Improved genetically encoded near-infrared fluorescent calcium ion indicators for in vivo imaging

Near-infrared (NIR) genetically encoded calcium ion (Ca2+) indicators (GECIs) can provide advantages over visible wavelength fluorescent GECIs in terms of reduced phototoxicity, minimal spectral cross talk with visible light excitable optogenetic tools and fluorescent probes, and decreased scattering and absorption in mammalian tissues. Our previously reported NIR GECI, NIR-GECO1, has these advantages but also has several disadvantages including lower brightness and limited fluorescence response compared to state-of-the-art visible wavelength GECIs, when used for imaging of neuronal activity. Here, we report 2 improved NIR GECI variants, designated NIR-GECO2 and NIR-GECO2G, derived from NIR-GECO1. We characterized the performance of the new NIR GECIs in cultured cells, acute mouse brain slices, and Caenorhabditis elegans and Xenopus laevis in vivo. Our results demonstrate that NIR-GECO2 and NIR-GECO2G provide substantial improvements over NIR-GECO1 for imaging of neuronal Ca2+ dynamics.


Introduction
Fluorescence imaging of intracellular calcium ion (Ca 2+ ) transients using genetically-encoded Ca 2+ indicators (GECIs) is a powerful and effective technique to spectral profile, peak maxima, extinction coefficient, quantum yield, and pKa, NIR-GECO2 and NIR-GECO2G are essentially identical to NIR-GECO1 (Supplementary Table 1).
One of the most pronounced changes in the biophysical properties is that the Ca 2+ affinities of NIR-GECO2 and NIR-GECO2G are higher than that of NIR-GECO1 with Kd values of 331 nM and 480 nM, respectively (Kd of NIR-GECO1 is 885 nM) ( Supplementary Fig. 2a). A parallel effort to construct a second-generation NIR-GECO1 by replacing the mIFP portion with the brighter and homologous miRFP 12 resulted in the functional indicator prototype. However further optimization was abandoned due to the apparent toxicity of the miRFP-based construct when expressed in E. coli ( Supplementary Figs. 3,4).

Characterization of new NIR-GECO variants
To compare the intracellular baseline brightness (that is, fluorescence in a resting state neuron) of NIR-GECO2 and NIR-GECO2G with the previously reported NIR-GECO1 and the NIR FP miRFP720 (Ref. 13), we expressed each construct in cultured neurons and quantified the overall cellular brightness 5 days after transfection. To correct for cell-to-cell variations in protein expression, we co-expressed GFP stoichiometrically via a self-cleaving 2A peptide 14 to serve as an internal reference for expression level.
For small numbers of APs (2 to 10), the responses of NIR-GECO2G were 2.5-to 3.3-fold larger than those of NIR-GECO1 and the responses of NIR-GECO2 were 1.8-to 2.8-fold larger than those of NIR-GECO1. At higher numbers of APs (20 to 80), the improvements of the new variants became less pronounced as the ΔF/F0 values of the three variants converged (Fig. 1c). The on (rise time, peak; Fig. 1d) and off (decay time, 1/2; Fig. 1e) kinetics of NIR-GECO2 and NIR-GECO2G, in response to field stimulation-evoked APs stimuli, remained similar to that of NIR-GECO1.
To investigate if NIR-GECO2 and NIR-GECO2G provide advantages over NIR-GECO1 for combined use with an optogenetic actuator, we co-transfected HeLa cells with the genes encoding Opto-CRAC and each of the three NIR-GECO variants. Opto-CRAC is an optogenetic tool that can be used to induce Ca 2+ influx into non-excitatory cells when illuminated with blue light 16 . Transfected HeLa cells were illuminated with 470 nm light at a power of 1.9 mW/mm 2 while the NIR fluorescence intensity of NIR-GECO variants was continuously recorded. Following 100 ms of blue light stimulation, the average -ΔF/F0 for NIR-GECO2G, NIR-GECO2, and NIR-GECO1 was 34.5%, 22.8%, and 12.1%, respectively (Fig. 2a, d). With 500 ms of illumination, the -ΔF/F0 values increased to 48.2%, 42.1% and 30.7%, respectively. At 1 s of illumination time, the -ΔF/F0 was 40.7%, 38.0%, and 33.3%, respectively (Fig. 2a, b, c). When expressed in acute brain slices, both NIR-GECO2 and NIR-GECO2G robustly reported Ca 2+ changes in neurons in response to either optogenetic (CoChR) or chemical (4-aminopyridine) stimulation ( Supplementary Fig. 5). These results support the conclusion that both NIR-GECO2G and NIR-GECO2 are more sensitive than NIR-GECO1 for reporting Ca 2+ transients at low Ca 2+ concentrations, and NIR-GECO2G is the best of the three. To further explore the use of NIR-GECO2G, we expressed it in human induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs). In iPSC-CMs, NIR-GECO2G enabled robust imaging of spontaneous Ca 2+ oscillations, caffeine induced Ca 2+ influx, and channelrhodopsin-2 (ChR2)-dependent activation (Supplementary Fig. 6).

In vivo imaging of Ca 2+ in C. elegans using NIR-GECO2
To determine if NIR-GECO2 was suitable for in vivo imaging of neuronal activity, we first sought to test it in Caenorhabditis elegans, a popular model organism in neuroscience. For this application, we chose to use NIR-GECO2 rather than NIR-GECO2G due to its higher Ca 2+ affinity, however NIR-GECO2G could be also readily expressed in C.elegans in neurons producing sufficient near-infrared fluorescence (Supplementary Figure 7d). As the internal BV concentration of C. elegans is quite low due to its inability to synthesize heme de novo (its main source of heme is from the ingestion of E. coli) 17,18 , we decided to coexpress heme-oxygenase1 (HO1) to increase the conversion of heme into BV 19 . We created C. elegans lines expressing NLS-NIR-GECO2-T2A-HO1 (where NLS is a nuclear localization sequence) and NLS-jGCaMP7s under the pan-neuronal tag-168 promoter in an extrachromosomal array. The resulting transgenic worms exhibited bright nuclear localized fluorescence from both NIR-GECO2 and jGCaMP7s. One notable advantage of the NIR-GECO series relative to the GCaMP series of indicators was the lower auto-fluorescence in the intestinal area of worms in the NIR fluorescence channel, as compared to the green fluorescence channel (Fig. 3a,   Supplementary Fig. 7a,d). Microfluidic chips 20 were used to deliver a high-osmoticstrength stimulus (200 mM NaCl) to individual worms, and the fluorescence was imaged simultaneously in the NIR and green fluorescence channels. Following exposure to a high concentration of NaCl, we detected synchronous but opposing fluorescent changes for jGCaMP7s (fluorescence increases) and NIR-GECO2 (fluorescence decreases) ( Fig.   3b). Quantitative analysis of 36 spikes from 3 neurons showed that the -ΔF/F0 of NIR-GECO2 was about half of the ΔF/F0 of jGCaMP7s following NaCl stimulation (ΔF/F0 = 0.39 ± 0.19 for jGCaMP7s; -ΔF/F0 = 0.19 ± 0.07 for NIR-GECO2; Fig. 2c).
We next attempted all-optical stimulation and imaging of neuron activity in C. elegans using the blue-light sensitive channelrhodopsin CoChR 21 and NIR-GECO2. We previously demonstrated that excitation wavelengths used to image NIR-GECO1 do not activate CoChR 8 . NIR-GECO2 (with co-expression of HO1) was expressed in AVA interneurons (involved in backward locomotion) under the flp-18 promoter, and CoChR-GFP was expressed in upstream ASH neurons under control of the sra-6 promoter.
Imaging of transgenic worms with confocal microscopy revealed two AVA neurons with expression of NIR-GECO2 and two ASH neurons with expression of CoChR (Fig. 3d).
Blue-light stimulation of CoChR in ASH neurons caused long-lasting (tens of seconds to a few minutes) fluorescent decreases in NIR-GECO2 fluorescence (-ΔF/F0 of 30% to 90%) in the downstream AVA interneurons (Fig. 3e). Collectively, this data indicates that the combination of NIR-GECO2 and CoChR provides a robust all-optical method to interrogate hierarchical circuits in C. elegans.

In vivo imaging of Ca 2+ in Xenopus laevis using NIR-GECO2G
To further evaluate the utility of NIR-GECO2G for in vivo imaging of neuronal activity in a vertebrate brain, we transiently expressed the genes encoding NIR-GECO2G (without co-expression of HO1) and GCaMP6s 1 in Xenopus laevis tadpoles by mRNA injection into early embryos. Light-sheet microscopy imaging of the olfactory bulb of live tadpoles revealed that individual neurons exhibited strong near-infrared fluorescent signals due to NIR-GECO2G expression ( Fig. 4a,

b, Supplementary Video 1).
Fluorescence from both NIR-GECO2G and GCaMP6s was observed to oscillate in a synchronous but opposing manner in response to spontaneous neuronal activity (Fig.   4c). These results demonstrate that NIR-GECO2G can be used to report dynamic Ca 2+ changes in vivo in Xenopus laevis in the absence of added BV or HO1 co-expression.

Discussion
In summary, we have developed two improved NIR fluorescent Ca 2+ indicators designated NIR-GECO2 and NIR-GECO2G. Of the two, NIR-GECO2 has higher response amplitudes but dimmer fluorescence compared to NIR-GECO1, based on characterization in neurons and HeLa cells. In contrast, NIR-GECO2G is improved relative to NIR-GECO1 in terms of both overall cellular brightness (~50% brighter than NIR-GECO1) and sensitivity (up to a ~3.7 fold improvement in -ΔF/F0 relative to NIR-GECO1 for single AP). As we have demonstrated in this work, these improvements make the new variants particularly useful for imaging Ca 2+ dynamics in small model organisms.
Specifically, NIR-GECO2 offers comparable sensitivity to jGCaMP7s in C. elegans and NIR-GECO2G enables robust imaging of Ca 2+ dynamics in the olfactory bulb of Xenopus laevis tadpoles. However, even with these improvements, NIR GECIs still face challenges like relatively low brightness, slower kinetics, and faster photobleaching compared to the state-of-art green and red fluorescent GECIs. Overcoming these challenges will undoubtedly require further directed molecular evolution and optimization of the NIR-GECO series, or the possible development of alternative NIR GECI designs based on brighter and more photostable NIR FP scaffolds.

Mutagenesis and molecular cloning
Synthetic DNA oligonucleotides used for cloning and library construction were purchased from Integrated DNA Technologies. Random mutagenesis of NIR-GECO variants was performed using Taq DNA polymerase (New England BioLabs) with conditions that resulted in a mutation frequency of 1-2 mutations per 1,000 base pairs.
Gene fragments for NIR-GECO libraries were then inserted between restriction sites XhoI and HindIII of pcDuex2 for expression. The DNA sequences encoding miRFP1 to 172, CaM-RS20 (from NIR-GECO1), and miRFP179 to 311 were amplified by PCR amplification separately and then used as DNA templates for the assembly of miRFP1 to 172 -CaM-RS20 -miRFP179 to 311 by overlap extension PCR. The resulting DNA sequence was then digested and ligated into the pcDNA3.1(-) vector for mammalian expression and into a pBAD-MycHisC (Invitrogen) vector for bacterial expression. Q5 high-fidelity DNA polymerase (New England BioLabs) was used for routine PCR amplifications and overlap extension PCR. PCR products and products of restriction digests were routinely purified using preparative agarose gel electrophoresis followed by DNA isolation with the For the construction of Opto-CRAC-EYFP, a synthetic double-stranded DNA fragment consisting of fused EYFP, LOV2 and STIM1-CT fragments (residues 336-486) 16 , flanked with NotI and XhoI restriction sites, was cloned into the pcDNA3.1(-) vector.

Protein purification and in vitro characterization
The genes for the miRFP-based Ca 2+ indicators NIR-GECO1, NIR-GECO2 and NIR-GECO2G, with a poly-histidine tag on the C-terminus, were expressed from a pBAD-MycHisC (Invitrogen) vector containing the gene of cyanobacteria Synechocystis HO1 as previously described 22,23 . Bacteria were lysed with a cell disruptor (Constant Systems Ltd) and then centrifuged at 15,000g for 30 min, and proteins were purified by Ni-NTA affinity chromatography (Agarose Bead Technologies). The buffer wasexchanged to 10 mM MOPS, 100 mM KCl (pH 7.2) with centrifugal concentrators (GE Healthcare Life Sciences). The spectra of miRFP-based Ca 2+ indicator prototype, with and without Ca 2+ , were measured in a 384-well plate. Briefly, purified proteins were loaded into 384-well plates and then supplied with either 10 mM EGTA or 5 mM CaCl2 before measuring emission spectra. The extinction coefficients (EC), quantum yield (QY) and pKa for NIR-GECO variants were determined as previously described 8 . Ca 2+ titrations of NIR-GECO variants were performed with EGTA-buffered Ca 2+ solutions. We prepared buffers by
Following 2 hours incubation, the media was changed to DEME (Gibco Fisher Scientific) with 10% fetal bovine serum (FBS) (Sigma), 2 mM GlutaMax (Thermo Fisher Scientific) and 1% penicillin-streptomycin (Gibco) and the cells were incubated for 48 hours at 37 °C in a CO2 incubator before imaging. Prior to imaging, culture medium was changed to HBSS. Wide-field imaging was performed on a Nikon Eclipse Ti microscope that was equipped with a 75 W Nikon xenon lamp, a 16-bit 512SC QuantEM EMCCD (Photometrics), and a 60× objective and was driven by a NIS-Elements AR 4.20 software package (Nikon). For time-lapse imaging, HeLa cells were treated with 4 mM EGTA (with 5 µM ionomycin) and then 10 mM CaCl2 (with 5 µM ionomycin). Images were taken every 5 seconds using a filter set with 650/60 nm excitation and 720/60 nm emission.
Optical stimulation was achieved with the 470 nm LED light at a power density of 1.9 mW/mm 2 . Fluorescence signals were recorded by a CMOS camera (ORCA-Flash4.0LT, Hamamatsu) and controlled by a software (HC Image).

Imaging of NIR-GECO2G in Human iPSC-derived cardiomyocytes
Human iPSC-derived cardiomyocytes (human iPSC cardiomyocytes -male | ax2505) were purchased from Axol Bioscience. The 96 well glass-bottom plate was first coated with fibronectin and gelatin (0.5% and 0.1%, respectively) at 37 °C for at least 1 hour. The cells were then plated and cultured for three days in Axol's Cardiomyocyte Maintenance Medium. IPSC-CMs were then transfected with pcDNA3.1-NIR-GECO2 with or without pcDNA3.1-hChR2-EYFP using Lipofectamine 3000 (Invitrogen) following the manufacturer's instructions.The medium was switched to Tyrode's buffer right before imaging. Imaging was performed with an inverted microscope (D1, Zeiss) equipped with a 63× objective lens (NA 1.4, Zeiss) and a multi-wavelength LED light source (pE-4000, CoolLED) using the same settings described above.

Imaging of NIR-GECO1, NIR-GECO2, NIR-GECO2G and miRFP720 in cultured neurons
For dissociated hippocampal mouse neuron culture preparation, postnatal day 0 or 1 Swiss Webster mice (Taconic Biosciences, Albany, NY) were used as previously
Five voltage pulses (50 V, 50 ms duration, 1 Hz) were delivered using round plate electrodes (ECM™ 830 electroporator, Harvard Apparatus). Injected embryos were placed back into the dam, and allowed to mature to delivery. Acute brain slices were obtained from Swiss Webster (Taconic) mice at postnatal day (P) P11 to P22, using standard techniques. Mice were used without regard for sex.

Imaging of NIR-GECO2 in C. elegans.
Worms were cultured and maintained following standard protocols 25  imaged using the same optical setup as above, using a microfluidic device that was described previously 20 .

Imaging of NIR-GECO2G in Xenopus laevis tadpoles.
NIR-GECO2G and GCaMP6s were cloned into the pCS2+ vector and the plasmid was linearized with NotI. Capped mRNA of NIR-GECO2G and GCaMP6s was transcribed with the SP6 mMessageMachine Kit (Ambion, Thermo Fisher). The RNA (500 pg of each sample) was injected in one blastomere at the 2-cell stage resulting in animals expressing NIR-GECO2G and GCaMP6s protein in one lateral half of the animal. The animals were kept at 20°C until stage 47. Immediately before imaging the tadpole was paralyzed with pancuronium bromide (1.5 mg/mL in 0.1× MSBH) and embedded in 1% low-melt agarose. The imaging data (Fig. 4) was acquired with 50 ms integration time and 2.39 seconds cycle time through the volume. The instrument has a short deadtime to home the axial position leading to a scanning frequency of 3.0 s for the entire volume. The raw data was corrected for drift and rapid movement with the ImageJ plugin TurboReg 26 . The image in Fig. 4a is a volume projection of a 45 µm thick volume capturing the spontaneous Ca 2+ responses in the olfactory bulb. The cells were manually selected if they showed a spontaneous Ca 2+ response as detected with both NIR-GECO2G and GCAMP6s (Fig.   4b). The fluorescence response was measured as mean fluorescence intensity per cell and normalized by Inorm = (Im-Imin)/(Imax-Imin). Im indicates the measured mean value per area and Imax and Imin the maximal and minimal value measured for the specific ROI. The normalized response of NIR-GECO2G and GCAMP6s of a cell is plotted over time in the graph in Fig. 4c.

Data and image analysis
All images in the manuscript were processed and analyzed using either ImageJ (NIH) or NIS-Elements Advanced Research software (Nikon). Traces and graphs were generated using GraphPad prism 8, Origin (OriginLab) and Matlab. Data are presented as mean ± s.d. or mean ± s.e.m, as indicated.

Data availability
Plasmids used in this study and the data that support the findings of this study are available from the corresponding author on reasonable request.  for NIR-GECO2G, NIR-GECO2 and NIR-GECO1 in a-c.