Molecular understanding of label-free second harmonic imaging of microtubules

Microtubules are a vital component of the cell’s cytoskeleton and their organization is crucial for healthy cell functioning. The use of label-free SH imaging of microtubules remains limited, as sensitive detection is required and the true molecular origin and main determinants required to generate SH from microtubules are not fully understood. Using advanced correlative imaging techniques, we identified the determinants of the microtubule-dependent SH signal. Microtubule polarity, number and organization determine SH signal intensity in biological samples. At the molecular level, we show that the GTP-bound tubulin dimer conformation is fundamental for microtubules to generate detectable SH signals. We show that SH imaging can be used to study the effects of microtubule-targeting drugs and proteins and to detect changes in tubulin conformations during neuronal maturation. Our data provide a means to interpret and use SH imaging to monitor changes in the microtubule network in a label-free manner.

indicating distinct GTP-tubulin sites within the microtubule lattice. (A') Regions within the fibroblast where the cytosolic space is constricted contain many densely packed microtubules, arranging the GTP tubulin sites in a dense and parallel manner, allowing efficient constructive interference for SH imaging. (B) Microtubules in non-neuronal cells such as fibroblasts generate detectable SH signals in constricted regions (arrowheads). The SH signal in a fibroblast colocalizes with a staining for GTP-bound tubulin dimers (MB11, red) and overall microtubule staining (α-tubulin, green).

Supplementary Figure 4.
Taxol increases GTP-bound tubulin dimers and not the total amount of tubulin.
SH signals from control (DMSO) and taxol (10 nM, 4h) treated cells where taxol acts as a microtubule-stabilizing agent by increasing the amount of GTP-bound tubulin dimer conformations. The corresponding α-tubulin staining (Intensity color coded) shows no difference in total number of microtubules between both groups (n=41 biologically independent cells; p>0.05 Mann Whitney test). GTP-bound tubulin (MB11 ; Intensity color coded) stainings show increased fluorescence intensity in taxol treated neurons compared to controls (n=41 biologically independent cells; ** p<0.01 Mann Whitney test). Neuronal cultures were imaged at 7 DIV. Bar plots presented as means ± standard error of the mean. All source data are provided as a Source Data file.

The SH signal from taxol-treated cells does not originate from taxol.
Second harmonic signals recorded in HEK 293T cells generate bright second harmonic signals upon taxol incubation, originating from tubulin and not taxol itself. (A) HEK 293T cells treated with DMSO (control) or taxol (10 nM, 4h). (B) The SH signal generated by microtubules in HEK293T cells and neuronal cultures treated with taxol (10 nM, 4h) disappears upon fixation with paraformaldehyde (PFA, 4%). As taxol is capable of generating SH signals when forming taxol asters in solution, the increase in SH signal intensity in taxol treated cells could be ascribed to the taxol molecules bound to the microtubule. Taxol asters generating SH in solution show no loss of SH signals before and after fixation (PFA, 4%). Therefore, the SH signal originating from taxol-treated cells is not related to the presence of taxol molecules whose density might be too sparse to generate a detectable signal compared to the densely packed asters formed in solution. Neuronal cultures were imaged at 7 DIV. As the neuronal cultures prepared from TAU deficient mice have GFP-encoding cDNA into exon 1 of TAU, a GFP fusion protein (amino acids 1-31 of TAU protein) is expressed. This can be used to identify TAU deficient neurons in a neuronal culture. As our frequency doubled SH signal has a wavelength (405 nm) that borders the absorption spectrum of GFP itself, the decrease in SH signal intensity in TAU deficient cultures could theoretically be linked to absorption of our SH signal by GFP. Therefore, apart from 810 nm excitation, we also recorded SH signals in wildtype (SHGWT), heterozygous (SHGHET) and knock-out (SHGKO) neurons using 850 nm excitation.

Supplementary
The frequency-doubled wavelength from the red shifted excitation is located closer to the GFP absorption maximum. If absorption of our SH light by GFP would occur, more absorption, thus decreased SH signal intensity, would take place for 850 nm excitation (425 nm SH) as compared to 810 nm excitation (405 nm SHG). Comparison of SH signal ratios of heterozygous (TAU +/-eGFP +/-) or knock out (TAU -/-eGFP +/+ ) over wild-type (TAU +/+ eGFP -/-) cells show no difference between 810 or 850nm excitation (n=27 biologically independent cells; p>0.05 Mann Whitney test). As no decrease in the ratio was found from 810 to 850 nm excitation, we can exclude the absorptive effect of GFP as the reason for decreased SH signal intensities in TAU deficient cultures. Neuronal cultures were imaged at 7 DIV. Bar plots presented as means ± standard error of the mean. All source data are provided as a Source Data file.