Near-infrared manipulation of multiple neuronal populations via trichromatic upconversion

Using multi-color visible lights for independent optogenetic manipulation of multiple neuronal populations offers the ability for sophisticated brain functions and behavior dissection. To mitigate invasive fiber insertion, infrared light excitable upconversion nanoparticles (UCNPs) with deep tissue penetration have been implemented in optogenetics. However, due to the chromatic crosstalk induced by the multiple emission peaks, conventional UCNPs or their mixture cannot independently activate multiple targeted neuronal populations. Here, we report NIR multi-color optogenetics by the well-designed trichromatic UCNPs with excitation-specific luminescence. The blue, green and red color emissions can be separately tuned by switching excitation wavelength to match respective spectral profiles of optogenetic proteins ChR2, C1V1 and ChrimsonR, which enables selective activation of three distinct neuronal populations. Such stimulation with tunable intensity can not only activate distinct neuronal populations selectively, but also achieve transcranial selective modulation of the motion behavior of awake-mice, which opens up a possibility of multi-color upconversion optogenetics.

Note: The construction of trichromatic upconversion nanoparticle required multistep engineering.
Each step must be repeat several times to completely block ion diffusion and surface exchange during synthesis.
Synthesis of β-NaErF4 core nanoparticles. The synthesis of β-NaErF4 core nanoparticles with the size of ~ 22 nm were similar to the previously reported thermolysis method 1 . ErCl3 (1.00 mmol), OA (6.00 mL) and ODE (15.00 mL) were mixed together and heated to 140 °C under vacuum until formed a clear solution, after that, the solution was cooled down to room temperature. A solution of NaOH (2.50 mmol) and NH4F (4.00 mmol) in methanol (10.00 mL) was added. The resultant mixture was stirred for half an hour. The mixture was then heated to 70 °C and maintained for half an hour to remove the methanol. Afterward, the solution was heated to 290 °C and maintained for 100 min under a gentle argon flow. Then, the solution was cool down to room temperature. The nanoparticles were centrifuged and washed twice with ethanol and finally dispersed in 10 mL of cyclohexane for further use.
Synthesis of β-NaErF4@NaYF4 Core-shell nanoparticles. The fabrication of core-shell nanoparticles were carried out by the one-pot successive layer-by-layer (SLBL) protocol, which was developed by our group previously 2 . The shell precursor was firstly obtained from mixing the Y-OA (0.10 M, 1.00 mL) and Na-TFA-OA (0.40 M, 0.50 mL). Then, 2.50 mL of the purified NaErF4 core nanoparticles cyclohexane solution was mixed with 4.00 mL of OA and 6.00 mL of ODE. The flask was pumped down at 100 °C for 30 min to remove cyclohexane and any residual air. Subsequently, the system was switched to Ar flow and further heated to 270 °C at a rate of ~ 20 °C/min. Then the shell precursors were introduced by dropwise addition at 270 °C with the rate of 3.00 mL/h. The shell thickness can be well tuned by changing the amount of the shell precursors. After injection, the reaction was kept 30 min for ripening. Finally, the obtained NaErF4/NaYF4 (C/S1) nanoparticles were precipitated and washed several times with ethanol and re-dispersed in cyclohexane for further use.
The above coating step was repeated 4 times to make sure a sufficient thickness with complete NaYF4 at surface. Otherwise diffusion and ion exchange during synthesis process would leak the inner erbium ion into outer layers and even very few erbium leakage would lead to cross-response under 980 nm excitation.
The flask was pumped down at 100 °C for 30 min to remove cyclohexane and residual air. Subsequently, the system was switched to Ar flow and further heated to 270 °C at a rate of ~ 20 °C/min.
Then the shell precursors were introduced by dropwise addition at 270 °C with the rate of 3.00 mL/h.
The reaction were kept 30 min after injection for ripening. The final product were precipitated and washed several times with ethanol and re-dispersed in cyclohexane for further use.

Supplementary Note 1: Fabrication and characterization of the trichromatic UCNPs
The obtained UCNPs show uniform morphology, discernible contrast for core-multishell nanostructure and crystalline hexagonal phase without any significant impurity. Layer by layer epitaxial growth strategy was used for construction of the core-shell structured UNCPs. Fabrication process repeat at least twice in construction of each layer, which aimed at blocking the diffusion of ions between layers as well as for a sufficient dissipation shell thickness. The seemingly heterogeneous growth of CS4 is originated from large seed/precursor ratio and the misfit strain in core-shell construction which do not affect the intrinsic optical property 4 . Blue 438 nm-480 nm, Green 507 nm-560 nm, Red 641 nm-688 nm. Single particle spots were first searched on a cleansed coverslip dispersed with dilute trichromatic UCNPs solution under a max laser power density about ~500 W/cm -2 . Laser is then switched to low power density of ~180 mW/mm -2 for collecting luminescence images and scanning spectra ranging from 400 nm to 700 nm. Source data are provided as a Source Data file.
Images were obtained in a darkroom with up to 60 seconds exposure time. Channel intensity of three color were derived from a calculation of average integral intensity of all spot scanning spectra images.
Histograms of each color channel intensity from over 100 luminescent spots show well fitted Gaussian distribution. The non-steep intensity curve suggest that these results are most from single UCNPs. When increasing Yb 3+ concentration to 80% with Tm 3+ doping, the excited state transition 1 G4 → 3 F4 of Tm 3+ related to emission peaked at 645 nm is enhanced (Supplementary Fig. 8b). This band would mix with the emission band of Er 3+ peaked at 651 nm even under a low excitation power density, which Source data are provided as a Source Data file.

Supplementary Note 5: Comparison of involved three dissipation process.
Three cases of dissipation process are involved in trichromatic UCNPs. Case #1 discussed in previous reports is corresponding to blue/green emission color, which is related to the same sensitizer Yb. The responsive emission is mainly derived from higher amount of Yb 3+ in outer shell gaining stronger absorption. Case #2 discussed in the main text is corresponding to blue/red emission which shows that with larger absorption cross-section of Yb than Er at 980 nm, enhancing energy consumption will be much effective. We can deduce the transmitted ratio from the change of inner emission intensity with or without dissipation shell. The relationship of excitation power and inner emission intensity ratio is: where 1 , 0 are inner emission intensity with or without dissipation shell, 1 , 0 are the excitation power density with or without dissipation shell. For ~651 nm red emission band from Er, the number of sensitizing steps is ~2 (CR involved). From experimental results, a shell structure of 49%Yb1%Tm will give 56.4% emission suppression, which means 66% ( 1 / 0 ) of excitation intensity was left after penetration. Similarly, a structure of 80%Yb1%Tm gives 60% intensity left.
Case #3 discussed in Supplementary Figure 11 is corresponding to red/green emission, which is related to sensitizer Nd and Er. Because the absorption cross-section of Nd is much larger than Er at 808 nm, the dissipation efficiency of middle part is much higher than other process. Dissipation of 808 nm is related to sensitizers Nd and Er. Since the trichromatic UCNPs used activator  Figure 11c), it's supposed to be the reason that Tm is not the energy extractor of sensitizer. On steady state, Yb in shell3 and shell4 is the excitation energy extractor that affect equilibrium energy density of sensitizer Nd other than emissive center Tm. We fabricated UCNPs of the same core and histograms with mean diameter and standard deviation were obtained from the statistical results of more than 30 particles. g, Luminescence spectra with corresponding red/green ratio plot of the obtained UCNPs under 1532 nm excitation. All the nanoparticles were dispersed in cyclohexane for the 24 collection of emission spectra at an excitation power density of 5.0 W·cm -2 . Source data are provided as a Source Data file.

Supplementary Note 8:Coupling cross-relaxation and surface passivation for pure red emission.
In the NaGdF4:x%Er@NaYF4 (C1-2, x= 5, 15, 25, 50, 100) UCNPs with same core size and shell thickness, it is found that red emission band at ~ 654 nm was enhanced and red/green ratio increased For a simplified multiple-step energy transfer upconversion process, the time-resolved rate equation of energy states of sensitizers Yb 3+ can be derived as follows: . This result is consistent with our experiment of 980 nm emission lifetime in main text.
In the excitation of core-shell UCNPs, the unconverted photons will penetrate the outer layers and get into the inner core. The emission of Er 3+ is positively correlative with the incident photons.
The rate of photons that entering the core is: where 0 is the rate of incident photons and 1 is the rate of converted photons in outer layer. The energy absorbing rate is , where is the absorption cross-section and is the excitation power density. But only sensitizers in ground state of outer layer are able to convert the photons, so that the photon converting rate of sensitizer in outer layer is: It means that lowering the outer layer sensitizer activated population can increase the converting rate 1 under the same excitation power and subsequently reducing the photons approaching the inner core ( ).
Tm 3+ , as the energy consumer, is able to extract energy from Yb 3+ in the activated state allowing Yb 3+ returns to ground state. When increasing the excitation power , the activated state population will increase to the total population . Doping of Tm delays this saturation process by adding extra energy transfer routes. Consequently, at a specified excitation power density , the population density of active state is decreased with the increasing coefficient A thus reducing the rate of photons approaching to inner core .

Simulation curves of sensitizer activated state population as a function of excitation power density with different Tm doping concentration.
These results indicate that excitation energy distribution can be tuned with altering the component of dissipation shell. A specific amount of sensitizers with more activators results to the reduction of active state population density favoring the energy trapping process. More excitation energy transfers from sensitizers to emissive centers leads to higher dissipation efficiency. As a result, excitation photons reach to inner core is reduced thus suppressed the inner core's emission. Red arrows indicate UCNPs. One sample tissue was processed. The light intensity used in chromatic selectivity experiments were 0.44 mW/mm 2 at 470 nm, 0.07 mW/mm 2 at 530 nm and 0.83 mW/mm 2 at 650 nm at the brain surface. To prevent the chromatic cross-26 talk of different optogenetic proteins in ChrimsonR-and C1V1-expressing neurons, we measured the 27 upconversion efficiency and kept the intensity of the 980, 808 and 1532 nm NIR excitations below 28 89.1 mW/mm 2 , 86.4 mW/mm 2 and 76.8 mW/mm 2 at the brain surface (Methods). Note that the visible 29 light intensity used is limited to the LED power, the permissible intensity of visible light for chromatic 30 selectivity could be higher. The brain surface temperature is much higher than inner local temperature with a fluctuation during 43 light exposure (Supplementary Figure 23a,b). So that we use temperature recorded from the brain 44 surface as representative for the highest brain tissue temperature increase. For standard one-45 dimensional model, the heat source surface is regarded as a surface directly contact with the brain 46 (Supplementary Figure 23d). In such arrangement, temperature as a function of time and distance from the surface can be described with an additional term of heat flow induced by blood perfusion 5 where and are the specific heat capacity and thermal conductivity of the brain respectively. 50 and are blood flow rate and perivascular tissue thermal conductivity respectively. ∆ is the 51 temperature difference compared with normal tissue, which is that we're interested in. In steady state,  38.90 mW/mm 2 for 980 nm, temperature was detected at every 10 s, total duration = 120 s).