Near-infrared light-controlled gene expression and protein targeting in neurons and non-neuronal cells

Abstract

Near-infrared (NIR) light-inducible binding of bacterial phytochrome BphP1 to its engineered partner QPAS1 is used for optical protein regulation in mammalian cells. However, there are no data on the application of the BphP1-QPAS1 pair in cells derived from various mammalian tissues. Here, we tested the functionality of two BphP1-QPAS1-based optogenetic tools, such as an NIR and blue light-sensing system for control of protein localization (iRIS) and an NIR light-sensing system for transcription activation (TA), in several cell types including cortical neurons. We found that the performance of these optogenetic tools often relies on physiological properties of a specific cell type, such as nuclear transport, which may limit the applicability of blue light-sensitive component of iRIS. In contrast, the NIR-light-sensing part of iRIS performed well in all tested cell types. The TA system showed the best performance in HeLa, U-2 OS and HEK-293 cells. Small size of the QPAS1 component allows designing AAV viral particles, which were applied to deliver the TA system to neurons.

Graphical Abstract

BphP1-QPAS1-based near-infrared optogenetic systems for control of protein localization (iRIS) and for transcription activation (TA) can be used in various cell types, however, require additional tuning for neuroblastoma cells and neurons. Small size of the QPAS1 component enables delivery of these optogenetic systems to primary neurons using AAV gene transfer.

Precise control of cell physiology with high spatial and temporal resolution is of high demand in basic studies and biomedical applications. One of the several existent approaches to this problem, namely optical control of protein activity and protein-protein interactions with genetically encoded constructs (also referred as non-opsin optogenetics), looks promising because of robust performance and high resulting precision ^[1]. Variety of proteins naturally sensitive to light was used to develop the tools for such optical manipulation. Small and well structurally characterized LOV domains were widely used for the design of optogenetic tools based on homodimerization, heterodimerization and structural rearrangements (caging-uncaging). Examples of biological applications of LOV domain-based tools include regulation of protein localization^[2], transcription activation in cells and in animals^[3,4] and control of receptor activity via caging^[5]. Similar schemes were implemented using the BLUF domains and cryptochromes. For instance, cryptochrome-derived heterodimerizers Cry2 and CIB were used for Cre recombinase-regulated gene expression^[6]. Although being successfully used in many promising applications all these tools sense blue light making them hardly compatible with popular probes for visualization and optical control excited in the same blue-green spectral range.

Unlike blue-light sensing LOV and BLUF domains and cryptochromes, phytochromes are proteins absorbing light in far red to near-infrared (NIR) spectral region. Phytochrome from Arabidopsis PhyB with its interacting partners were used for the development of optogenetic tools based on light-controlled heterodimerization. It was applied for transcription activation^[7], manipulation of cell morphology^[8] and precise protein targeting in living fish embryo^[9]. Being derived from plant, PhyB requires phycocyanobilin for functioning, which is not produced by animal cells and therefore must be added exogenously, or synthesized within the cells using additionally overexpressed enzymes.

Recently, several optogenetic systems were developed using bacterial phytochromes as templates. Bacterial phytochromes incorporate biliverdin (BV) as a chromophore. BV is found in mammalian cells as a product of heme degradation, and, therefore, bacterial phytochromes work in most of the animal cells without additional non-native chemical compounds. Bacterial phytochrome from Rhodopseudomonas palustris RpBphP1 (hereafter BphP1) was used for light-controlled cell signaling manipulation and transcription activation^[10]. Subsequently, the engineered version of its interacting partner PpsR2, termed QPAS1, was used to design the dual-color (NIR and blue) light-sensing system for control of protein localization (iRIS) ^[11].

Despite a vast number of non-opsin optogenetic tools applications in mammalian cells and whole body applications^[12], their performance may vary dramatically in different cell types^[7]. At the same time, the applicability of the tool to a certain experimental model relies strongly on specific features of cell physiology.

To explore this issue, in this study, we tested the performance of BphP1-QPAS1 optogenetic system in mammalian cells of different origins, including neuroblasts and primary neurons. To analyze the protein transport machinery, which is crucial for the functionality of intracellular localization optical controllers, we utilized the tool combining BphP1 and AsLOV2, called iRIS. Then we tested the BphP1-QPAS1 system for the ability to activate the transcription in the same cells. Lastly, we took advantage of QPAS1 small size to address the problem of optogenetic system delivery to the neuronal cells. For this, we used the AAVs to deliver BphP1-QPAS1 system for gene regulation to activate the transcription in primary neurons.

RESULTS

Optical control of protein localization.

To assess the functionality of BphP1-QPAS1-based system in protein targeting applications, a series of mammalian cells of different origins were transfected with iRIS construct (Figure 1). In NIR light-controlled component of iRIS, CAAX motif (-CVIM) is utilized to target the protein of interest to the plasma membrane. In blue light controlled component, nuclear localization of the protein of interest is achieved via the balanced action of nuclear localization (c-Myc NLS) and nuclear exclusion signals (Smad4 NES). mCherry tag is used for visualization.

Figure 1.

Schematic representation of light-controllable protein targeting using the near-infrared-blue-light-inducible shuttle (iRIS). In cells transfected with iRIS, under 740 nm light, NES-mCherry-QPAS1-AsLOV2cNLS is relocalized to plasma membrane-bound BphP1-mVenus (NIR light-controlled component). Under 460 nm light, the same construct is relocalized to nucleus, due to uncaged nuclear localization signal in AsLOV2 (blue light-controlled component). Localization of protein of interest is shown in red.

As observed by epifluorescence microscopy, in darkness, iRIS-controlled protein was localized in cytoplasm, showing slightly lower fluorescence intensity in nucleus (Figure 2). Under NIR light of 740 nm (1 mW cm⁻²) all tested cell types showed robust protein relocalization from cytoplasm to the plasma membrane where it decorated cell periphery and filopodia (Figure 2, Supplementary Figure 1). Under blue light of 460 nm (1 mW cm⁻²), the nuclear accumulation of mCherry was observed in all cell types studied, except neurons and neuroblastoma-derived Neuro-2a cells. In Neuro-2a cells under blue illumination, protein of interest was observed on plasma membrane, decorating the neurites, and in juxtanuclear compartments, which presumably represented aggresomes^[13]. Interestingly, another neuroblastoma-derived cells, SH-SY5Y, showed normal nuclear mCherry localization under blue light. In rat primary cortical neurons, similarly to Neuro-2a, under blue light mCherry fluorescence decreased in cytoplasm but no nuclear fluorescence was observed. In some cell types, such as HEK293 (Figure 2, bottom panel) under 460 nm illumination the considerable amount of protein was observed at the plasma membrane. Likely, this was caused by rather high iRIS expression level, and should be corrected either by reducing of plasmid amount in transfection or by using weaker promoter for iRIS expression.

Figure 2.

Tri-directional protein targeting in living mammalian cells transfected with iRIS, controlled with 740 nm and 460 nm light. Under 740 nm illumination, the protein is observed on plasma membrane, while under 460 nm light the protein is expected to be localized in nucleus. Scale bar: 10 μm.

Quantification of fluorescence intensity showed the 35-50% signal decrease in cytoplasm upon 740 nm light in all studied cell types (Figure 3A), which reflects the cytoplasm to membrane relocalization^[11]. In accordance to visual observations, quantitative analysis of mCherry distribution under 460 nm showed no significant increase of nucleus to cytoplasm ratio in Neuro-2a neuroblasts and insignificant increase (up to 1.1 ratio) in primary neurons while the other cell types showed 1.3-1.5-fold ratio (Figure 3B).

Figure 3.

Change of mCherry fluorescence levels of iRIS in different compartments under 740 nm and 460 nm illumination. A) mCherry fluorescence levels of iRIS as measured in the cytoplasm in dark (gray) and after 10 min of 740 nm illumination (pink). Error bars represent s.e.m., n=5 cells for each bar. B) Nucleus to cytoplasm ratio of mCherry fluorescence of iRIS in dark (gray) and after 10 min of 460 nm illumination (cyan). Error bars represent s.e.m., n=5 cells for each bar. C) mCherry fluorescence levels of iRIS as measured in the cytoplasm of HeLa cells in cycles of 5 min of 740 nm illumination followed by 20 min incubation in darkness. Error bars represent s.e.m., n=5 cells for each bar. Images were acquired using an epifluorescence microscope.

To further characterize the system for light-controlled subcellular protein targeting, we tested how long iRIS could cause protein relocalization. Since the reversibility of LOV-based part of iRIS was previously characterized^[3], we performed time-lapse imaging under NIR illumination, followed by thermal relaxation in darkness in sequential cycles. To achieve maximal number of cycles within reasonable duration of time-lapse imaging session, duration of 740 nm illumination was reduced to 5 min, which led to about 20% decrease of fluorescence level in cytoplasm, while longer incubation caused 40% decrease in epifluorescence microscopy. We found that at least 7 cycles of the iRIS relocalization from cytoplasm to the plasma membrane could be performed (Figure 3C).

In addition to epifluorescence microscopy, we employed confocal imaging to better characterize the iRIS protein targeting to the plasma membrane. When imaged by epifluorescence microscopy, the signal from membrane is often mixed with the unspecific signal from the neighbouring cellular environment. Moreover, blurred signal from membrane, especially from filopodia, can appear in the cytoplasm, sometimes hardly distinguishable from protein aggregates. Confocal microscopy allows avoiding these effects, providing clear images of a particular optical section. In confocal images of HeLa cells expressing iRIS, after 10 min of 740 nm light, cell perimeter was clearly visible and highlighted with signal from filopodia (Figure 4A). The cytoplasm showed an even distribution of mCherry signal, which was approximately 60% lower than signal at the plasma membrane after illumination (Figure 4B). Similarly, in primary cortical neurons expressing iRIS, under 740 nm light, mCherry signal was localized at the cell perimeter (Figure 4C). In fluorescence intensity profile plasma membrane appeared as sharp peaks, with the decrease in cytoplasmic fluorescence of about 50% (Figure 4D).

Figure 4.

NIR light-driven relocalization of the iRIS protein imaged with confocal microscopy. A) HeLa cells expressing iRIS. Scale bar: 10 μm. Dashed line marks the region used for profile plotting. B) Intensity profiles of mCherry fluorescence in cells shown in panel A. C) Cortical neuron expressing iRIS. An asterisk marks the neuron body, and the inset shows the body enlarged. Scale bar: 10 μm. Dashed line marks the region used for profile plotting. D) Intensity profiles of mCherry fluorescence in cell shown in panel C.

Altogether, these data suggest that the BphP1-QPAS1 pair can be used in all tested cell types, including neurons, however, further optimization of the blue-light sensing part of iRIS system is needed to efficiently work in neuronal cells.

Light-controlled transcription regulation.

We further analyzed the performance of BphP1-QPAS1 optogenetic pair applied to transcription control in different mammalian cell types. For this, the set of cells, corresponding to those tested with iRIS, was cotransfected with pQP-T2A plasmid and luciferase reporter, controlled by GAL4 upstream activation signal sequences, GAL4 UAS (Figure 5). pQP-T2A plasmid encodes the BphP1 fused to VP16 transactivator and QPAS1 fused to GAL4 DNA-binding domain and tagged with SV40 NLS (PKKKRK). Under NIR illumination, GAL4 DNA-binding domain and VP16 transactivator associate due to light-induced BphP1-QPAS1 interaction. Being in the nucleus, reconstituted transcription factor drives the expression of GAL4-UAS controlled reporter gene.

Figure 5.

Schematic representation of light-controllable gene transcription activation. BphP1-VP16 fusion is localized in cytoplasm in darkness. Under NIR light it interacts with QPAS1 fused to GAL4 DNA-binding domain. The complex is transported to the nucleus because of the presence of nuclear localization signal in QPAS1-GAL4 construct, leading to activation of reporter gene expression (shown as green highlighting).

Most of the tested cell lines showed several fold elevation of luciferase signal under NIR light as compared to signal in darkness (Figure 6). The highest light-to-dark contrast was observed in HeLa, HEK-293 and U2 OS cells (28.0, 8.9 and 6.1-fold, respectively). Lower light-to-dark contrast was observed in COS-7, SH-SY5Y and Neuro-2A cells, which is in accordance with data on optical control of protein targeting (Figure 3B), as transcription activation partly depends on relocalization of complex of BphP1-VP16 and NLS-GAL4-QPAS1 driven by nuclear localization signal.

Figure 6.

Change in the luciferase expression under NIR light in mammalian cells of different origin cotransfected with pQP-T2A and luciferase reporter plasmids. Cells were kept in darkness or under 740 nm illumination. For each cell type, the luciferase induction level was normalized to the level in cells kept in darkness. Error bars represent s.e.m., n=3. Luciferase levels for cells kept in darkness, 740 nm light illuminated cells, and cells transfected with reporter construct only can be found in Supplementary Figure 2.

Viral gene delivery of BphP1-QPAS1 system to neurons.

Primary cell cultures are often referred as hard to transfect^[14] and primary hippocampal and cortical neurons are sometimes considered as one of the hardest targets for gene transfer^[15]. Electrical transfection methods are reported to irreversibly affect cellular physiology and reduce neurons viability, while chemical transfection methods (such as calcium phosphate transfection and lipofection) usually provide low transfection efficiency, especially in non-dividing cultures^[16]. To overcome these limitations we decided to use the viral system for the delivery of BphP1-QPAS1 pair to primary neuronal cultures.

We chose the adeno-associated virus (AAV) serotype 9 because it was shown that this serotype transduces effectively and specifically neurons in mice^[17] and rats^[18]. Since mammalian promoters in some cases are advantageous comparing to viral promoters^[19], we decided to use CamKII promoter to drive the BphP1-QPAS1 expression in neurons. Together with shortened WPRE-PolyA expression cassette^[20], QPAS1 small size (519 base pairs) and the general optimization of vector architecture^[11] allowed us to pack the whole BphP1-QPAS1 system (3684 base pairs) for transcription activation into a single AAV-vector (Figure 7A). UAS-driven luciferase reporter, with junk DNA added to reach the length optimal for AAV packaging, was delivered in separate AAV particles. In transduced primary rat cortical neurons, luciferase signal was 2.5-fold higher under NIR light than in darkness (Figure 7B). Thus, we have applied, for the first time, AAV gene transfer system for delivery of bacterial phytochrome-based optogenetic system to primary neurons. Among the viral gene delivery approaches, AAV-based systems are generally regarded as superior for neuronal cultures because of low toxicity, high efficiency and safety^[21].

Figure 7.

Viral gene delivery of BphP1-QPAS1 system for transcription activation to primary cortical neurons. A) Constructs used for AAV production. B) Luciferase expression level in neurons transduced with AAV carrying constructs shown in A) and kept in darkness and under 740 nm light. Error bars represent s.e.m., n=3.

DISCUSSION

In this study, we tested the applicability of optogenetic systems for protein targeting (iRIS) and transcription activation (TA) developed from bacterial phytochrome to mammalian cells of different origins. We found that the performance of BphP1-QPAS1-based systems varied significantly in different cells (Table 1). Particularly, in experiments with iRIS, poor efficiency of cytoplasm to nucleus relocalization was observed under blue light in Neuro-2A cells and in neurons. This shifted NLS/NES equilibrium in neuroblasts and neurons may be explained by the difference in active nuclear transport mechanisms between the cells of neuronal and non-neuronal origins. So that, in non-neuronal cells, in the cytoplasm, proteins carrying an NLS sequence bind importin-β and importin-α complex. After the transition through the nuclear pore, NLS-tagged proteins are released in the nucleus and the importins are recycled^[22]. On the contrary, in cells of neuronal origin, due to polar phenotype, importins are involved not only in cytoplasm-to-nucleus transport, but also in synapse to nucleus signaling^{[23, 24]}. Thus, nuclear import machinery components are distributed in the cell in a different way, being localized not only in somal parts and nucleus, but also in neurites and partly immobilized by cytoskeleton^{[25, 26]}.

Table 1.

Performance of the BphP1-QPAS1-based iRIS and TA optogenetic systems in cells of different origin.

Cell type	Organism	Tissue	Disease	iRIS		TA light-to-dark signal ratio under 740 nm 3
				% of change under 740 nm 1	Change of nucleus-to-cytoplasm ratio under 460 nm 2
HeLa	H. sapiens	cervix	adenocarcinoma	39	0.48	28.0
U-2 OS	H. sapiens	bone	osteosarcoma	49	0.43	6.1
HEK-293	H. sapiens	kidney	normal condition	45	0.49	8.9
COS-7	Cercopithecus aethiops	kidney	SV40 transformed	46	0.43	2.6
SH-SY5Y	H. sapiens	bone marrow	neuroblastoma	36	0.33	3.0
Neuro-2a	Mus musculus	brain	neuroblastoma	29	0.05	2.5
Primary cortical neurons	Rattus norvegicus	brain, cortex	normal condition	53	0.29	2.5*

¹Decrease of mCherry fluorescence in cytoplasm in percentage as a result of protein relocalization from cytoplasm to plasma membrane; indicates the performance of NIR-sensitive component of iRIS.

²Change in nucleus to cytoplasm signal ratio as a result of protein relocalization from cytoplasm to nucleus; indicates the performance of blue-sensitive component of iRIS.

³Light-to-dark signal ratio in cells transfected with pQP-T2A plasmid for transcription activation measured using luciferase assay.

^*Detected in primary neurons transduced with AAV particles carrying the TA system.

Transcription activation system also showed remarkably various results among the tested cells in transcription activation, which is in accordance with the studies of other chemically^[27] and optically^{[7, 28]} controlled systems tested in different cell types. Notably, the performance of the TA system depends not only on transcription factor cellular localization controlled by nuclear import machinery, but also on BphP1-QPAS1 interaction, which adds the additional level of control to the system. In the darkness, dissociated GAL4 and VP16 do not activate the transcription. This may explain the available increase in reporter expression despite the reduced relocalization functionality in certain cell types. Performance of BphP1-QPAS1 pair in the both described applications could be improved by further engineering of BphP1, aimed in reduction of undesirable binding of QPAS1 in darkness. For this, similarly to other light-sensing proteins ^[12], directed molecular evolution and structure-based protein engineering of BphP1 can be used.

Our data suggest that the NIR protein targeting and transcription activation systems can be used in various cell types, however, require additional adjustment for neuroblastoma cells and neurons. Although a considerable number of optically controlled tools have been proposed to study neuronal functions^[29], the NIR-light controlled tools are represented rather poorly in neuroscience. We anticipate that our results will be a starting point for further development and use of BphP1-based systems in neuronal tissues. Together with the NIR fluorescent proteins used for advanced imaging in neurons^[30–32], NIR optogenetic tools should be superior for multiplexing with blue-green light sensing constructs including opsin-based actuators and inhibitors.

EXPERIMENTAL SECTION

Mammalian cell culture.

Human epithelioid cervix carcinoma HeLa cells, human bone osteosarcoma epithelial cells U-2 OS, human embryonic kidney cells HEK-293, African green monkey fibroblast-like cell line COS-7, human bone marrow neuroblastoma cells SH-SY5Y, mouse neuroblastoma cell line Neuro-2a, and mouse embryonic fibroblast cells NIH/3T3 were maintained in Dulbecco’s modified Eagle’s medium (Gibco) supplemented with 10% FBS (Gibco) and Antibiotic-Antimycotic (Gibco). Effectene transfection reagent (Qiagen) was used for transfection. The culture medium was changed 6 h after the transfection with the new one containing 25 μM of BV.

Cell light activation and imaging.

Epifluorescence microscopy was performed using an Olympus IX83 equipped with a 200 W metal halide arc lamp (Lumen 220PRO, Prior) and an optiMOS sCMOS camera (QImaging). Cells transfected with pQP-iRIS^[11] (Addgene #102584) were imaged using 60×, 1.35 NA oil objective lens (UPlanSApo, Olympus). During imaging, cells were kept in a live cell imaging solution (Life Technologies-Invitrogen) at 37 °C. The data were analyzed using a SlideBook v. 6.0.8 (3i), and ImageJ v. 1.50b software. For relocalization assay, blue and NIR illumination was applied using the 460/20 nm and 740/25 nm custom-assembled LED arrays (LED Engin), respectively. First, cells were illuminated for 10 min with 740/25 nm LED (1 mW cm⁻²), then after 30 min in darkness for relaxation, the dish was illuminated with 460/20 nm LED (1 mW cm⁻²). Focusing of the microscope was performed in a mCherry excitation channel to prevent unspecific activation of blue and NIR light-sensing components. For imaging CELLview glass bottom dishes (Greiner Bio-One) were used. Confocal imaging was performed using a Leica TCS SP8 X microscope equipped with 63×, NA 1.4 and HC PL APO CS2 objective and a white light laser (470-670 nm).

After image acquisition, the images were analyzed using a Fiji software (version 1.50b). First, the background fluorescence was subtracted. Then 5 circular regions of interest (ROI) were placed randomly in cytoplasm and nucleus of the imaged cells while avoiding aggregates, filopodia and nucleoli. The mean fluorescence intensity was calculated for each ROI. After compensation for photobleaching, the mean values were calculated for 5- 10 cells in each experimental group. The nucleus-to-cytoplasm ratio was calculated for cells illuminated with 460 nm light, and normalization to level in darkness was performed for cells kept under 740 nm illumination. Data were plotted using OriginPro (Origin Labs; version 8.6)

Firefly luciferase (Fluc) assay.

For light-controlled transcription activation, cells were cotransfected with the constructs pQP-T2A^[11] (Addgene #102583) and pFR-Luc (Agilent Technologies) in a 3:1 ratio in 24-well plates (Greiner Bio-One). Illumination of plates was applied directly in CO₂ incubator with 740/25 nm light (0.2 mW cm⁻²) in alternating cycles: 30 s light and 180 s darkness. To measure Fluc activity, transfected cells were lysed 48 h after transfection. Cells were washed with PBS, and then 100 μl of lysis buffer (20 mM Tris–HCl pH 8.0, 10% glycerol, 0.1% β-mercaptoethanol, 0.1% Triton X-100, 1 mM PMSF) was added to each well of 24-well plate and incubated on ice for 30 min. 10 μl of cell lysates were mixed with 20 μl of Firefly Luc Assay reagent (NanoLight Technology) in 96-well half-area white plates (Costar). Bioluminescence signal was measured using a Victor X3 multilabel plate reader (PerkinElmer).

Neuronal culture and transfection.

Primary rat neuronal cultures were prepared in Neuronal Cell Culture Unit, University of Helsinki. All animal work was performed in accordance with the ethical guidelines of the European convention and regulations of an Ethics committee for animal research of the University of Helsinki. Dissociated cortical neurons were plated to 24-well plate at a density of 10⁵ cells per well, coated with Poly-L-Lysine (0.1 mg/ml) (Sigma-Aldrich), in a neurobasal medium (Gibco) supplemented with B27 (Life Technologies/Invitrogen), L-glutamine (Invitrogen), and penicillin-streptomycin (Lonza). Neurons were transduced on DIV3 (days in vitro); each AAV in the concentration of 10⁹ of genome copies per well. 740/25 nm illumination (0.2 mW cm⁻², in alternating cycles: 30 s light and 180 s darkness) started on DIV6 and cells lysed on DIV9 for Fluc assay. 25 μM of BV was added to the culture medium on DIV3. Fluc assay was performed as described above.