Cyril Barsua, Rouba Cheaibbc, Stéphane Chambertbc, Yves Queneaubc, Olivier Maurya, Davy Cottetd, Hartmut Weged, Julien Douadyd, Yann Bretonnière*a and Chantal Andraud*a
aUniversité de Lyon, Laboratoire de Chimie de l'ENS Lyon, UMR 5182 CNRS-ENS Lyon, 46 allée d'Italie, 69364, Lyon, France. E-mail: Chantal.Andraud@ens-lyon.fr
bINSA Lyon, Laboratoire de Chimie Organique, Bâtiment J. Verne, 20 av A. Einstein, 69621, Villeurbanne, France
cCNRS, UMR 5246, ICBMS, Université Lyon 1, INSA-Lyon, CPE-Lyon, Bâtiment CPE, 43 bd du 11 novembre 1918, 69622, Villeurbanne, France
dLaboratoire de Spectrométrie Physique, UMR 5588 CNRS-Université Joseph Fourier (Grenoble I), 140 avenue de la physique, 38402, Saint Martin d'Hères, France
First published on 30th October 2009
A new class of push-pull molecules was recently identified, based on pyridine dicarboxamide as an electron acceptor group and bearing a fluorenethynyl π-conjugated bridge. The molecules present good second and third order nonlinear properties, with a static quadratic hyperpolarisability μβ (at 1.907 μm) of 320 × 10−48 esu (μβ0 = 249 × 10−48 esu) and a maximum two-photon absorption cross-section of 1146 GM. Starting from this generic structure, we designed a series of eight amphiphilic nonlinear probes, varying the length of the lipophilic tail and the nature of the polar head, and tested their cell membrane affinity by nonlinear optical imaging. A good membrane affinity was observed with probes bearing short alkyl chains and carbohydrate moieties as the polar head, emphasizing the importance of the lipophilic–hydrophilic balance, as well as the role of the polar head. This original approach based on carbohydrate head groups is an important advancement in the design of membrane probes, since these highly versatile functional groups confer adequate hydrophilicity and yet conserve overall molecular neutrality.
![]() | ||
Chart 1 Chemical structures of some voltage-sensitive fast response probes. |
The same molecules have large second order nonlinear coefficients,7 and have set the basis of the recent development of new membrane potential imaging techniques based on nonlinear optics: two-photon excited fluorescence microscopy (TPM) combined with second-harmonic imaging microscopy (SHIM). Both microscopy techniques originate from fundamentally different nonlinear optical phenomena, two-photon excited fluorescence (TPEF) and second harmonic generation (SHG), respectively. TPEF involves the simultaneous absorption of two photons (TPA) of lower frequency to trigger the fluorescence of the molecule, whereas in SHG, two incident photons are scattered into one coherent output photon with exactly twice the incident frequency. As the amplitude of both nonlinear phenomena are proportional to the square of the incident light intensity, they only occur at the focal point of a tightly focused laser, providing an intrinsic confocality, which results in a high z resolution. Furthermore, since the incident laser wavelengths are in the near infrared (NIR, typically in the tissue transparency region between 750–850 nm), lower tissue autofluorescence or self-absorption, increased penetration depth, reduced out-of-plane photobleaching and phototoxicity can be achieved.8,9 The SHG signal requires a non-centrosymmetric local environment, and only molecules asymmetrically distributed in the cell membrane will contribute to the observed signal,9–11 without any background signal originating from isotropic distribution in the surrounding medium. On the other hand, TPEF is an incoherent process and imposes no restriction on symmetry. The observation of TPEF will only depend on the presence of an emissive molecule, giving information about molecular distribution.
Some membrane-bound dyes of the ANEP family were found to engender a SHG signal highly sensitive to membrane potential,9,12 and others to present sub-millisecond response times to voltage change, which was shown to arise from electron redistribution or molecular tilt mechanisms.13 These assets make combined TPM/SHIM a unique and specific tool for the optical imaging of cellular membranes providing complementary information, and for direct time-resolved measurement of membrane potential in living cells.14 These studies have shown the proof of concept, but also pointed out the need for new chemical probes specifically designed for that purpose.15 In a recent report, Anderson, Clays and co-workers specifically designed a series of amphiphilic porphyrin-based second order nonlinear chromophores for SHG imaging of cell membranes, but still bearing a charged pyridinium polar head.16 Although a very strong SHG signal was observed in lipid vesicles for all compounds, the in vitro membrane localization strongly depends on the nature of the polar head. Therefore, despite the push in establishing pyridinium dyes as effective SHG sensors of membrane potential, there is room for improved molecules presenting largely different structures.
![]() | ||
Chart 2 Structure of new pyridine dicarboxamide nonlinear probes. |
n | CH2Cl2a | Water | s/cm−1e | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
λmax/nm | ε(λmax)/103 M−1 cm−1 | λem/nmb | Δṽ/cm−1 | ΦFc | λmax/nm | ε(λmax)/103 M−1 cm−1 | λem/nmb | Δṽ/cm−1 | ΦFc | ||
a Except for 1 in CHCl3.b λexc = 410 nm.c λexc = 410 nm, reference Coumarin 153 in MeOH (ΦF = 0.45).26d Ref. 17.e Excluding water and methanol.f Excluding cyclohexane. | |||||||||||
1d | 411 | 37.0 | 525 | 5438 | 0.89 | — | — | — | — | — | 19385 |
2a | 389 | 12.2 | 543 | 7290 | 0.69 | 385 | 10.1 | 570 | 8368 | 0.05 | 19700 |
3a | 396 | 50.0 | 530 | 6348 | 0.75 | 402 | 30.0 | 575 | 7824 | 0.09 | 15552 |
4a | 391 | 24.8 | 529 | 6808 | 0.60 | 390 | 11.9 | 598 | 8720 | 0.07 | 24819f |
5a | 390 | 18.4 | 542 | 7190 | 0.49 | 397 | 10.1 | 584 | 8129 | 0.08 | 20718f |
2b | 390 | 35.7 | 537 | 7019 | 0.79 | 393 | 24.7 | 561 | 7620 | 0.16 | 20879 |
3b | 394 | 50.0 | 531 | 6477 | 0.78 | 397 | 27.9 | 552 | 7458 | 0.20 | 31497 |
4b | 388 | 32.9 | 525 | 5884 | 0.69 | 395 | 27.5 | 574 | 2299 | 0.17 | 24779f |
5b | 387 | 21.8 | 529 | 6936 | 0.50 | 388 | 18.8 | 561 | 7783 | 0.14 | 30588f |
Fluorescent probes with charged groups were found to occupy only a shallow location within the polar headgroup region of the bilayer, no matter how lipophilic the molecule to which they are linked.18 Therefore, in order to obtain a deeper membrane penetration, we incorporated two carbohydrate entities (glucosyl or lactosyl residues) into the probe, with the aim of providing an appropriate hydrophilicity while keeping overall neutrality. Carbohydrates are major constituents of cell membrane surfaces as glycoproteins or glycolipids. Carbohydrates mediate cell–cell contacts through sugar–protein specific interactions, and as such, are involved in many significant biological processes at the cell surface such as cell adhesion, cell recognition and differentiation, development of the neuronal network, inflammation, metastasis, and viral or bacterial infection.19 By grafting carbohydrate residues into the precedent nonlinear chromophores, we sought to mimic the overall structure of natural glycolipids and give access to neutral probes with good affinities for the cell membrane. To the best of our knowledge, no specially designed non-ionic carbohydrate-containing probes have been developed to date for nonlinear membrane imaging. The only example is a ribose-containing ANEP dye (JPW1259, Chart 1) in which the sugar cycle has been introduced as a chiral group to help assure a non-centrosymmetric environment.12 Carbohydrates are highly versatile scaffolds offering huge structural variation and can easily be introduced to primary amine groups through the use of carboxymethylglycoside lactones (CMGLs).20 In this article, we describe the synthesis and appraisal in nonlinear optical cell imaging of eight new amphiphilic compounds 2a/2b–5a/5b (Chart 2) bearing either neutral carbohydrate entities (two glucosyl or lactosyl residues) or simple amino and polyethyleneglycol (PEG) motifs for comparison. These different moieties enable considerable variation in hydrophilicity, whereas the length of the alkyl chains on the pyridinecarboxamide groups and the presence of two n-hexyl chains on the fluorenyl ring served to modulate the lipophilic character.
![]() | ||
Scheme 1 Synthesis of the diamino compounds (2a) and (2b). Reagents and conditions: (i) 1-bromoethanol, NaI, Na2CO3, DMF, 95 °C, 48 h, 66%; (ii) N-phthalimide, PPh3, DEAD, THF, rt, 18 h, 95%; (iii) R = ethyl: SOCl2, DMF cat., reflux, 2 h, then Et2NH, CH2Cl2, reflux, 2 h, 93% or (iv) R = n-C8H17: SOCl2, DMF cat., reflux, 2 h, then NH(n-C8H17)2, Et3N, CH2Cl2, reflux, 2 h, 90%; (v) 57% HI, H3PO3, 80 °C, 3.5 h, 72% (11a)/76% (11b); (vi) trimethylsilylacetylene, CuI, PdCl2(PPh3)2, THF, Et3N, rt, 4 d; (vii) K2CO3, MeOH, rt, 2 h, for the two steps 86% (12a)/61% (12b); (viii) CuI, PdCl2(PPh3)2, THF, Et3N, 50 °C, 48 h, 60% (13a)/65% (13b); (ix) NH2NH2·H2O, crotyl alcohol, THF, reflux, 18 h, 90% (2a)/95% (2b). |
![]() | ||
Scheme 2 Synthesis of the target probes 3–5. Reagents and conditions: (i) DiC, HOBt, CH2Cl2, rt, 2 d, 51% (3a)/58% (3b); (ii) CH2Cl2, rt, 12 h; (iii) MeONa cat., MeOH, rt, 16 h, 45% (4a)/74% (4b), 30% (5a)/63% (5b) overall yield for the 2 steps. |
The molecules varied considerably in their water solubility. The six PEG groups in molecules 3a–b for instance provided very high solubility. Short chain 3a is soluble at 10−2 M and a clear solution was obtained, whereas the n-octyl analogue at 10−3 M in water gave a cloudy solution but no precipitate, suggesting the formation of micelles. On the other hand, the solubility in water of the sugar compounds 4a–b and 5a–b remains very low (a precipitate was observed in water at 10−6 M) even for the short-chain lactose compound 5a possessing four sugar residues.
The absorption spectra of all compounds displayed an intense broad band in the range 350–450 nm with maxima at around 390 nm in dichloromethane. It is worth noting that the position of the absorption band allows an enhancement of the SHG signal by resonance at around 800 nm. The 20 nm blue shift of the linear absorption of 2–5 with respect to that of 1 could be related to the decreased donor strength due to the existence of hydrogen-bonding between the aniline nitrogen and the primary amine protons (compounds 2a–2b) or the amide hydrogen (3–5), as suggested for push-pull chromophores where only two or three carbon atoms separate the nitrogen donor atom from a hydrogen donor functionality such as a hydroxyl or an amide group.27 The molar extinction coefficients, at the wavelength of maximum absorption (εmax) ranged from 12.2 × 103 M−1 cm−1 for the diamine 2a to 50.0 × 103 M−1 cm−1 for the PEG compounds 3a and 3b in dichloromethane. A substantial decrease is observed in water for all systems with values decreasing from 50.0 × 103 M−1 cm−1 in CH2Cl2 to 30.0 and 27.9 × 103 M−1 cm−1 in water for 3a and 3b, respectively, for instance (Table 1).
All the studied compounds exhibited an intense emission with a structureless band centered at 525–545 nm in dichloromethane (Fig. 1). No residual emission was observed at 400 nm for each compound, suggesting an unambiguous separation of TPEF and SHG signals, which is a relevant condition for nonlinear imaging. The quantum yields measured in dichloromethane were high (ΦF = 0.50 for 5a and 5b to ΦF = 0.79 for 3b and 2b) and similar to that measured for the reference 1 (ΦF = 0.89), independent of the nature of the polar head. No differences were observed between short chain and long chain amides. Only the presence of four sugar groups (lactose compounds) resulted in a decrease of the quantum yield. In water, however, long chain compounds displayed a much higher quantum yield than the short chain analogs, suggesting a different organization in solution for both cases.
![]() | ||
Fig. 1 Normalized absorption and emission spectra for compounds 2a (a), 3a (b), 4a (c) and 5a (d) in different solvents: (—) cyclohexane, (![]() ![]() ![]() ![]() ![]() ![]() |
The influence of the functional changes on the chromophores’ optical properties was studied by recording the absorption and fluorescence spectra in solvents of different polarity and proton donating ability at room temperature. Fig. 1 shows the absorption as well as the fluorescence spectra of the short chain compounds 2a–5a in the various solvents. Data for other probes are given in detail in the ESI.† An increase in solvent polarity induced a marked red shift of the emission band. The positive solvatochromic properties were analyzed following the Lippert–Mataga model.28 The solvent polarity parameter is thereby defined by the factor , where ε and n are the dielectric constant and the refractive index of the solvent, respectively. The Stokes shift (
in cm−1) is correlated to Δf according to eqn (1), where K is a constant, c and h are the light speed and the Planck constant, and a is the radius of the spherical cavity of the solute.
![]() | (1) |
The solvatochromic slopes s derived from the Lippert–Mataga model are reported in Table 1. Even if this model is limited to a point dipole approximation, which is not the case here, a good linear correlation with relationship (1) was found if hydrogen bonding solvents (methanol and water) and, in the case of sugar-based molecules, cyclohexane are excluded (Fig. 2, and S2–S10 in the ESI†). This points out that the emission process occurs from a dipolar excited state, and is indicative of a rather high variation of the dipole moment between the excited and the ground state (Δμ). It is interesting to note that the emission wavelength is more sensitive to solvent than the absorbance, which indicates that the relaxed excited state may have a significantly different structure to that of the ground state. The solvatochromism was not analyzed further, as the shape and the dimension of the solute cavity a can hardly be simply estimated for such chromophores. More importantly, very close values have been found for all molecules 2–5, which are similar to that of the reference compound 1. This led us to think that the substitution does not significantly modify the chromophore’s electrochromic properties, and by extension, the nonlinear properties. The ETN scale29 emphasized that the points for protic solvents are outside the linear regression zone (see the ESI, Fig. S11–S12†). This was observed for all compounds including the reference 1, showing the existence of a hydrogen bond involving the electron acceptor group (pyridine dicarboxamide) of the chromophores and the solvent.
![]() | ||
Fig. 2 Variations of the Stokes shift (cm−1) Δṽ with Δf for compounds 4a–5a and reference compound 1. |
Cell staining was observed by TPEF imaging immediately after dye injection, without washing. Experiments were run on a modified upright microscope adapted for two-photon imaging with an external scan head and pulsed laser illumination, with direct immersion of the objective into the cell culture medium. Irradiation was carried out at 810 nm in the TPA band17 with a mode-locked Ti:sapphire laser, characterized by an output pulse width of ∼100 femtoseconds and a repetition rate of 78 MHz.
Surprisingly, the most lipophilic long chain compounds (2b–5b) did not stain the cell membranes. No intracellular fluorescence could be observed at any concentration, but a high background signal arising from the surrounding liquids suggested that the dyes stayed in the extracellular media, even after long incubation times (overnight). For the short chain compounds 2a–5a, marked differences were observed depending on the nature of the polar head, as shown in Fig. 3 and 4. The diamino compound 2a (protonated at physiological pH) and compound 3a bearing six PEG groups quickly entered the cell and accumulated in the cell cytoplasm where fluorescence was observed, as shown in Fig. 3. The cell nucleus remained unstained. The exact localization of the dyes, which did not evolve with time, has not been further investigated.
![]() | ||
Fig. 3 TPEF images of CHO cells stained externally (0.1 μM) by 2a (left) and 3a (right). The average excitation power was 10 mW at the sample and the total acquisition time was 0.9 s. Scale bars are 20 μm. |
![]() | ||
Fig. 4 TPEF images of CHO cells stained externally by 4a (10 μM) showing the progressive internalization of the dye: immediately after dye injection (left) and after 10 min (right). The average excitation power was 10 mW at the sample and the total acquisition time was 0.9 s. Scale bars are 10 μm. |
Conversely, showing the role of the peripheral substituents and of the balance between hydrophilicity and lipophilicity, a good membrane staining was achieved with dyes 4a and 5a bearing two and four pyranoside structures, respectively. It was observed immediately after dye injection and remained as long as the extracellular medium was not washed. The cytoplasm, where no fluorescence was observed immediately after injection, was progressively stained, as shown in Fig. 4. This indicates that the dyes were able to cross the membrane bilayer and diffuse into the cytoplasm. As for the PEG compound, the cell nucleus was not stained. When stained cells were washed by fresh culture medium after a few minutes of incubation, fluorescence from the membrane vanished as the dyes progressively internalized. Focusing the laser beam on a small area of the cell resulted in too strong absorption and rapid cell death. These encouraging results prompted us to try to simultaneously record TPEF and second harmonic generation signals. In SHG, being a coherent process, the signal propagates roughly in the same direction as the incident beam9,10 and was therefore collected in a transmitted light configuration. With incident light at 810 nm in the TPA band, the SHG was considerably enhanced by resonance, and the signal was also spectrally separated from the fluorescence which allowed for its selective detection by means of a 405 nm narrow band-pass filter. The TPEF signal was collected in epi-configuration. We also used commercially available di-4-ANEPPS as a reference membrane-staining compound to assess instrument performance and to adjust the recording configuration.
Fig. 5 shows the compared TPEF (left)–SHG (right) images recorded with compound 5a on F98 glial cells. The TPEF signal of the stained membrane was very strong and a single scan was sufficient for recording a good quality image, while the SHG signal observed at 405 nm, arising only from molecules in the cell membrane, was obtained after averaging of 3 to 10 successive scans, due to the weak intensity of the signal. It is worth noting that similar results were obtained with CHO cells. Although the SHG signal strength depends quadratically on incident intensity, it could not be improved by increasing the laser power, due to a simultaneous enhancement of the fluorophore TPA, resulting in increased photo-induced damage in the cells, and limiting the experiment to a laser power of 5–15 mW under the objective. Focusing and scanning a reduced area of the cell to assess the kinetics of internalization also resulted in cell death for the same reasons.
![]() | ||
Fig. 5 Simultaneous TPEF (left) and SHG (right) images of F98 cells stained externally with 5a (2 μM) and excited at 810 nm. The SHG image has been contrasted for better visualization. The average excitation power was 15 mW at the sample, and the total acquisition time was 0.9 s for the TPEF image and 3.6 s for the SHG image. Scale bars are 20 μm. |
Similar results were obtained for both compounds 4a and 5a, bearing a glucosyl and a lactosyl structure, respectively. On the other hand, although quantification could not be obtained, membrane staining with compounds 4a and 5a was long-lasting and the signal was observed for a long time span by comparison with di-4-ANEPPS. In the same imaging experiment, the latter showed a stronger signal, almost immediately seen, but it vanished quickly as the dye was internalized.
Good membrane staining was observed on CHO and F98 glial cells for the dyes 4a and 5a bearing short alkyl chains and containing two and four pyranoside structures, respectively, while no significant difference was found between glucose- and lactose-based groups. The approach based on carbohydrate head groups is an important advancement in the design of membrane probes, since these highly versatile functional groups confer adequate hydrophilicity and yet conserve overall molecular neutrality. Membrane specificity of the most promising dyes is currently being tested in neuronal cells and brain slices in order to test the voltage dependence in physiological conditions. With the huge structural variety offered by carbohydrate scaffolds, more selective staining, as well as nonlinear imaging of specific sugar receptors on the cell membrane surface can be envisaged.
![]() | ||
Fig. 6 Schematic representation of the microscopy set-up. |
Footnote |
† Electronic supplementary information (ESI) available: All experimental procedures and characterization. Spectroscopic data for all compounds 2–5, including Lippert–Mataga plot and fit. See DOI: 10.1039/b915654b |
This journal is © The Royal Society of Chemistry 2010 |