Tissue effects of intra-tissue refractive index shaping (IRIS): insights from two-photon autofluorescence and second harmonic generation microscopy

: Intra-tissue refractive index shaping (IRIS) is a novel, non-ablative form of vision correction by which femtosecond laser pulses are tightly focused into ocular tissues to induce localized refractive index (RI) change via nonlinear absorption. Here, we examined the effects of Blue-IRIS on corneal microstructure to gain insights into underlying mechanisms. Three-layer grating patterns were inscribed with IRIS ~180 µm below the epithelial surface of ex vivo rabbit globes using a 400 nm femtosecond laser. Keeping laser power constant at 82 mW in the focal volume, multiple patterns were written at different scan speeds. The largest RI change induced in this study was + 0.011 at 20 mm/s. After measuring the phase change profile of each inscribed pattern, two-photon excited autofluorescence (TPEF) and second harmonic generation (SHG) microscopy were used to quantify changes in stromal structure. While TPEF increased significantly with induced RI change, there was a noticeable suppression of SHG signal in IRIS treated regions. We posit that enhancement of TPEF was due to the formation of new fluorophores, while decreases in SHG were most likely due to degradation of collagen triple helices. All in all, the changes observed suggest that IRIS works by inducing a localized, photochemical change in collagen structure. in Blue-IRIS treated regions of the cornea. TEM revealed that both the pseudo-hexagonal arrangement of collagen fibrils and collagen D-banding were lost in Blue-IRIS


Introduction
Following the success of inducing phase changes in ophthalmic hydrogels in 2006 [1,2], we proposed a new paradigm for refractive vision correction, in which corneal refractive properties can be modified without flap cutting or tissue ablation. This technique, termed Intra-tissue Refractive Index Shaping (IRIS), relies on tightly focused, low-energy femtosecond laser pulses to locally induce phase and refractive index (RI) changes in the corneal stroma via multiphoton absorption. While dopants were needed for 800 nm IRIS to enhance two-photon absorption and attain meaningful RI changes [3], Xu et al. [4] demonstrated that Blue-IRIS can more readily induce larger RI changes because of the significant, native two-photon absorption of cornea at 400 nm. Demonstrating feasibility of this approach, Blue-IRIS was then used to inscribe cylinder lenses in the corneas of adult cats in vivo; the change in refractive power of their eyes was verified with wavefront sensing and shown to persist for a year [5]. Histological analysis of Blue-IRIS in the cat cornea also showed markedly less cell death compared with femtosecond-laser in situ keratomileusis However, whether the mechanisms by which Blue-IRIS induces RI changes in cornea are analogous to those in hydrogels remains unclear. Because Blue-IRIS appears to be a promising technique for vision correction, it has become necessary to thoroughly and systematically investigate the dependence of RI change on IRIS laser scanning and other parameters, and to characterize the tissue effects underlying these RI changes.
Although thinner than human corneas (550 ± 40 µm), rabbit corneas (420 ± 20 µm) share many similarities with human corneal tissue, including microanatomy, biomechanical strength and molecular composition [8,9]. This makes the rabbit cornea a suitable model for vision correction studies. To explore the tissue effects of Blue-IRIS and its underlying mechanisms, a RI profile was built by inscribing IRIS patterns of different phase change magnitudes into the corneal stroma of ex vivo rabbit globes. Local structural changes were imaged and quantified using 810 nm-excited two-photon excited autofluorescence (TPEF) and second harmonic generation (SHG) microscopy. Multiphoton imaging techniques like TPEF and SHG have been shown to be sensitive probes for thick tissues and live animals due to their high axial resolution, large penetration depth and minimal photo-toxicity [10][11][12]. Despite the extremely low two-photon absorption probabilities, tissue and live animals normally have intrinsic fluorescent components with large effective cross-sections in the green and cyan parts of the spectrum, making TPEF feasible for in-depth and label-free tissue imaging [13]. While SHG can be used for similar spatial "optical sectioning" of thick tissues as TPEF, it can provide additional information on the amount and organization of fibrillary collagen in samples, as SHG signals originate from intrinsic, non-centrosymmetric molecular structures [14,15]. SHG sources within the focal volume generally remain phase-matched and emit coherently; however, phase-matching for backward propagation only occurs when the sources are aligned at a specific spatial frequency, thus resulting in much weaker backward-SHG (B-SHG) signals [16,17]. Although there remains some controversy in explaining the complex second harmonic response from inhomogeneous SHG sources, the magnitude ratio of forward-SHG (F-SHG) and B-SHG, mostly denoted by F/B ratio, has been used to estimate the orientation and axial size of collagen fibrils [18,19], as well as distinguishing reactive or diseased tissues from normal ones [20]. The capability of backward detection is also noteworthy, considering the difficulties of collecting F-SHG signals in many in vivo tissue imaging cases. Since the majority of corneal lamellae are highly-hyperpolarized type I collagen, which can emit stronger SHG signals than most SHG generators [16], TPEF and SHG together are ideal techniques to use for visualizing structural modification induced by Blue-IRIS in the corneal stroma.

Blue-IRIS instrumentation and procedure
The femtosecond laser system used to perform IRIS ( Fig. 1(a)) was comprised of a modelocked Ti:Sapphire oscillator (Vitesse; Coherent Corporation, Santa Clara, CA, USA) emitting 800 nm, 100 fs pulses at 80 MHz. The 800 nm beam was first frequency doubled with a second harmonic generator, passing through a metallic variable neutral density filter, and finally focused into each sample with a high numerical aperture (N.A. = 1.05), waterimmersion microscope objective (XLPLN25XWMP2, Olympus, Center Valley, PA, USA). With the laser beam under-filling the objective, the laser beam profile was measured using a knife-edge method [21,22] and the effective N.A. was calculated to be 0.26. Eye globes were mounted on a single-axis, nano-positioning stage on top of a two-axis, linear stage (ANT95-L-Z and ANT95XY; Aerotech Inc., Pittsburgh, PA, USA) to enable precise, large-scale motion control. During IRIS, the objective was stationary while the phase gratings were written along the x-axis with different scan speeds.
Six Dutch-belted rabbit eyeballs were obtained from an approved supplier (Technical Services Specialists, Inc. Walterboro, SC, USA). Post-mortem, the globes were immediately immersed in Optisol-GS (Bausch & Lomb Inc., Rochester, NY, USA) and shipped to our laboratory ov guidelines of statement for the Care and medium used globe was flat previous publ µm-thick roun vacuum sucti patterns were power of 82 speeds rangin consisted of t surface. With changes were patterns were for phase cha 0.3 mm separ fiducial marke

Tissue p
Immediately a in the refriger cornea to retu be much sma under longer was likely du energy depos quantification eyeball was b containing the fully submers mimic intraoc the intact eye.  Globes used for microscopy were prepared by first dissecting around the scleral rim, then immersion-fixing the cornea-containing piece in 4% paraformaldehyde in Phosphate Buffered Saline (PBS, 0.1 M, pH = 7.4) for 10 minutes. The tissue was then transferred to 30% sucrose in PBS (0.1 M, pH = 7.4) at 4 °C for another 24 hours for cryo-protection before sectioning. After that, each cornea was mounted in OCT compound (Tissue Tek; Sakura Finetek, Torance, CA), frozen and sectioned into 20 µm-thick slices using a cryostat (2800 Frigocut E; Leica, Bannockburn, IL). Slide-mounted sections were then rinsed in 0.1 M PBS solution, lightly dried in an oven and cover-slipped with antifade mounting medium (VECTASHIELD H-1000, Vector Laboratories, Burlingame, CA).

Measurement of IRIS-induced phase changes
Phase change was measured on a custom-built MZI using a 633 nm He-Ne laser ( Fig. 1(b)), with the mounted cornea placed on a XYZ translation stage in the test arm. An interferogram and its co-located bright field image (taken by blocking the reference arm), were collected for each IRIS pattern. Similar to the method we have used earlier to quantify phase change in hydrogels [25], the phase map corresponding to each corneal interferogram was calculated using the carrier fringe methods [26], and the induced phase change was unwrapped with the Goldstein's branch cut unwrapping algorithm [27]. To reduce the effect of low-frequency phase variations across the interferogram caused by residual aberrations and wrinkles in the cornea, each IRIS pattern was measured at 10 different positions to collect 10 pairs of phase change values with corresponding standard deviations. All these calculations were conducted automatically using a MATLAB (MathWorks, Natick, MA) program, and the same procedure was repeated on three different eyes.

TPEF and SHG microscopy
Image acquisition was performed with a Fluoview FV300 confocal scanning system interfaced with a BX61WI upright microscope (Olympus, Center Valley, PA), and the 810 nm circular-polarized beam was generated by a MaiTai Ti:Sapphire laser (Spectra Physics, Santa Clara, CA) at 100 fs, 80 MHz, passing through a Berek's compensator ( Fig. 1(c)). Backscattered signal was collected by a XLUMPLFL20XW water-immersion objective (N.A. = 0.95, Olympus), while another Olympus 0.9 N.A. condenser lens was used to collect the forward scattered signal simultaneously. On the forward channel, the emission signal was separated from the excitation beam by a long-pass dichroic mirror (565 DCSX, Chroma, Rockingham, VT), passing through a 405/30 nm bandpass filter (HQ405/30 m-2P, Chroma), and was collected by a photomultiplier tube (PMT, HC125-02, Hamamatsu Corporation, Japan). Besides a short-pass dichroic mirror (670 DCSX, Chroma) used for beam separation, a 525/100 nm bandpass filter (HQ525/100 m, Chroma) was used before PMT to collect the autofluorescence signal on the backward channel. This filter was replaced with a 405/30 nm filter when collecting backward SHG. Laser power, PMT voltage, gain and offset were monitored and kept constant throughout the experiment. Using the surface scan mode, each scan took 5.36 s with a 2 µm step size, obtaining two image stacks on both channels simultaneously. Each image stack was maximum-intensity projected in ImageJ, in order to "autofocus" each stack into one comparable image. Background was defined as the average pixel counts of two equivalent laser-excited image stacks captured from an area without tissue on both channels, and maximum-intensity projected background images were subtracted from all original images to create background-subtracted TPEF, F-SHG and B-SHG images used for subsequent analysis described below. Image stacks of each IRIS pattern were collected sequentially from 4 samples of each eye, and 324 projected images were analyzed in total.
A MATLAB program was written to quantify the TPEF intensity of the IRIS patterns by manually selecting a region of interest (ROI) around IRIS patterns and extracting the histogram of each ROI ( Fig. 2(a)). Since the histogram is a distribution of the number of pixels with respect to their intensity, the two peaks in the histogram correspond to the absolute TPE circled in the Eq. (1) Fig. 2(b)). d an IRIS tting was sced only ctions, we S-induced EF images and pixels mage was in which otherwise ixel value intensity the IRIS asured by neal depth cent ROIs non-IRIS intensity ed by the ence in B-  (Fig. 4). T ge of the corne ween individua n speed, with ã lly, scan spee ber = 0.018), ho the induced ph with the interf front (Fig. 3) affecting the pulse energy deposited at each point within IRIS lines. Because of the eye's exposure limit, the entire IRIS process needs to be as fast as possible. However, sufficient power is also needed to induce localized nonlinear absorption. Therefore, with the laser power fixed, a relationship between phase change and scan speed acts as a useful calibration function for the induction of refractive structures in the cornea. The effects of Blue-IRIS on collagen fibrillar organization were studied using both TPEF and SHG microscopy. Compared to the native corneal stroma, IRIS treated regions were characterized by a dramatic increase in TPEF. Because phase change is a monotonically decreasing function of scan speed, it can be deduced from the quantification results of fluorescence intensity that below the optical breakdown threshold, the higher the induced RI change, the stronger the autofluorescence in treated zones. For the patterns written at 100 mm/s, whose phase changes were too low to be resolved on the MZI, a ~115% increase in TPEF was still observed. Therefore, at 100 mm/s, the IRIS process may have altered protein structure in the corneal extracellular matrix in a manner sufficient to change its fluorescence but insufficient to alter the material's RI. Corneal fluorescence originates from the endogenous fluorophores physiologically present in corneal structures [29], and this enhancement was associated with possible photochemical reactions in macromolecules following laser irradiation. Vogel et al. and Wang et al. reported that free electrons can be generated via multiphoton ionization even when the femtosecond irradiation energy was below the optical breakdown threshold. This ionization field could trigger molecular reactions in the focal volume without introducing damaging thermoacoustic and shock waves [24,30]. Hovhannisyan et al., reported a substantial autofluorescence increase in type I collagen from rat tail tendon and bovine Achilles' tendon under 780 nm, 120 fs laser pulse illumination. They hypothesized that the observed increase of autofluorescence was due to the formation of bi-tyrosine structures, a laser-induced photo-modification affected by both the concentration of native collagen and the amount of photoproducts present [31,32]. Tyrosine is a hydrophobic aromatic amino acid and the main chromophore of collagen molecules [33]. A spectroscopic study from Wisniewski et al. stated that tyrosine exhibited a higher quantum yield of fluorescence compared with other amino acids in collagen polypeptide chain [34]. Manickavasagam et al. studied the laser irradiated collagen degradation of acellular type I collagen gels and noticed a similar phenomenon when gels were point-irradiated using infrared femtosecond lasers [35]. Drawing from previous studies, it may be that the increased autofluorescence in IRIS regions can be attributed to the photo-degradation of collagen fibrils, and the formation of new fluorophores -perhaps tyrosine dimers -in Blue-IRIStreated zones [36].

IRIS-indu
Besides changes in molecular composition suggested by TPEF, SHG offers another perspective on potential alterations in collagen structure at the fibrillar level. As can be seen from Fig. 5, IRIS patterns, especially those with higher phase change, can be easily distinguished from the native stroma in SHG images, with a noticeable suppression of SHG signal in IRIS-treated regions. Quantitative analysis of SHG showed that larger phase changes were associated with larger decreases of F-SHG compared to B-SHG intensity. At the macromolecular level, the elementary source of efficient second harmonic emission in collagen molecules results from the peptide bonds along the three helical chains, even though individual molecules are un-observable by SHG [37,38]. Weakening of F-SHG suggests a loss of the non-centrosymmetric collagen organization in IRIS-treated regions of the stroma, likely caused by alteration of the triple helix structure. We have done an extensive study of the changes in the collagen fibrillar organization using transmission electron microscopy (TEM). This study was done under similar conditions and supports our interpretation of the decrease in SHG [39,40]. In this study TEM, chemical hybridization and immunohistochemistry were used to show significant disruption and alteration of collagen organization in Blue-IRIS treated regions of the cornea. TEM revealed that both the pseudohexagonal arrangement of collagen fibrils and collagen D-banding were lost in Blue-IRIS regions. The TPEF/SHG results in the present manuscript represent an additional "signature" of these ultrastructural changes, confirming with a different approach that IRIS causes highly localized loss of fibrillar structure.
Analysis of B-SHG revealed a lack of significant change as a function of IRIS scan speed. B-SHG signals can be detected if the axial distance between adjacent sources is smaller than the coherence length of the scattering process, which is a function of the excitation and emission wavelengths [19,41,42]. It is known that corneal collagen fibrils are built from about 4 -8 nm wide microfibrils that are stabilized by covalent intermolecular and intramolecular crosslinking [43]. With microfibrils aligned to the fibril long axis, this super-helical fibrillar structure exhibits a ~65 nm axial periodicity and each fibril acts as a scatter of incoming waves [44,45]. The lack of a statistically significant change in difference in B-SHG intensity as a function of IRIS scan speed suggests that there was no significant change in the spaces between neighboring fibrils. This, combined with the significant change in F-SHG further suggests that spatial disorder of fibrils, but not interfibrils spacing, is likely the key contributor to the observed SHG suppression.
The ratio of F-SHG to B-SHG ranged between 0.79 and 1.19 in IRIS-treated regions and did not vary significantly with IRIS scan speed, nor compared to the natural stroma. Although F/B ratio can provide information on the length scale of sources along the optical axis, different properties such as fibril thickness, ratio of different collagen components, and regularity versus irregularity are all possible causes for alterations of F/B ratio [17]. Therefore, at this stage, based on the F/B ratio results, we cannot determine which of these structural properties, if any, were altered after IRIS.
Depending on the deposited laser energy and tissue optical properties, laser-tissue interactions can result in either photochemical or thermal reactions. Under the breakdown threshold, chemical effects or reactions can be induced within tissues in the photochemical regime; in contrast, thermal modification features the denaturation of proteins or tissue removal after moderate temperature increase [46]. Thermal denaturation normally involves heat generation and dissipation during the interactions between tightly focused laser pulses and collagen. Thermal effects could disrupt the collagen matrix via conduction. However, under the damage threshold, this generally cannot introduce significant change in SHG signal since it does not alter the interatomic bonds in collagen molecules [35]. Additionally, an earlier study showed that during collagen thermal denaturation, SHG suppression was accompanied by a decrease in fluorescence, which was caused by the thermal decomposition of the fluorophores [47]. Since the change in IRIS-induced SHG intensity was opposite to that of autofluorescence, Blue-IRIS most likely leads to a photochemical (rather than thermal) degradation of stromal collagen fibril organization.
When analyzing how SHG intensity correlates with IRIS-induced RI change, one limitation lies in the quantification of the absolute SHG intensity of regions outside the IRIS zones. As illustrated by the F-SHG and B-SHG images in Fig. 5, collagen fibers in the anterior stroma are more randomly organized and more densely-packed than those in the posterior stroma; this inhomogeneous distribution can lead to variation of the SHG intensity at different optical planes. Therefore, instead of measuring the average pixel intensity of the entire image, similar size of ROIs at the same axial depth were quantified and defined as the SHG intensity from regions outside the IRIS zones. However, partial corneal tissue was lost in some samples during the cryostat sectioning, which introduced a larger deviation of the SHG intensity change than that of the fluorescence intensity. Nevertheless, compared with the qualitative approach generally used in multiphoton microscopy imaging studies, the present study illustrates that it is possible to use quantifiable SHG parameters to discriminate different refractive phase structures, and that the dual-modal imaging method hereby described can provide useful, complementary insights into structural changes of the corneal collagen matrix.

Conclusion
The present set of experiments allowed us to build a phase change profile for Blue-IRIS as a function of scan speed in the living cornea, using it to calibrate changes in SHG images for the first time. Approximately 0.51 waves of phase change could be obtained from IRIS when writing at 20 mm/s with 82 mW laser power, corresponding to a RI change of + 0.011. The effects of Blue-IRIS on corneal collagen organization were studied using simultaneous TPEF and SHG microscopy. By quantifying the intensity difference between IRIS-treated and untreated regions of the cornea, a significant increase in autofluorescence and decrease of SHG signals were observed in proportion to the increased phase change. This work provides compelling evidence that considering F-SHG as well as B-SHG in a quantitative manner can give useful insights into corneal collagen structure. We believe this particular combination of TPEF and SHG signatures indicates that Blue-IRIS modified the organization of collagen triple helices at a molecular level, and that this caused degradation of collagen fibril arrangement. Ongoing research is seeking to ascertain more specifically what features of the collagen microstructure, e.g. fibril density, diameter, direction, and molecular composition, are modified by Blue-IRIS. These findings provide valuable insights into potential photophysical mechanisms by which Blue-IRIS induces a RI change in the cornea.