Bottom-up approach to nonlinear optical imaging of biomaterials

Biological tissues are complex systems. Conventional microscopy, ultimately limited to a spatial resolution of roughly half the wavelength of light, is often unable to see molecular details unless sophisticated and invasive methods are employed. However, recent advances in nonlinear optical-imaging techniques for biological tissues have resulted in both higher spatial resolution and better tissue contrast than achieved before.¹ As a result, a great deal of focus has been placed on engineering nonlinear optical probes for use with two-photon excited-fluorescence and second-harmonic-generation (SHG) imaging. Both employ femtosecond-laser microscopy to measure nonlinear optical responses.

In SHG, two photons at a fundamental frequency are converted into one at once or twice the harmonic frequency. In addition to the numerous organic and inorganic optical probes that have been developed for labeling tissues and cell membranes, certain cellular components, such as collagen, have intrinsic SHG properties.² This makes SHG a favourable method for noninvasive imaging of large structures and cellular processes. However, it still lacks resolution at the molecular level. In addition, for smaller components the relationship between the intensity and polarization state of the nonlinear optical response and the molecule's conformation remains elusive. Here, we provide preliminary insights into this dependence.

Figure 1. Quadratic hyperpolarizability of tryptophan (W)-rich peptides as a function of the number of these amino acids. Dots: Experimental data. Empty circles and squares: coherent and incoherent models, respectively.

In addressing this problem, the focus is usually placed on the imaging potential of the techniques. Very few studies have tackled the problem in a bottom-up approach, starting from the elementary bricks of the proteins—the individual amino acids—and working up to the protein itself. The nonlinear optical properties of these molecules can be observed using hyper-Rayleigh scattering (HRS), i.e., the incoherent nonlinear scattering of light observed when an isotropic liquid solution of biomolecules is irradiated with a beam of photons of a specific frequency.³ HRS relies on instantaneous fluctuations of an otherwise homogeneous solution to produce frequency-doubled light. With an experimental setup capable of single-metallic-particle sensitivity,⁴ we measured (for the first time) the quadratic hyperpolarizability (QH) of amino acids. Effectively, we determined their efficiency to scatter light at the second-harmonic frequency. Subsequently, we investigated how the QH of a repetitive amino-acid sequence varies as a function of their number in a peptide chain. We have performed similar work with collagen, a biomolecule with a strong response but containing individual amino acids lacking a noticeable QH.

To determine the hyperpolarizabilities of amino acids, we used the dilution method, in which the HRS intensity is recorded as a function of the concentration of amino acids dispersed in a buffered aqueous solution. The amino acids tryptophan (W) and tyrosine have QHs of (29.6±0.04)×10⁻³⁰ and (25.7±0.03)×10⁻³⁰electrostatic units (esus), respectively, while phenylalanine and lysine (K), a nonaromatic amino acid, have QHs of less than, respectively, 8.5×10⁻³⁰ and 1.5×10⁻³⁰esu.

We synthesized the peptides KWK, KWWK, KWWWK, and KWWKWWK and measured their QHs. Our results indicate that tryptophan exhibits an extremely large variability, depending on its environment. In addition, the QH dependence of W on the number of Ws in the peptide is much stronger than expected if we assume a coherent response model (see Figure 1). A coherent response would arise if the amino-acid residues were considered as independent elementary bricks with the same spatial orientation but without any interactions. Thus, our results unambiguously demonstrate that there must be strong interactions between the residues in these peptides, which in turn must depend strongly on the conformation of these small peptides. It is also possible that the peptide bond itself may be a factor in the resulting scattering.

We further investigated this role using the protein collagen I extracted from rat tail. Collagen I is a rigid, rod-like molecule in the form of a 290nm-long helix composed of three interwoven αchains. The primary sequence is a repeating GPP motif, where G and P are the one-letter codes for glycine and proline, respectively. These proteins produce very strong signals in SHG imaging microscopy.² However, the QH of the peptide GGG is only (6±2)×10⁻³⁰esu, indicating at most an effective QH per glycine similar to that of lysine. Likewise, proline as a monomeric peptide (PPP…) lacks a strong response. Using a microscopic model that considers the nonnegligible length of the collagen I protein with respect to the wavelength of light,⁴ we showed that the strong collagen I value of (1250±20)× 10⁻³⁰esu results largely from the arrangement of microscopic units along the three α-chain backbones. It is then questionable whether the QH response is associated with peptide bonds. We will address this problem in the future.

These early studies are the first steps in linking the microscopic quadratic nonlinear optical response of the elementary building blocks (the amino acids and their peptide bonds) with that of proteins. We are applying this technique to other proteins, in particular those incorporating both strong and weak amino acids, to formulate general rules describing this link.

The author would like to acknowledge discussions and collaboration with Julien Duboisset, Gladys Matar, Françoise Besson, Claire Loison, and Emmanuel Benichou at the University Claude Bernard Lyon 1 (France) and Ariane Deniset-Besseau and Marie-Claire Schanne-Klein at the École Polytechnique (France).