Modulation of fluorescence signals from biomolecules along ...
Supplementary information
Modulation of fluorescence signals from biomolecules along nanowires due to interaction of light with oriented nanostructures
Rune S. Frederiksen1, Esther Alarcon-Llado2, Morten H. Madsen3, Katrine R. Rostgaard1, Peter Krogstrup4, Tom Vosch5, Jesper Nygård4, Anna Fontcuberta i Morral2, Karen L. Martinez1
Material and Methods InAs NW array fabrication. NWs were grown by the gold-assisted vapor–liquid–solid growth mechanism using molecular beam epitaxy (MBE) and the gold catalyst particles were defined using electron beam lithography (EBL) on InAs (111)B substrate as previously described 1. Selfcatalysed NWs were grown on silicon2, see SI below. NW functionalization with BSA-biotin and SA. NW arrays were functionalized with 0.1mg/ml BSA-biotin (Sigma-Aldrich) and 1 M Alexa Fluor 488/555/633 streptavidin conjugate (SA488, SA-555, SA-633) in 0.01 M PBS buffer (Life Technologies) as described in Rostgaard et al.3 Cell data. HEK293 cells have been interfaced with arrays of InAs NWs and fluorescently labeled with a Dyomics dye DY-547 (ex546 nm: em560-600 nm) upon interface as described by Berthing et al.4 Fluorescence imaging. CLSM imaging was performed on an inverted confocal laser-scanning microscope (Leica TCS SP5) using a 63x magnification, water-immersion objective with 1.2
1
numerical aperture. The fluorophores were excited and the emission was collected by a photomultiplier tube at following excitation: emission wavelengths: ex488: em508–600 nm (SA488), ex546 nm: em560-600 nm, (SA-555) and ex633nm: em650-800nm (SA-633). Complying with the Nyquist rate for optimal imaging conditions, Z-stacks were collected with a pinhole of one airy unit, and a pixel- and step size of approximately 50 x 50 nm2 and 170 nm respectively. Quantification of fluorescence signals. Quantification of fluorescence signals was performed with ImageJ software. In case of the Z-stack profiles, the NW signals were measured with ROI with 10 pixels around each NW. The background signal, measured as the signal from the buffer solution above the NWs, was subtracted. In case of the line scans of horizontal NWs, they were measured with a 5-pixel average. Time-correlated single photon counting experiments. The experiments were carried out on a home-built scanning fluorescence confocal microscopy system based on an Olympus IX71 inverted microscope. A piezo-driven scanning stage (Physik Intrumente P5173CL), allowed for imaging the sample point by point in a raster scanning fashion in a range up to 100 µm x 100 µm. Upon laser illumination, the fluorescence emission signal from the sample was collected by a 100x oil immersion objective (Olympus UPLFLN 100x), and directed to an avalanche photodiode (APD, Perkin Elmer CD3226) for recording the fluorescence intensity and for performing time-correlated single photon counting experiments. The laser light was blocked in the emission path by appropriate long-pass filters (Semrock LP02-488RU-25). A 480 nm pulsed laser (Picoquant LDH-P-C-470) was used as excitation source. Excitation was performed with linear polarized light and no polarizers were inserted before the APD. SEM Imaging. To evaluate the protein density along the InAs NW, we used InAs NWs
2
functionalized with BSA-biotin (as described above) and evaluated by SEM the distribution of streptavidin fused to 20nm SA-NP (Invitrogen) along the NWs. The InAs NW was firstly premodified with BSA-biotin incubation secondly modified with 0.04 nmol/L (SA-NP) for 30 min, followed by incubation overnight in vacuum chamber and finaly imaged in a SEM (JEOL JSM- 6320F). Simulations The InAs NW was set to 90nm in diameter and 4.5 m long vertically standing on InAs substrate. A gold semi-sphere was included at the NW tip and the whole system was embedded in water. Simulations were performed with Meep (http://ab-initio.mit.edu/wiki/index.php/Meep), a free finite-difference time-domain (FDTD) software package to model electromagnetic systems. For simulating the excitation field distribution profiles, a plane monochromatic wave was used as light source propagating in the vertical direction. Perfectly matching layers around the cell were included to avoid boundary conditions effects. Simulations were performed at a 5 nm distance from the NW to account for the size of the intermediate protein layer (BSA-biotin) onto which fluorescent SA is bound. Self-catalyzed NWs. Gold NPs are known to amplify fluorescence signals5. To evaluate if the gold catalyst particle present at the NW tip causes the alterations in the fluorescence profile, we studied the Z-profiles from SA-488 immobilized on self-catalyzed NWs (without the gold-NP tip). The self-catalyzed InAs NWs were grown on a Si(111) wafer by Molecular Beam Epitaxy using a thin native oxide layer to initiate the growth, see2 for details on the growth method. The gold-NP does not seem to be the main source of the tip effect, as the tip effect is still present on the self-catalyzed NWs, as seen in Figure S1.
3
7
Z-stack (um)
6 5 4 3 2 1
0
50
100 150 200
Fluroescence Intensity (a.u.)
Figure S1: Self-catalyzed NWs functionalized with BSA-biotin and SA-488
Fluorescence photobleaching along the InAs NW To evaluate if the fluorescence modification originates from modification of the excitation light flux along the NW, we studied the photobleaching rate of the biomolecules along the NW. This data suggest that at red wavelengths (633 nm) the incoming light is reduced at the NW tip – fitting data to a biexponential function shows a 20% decrease at the tip – where the light flux is distributed more equally along the NW at lower wavelengths.
Figure S2: Fluorescence photobleaching curves (averaged over three measurements) at three
4
different locations of the NWs and on a planar part of the wafer, to evaluate the excitation light flux along the NW Z-profile at different wavelengths (a) blue wavelength (alexa488) (b) red wavelength (alexa633).
Emission spectra recorded along the InAs NW We measured the emission spectra at three locations along the NW (tip, middle and base). Figure S3, shows that the fluorescence spectra are identical at the various positions tested along the NW.
Figure S3: Fluorophore emission spectra from SA-633 at the tip, middle and base of the NW.
Simulations of NW geometry impact on the fluorescence signal. We investigated whether the increased area at the NW tip (i.e. the increased protein quantity at this location of the NW) influences the fluorescence signal. This investigation was done by simulating the detected fluorescence signal – only considering the confocal volume and the protein concentration on the NW side and at the NW tip. The confocal volume was described by a two-component Gaussian point-spread function, expressing the detected fluorescence pixel intensity at the NW:
5
𝐼𝑝 = 𝐼0 ∫
𝑒
−
(𝑥+𝑑𝑥)2 +(𝑦+𝑑𝑦)2 (𝑧+𝑑𝑧)2 − 2𝜔𝑥𝑦 2 𝑒 2𝜔𝑧 2 𝛿𝑑𝑉
𝑉𝑜𝑙𝑢𝑚𝑒
Where the pixel intensity signal (Ip) of the fluorescent object was estimated at the coordinates (dx,dy,dz) compared to the protein location, the PSF fwmh was respectively wz=840nm and wxy=240nm (measured PSF from our CLSM using a 488nm laser), and I0 is the original signal from the proteins. The volume integral is an integral over the fluorescence object defined by the boundary conditions of the weight function (δ) describing the protein locations at the NW surface. Here the NW was set to a cylinder with a diameter and height of respectively 90nm and 4.5m. Quantification of the single dipole emission intensity Simulations were performed for a single dipole emitting at 5 nm from an InAs NW surface. In this regard, Figure S4 is a representation of the field energy distribution for a single dipole emitting at different positions along the NW surface, from 50 to 2000 nm from the tip. The bottom plots map the area that lies within the detection resolution (~1 µm), while the tip area is represented above them. For dipoles polarized in the radial direction, part of the energy is also guided and dissipated along the NW surface. We have also investigated dipoles emitting in the axial and transverse directions, but in those cases we observe a poorer light coupling, thus we only show the radial case.
6
Figure S4: Single dipole. Radial polarization. 515 nm emission For the quantification of the detected light emitted by a fluorophore, the Poynting vector was analyzed at a fixed plane in the simulation cell. We integrated the Poynting vector projected for a light cone with its center being the dipole position. The integral gives basically the energy flux through a surface with a maximum angular radius of 40o, due to cell size restrictions (See Figure S5). Reference simulations were performed for dipoles emitting at different depths without the presence of the InAs NW. As expected, the Poynting vector norm shows a circular shape around the center (see Figure S5a-b) and the calculated energy flux is independent of dipole position. By introducing the InAs NW, the radiation pattern is modified (Figure S5c-d) and is highly dependent on the dipole position.
7
Figure S5: Representation of the simulations performed for the quantification of detected light coming from a fluorophore. The projected emission profile is dependent of the dipole position in the case of the presence of a NW, while it is not the case without.
Substrate effects The presence of an InAs substrate underneath the InAs nanowire is responsible for the interference pattern observed in Figure 4a (main text). In order to better identify the main physical changes with excitation energy, the local modulation of the field energy was averaged between neighboring points (see figure below). The averaging was set to 14, 17 and 19 neighboring points for the blue, green and red excitations, respectively. The light reflected by the substrate is only affecting the local intensity profile coming from interferences. In fact, Figure S6 also compares the profile that would be obtained for a free standing NW. The averaged profile and that for the free-standing are in good agreement.
8
Figure S6: Example of the data treatment for simulated excitation energy profiles, for the blue light excitation. On the left, we show the crossectional map of the steady state energy density distribution in the NW on a substrate. On the right, the energy density profile along the vertical direction is represented before and after neighbor averaging. The profile obtained for the free standing nanowire is also represented. The profile is taken at 5nm distance from the NW distance along the radial direction.
Dependence of NW-light interaction with geometrical parameters We have simulated the excitation profiles in 3µm long NWs with different diameters (Figure S7). The simulations clearly show the evolution from an almost uniform excitation along the NW length for thin NWs, to a tip enhanced effect for larger diameters. The tip effect is first given at
9
short-wavelength radiation, leaving a wavelength dependent profile in intermediate diameters (90-100nm). Furthermore, no influence of the NW length is predicted from these simulations (Figure S7 bottom)
Figure S7: (top) Electric field energy density profiles (averaged) at InAs NW surface for three different NW diameters of 50, 100 and 120nm. (bottom) Electric field energy density profiles (with and without averaging) at InAs NW surface for NW of two different lengths (2 and 4.5 µm)
Comparison between horizontal and vertical NWs
10
As explained in the text, for dipoles emitting close to the NW tip, the energy guided along the surface is released through the gold tip. In case where no gold tip is present, we observe a similar antenna effect at the NW edges. Interestingly, there is almost no electric field energy at the tip when the dipole is below ~1 m, which results in a slightly stronger signal being detected for fluorophores emitting around the first micron from the tip. This is what is given in the main manuscript. On the other hand, if this was the only reason affecting the detected signal, one should also expect a light intensity increase at the NW edges when lying horizontally on a substrate. Simulations were thus also performed for fluorophores on an InAs NW that is horizontally lying on a glass substrate. Figure S8 represents the simulated local excitation and emission profiles in such a case. Similarly to the vertical case, there is an increase in intensity at the NW tip, especially for the presence of gold. However, the enhancement is extremely confined at the edges, since the light coupling is transverse to the detection. Due the confined nature of the enhancement, the total signal is very uniform after convoluting for the Gaussian response typical of the experimental set-up.
11
Figure S8. Simulated excitation, emission and convoluted light distribution for fluorophores attached to a horizontally lying nanowire. References (1)
Madsen, M. H.; Krogstrup, P.; Johnson, E.; Venkatesan, S.; Mühlbauer, E.; Scheu, C.; Sørensen, C. B.; Nygård, J. J. Cryst. Growth 2013, 364, 16–22.
(2)
Madsen, M. H.; Aagesen, M.; Krogstrup, P.; Sørensen, C.; Nygård, J. Nanoscale Res. Lett. 2011, 6, 516.
(3)
Rostgaard, K. R.; Frederiksen, R. S.; Liu, Y.-C. C.; Berthing, T.; Madsen, M. H.; Holm, J.; Nygård, J.; Martinez, K. L. Nanoscale 2013, 5, 10226–10235.
(4)
Berthing, T.; Bonde, S.; Rostgaard, K. R.; Madsen, M. H.; Sørensen, C. B.; Nygård, J.; Martinez, K. L. Nanotechnology 2012, 23, 415102.
(5)
Kang, K. a; Wang, J.; Jasinski, J. B.; Achilefu, S. J. Nanobiotechnology 2011, 9, 16.
12
Modulation of fluorescence signals from biomolecules along nanowires due to interaction of light with oriented nanostructures
Rune S. Frederiksen1, Esther Alarcon-Llado2, Morten H. Madsen3, Katrine R. Rostgaard1, Peter Krogstrup4, Tom Vosch5, Jesper Nygård4, Anna Fontcuberta i Morral2, Karen L. Martinez1
Material and Methods InAs NW array fabrication. NWs were grown by the gold-assisted vapor–liquid–solid growth mechanism using molecular beam epitaxy (MBE) and the gold catalyst particles were defined using electron beam lithography (EBL) on InAs (111)B substrate as previously described 1. Selfcatalysed NWs were grown on silicon2, see SI below. NW functionalization with BSA-biotin and SA. NW arrays were functionalized with 0.1mg/ml BSA-biotin (Sigma-Aldrich) and 1 M Alexa Fluor 488/555/633 streptavidin conjugate (SA488, SA-555, SA-633) in 0.01 M PBS buffer (Life Technologies) as described in Rostgaard et al.3 Cell data. HEK293 cells have been interfaced with arrays of InAs NWs and fluorescently labeled with a Dyomics dye DY-547 (ex546 nm: em560-600 nm) upon interface as described by Berthing et al.4 Fluorescence imaging. CLSM imaging was performed on an inverted confocal laser-scanning microscope (Leica TCS SP5) using a 63x magnification, water-immersion objective with 1.2
1
numerical aperture. The fluorophores were excited and the emission was collected by a photomultiplier tube at following excitation: emission wavelengths: ex488: em508–600 nm (SA488), ex546 nm: em560-600 nm, (SA-555) and ex633nm: em650-800nm (SA-633). Complying with the Nyquist rate for optimal imaging conditions, Z-stacks were collected with a pinhole of one airy unit, and a pixel- and step size of approximately 50 x 50 nm2 and 170 nm respectively. Quantification of fluorescence signals. Quantification of fluorescence signals was performed with ImageJ software. In case of the Z-stack profiles, the NW signals were measured with ROI with 10 pixels around each NW. The background signal, measured as the signal from the buffer solution above the NWs, was subtracted. In case of the line scans of horizontal NWs, they were measured with a 5-pixel average. Time-correlated single photon counting experiments. The experiments were carried out on a home-built scanning fluorescence confocal microscopy system based on an Olympus IX71 inverted microscope. A piezo-driven scanning stage (Physik Intrumente P5173CL), allowed for imaging the sample point by point in a raster scanning fashion in a range up to 100 µm x 100 µm. Upon laser illumination, the fluorescence emission signal from the sample was collected by a 100x oil immersion objective (Olympus UPLFLN 100x), and directed to an avalanche photodiode (APD, Perkin Elmer CD3226) for recording the fluorescence intensity and for performing time-correlated single photon counting experiments. The laser light was blocked in the emission path by appropriate long-pass filters (Semrock LP02-488RU-25). A 480 nm pulsed laser (Picoquant LDH-P-C-470) was used as excitation source. Excitation was performed with linear polarized light and no polarizers were inserted before the APD. SEM Imaging. To evaluate the protein density along the InAs NW, we used InAs NWs
2
functionalized with BSA-biotin (as described above) and evaluated by SEM the distribution of streptavidin fused to 20nm SA-NP (Invitrogen) along the NWs. The InAs NW was firstly premodified with BSA-biotin incubation secondly modified with 0.04 nmol/L (SA-NP) for 30 min, followed by incubation overnight in vacuum chamber and finaly imaged in a SEM (JEOL JSM- 6320F). Simulations The InAs NW was set to 90nm in diameter and 4.5 m long vertically standing on InAs substrate. A gold semi-sphere was included at the NW tip and the whole system was embedded in water. Simulations were performed with Meep (http://ab-initio.mit.edu/wiki/index.php/Meep), a free finite-difference time-domain (FDTD) software package to model electromagnetic systems. For simulating the excitation field distribution profiles, a plane monochromatic wave was used as light source propagating in the vertical direction. Perfectly matching layers around the cell were included to avoid boundary conditions effects. Simulations were performed at a 5 nm distance from the NW to account for the size of the intermediate protein layer (BSA-biotin) onto which fluorescent SA is bound. Self-catalyzed NWs. Gold NPs are known to amplify fluorescence signals5. To evaluate if the gold catalyst particle present at the NW tip causes the alterations in the fluorescence profile, we studied the Z-profiles from SA-488 immobilized on self-catalyzed NWs (without the gold-NP tip). The self-catalyzed InAs NWs were grown on a Si(111) wafer by Molecular Beam Epitaxy using a thin native oxide layer to initiate the growth, see2 for details on the growth method. The gold-NP does not seem to be the main source of the tip effect, as the tip effect is still present on the self-catalyzed NWs, as seen in Figure S1.
3
7
Z-stack (um)
6 5 4 3 2 1
0
50
100 150 200
Fluroescence Intensity (a.u.)
Figure S1: Self-catalyzed NWs functionalized with BSA-biotin and SA-488
Fluorescence photobleaching along the InAs NW To evaluate if the fluorescence modification originates from modification of the excitation light flux along the NW, we studied the photobleaching rate of the biomolecules along the NW. This data suggest that at red wavelengths (633 nm) the incoming light is reduced at the NW tip – fitting data to a biexponential function shows a 20% decrease at the tip – where the light flux is distributed more equally along the NW at lower wavelengths.
Figure S2: Fluorescence photobleaching curves (averaged over three measurements) at three
4
different locations of the NWs and on a planar part of the wafer, to evaluate the excitation light flux along the NW Z-profile at different wavelengths (a) blue wavelength (alexa488) (b) red wavelength (alexa633).
Emission spectra recorded along the InAs NW We measured the emission spectra at three locations along the NW (tip, middle and base). Figure S3, shows that the fluorescence spectra are identical at the various positions tested along the NW.
Figure S3: Fluorophore emission spectra from SA-633 at the tip, middle and base of the NW.
Simulations of NW geometry impact on the fluorescence signal. We investigated whether the increased area at the NW tip (i.e. the increased protein quantity at this location of the NW) influences the fluorescence signal. This investigation was done by simulating the detected fluorescence signal – only considering the confocal volume and the protein concentration on the NW side and at the NW tip. The confocal volume was described by a two-component Gaussian point-spread function, expressing the detected fluorescence pixel intensity at the NW:
5
𝐼𝑝 = 𝐼0 ∫
𝑒
−
(𝑥+𝑑𝑥)2 +(𝑦+𝑑𝑦)2 (𝑧+𝑑𝑧)2 − 2𝜔𝑥𝑦 2 𝑒 2𝜔𝑧 2 𝛿𝑑𝑉
𝑉𝑜𝑙𝑢𝑚𝑒
Where the pixel intensity signal (Ip) of the fluorescent object was estimated at the coordinates (dx,dy,dz) compared to the protein location, the PSF fwmh was respectively wz=840nm and wxy=240nm (measured PSF from our CLSM using a 488nm laser), and I0 is the original signal from the proteins. The volume integral is an integral over the fluorescence object defined by the boundary conditions of the weight function (δ) describing the protein locations at the NW surface. Here the NW was set to a cylinder with a diameter and height of respectively 90nm and 4.5m. Quantification of the single dipole emission intensity Simulations were performed for a single dipole emitting at 5 nm from an InAs NW surface. In this regard, Figure S4 is a representation of the field energy distribution for a single dipole emitting at different positions along the NW surface, from 50 to 2000 nm from the tip. The bottom plots map the area that lies within the detection resolution (~1 µm), while the tip area is represented above them. For dipoles polarized in the radial direction, part of the energy is also guided and dissipated along the NW surface. We have also investigated dipoles emitting in the axial and transverse directions, but in those cases we observe a poorer light coupling, thus we only show the radial case.
6
Figure S4: Single dipole. Radial polarization. 515 nm emission For the quantification of the detected light emitted by a fluorophore, the Poynting vector was analyzed at a fixed plane in the simulation cell. We integrated the Poynting vector projected for a light cone with its center being the dipole position. The integral gives basically the energy flux through a surface with a maximum angular radius of 40o, due to cell size restrictions (See Figure S5). Reference simulations were performed for dipoles emitting at different depths without the presence of the InAs NW. As expected, the Poynting vector norm shows a circular shape around the center (see Figure S5a-b) and the calculated energy flux is independent of dipole position. By introducing the InAs NW, the radiation pattern is modified (Figure S5c-d) and is highly dependent on the dipole position.
7
Figure S5: Representation of the simulations performed for the quantification of detected light coming from a fluorophore. The projected emission profile is dependent of the dipole position in the case of the presence of a NW, while it is not the case without.
Substrate effects The presence of an InAs substrate underneath the InAs nanowire is responsible for the interference pattern observed in Figure 4a (main text). In order to better identify the main physical changes with excitation energy, the local modulation of the field energy was averaged between neighboring points (see figure below). The averaging was set to 14, 17 and 19 neighboring points for the blue, green and red excitations, respectively. The light reflected by the substrate is only affecting the local intensity profile coming from interferences. In fact, Figure S6 also compares the profile that would be obtained for a free standing NW. The averaged profile and that for the free-standing are in good agreement.
8
Figure S6: Example of the data treatment for simulated excitation energy profiles, for the blue light excitation. On the left, we show the crossectional map of the steady state energy density distribution in the NW on a substrate. On the right, the energy density profile along the vertical direction is represented before and after neighbor averaging. The profile obtained for the free standing nanowire is also represented. The profile is taken at 5nm distance from the NW distance along the radial direction.
Dependence of NW-light interaction with geometrical parameters We have simulated the excitation profiles in 3µm long NWs with different diameters (Figure S7). The simulations clearly show the evolution from an almost uniform excitation along the NW length for thin NWs, to a tip enhanced effect for larger diameters. The tip effect is first given at
9
short-wavelength radiation, leaving a wavelength dependent profile in intermediate diameters (90-100nm). Furthermore, no influence of the NW length is predicted from these simulations (Figure S7 bottom)
Figure S7: (top) Electric field energy density profiles (averaged) at InAs NW surface for three different NW diameters of 50, 100 and 120nm. (bottom) Electric field energy density profiles (with and without averaging) at InAs NW surface for NW of two different lengths (2 and 4.5 µm)
Comparison between horizontal and vertical NWs
10
As explained in the text, for dipoles emitting close to the NW tip, the energy guided along the surface is released through the gold tip. In case where no gold tip is present, we observe a similar antenna effect at the NW edges. Interestingly, there is almost no electric field energy at the tip when the dipole is below ~1 m, which results in a slightly stronger signal being detected for fluorophores emitting around the first micron from the tip. This is what is given in the main manuscript. On the other hand, if this was the only reason affecting the detected signal, one should also expect a light intensity increase at the NW edges when lying horizontally on a substrate. Simulations were thus also performed for fluorophores on an InAs NW that is horizontally lying on a glass substrate. Figure S8 represents the simulated local excitation and emission profiles in such a case. Similarly to the vertical case, there is an increase in intensity at the NW tip, especially for the presence of gold. However, the enhancement is extremely confined at the edges, since the light coupling is transverse to the detection. Due the confined nature of the enhancement, the total signal is very uniform after convoluting for the Gaussian response typical of the experimental set-up.
11
Figure S8. Simulated excitation, emission and convoluted light distribution for fluorophores attached to a horizontally lying nanowire. References (1)
Madsen, M. H.; Krogstrup, P.; Johnson, E.; Venkatesan, S.; Mühlbauer, E.; Scheu, C.; Sørensen, C. B.; Nygård, J. J. Cryst. Growth 2013, 364, 16–22.
(2)
Madsen, M. H.; Aagesen, M.; Krogstrup, P.; Sørensen, C.; Nygård, J. Nanoscale Res. Lett. 2011, 6, 516.
(3)
Rostgaard, K. R.; Frederiksen, R. S.; Liu, Y.-C. C.; Berthing, T.; Madsen, M. H.; Holm, J.; Nygård, J.; Martinez, K. L. Nanoscale 2013, 5, 10226–10235.
(4)
Berthing, T.; Bonde, S.; Rostgaard, K. R.; Madsen, M. H.; Sørensen, C. B.; Nygård, J.; Martinez, K. L. Nanotechnology 2012, 23, 415102.
(5)
Kang, K. a; Wang, J.; Jasinski, J. B.; Achilefu, S. J. Nanobiotechnology 2011, 9, 16.
12
