered on the protein when it was denatured on the SDS-PAGE gel. Because this FMN 674

DOI: 10.1021/acssynbio.6b00284 ACS Synth. Biol. 2017, 6, 667−677

Research Article

ACS Synthetic Biology Mammalian Cell Culturing and Transfection. HEK293T cells obtained from the American Type Culture Collection (passages