www.nature.com/scientificreports OPEN Received: 4 August 2017 Accepted: 1 March 2018 Published: xx xx xxxx Tailoring light delivery for optogenetics by modal demultiplexing in tapered optical fibers Marco Pisanello1,2, Filippo Pisano1, Leonardo Sileo1, Emanuela Maglie1,2, Elisa Bellistri1, Barbara Spagnolo1, Gil Mandelbaum3, Bernardo L. Sabatini 3, Massimo De Vittorio1,2 & Ferruccio Pisanello 1 Optogenetic control of neural activity in deep brain regions ideally requires precise and flexible light delivery with non-invasive devices. To this end, Tapered Optical Fibers (TFs) represent a versatile tool that can deliver light over either large brain volumes or spatially confined sub-regions, while being sensibly smaller than flat-cleaved optical fibers. In this work, we report on the possibility of further extending light emission length along the taper in the range 0.4 mm-3.0 mm by increasing the numerical aperture of the TFs to NA = 0.66. We investigated the dependence between the input angle of light (θin) and the output position along the taper, finding that for θin > 10° this relationship is linear. This mode-division demultiplexing property of the taper was confirmed with a ray tracing model and characterized for 473 nm and 561 nm light in quasi-transparent solution and in brain slices, with the two wavelengths used to illuminate simultaneously two different regions of the brain using only one waveguide. The results presented in this manuscript can guide neuroscientists to design their optogenetic experiments on the base of this mode-division demultiplexing approach, providing a tool that potentially allow for dynamic targeting of regions with diverse extension, from the mouse VTA up to the macaque visual cortex. Over the past decade, optogenetic stimulation has had a remarkable impact on neuroscience research as it permits millisecond precision manipulation of genetically-targeted neural populations1–3. However, harnessing the full potential of the optogenetic approach requires light to be delivered to the opsin-transfected neuron under tight spatiotemporal control. This task is even more arduous when targeting brain regions beyond 1 mm depth, as common light-delivery methods (cleaved fiber optics and two-photon excitation) fail in penetrating such depths or are highly invasive. Driven by the experimental need for novel approaches, research in innovative technologies for deep brain light delivery has produced several promising solutions such as multi-dimensional wave-guides or high-density µLED probes4–20. In a recent work21 we demonstrated a simple and cost-effective device based on a thin Tapered Optical Fiber (TF) that can perform both homogeneous light delivery and dynamically-controlled spatially-restricted illumination over brain regions extending up to ~1.8 mm in the dorso-ventral direction. This was done by exploiting TFs light-emission properties that are determined by the subset of guided modes that propagates in the taper22. Experiments from two independent research groups have shown in the last year the great advantages of TFs for in vivo control of neural activity in different animal models. For example, wide-volume illumination was obtained in the motor cortex and in the striatum of both free-moving and head restrained mice21, as well as in the Frontal Eye Field of non-human primates23. TFs also allowed for site selective light delivery in the striatum to control specific locomotion behavior, on the base of different modal subsets injected into the fiber to illuminate ventral or dorsal striatum using only one implant21. 1Istituto Italiano di Tecnologia (IIT), Center for Biomolecular Nanotechnologies, 73010, Arnesano (LE), Italy. 2Dipartimento di Ingegneria dell’Innovazione, Università del Salento, Lecce, Italy. 3Department of Neurobiology, Howard Hughes Medical Institute, Harvard Medical School, Boston, 02115 MA, USA. Marco Pisanello and Filippo Pisano contributed equally to this work. Correspondence and requests for materials should be addressed to F.P. (email: ferruccio.pisanello@iit.it) Scientific RePorTS | (2018) 8:4467 | DOI:10.1038/s41598-018-22790-z 1 www.nature.com/scientificreports/ Building on this background, here we analyze the use of high-NA fibers to obtain wide volume or spatially-selective light-delivery over ~3 mm, a light emission length that is suitable to cover the extent of brain regions of several animal models, from mice to non-human primates. We demonstrate that a linear relationship exists between the position of the emission region along the taper and the input angle at which light is injected into the fiber. In addition, we characterize mode-division demultiplexing from 0.66NA TF at 473 nm and 561 nm in fluorescent solution and in mouse brain slices. Assessing TFs light-delivery performances at longer wavelengths is indeed a crucial aspect for using this technology with a broad set of opsins24–26. The versatility of TFs as customizable light-delivery tools may be seen as a promising resource for neuroscientists seeking to obtain bi-directional optogenetic modulation of neural activity in shallow or deep brain regions, with the results reported in this paper giving access to a set of quantitative characterization useful for experimental design with a wider range of opsins in multiple regions of the brain. Theoretical Background When injecting light in the whole fiber numerical aperture (NA) with a Gaussian beam, light-delivery devices based on TFs, schematically displayed in Fig. 1a, can illuminate large brain volumes by emitting radiation from a long segment of the taper surface21,23. This occurs because of a gradual loss of light along the taper generated by pthreoppargoagtrieosnsicvoennsatarrnotswkint ogfotfhtehgeuwidaevdegmuoiddeesdiianmcreetaesreas (fzro).mGitvheeninaitfiibalevrawluitehkitn(ait0i)aflodllioamwientgerthae0,rtehleattiroann2s7versal kt(a) ≈ a0 a kt(a0). (1) The can sbuephpaovriot rlioghf tsopmroepkatgvaatilounesoanslya function of a is shown for modes with27 in Fig. 1a as a reference for the reader. Since the taper kt(a) < kt, max = k0NA = 2π λ NA, (2) where λ is the propagating wavelength, each mode that does not fulfill this constraint is out-coupled. If the radia- tion injected into the condition in the tapered equation 2 region of the waveguide is composed by guided will be gradually broken along the taper. This misosdheoswwnitihnkFt iing.t1hae iwnhteerrveatlr[a0n, ksvt,meraxs]e, pmroodpeasgawtiitohnhciognhsetratnrtasn(scvoelrosraeldplrionpeas)gaotvioerntackoensthtaenktst,mteaxntdhtroesbheooldut(-dcaosuhpeldedlinate)la. rAgsera consequence, higher taper diameters. Low order order tmkot,omoadpxetasincwadliltsyhuislnotatweinrkfitanacgreeawihniitsghtheteahrdenoeuunmtv-icbroeournpomlfeemdncot.ldoFesoseliltsoowetxhipneegtcaStpenedyrdttoeiprp.erItotadfolu.l2cl7oe, wtthaseptehtoraettdaalsnneuoctpmitoibnceasrlwofiifbtgheuraisduloepndpgomerrotirdneeggsiao(nMlara(gbze)ler) for a section of diameter a(z) can be approximated as M(z) ≈ 1 2 π λ a(z)NA2 . (3) As a direct consequence, fibers with higher core size and higher NA can sustain a larger number of guided modes (Fig. 1b) and the number of guided modes diminishes when approaching the fiber tip (Fig. 1c). These two parameters are reported in Fig. 1b and c, respectively, as a support for the following description. Results Effects of NA and maximum transverse propagation constant on the emission properties of tapered optical fibers. Building on the well-known properties described in Section 2, we considered three optical fibers that support an increasing number of modes distributed over widening intervals of trans- vwNeiArthsT =eha ce00o .=e3nx9 s2teta0ann0ntd µtosamf0[t 0=ha,ne 2kdt0ta,0mcpoaeµxmr]mswpe21agi.trmhedesnitmht eeimmlaritwttiiantphgeplrirgaehnvtgi(olheuessr(erψeasf~tue3lrt.7sr°ef)of.errNrfieabdmerteoslywa,siwt“heemNiniAtvti e=nsgt 0igl.e2an2tgeadtnh”fd,ibEaeL0r =)sw w5a0ist µhmmNeaaAsnu =dre w0d.i6tah6s the distance from the tip at which the detected fluorescence decreases to half of the intensity at the fiber tip (see emission profiles displayed in Fig. 1d,e and detailed definition in Methods). For the same taper angle (ψ~3.7°), TFs with higher NA are optically active over a longer taper region (Fig. 1e), as relation 2 starts to be broken farther from the taper tip. In detail, 0.22 NA fibers show EL~484 µm, for 0.39 NA TFs EL~1220 µm, and EL~2038 µm for 0.66 NA. Data pertaining 0.22 NA and 0.39 NA fibers were reused from Pisanello et al.21, with the goal to com- pare higher NA TFs performance with known TFs. It is interesting to notice that NA = 0.66 TF have a shorter EL (1874 µm) when injected with 561 nm light (Fig. 1e, yellow curve), as expected due to the lower number of guided moAdeltsh(oaungdhlotwheernkutm,mabxe)raot fregdu-isdheidftemdowdaevsedleencrgetahsse(ssaese Fig. 1c). soon as the waveguide narrows (Fig. 1c) and relation (2) is broken just after the taper entrance the fluorescence measurements only for at a mFiorsdteEs mwiitshsiohnigDh ikatm(Feitge.r1(aF)E, lDig)httheamt disespioenndstsaortns to be appreciable in fiber NA and wave- guide size (Fig. 1f). This can be ascribed to the fact that high order modes have higher propagation losses with respect to low order modes and they are excited with lower power efficiency by a Gaussian beam focused on the fiber core27,28. As shown in Fig. 1f, fibers with large core size and large NA (e.g. supporting a larger number of fmiboedrsesinavnedstaighaitgehderhekrt,em,axt)hilesaldeatdoshtioghtheer values of the FED, which do observation that the higher not depend the NA, the on the higher taper angle ψ. For the ratio between the the FED and the fiber diameter at the taper entrance: 55/125 = 0.44 for NA = 0.22, 126/225 = 0.56 for NA = 0.39 and Scientific RePorTS | (2018) 8:4467 | DOI:10.1038/s41598-018-22790-z 2 www.nature.com/scientificreports/ Figure 1. Tailoring TF devices emission lengths. (a) (top) Schematic representation of a tapered optical fiber. (bottom) Evolution of the transversal propagation constant of four different guided modes as a function of the taper diameter. (b) Number of modes sustained by fiber optics with increasing core size and NA. (c) Number of guided modes at increasing diameters of the tapered section for fibers with 50 µm core/0.22 NA at 473 nm (red curve), 200 µm core/0.39 NA at 473 nm (green curve), and 200 µm core/0.66 NA at 473 nm (blue curve) and 561 nm (yellow curve). (d) Fluorescence image of light emission for a fiber with 50 µm core/0.22 NA (left), 200 µm core/0.39 NA (center), and 200 µm core/0.66 NA (right), ~3.7° taper angle, injected with 473 nm light; scale bars are 1 mm. (e) Normalized intensity profile measured along the taper surface for the three fibers in panel (d) injected with 473 nm light; the yellow curve represents the emission profile of a NA = 0.66 TF injected with 561 nm light. (f) First Emission Diameters (left) and Emission Lengths (right) for different fiber types with increasing taper angle. (g) Diagram of dorso-ventral depth of brain regions targetable with optogenetic stimulation in mouse, rat and macaque. The shaded areas in the background represent the dorso-ventral extension of each region as obtained from on-line brain atlas. Colored bars display the maximum emission length provided by TFs with NA = 0.22 (dark blue), NA = 0.39 (orange), and NA = 0.66 (pale blue). M1, primary motor cortex; HP hippocampus; CTX, cortex; STRd, dorsal striatum; VTA, ventral Tegmental Area; V1, primary visual cortex. For panels d–g, data pertaining 0.22 NA and 0.39 NA fibers were reused from Pisanello et al.21. Scientific RePorTS | (2018) 8:4467 | DOI:10.1038/s41598-018-22790-z 3 www.nature.com/scientificreports/ Figure 2. Spatially selective light delivery using high-NA TF. (a) Diagram of the optical setup used for mode division demultiplexing (L1, L2, L3 lenses; GM galvanometric mirror). Laser light enters the fiber patch cord at an angle and is out-coupled from the taper surface immersed in fluorescein. A sCMOS camera acquires fluorescence images of the light emission. The GM motion is synchronized with the camera acquisition via custom software (see Methods). (b) Left, emission images for a NA = 0.66, 200 µm core size fiber with increasing input angle. Scale bars are 250 µm. Center, diagram of the tapered section. The blue line indicates the profile along which the intensity was measured. Right, intensity profiles measured at different input angles with respect to the distance from the tip. Four selected regions are selected by varying the input angle. (c) Position of the intensity centroid of the emission region (blue circles) and positions of the FWHM values of the emission region (pale blue dots) with respect to the input angle for a NA = 0.66 TF. The red line indicates a linear fit performed on the centroid position. Ray tracing simulations results are displayed as black diamonds, with a linear fit shown as a black dashed line. (d) As in panel (c) for a NA = 0.39 TFs. (e) Far field images of the patch cord injected with θ = 10°, 20° at versus input angle. dashed blue line. 473 nm. Bottom, Scale bars are 0.15∙2π/λ. (f) centroid position versus the Top, centroid and boundaries of the injected kt for 0.66 NA, orange line, injected and 0.39 kNt-Asu, bset 154/230 = 0.67 for NA = 0.66. At the same time, we observed a diminished FED (130 µm) for NA = 0.66 TFs when injected with 561 nm light. This is that in turn translates to a smaller consistent with FED. In Fig. 1f, equations 1–2, data pertaining as a longer wavelength implies a smaller 0.22 NA and 0.39 NA fibers were reused fkrto,mmax Pisanello et al.21, with the goal to compare higher NA TFs performance with known TFs. Therefore, 0.66NA fibers are expected to support the longest ELs. Given the independence of FED on ψ, EL is expected to depend on the taper angle for a linear taper profile following the relation EL = 2 α ⋅ ⋅ FED tan(ψ/2) , (4) Scientific RePorTS | (2018) 8:4467 | DOI:10.1038/s41598-018-22790-z 4 www.nature.com/scientificreports/ Figure 3. Characterization of light coupling efficiency and output power density for NA = 0.66 TFs. (a) Light coupling across the NA = 0.66 angular acceptance. The input power (red line) was measured at the entrance of the tapered region. The output power (black line) was measures in close proximity to the taper tip, as shown in the inset. A drop in power output was observed for low input angles. (b) Power coupling efficiency, calculated as the ratio between output and input power, versus light input angle. The coupling efficiency is approximately choonmstoagnetnfeooruθsinp >o r1t0io°.n(co)f Scanning the taper electron microscope images region, scale bar 400 µm, the of a NA 0.66 TF. While the close-up in the right panel left panel shows a shows degradation of the taper surface in proximity of the tip, scale bar is 20 µm. (d) The three panels show the distribution of power density on the taper surface when 12 mW/mm2 are emitted As light is out-coupled from a wider region when the input afonrgltehrineecrienapsuets,atnhgeleins,pnuatmpoelwyeθrinh =as 1t5o°,b2e4v°a, r3i5e°d. accordingly to maintain a constant output power density. as shown by the good agreement with experimental data displayed Fig. 1f. The scale factor α was defined as the ratio between the EL and the FED distance from the taper tip for ψ = 3.7°, and was similar across NAs f(siαabmNeAre.=Iψ0n.6i6ps =atrh 0tei.c8ou6nl5ae2rt,,hαinaNtcArd=eia0s.ts3r9in i=bgu 0nt.eu8s9mt5he6eriawcnaidldaeαpsNteArr=taun0.r2ge2e =too 0f0.k8.6t8o64n,8a)tlh.loTewhweeiddfierbesetarcnahulilmnogwbeeinrmgoifsfosgiruotinhdleeedlnomgntghoedsseotsfe,~mi.3ei .smsthimoen,0af.go6a6riNtnhAset the ~1.8 mm obtained previously21. In light of these results, the FED can be identified as a chief design parameter for TFs, taking into account the effects on taper emission of several constitutive parameters of the fiber: the num- ber of guided modes that propagates in The relation linking TFs emission the taper length to and NA oanf tdhteairpkert values angle in relation (equation 4w)itahllkotw,msaxc. hoosing the correct device to match the size of targeted structures in the living brain. As schematically shown in Fig. 1g, probes with NA = 0.22 the ventral atengdmliemntiatel darceoar(eVsTizAe)(.aW0 =hi 5le00 µ.m22)NarAe/s5u0i µtemd to target regions in mouse core fibers can illuminate a cortex or deep nuclei such as relevant portion of the mouse hippocampus, devices with larger NA and core, such as 0.39 TFs with 200 µm core, are more appropriate for this task as they can be engineered to illuminate up to 1.8 mm in the dorso-ventral axis. These TFs can be applied to target both dorsal and ventral, for instance. However, this emission length cannot cover both a cortical-striatal Scientific RePorTS | (2018) 8:4467 | DOI:10.1038/s41598-018-22790-z 5 www.nature.com/scientificreports/ Figure 4. High NA TFs: light delivery at 561 nm and in brain tissue. (a) Spatially selective light delivery observed for a NA = 0.66, 200 µm core size fiber injected with 561 nm light and submerged in a PBS-eosin sinolpuatinoenl .(Sac).a(lec)bPaorssiatrioen20o0f µthme (b) Emission profiles measure along the taper surface for the intensity centroid of the emission region (orange circles) and input angles positions of shown the FWHM values of the emission region (pale orange dots) with respect to the input angle. The red line indicates a linear fit on the centroid position. (d) Large volume illumination obtained by injecting 473 nm light over the fiber full NA once inserted in mouse striatum; scale bar is 500 µm. (e) False color overlay of adjacent regions iclelunmtroinidatoefdtwheitehmspisastiioanllyresgeiloencti(vbeluliegchitrcdleelsi)vearnydapto4s7i3ti onnms,osfcathleebFaWr iHs 5M00v µalmue. s(fo)fPtohseiteimonisosfiotnhereingtioenns(iptyale blue dots) with respect to the input angle. (g) Large volume illumination obtained by injecting 561 nm light over the fiber full NA once inserted in mouse cortex and hippocampus; scale bar is 500 µm. (h) False color overlay of adjacent regions illuminated with spatially selective light delivery at 561 nm, the intensity centroid of the emission region (yellow circles) and positions of stchaeleFWbarHiMs 5v0a0l µumes.o(fgt)hPeoesmitiiosnsioonf region (yellow dots) with respect to the input angle. projecting region and the dorsal ventral axis of mouse striatum. In addition it cannot cover the full extent of rat motor cortex (M1) or the macaque visual cortex (V1) depth. The introduction of 0.66NA TFs is therefore essential to efficiently deliver light in several functional regions, as shown in Fig. 1g. These TFs can also potentially cover the full dorso-ventral extension of the above-mentioned regions and might be exploited to simultaneously reach cortical and subcortical regions in the mouse and the rat (Figs 1g, 4). The shaded areas in the background represent the dorso-ventral extension of each region as obtained from on-line brain atlases29–31. Linear control of TFs emission regions. As previously demonstrated, the light emitting portion of TFs can be dynamically controlled by remotely adjusting the light input angle21. In fact, coupling light into TFs with an angle-selective launching system selects well-defined subsets of bounded modes that propagate into the optical fiber, which are then out-coupled at different taper sections21. In the following the relationship between the input angle and the out-coupling position along the taper is quantitatively estimated. To this purpose the light redirection system displayed in Fig. 2a was implemented: a lens L1 focuses a Gaussian laser beam onto the rotation axis of a galvanometer mirror (GM). The beam is deflected by the GM of an angle θGM, collimated by lens L2 and focused onto the fiber by lens L3 at an angle θin. The taper was submerged in a Scientific RePorTS | (2018) 8:4467 | DOI:10.1038/s41598-018-22790-z 6 www.nature.com/scientificreports/ PBS:fluorescein bath to analyse the light emission geometry through an epi-fluorescence microscope as a func- taicoqnuiosfitθioinn. The GM deflection was synchronized with the of closely spaced emission profiles (~0.5° step in acquisition camera (see Methods) to speed up the input angle). 0.39 NA and 0.66 NA TFs with 200 µm core diameter were tested, both with ψ~3.7°. As discussed in the previous paragraph, 0.66 NA fibers sustain a high number of guided modes (~4.0 × 105) distributed over a large ensemble of transverse propagation constants i(nkt,Fmiagx .=2 b8..7A ×s s 1h0o−w3 nnmby−1thfoereλm =is s4i7o3n npmro)f.iRleespirnesthenetpatriovxeimimitaygeosfothf eliglihnteeamr tiasspieornsfuorrfafocuer(Fdiigff.e2rebn, rtiθginhat)r,eitdiisspploaysseid- ble to dynamically redirect light output over four different emission regions with equally spaced emission peaks. By extracting the centroid of the emission region (c) and the relative FWHM (Δc) from the emission profile at each input angle (definitions of c and Δc in Fig. 2b and in Methods), we found that c depends linearly on the input angle θ for θ > 10° (with a slope of ∼58 µm/°, fit RMSE of 11 µm) and can be dynamically moved along ~1.8 mm (Fig. 2c). Moreover, the FWHM was found to be approximately constant at Δc = 550 ± 5 µm for θ > 10°. This behaviour was confirmed by ray tracing simulations that, as displayed in Fig. 2c, highlight the linear relation between the position of the emission region and θ. Experimental and simulated data are in good agreement (Fig. 2c). For 0.39 NA fibers with the same taper angle, supporting a lower number of modes (~1.3 × 105 at 473 nm) and a narrower position of trhaencgeenotfrokitd(kslt,imgahxt =ly 5d.e2v ×ia 1te0s−f3ronmm−a1lifnoeraλr =de 4p7e3n dnemn)c,ec(csalonpbeefimt oofv∼ed50a µlomn/g°,~f1it.4R mMmSE, although the of 76 µm), as shown in Fig. 2d. Building on these measurements, we analyzed the relationships between guided modes and emission prop- erties of TFs with different numerical apertures. To confirm that the observed difference between geometrically atidhneegnlmeti.ecBtahlroi0e.d3fl9py,rNowApeoaisnnejddec0itn.e6dr6eNlfi.Ag22hTttoFomsniesaarpseualatrcteehtdhcteoorstduhbefisbdeetifsrfeoarfnekdntttienhxjeetnecntiemtdoafignkettodva0fal.u6re6fsiNesluAdstoaafnintdhe0de.3fbi9ybeNthrAeemtTwFiosssfifioobnrereosan,cwhaeiCnuCpseuDdt camera (Supplementary Figure 1). As shown in Fig. 2e, this produced light rings whose diameter is proportional ttfi0nooi.o6rptn6uh0to.eN3afiA9nnmapgTonlueFddt(aia0sslne.cs6egul6loMebsNsaeeeArntthdtwofoicbidtatheshnr)e.sabfTiaekbhrtlee(eiθrnar)tkle i±mespud oΔ(lFtstsoktigtoo(t.hfθv2)etefhrc,tleurbaapmornvptteseiovan, meswgur)esfroeedermpuxθrteer o>ntapoct 1atatg0ehrad°ee.titrFhoheoinpegrohoctreohutrneetprdssetuaafitnmrnartFiecnktiigigtn.v.jIed2eniifcanot(medtroddeepxetke),rt.ratvAnoaandslcudehleixagi,ptrhesaemtccFettWieemsrdsHiii,zsokseMnit(otθfhnfr)oeocvrmiaanenljautehcbechees- ttmhhoaenvAdeniddfoofeotrlhnoeewnartwioamritddtpeeeonrrrutmtaaaptoniedotrneaessxo2p7tfe,e2lcn8i.tgtTthbohtyieisvnviijarsetlcuvutiaeestidoebfalteththeideniihfntfhiefglreuheemennrteckθaetis,nmouaafrxseisnmhujipeegcnphttoesordrtdedkidsetprvblamyalyutoeheddesei0osn.n6hF6aTivgNFe.sA3hlaiiTgg(Fhhr.etedremactuitsersvnieuo)na:tbipoernfoopcreeornetnsiettasenris-t cienfufgircvtihee)ne.ctIaynptbeeerrt,ewslitegienhngt lptyho, eiwnteatrphseelrirgiahnntpgluyetdoaefncidrneptahuseet satanapsgelaerfofuounrtcpwtuihotinrcehomfcaθdiinnepasneandpdptshroelixnsiamemaaretlyehloyanpcpoθneinns(tsia.anet.t,tθhainse >stha 1po0ew°r)n,otiuhntepFuciogt u.(b3pbllai.ncAkg drop in coupling efficiency was observed for low injection angles. This effect was explained by SEM inspection, which revealed lower surface quality at the very tip of the taper (Fig. 3c). On the basis of these considerations, the output power density distribution around the taper for different input angles was assessed from the measured emission profiles (see Methods for details on the performed calculations). in the linear range are shown in Fig. 3d. The average power density can be Sample diagrams for maintained constant othvreerethdeiffdeirfefenrteθnint eomf tihsseioonptriecgailolynascbtyivselisguhrtflaycienocrfeeamsiinttginthgeretogtiaolnosuattphuitgphoewr edriaams aetfeurns.cFtioorntohfeθpina,rctoicmuplaerncsaastiensghfoowr nthieniFnicgr.e3asde, an increase of ~3 fold, roughly similar to the increase in optically active surface, holds the average power density ocopntosgtaennte(ti1c2c monWtr/oml omf2n)efuorraθlina =ct i1v5it°y, 24°, 35°. This is well above with ChR2 (1 mW/mm2)26 the widely recognized threshold and obtained with relatively low for obtaining input powers (ranging from 0.8 to 2.1 mW). Yellow and dual-wavelength light delivery with high-NA TFs in brain tissue. New optogenetic actuators have been developed in the last years, with activation peaks spanning the visible spectrum24–26. In par- ticular, optogenetic inhibition of neural activity has been demonstrated by delivering yellow light over a neural population transfected with inhibiting opsin probes, such as Halorhodopsin26. Moreover, a recent work used a TF-based device to achieve neuronal inactivation over a large volume in the non-human primate cortex by tar- geting a red light-sensitive halorhodopsin (Jaws) with 635 nm light23. Since both the number of guided modes and the width of 473, 561 nm and t6h3e5k nt mvalrueesspseucsttivaienlye,dsebey TFs decrease at higher wavelengths equations 2 and 3), it is important (M~4 × 105, 3 × 105, 2 × 105 to assess TFs performances for for site-selective optogenetic inhibition. Fluorescence emission profiles for angle-selective injection (Fig. 4a,b) were acquired in a water:eosin solution (see Methods) to test fiber response at 561 nm light. As observed for blue wave- lengths, we found that the position of the emission region centroid moves linearly as a function of the input angle for Iθnin o>r d1e0r° (Fig. 4c) to verify (fit slope ∼60 µm/° and RMSE 25 µm). that the emission features described so-far are not modified by tissue absorption and scatter- ing, light delivery performances were tested in brain tissue by inserting 0.66 NA TFs in fixed mouse brain slices previously stained with SYBR green and MitoTracker deep red to test fibers properties at 473 nm and 561 nm, respectively. Figure 4d shows the full NA emission in the mouse striatum with 473 nm light exciting SYBR green. Spatially selective light delivery was characterized by stimulating fluorescence emission from the taper while remotely varying the input angle with a step of 0.5°. Figure 4e shows a false color overlay of the fluorescent emis- sion excited over four adjacent brain regions with 473 nm light. Images 4d-e were acquired with a green fluores- cence filter (525/50). As for the characterization in fluorescein, we used the emission profiles measured along the Scientific RePorTS | (2018) 8:4467 | DOI:10.1038/s41598-018-22790-z 7 www.nature.com/scientificreports/ Figure 5. Simultaneous dual-wavelength mode-division demultiplexing in brain tissue with high NA TF. (a) Brightfield image of the location of a NA = 0.66 TF inserted in a mouse coronal brain slice with the tip reaching the red (b2) and hippocampus. Scale bar overlay (b3) channels of issim50u0lt µamneaonuds is common to all micrographs in the figure. (b) Green (b1), excitation of SYBR green fluorescence in the hippocampus and MitoTracker deep-red fluorescence in the cortex. Panel (b4) schematically shows the light injection configuration, with 473 nm entering at ∼4° and 561 nm entering at ∼20°. (c) Green (c1), red (c2) and overlay (c3) channels of simultaneous excitation of SYBR green fluorescence in the cortex and MitoTracker deep-red fluorescence in the hippocampus. Panel (c4) schematically shows the light injection configuration, with 473 nm entering at ~20° and 561 nm entering at ~5°. taper surface to extract centroid c and FWHM Δc of the emission region for each input angle. As shown in Fig. 4f, scattering from brain tissue slightly widens the distribution of the light emitted for a given modal subset, but the due overall to the nboenh-ahvoiomr oogfecnvesitθyinoifstnisostuaefsfeccattetedr,iwngit.hWseomtheendepveiraftoiornmsefdroamn the linearity measured in fluorescein analogous measurement by inserting a NA = 0.66 TF of similar taper angle in a brain slice stained with MitoTracker deep red. Figure 4g shows the full NA emission distributed between cortex and hippocampus with 561 nm light. Afterwards, we repeated the evaluation of site selective mode-division demultiplexing. Figure 4h shows a false color overlay of the fluorescent emission excited over four adjacent brain regions with 561 nm light. Images 4g-h were acquired with a red fluo- rescent filter (716/40). The position of the centroid c and FWHM Δc were extracted and reported in the graph in Fig. 4i. Also in this case, a wider Δc was observed due to tissue scattering, with an almost-preserved linear behSaivmiourltaasnaeofuunscmtioodneo-fdθivini.sion demultiplexing at 473 nm and 561 nm was demonstrated by independently rout- ing two laser beams into a TF inserted in a coronal mouse brain slice co-stained with SYBR-green and MitoTracker deep-red. Figure 5a shows a NA = 0.66 TF inserted in a coronal brain slice with the tip reaching the hippocampus. We simultaneously confined light emission from SYBR-green in the hippocampus and MitoTracker deep-red in the cortex (Fig. 5b), and then swapped the excitations (Fig. 5c). Red and green channels images were acquired by changing the flu- orescence filters during simultaneous excitation at 473 nm and 561 nm (see Methods). Figures 5b3–c3 show overlays of the two channels, respectively acquired for each configuration, as schematically depicted in fig. 5b4–c4. Supplementary Figure 3 shows co-localized light stimulation at 473 nm and 561 nm both in the cortex (Supplementary Figure 3b) and Scientific RePorTS | (2018) 8:4467 | DOI:10.1038/s41598-018-22790-z 8 www.nature.com/scientificreports/ the hippocampus (Supplementary Figure 3c). These measurements confirm that a single TF device can perform independent dual-wavelength light delivery over two functionally distinct brain regions. Discussion Delivering light in a controlled fashion is required in order to fully exploit the potential of optogenetic techniques in controlling and monitoring neural circuits32. To this end, light delivery based on TFs offers an innovative tool to reach deep brain regions that are precluded to ordinary optogenetic stimulation protocols. By using 0.22, 0.39 and 0.66 TFs the possibility of tailoring emission length of TFs devices is extended to the interval ~0.4 mm to ~3 mm, with a simple relationship (equation 4) to define taper angle and fiber NA. This tuning range well matches with the extent of brain structures of different animals including mice, rats and non-human primates, as shown in Fig. 1g. mroeuvetenpBatuyloteidfpntojahewcleitendirneirargeerdmcdteeafrpiienneleasntdadiolesmnunsbcohessietfptocsrobonienftswpgtauuenteitdna,enwtdghhlmeeilsleoiθgdtinhhe >tes owe 1um0itt°pihtfuitonetrdpsbpopoesotichwtiifeobircnludkaeetlanoinnnstdigetyrytvhesallelilogstw,hafptalleyirgrdfhaieetn.clddWretaaihntsheeaislinyndsjtuiehsceeattelloildnotewkhateervdiranemlucgrieeoesaan.ssTu,ethrhoeies-f waveguide diameter when the emitting region moves farther from the taper tip. TFs represent a unique approach to deliver light into the living brain with reduced invasiveness, as they allow for site-selective stimulation over wide volumes without the implantation of electric devices or of multiple waveguides. Although uncoated TFs cannot produce an illumination spot as tight as microfabricated TFs22,33 or cylindrical fibers34, they provide the user with dynamic re-configuration between wide-volume and site-selective light emission. Moreover, there is no need to determine the illumination area a priori, as light emission can be moved continuously across the entire optically active range21. Therefore, we view these approaches as complementary rather than exclusive. All these features let us envision that the here-presented results can help neuroscientists in improving experimental protocols targeting sub-cortical regions of the mouse brain, functional structures of larger rodents and non-human primates, as well as analysis methods not compatible with conducting metals (such as opto-fMRI), placing TFs as a great complement to available methods to deliver light into brain. Materials and Methods Device structure and fabrication process. Tapered optical fibers with taper angles ranging from 2° to 8°, fabricated from fiber cords with NA = 0.22 core/cladding = 50/125 µm (Thorlabs FG050UGA), NA = 0.39 core/ cladding = 200/225 µm (Thorlabs FT200UMT), NA = 0.66 core/cladding = 200/230 µm (Plexon PlexBright High Performace patch cable) were obtained from OptogeniX (www.optogenix.com). Tapers were fabricated by heat- afonrdc-eptuelml mpoertahlopdatutseirnnghaalvaesebreepnuloleprt.iOmuiztepduttopoowbtearinofththeedCesOir2eldastearp,esrcashnanpinesg. length over the fiber, and pulling Details of the procedure are pro- vided in previous works21,35. Fibers were connected to ceramic or metallic ferrules with a diameter of 1.25 mm. Supplementary Figure 4 shows cross section images of the tapered region for NA = 0.66 and NA 0.39 fibers. These were obtained by cutting and polishing the 0.39 and 0.66 TFs at diameters of ~85 µm and ~125 µm respec- tively with the aid of a 127 µm bare ferrule. Interestingly, while no sign of polymeric cladding was observed for NA = 0.39 fibers, the boro-silicate cladding in NA = 0.66 TFs can be distinguished. The ratio between the core and cladding diameter at the investigated section (110/125 = 0.88) and at the input facet of the fiber (200/230 = 0.87) is constant, suggesting that core and cladding reduce proportionally in the NA = 0.66 fibers used. Emission properties characterization. Data acquisition. The emission properties of the devices were characterized with two different light coupling systems: (i) injection of the whole numerical aperture accepted by the fiber and (ii) injection of a defined 473) and 561 nm (Coherent OBIS 561 nm iLnSp)u. tInanbgoltehθsiny.sLteamsesr, light was the taper injected at 473 nm (Laser Quantum Ciel was coupled to a patch fiber with match- ing NA and core size by a ferrule to ferrule butt-coupling. In the full NA injection configuration of case (i) light was coupled to the fiber patch cord with an Olympus objective AMEP-4625 (focal length 4.5 mm, N.A. = 0.65), or with fiberports (Thorlabs PAF-SMA-5-A focal length 4.6 mm, N.A. = 0.47, PAF-SMA-7-A focal length 7.5 mm, N.A. = 0.29). To fill the entire clear aperture of the coupling lens, the laser beam was expanded by a proper factor through a beam expander realized with off-the-shelf lenses. The angle-selective launch system of case (ii) was implemented using a galvanometric-mirror based scanning l(seSynusgttettemhrf.R2T =EwS 1oS0Cl0eA mnNsme-sG, LrEe1Nspa)enicdnttiLvoe2aly(dT)ihrsepollraalycaeebdms tLehnAet1lta0ps5ee0rr-pbAeenawdmiitchaunlfadorcctaoolntlvehenergtotephdtiftc1h a=el aa1xn0ig0su molfamrthdeaensflydesctAteimLo5n.0Lo1ef0nt0hs-eLA3Gw(MTithhmofriorlarcboasrl aAcLtu4a5l3i2n-pAuwt aitnhglfeocθawl laesnmgtehafs3u(cr)e =d 3b2y mremgisatnedrinag4t5h me mdisapplaecrteumree)nftoocfutsheedltahseerlisgphottoonntoa the patch fiber digital camera core. The placed in front of the coupling lens for a known mirror deflection. The tapered fibers were immersed in a PBS:fluorescein or PBS:eosin solution to image light emission patterns with input wavelengths of 473 nm or 561 nm, respec- tively. Images were acquired using an upright epi-fluorescence upright microscope (Scientifica Slicescope) equipped with a 4 × objective (Olympus XLFLUOR4X/340 with immersion cap XL-CAP), fluorescence emis- sion filters (525/50 for fluorescein emission and 605/70 for eosin emission) and a sCMOS camera (Hamamatsu ORCA-Flash4.0 V2). Optical output power was measured in air with a power meter (Thorlabs PM100USB with S120VC sensor head). Power coupling efficiency was measured as the ratio between taper and patch fiber optical power output. Tapered fiber emission in brain slices was measured by inserting the light emitting region of the taper in coronal mouse brain slices fixed in PFA and stained with SYBR green (S9430, Sigma Aldrich) or both SYBR green and Mitotracker deep-red (M22426, Thermo Fisher Scientific). Excitation laser light at 473 nm and/or Scientific RePorTS | (2018) 8:4467 | DOI:10.1038/s41598-018-22790-z 9 www.nature.com/scientificreports/ 561 nm was coupled with system (i) and (ii) into the tapered fiber. Detection of fluorescence was performed through an epi-fluorescence upright microscope (Scientifica Slicescope) equipped with a 4 × objective (Olympus XLFLUOR4X/340 with immersion cap XL-CAP) and a sCMOS camera (Hamamatsu ORCA-Flash4.0 V2). Simultaneous light injection at 473 nm and 561 nm was obtained by combining two independent scan path in front of lens L3 using a beam splitting cube mounted on a translation stage (Supplementary Figure 5). Images were acquired by subsequently filtering SYBR-green and MitoTracker deep-red emission with fluorescence filters (525/50 and 716/40 respectively). Supplementary Figure 6 shows a proposed setup to independently perform simultaneous mode division demultiplexing with three different wavelengths by combining three different galvanometric scanners by means of dichroic mirrors. Data analysis. A set of intensity profiles was extracted from the fluorescence images collected with coupling scheme (i) and (ii). Profiles were measured along a line drawn just outside the taper, parallel to the waveguide surface. Profiles given by full N.A. excitation were used to quantify both the First Emission Diameter (FED) and the total length of the emission region (Emission Length, EL). The FED was defined as the position at which the intensity recorded by the sCMOS sensor is half of the average intensity detected in the pixel with a recorded intensity exceeding 90% of the maximum recorded intensity (Supplementary Figure 2). The same intensity profiles were used to estimate the power density distribution along the taper when the coupling scheme (ii) was used. Assuming a rotational symmetry around the taper axis of the power density distribution p, a Matlab code has been developed to calculate p starting from the total output power and the intensity profile extracted from the image. Spatial selectivity in light out-coupling from the taper was quantified from the profiles measured with angle-selective excitation. The centroid and Full Width at Half Maximum (FWHM) positions of the emission region were measured for each input angle θ with respect to the fiber tip. Linear fits as function of the input angle θ were computed on the centroid position and both the FWHM boundaries positions. Data points for which the lower bound of the emission region was coincident with the fiber tip were excluded from the fit computation because the presence of the taper tip affects these measurements by breaking the shape symmetry. Ionpjteiccatlefidbekrtams aefausnucrteiomn eofntthse. inpTuhteavnagllueewoafstmheetarsaunrsevdebrsyailmpraogpinaggatthieonfacrofinelsdtapnattktetronf the light guided by the of the light outcoupled by the facet of an optical fiber stub while changing θ at the input facet, as previously described22,36. In particular, Olympus objective AMEP-4625 (focal length 4.5 mm, N.A. = 0.65) was used to generate the farfield pattern, and a two-lens imaging system (realized with Thorlabs AC254-040-A-ML and AL50100-A) was used to magnify the pattern over the chip of the sCMOS sensor (Hamamatsu ORCA-Flash4.0 v2). The signal detected at distance r from the center of the chip is associated to kt by the equation r = 4.5 mm ⋅ 100.0 40.0 mm mm tansin−1 λ 2π kt. (5) For each input were extracted afrnogmle,ththeedkist ka-t which the maximum signal is detected or ring-shaped images recorded by the and the sCMOS half-prominence sensor. width of the peak Ray tracing simulations. Ray tracing simulations were performed with the commercial optical ray tracing software Zemax-OpticStudio (http://zemax.com). 0.66NA TFs were modeled as straight core/cladding nested cylinders followed by a conical taper section with 3.7° angle. Core/cladding refractive indexes were set as core = 1.63 and cladding = 1.49 as retrieved from ref.37. Tapered sections were modeled by nesting two cones with a common vertex. Core/cladding diameters were set as the manufacturer nominal values of 200/230 µm. The length of the core/cladding non-tapered segment was set to 50 mm. The source (5 M rays) was modeled as a circular homogeneous light distribution coupled in the fiber with an ideal paraxial lens. Different input angles were simulated by modulating the tilt of the source-lens system with respect to the fiber axis. The irradiance profiles of the emitted light were detected by placing a rectangular, pixelated detector (6000 × 41) in close proximity (~50 µm) of the taper surface. Data availability. Data are available from the corresponding author on reasonable request. References 1. Häusser, M. Optogenetics: the age of light. Nat. Methods 11, 1012–1014 (2014). 2. Deisseroth, K. Optogenetics. Nat. Methods 8, 26–29 (2011). 3. Miesenböck, G. The Optogenetic Catechism. Science 326, 395–399 (2009). 4. Scharf, R. et al. Depth-specific optogenetic control in vivo with a scalable, high-density μLED neural probe. Sci. Rep. 6, 28381 (2016). 5. Warden, M. R., Cardin, J. A. & Deisseroth, K. Optical Neural Interfaces. Annu. Rev. Biomed. Eng. 16, 103–129 (2014). 6. Kim, T.-I. et al. Injectable, Cellular-Scale Optoelectronics with Applications for Wireless Optogenetics. Science 340, 211–216 (2013). 7. Jeong, J.-W. et al. Wireless Optofluidic Systems for Programmable In?Vivo Pharmacology and Optogenetics. Cell 162, 662–674 (2015). 8. McCall, J. G. et al. Fabrication and application of flexible, multimodal light-emitting devices for wireless optogenetics. Nat. Protoc. 8, 2413–2428 (2013). 9. Park, S. I. et al. Soft, stretchable, fully implantable miniaturized optoelectronic systems for wireless optogenetics. Nat. Biotechnol. 33, 1280–1286 (2015). 10. Park, S. I. et al. Stretchable multichannel antennas in soft wireless optoelectronic implants for optogenetics. Proc. Natl. Acad. Sci. 113, E8169–E8177 (2016). 11. Goßler, C. et al. GaN-based micro-LED arrays on flexible substrates for optical cochlear implants. J. Phys. Appl. Phys. 47, 205401 (2014). 12. Wu, F. et al. Monolithically Integrated µLEDs on Silicon Neural Probes for High-Resolution Optogenetic Studies in Behaving Animals. Neuron 88, 1136–1148 (2015). Scientific RePorTS | (2018) 8:4467 | DOI:10.1038/s41598-018-22790-z 10 www.nature.com/scientificreports/ 13. Wu, F. et al. An implantable neural probe with monolithically integrated dielectric waveguide and recording electrodes for optogenetics applications. J. Neural Eng. 10, 056012 (2013). 14. Zorzos, A. N., Scholvin, J., Boyden, E. S. & Fonstad, C. G. Three-dimensional multiwaveguide probe array for light delivery to distributed brain circuits. Opt. Lett. 37, 4841 (2012). 15. Lee, J., Ozden, I., Song, Y.-K. & Nurmikko, A. V. Transparent intracortical microprobe array for simultaneous spatiotemporal optical stimulation and multichannel electrical recording. Nat. Methods 12, 1157–1162 (2015). 16. Segev, E. et al. Patterned photostimulation via visible-wavelength photonic probes for deep brain optogenetics. Neurophotonics 4, 011002 (2016). 17. Canales, A. et al. Multifunctional fibers for simultaneous optical, electrical and chemical interrogation of neural circuits in vivo. Nat. Biotechnol. 33, 277–284 (2015). 18. Park, S. et al. One-step optogenetics with multifunctional flexible polymer fibers. Nat. Neurosci. 20, 612–619 (2017). 19. Pisanello, F., Sileo, L. & De Vittorio, M. Micro- and Nanotechnologies for Optical Neural Interfaces. Front. Neurosci. 10, (2016). 20. Cho, Y. K. et al. Roadmap on neurophotonics. J. Opt. 18, 093007 (2016). 21. Pisanello, F. et al. Dynamic illumination of spatially restricted or large brain volumes via a single tapered optical fiber. Nat. Neurosci.. https://doi.org/10.1038/nn.4591 (2017). 22. Pisanello, M. et al. Modal demultiplexing properties of tapered and nanostructured optical fibers for in vivo optogenetic control of neural activity. Biomed. Opt. Express 6, 4014 (2015). 23. Acker, L., Pino, E. N., Boyden, E. S. & Desimone, R. FEF inactivation with improved optogenetic methods. Proc. Natl. Acad. Sci. 113, E7297–E7306 (2016). 24. Fenno, L., Yizhar, O. & Deisseroth, K. The Development and Application of Optogenetics. Annu. Rev. Neurosci. 34, 389–412 (2011). 25. Guru, A., Post, R. J., Ho, Y.-Y. & Warden, M. R. Making Sense of Optogenetics. Int. J. Neuropsychopharmacol. 18, pyv079 (2015). 26. Yizhar, O., Fenno, L. E., Davidson, T. J., Mogri, M. & Deisseroth, K. Optogenetics in Neural Systems. Neuron 71, 9–34 (2011). 27. Snyder, A. W. & Love, J. Optical waveguide theory. (Springer Science & Business Media, 2012). 28. Botta, E. & Bressani, T. Elementi di elettromagnetismo avanzato. (Aracne editore, 2009). 29. Lein, E. S. et al. Genome-wide atlas of gene expression in the adult mouse brain. Nature 445, 168–176 (2007). 30. Paxinos, G. & Watson, C. The rat brain in stereotaxic coordinates. (Elsevier, 2007). 31. Mikula, S., Trotts, I., Stone, J. M. & Jones, E. G. Internet-enabled high-resolution brain mapping and virtual microscopy. NeuroImage 35, 9–15 (2007). 32. Packer, A. M., Roska, B. & Häusser, M. Targeting neurons and photons for optogenetics. Nat. Neurosci. 16, 805–815 (2013). 33. Pisanello, F. et al. Multipoint-Emitting Optical Fibers for Spatially Addressable In Vivo Optogenetics. Neuron 82, 1245–1254 (2014). 34. Nguyen, H. et al. Fabrication of multipoint side-firing optical fiber by laser micro-ablation. Opt. Lett. 42, 1808 (2017). 35. Sileo, L., Pisanello, M., De Vittorio, M. & Pisanello, F. Fabrication of multipoint light emitting optical fibers for optogenetics. in (eds Hirschberg, H. et al.) 93052O, https://doi.org/10.1117/12.2075819 (2015). 36. Della Patria, A., Pisanello, M., Sileo, L., De Vittorio, M. & Pisanello, F. Influence of laser beam quality on modal selection in tapered optical fibers for multipoint optogenetic control of neural activity. In 1–4 (IEEE, https://doi.org/10.1109/ICTON.2016.7550329 2016). 37. Refractive index database. Available at http://www.filmetrics.com/refractive-index-database (2016). Acknowledgements The authors thank Andrea Della Patria for providing guidance and useful insights with Ray Tracing simulations. F. Pisanello, F. Pisano, E.B. and E.M acknowledge funding from the European Research Council under the European Union’s Horizon 2020 research and innovation program (#677683); B.S. and M.D.V. acknowledge funding from the European Research Council under the European Union’s Horizon 2020 research and innovation program (#692943). L.S., M.P., M.D.V. and B.L.S. are funded by the US National Institutes of Health (U01NS094190); G.M. and B.L.S. are funded by the Simons Collaboration on the Global Brain. Author Contributions M.P. and F. Pisano equally contributed to this work. M.P. and F. Pisano developed the experimental setup, performed the experiments and analyzed the data. L.S. engineered the devices. E.M., M.P., F. Pisano and F. Pisanello implemented ray tracing simulations. E.M., E.B. and B.S. stained brain slices. M.P., F. Pisano, L.S., G.M., B.L.S., M.D.V. and F. Pisanello developed and optimized the tapered fiber design. B.L.S, M.D.V. and F. Pisanello conceived and supervised the research. M.P., F. Pisano and F. Pisanello wrote the manuscript with contributions from all authors. Additional Information Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-018-22790-z. Competing Interests: M.D.V., F. Pisanello, B.L.S., and L.S. are co-founders of Optogenix LLC, a company based in Italy that produces and markets the tapered fibers described here. Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2018 Scientific RePorTS | (2018) 8:4467 | DOI:10.1038/s41598-018-22790-z 11