Spatially selective multiphoton photothermolysis guided by reflectance confocal microscopy for precise vessel closure in the mouse eye limbus

Abstract

Corneal neovascularization (CNV) is characterized by abnormal vessel growth into the cornea, often impairing vision. Conventional laser therapies for CNV lack precision and risk collateral damage. We developed a multiphoton photothermolysis (MPP) approach using femtosecond (fs) laser-based two-photon absorption guided by real-time reflectance confocal microscopy. Two MPP laser treatment configurations were implemented: dual lasers for imaging (785 nm CW) and treatment (830 nm fs), and a single fs laser for both functions. We demonstrated that in the mouse eye limbus, MPP achieved selective vessel closure without collateral tissue damage. Results also indicate therapeutic effects are mediated by two- rather than one-photon absorption, highlighting MPP’s potential as a precise and safe CNV treatment.

1. Introduction

Corneal neovascularization (NV) is a proliferative vascular disease characterized by the abnormal ingrowth of new blood vessels from the limbal vascular plexus into the normally transparent cornea [1]. It arises when the balance between pro- and anti-angiogenic factors is disrupted and is frequently associated with conditions such as hypoxia (e.g., prolonged contact lens wear), inflammation, mechanical or chemical injury, and various corneal diseases [2–4]. Severe corneal NV can compromise corneal transparency, resulting in edema, scarring, chronic inflammation, and poorer outcomes following keratoplasty. Globally, corneal NV represents a significant public health concern, with an estimated 1.4 million new cases annually in the United States alone, approximately 12% of which progress to vision loss [5]. The prevalence of corneal NV also increases with age, making it a common clinical condition in elderly populations [6]. Moreover, vascularization is observed in about 20% of corneal transplants [5]. These findings underscore the urgent need for effective therapeutic methods to manage corneal NV.

Various treatment methods for corneal neovascularization (NV) have been reported, with the mainstay being topical corticosteroids and anti-VEGF agents [7]. Although these medical therapies can be effective, they are often associated with adverse side effects and show limited efficacy in patients with mature, pre-existing vessels [8,9]. Photonic methods, such as photodynamic therapy (PDT) [10,11] and conventional photocoagulation (laser therapy), have been developed to selectively destroy these existing vessels. However, PDT requires a two-step process that involves the injection of a photosensitizer followed by laser irradiation of the targeted area; this makes the treatment both costly and time-consuming [7].

Laser therapy has been widely used for treating various eye diseases over the past decades due to its safety, efficiency, and relatively low cost [12–14]. Its selectivity typically relies on conventional chromophore-selective photothermolysis. However, it can sometimes result in unintended normal tissue alterations in both lateral and axial directions due to non-specific peripheral tissue heating, especially when the absorption difference is insufficient or when the laser power is set too high. Reported complications of conventional photocoagulation include iris damage, temporary peripheral corneal hemorrhage, and severe corneal thinning, all of which can impair normal physiological function [15–17]. Also, in conventional laser photocoagulation, the table-mounted slit lamp is the most common delivery system for guidance and for transmitting the treatment laser beam. However, its limited resolution, especially in the axial direction, leads to questionable accuracy. Thus, in some cases, fluorescein angiography is required as an aid [18]. In addition, it is hard to record and document the treatment process and parameters, which are essential for future reference and procedure improvement.

Recently, our group has introduced a spatially selective multiphoton photothermolysis (MPP) method that is based on two-photon absorption [19–21]. It was successfully used to precisely close single blood vessels in a mouse ear model [21]. Unlike conventional laser-based techniques, this new method utilizes a near-infrared (NIR) femtosecond (fs) laser that is precisely focused on the target vessel to achieve the treatment effect. The intense two-photon absorption process occurs within the target location because the equivalent absorption wavelength (half of the incident wavelength) coincides with the largest major absorption band of hemoglobin (415 to 430 nm). Outside the focal volume, two-photon absorption drops precipitously due to the low photon density. As a result, the treatment effect is confined to the focal volume, ensuring minimal damage to surrounding tissues.

The aim of this study is to develop a specialized opto/mechanical system suitable for MPP treatment of corneal NV. The system employed uses reflectance confocal microscopy (RCM) to guide MPP treatment of the mouse cornea. The treatment fs laser was focused within the RCM imaging plane, enabling precise optical-sectioning imaging of the cornea to identify and remove target blood vessels at their precise depths. A white-light imaging channel was also incorporated to provide a macroscopic view of the corneal blood vessels, enabling the initial localization of target vessels for RCM imaging. For demonstration purposes, limbal vessels at the edge of the normal mouse cornea were chosen as the treatment targets in this study. Two system configurations were implemented. Both were demonstrated to be able to precisely close single blood vessels in the cornea. One is similar to the point treatment system as in our previous work, which uses two lasers with different wavelengths for RCM imaging and laser treatment, respectively. In the second configuration, the system is updated to use a single fs laser for simultaneous imaging and treatment. Adding another CW laser to the system enabled direct comparison of the treatment effects of the continuous-wave (CW) and fs laser under identical conditions. The comparison demonstrated the predominance of the two-photon absorption therapy mechanism over one-photon absorption.

2. Methods

2.1. Animals

Normal C57BL/6J mice from Jackson Laboratory were bred at the BC Cancer Research Centre (BCCRC) animal facility and used at 3–4 months of age. Mice were anesthetized with chloral hydrate (3%, 1 ml/100 mg) via intraperitoneal (IP) injection. All animal experiments were performed in accordance with a protocol approved by the University of British Columbia Animal Care Committee (certificate number: A15-0234).

In total, nine mice were included in this study: five were used for system optimization and treatment condition exploration; and four mice (eight eyes) were used for treatment experiments, during which 41 limbal vessels were treated.

2.2. Head holder and stabilizer

For convenient and stable imaging and treatment of the mouse eye's limbus region, a mouse holder and eye stabilizer were used to minimize involuntary eye movements following Ref. [22] and with further modifications. As shown in Fig. 1, the head-holding adaptor (SG-4N, Narishige, Japan) is made of a mouse/nose clamp (Fig. 1(a)) and two ear bars (Fig. 1(b)). The eye stabilizer consists of a piece of polyethylene tubing (Fisher Scientific, item no. 22204008) held by the tips of Dumont No. 5 forceps (Fine Science Tools, item no. 11251-10). The head holder and the eye stabilizer were mounted on a custom-built stage and could be moved together with it. The stage could be freely rotated by up to 360°, ensuring that the selected area of the limbus for imaging and treatment could be placed parallel to the focal plane of the objective and the treatment laser beam focus. Transparent tear gel (Alcon) was applied to the eye surface, serving as the immersion medium for the objective lens to match the refractive index and to prevent corneal dehydration. The body and limbs of the mice were fixed on the stage using adhesive tape.

Fig. 1.

Head holder and eye stabilizer for non-invasive imaging and MPP laser treatment of the limbus in the mouse eye. The head of the mouse is restrained in the head holder, and its eye is kept open and stable using the eye stabilizer. a: mouse/nose clamp, b: ear bars, c: polyethylene tube.

2.3. RCM Imaging guided MPP treatment system set-up: configuration 1

The first schematic configuration of the imaging-guided MPP treatment system is shown in Fig. 2, which utilizes two different lasers for in vivo mouse corneal imaging and limbal vessel treatment, respectively. The CW laser (Star Bright 785 XM, Star Bright Laser) with a wavelength of 785 nm is used for performing RCM imaging for real-time guidance. The CW laser beam passes through a spatial filter (lenses L1, L2 and a pinhole in between), a half-wave plate (HWP1), a polarization beam splitter (PBS1), and a quarter-wave plate (QWP1) before entering the scanning unit. The scanning unit then redirects the laser beam through a relay lens pair (L3 and L4), focusing it onto the cornea surface using a 60X water-immersion objective (LUMPLFLN60X/W, Olympus Canada). The backscattered signal from the tissue is collected by the same objective and focused into an avalanche photodiode (APD, C10508, Hamamatsu Corp.) through a 20 µm pinhole for RCM imaging. RCM imaging can be acquired at half video-rate (15 frames per second) with subcellular resolution (1 µm lateral and 3 µm axial). This CW laser illumination RCM imaging mode is designated as cwRCM.

Fig. 2.

Schematic configuration of the cwRCM imaging-guided MPP treatment system. L1-L8: lens, HWP1-HWP2: half-wave plate, PBS1-PBS2: polarized beam splitter, QWP1: quarter wave plate, M1-M4: Ag mirror, D1-D2: dichroic mirror, APD: avalanche photodiode.

A white-light imaging modality was also added to help localize the target vessels before RCM imaging. It is achieved using white light LED illumination and a digital camera (1280 × 1024 resolution, AD4013T-FVW, Dino-Lite) together with two relay lenses. It shares the same objective as the RCM channel and operates in two modes: a large field-of-view (FOV) mode and a high-magnification mode. The optical pathway, highlighted in Fig. 2, illustrates the detection path of the white-light imaging channel and the mechanism for switching between the two channels. In the large FOV mode, the objective is positioned 5-10 mm from the corneal surface to capture the entire limbal region and the entire area accessible for zoom-in RCM imaging. This mode provides initial guidance for target selection before treatment. After the target vessel is selected, the stage is raised to bring the objective closer to the corneal surface, reducing the field of view to match that of the RCM channel closely. This configuration enables simultaneous recording of a high-magnification colored white-light image and an RCM image before and after treatment.

For MPP treatment, the fs laser is set at 830 nm to selectively close the blood vessels in the limbus of the mouse eye. The 785 nm CW laser beam for imaging guidance and the fs laser beam for MPP treatment are merged using a short-pass dichroic mirror (DMSP805, Thorlabs) because they have different wavelengths. The power of the fs laser for treatment is adjusted using a combination of a half-wave plate (HWP2) and a polarization beam splitter (PBS2).

To achieve precise spatially selective photothermolysis, fs laser exposure is modulated using a programmable shutter. By presetting the on-time and off-time of the shutter, the 80 MHz fs laser pulses are chopped to generate treatment laser pulse packets [19]. All the parameters of the laser pulse packet, including duration, period, and packet number, are preset depending on the size and depth of the vessels. The laser beam is automatically blocked after delivering the last laser pulse packets.

2.4. Treatment point calibration for the fs laser beam

Before performing the treatment, a calibration procedure is carried out to determine the treatment point in the field of view of the cwRCM imaging channel. A piece of paper with black ink-printed words was placed on the stage as the sample. It was imaged in the cwRCM channel, as shown in Fig. 3(a); the inked site is bright under 785 nm illumination. When the fs treatment beam is turned on, a spot near the center of the FOV was immediately burned, appearing as a bright white spot with dark edges (red arrow in Fig. 3(b)). The location of this spot represents the site where the fs laser beam was focused, and it was marked by a white crosshair on top of the cwRCM images through the imaging software as an indication of the treatment point.

Fig. 3.

Treatment point calibration for the fs laser beam. The cwRCM imaging of a piece of paper with black ink before (a) and after laser irradiation (b). The red arrow points to the burning point at the center of the FOV indicating the focal point of the fs laser. Scale bar, 50 µm.

2.5. RCM Imaging guided MPP treatment system set-up: configuration 2

Unlike the two-laser configuration, which uses separate lasers with different wavelengths for imaging and treatment, the second configuration utilizes a single laser for simultaneous RCM imaging and MPP treatment. This setup enables the use of a 785 nm fs laser for RCM imaging, designated as fsRCM, as well as for MPP. Adding another 785 nm CW laser beam for treatment also facilitates direct comparison of treatment effects between a 785 nm CW laser and a 785 nm fs laser. This comparison is crucial for confirming that the treatment effect is primarily driven by two-photon absorption rather than one-photon absorption in the near-infrared range.

The single-laser configuration was initially created for imaging-guided confocal Raman measurements by our group, where a single laser is used for simultaneous confocal imaging and point Raman excitation [23]. Herein we apply a similar concept/configuration for imaging-guided MPP treatment. Unlike systems that use separate lasers for imaging and treatment, a single-laser setup presents the challenge of effectively integrating the imaging and treatment beams/pathways while ensuring proper separation of the imaging illumination and detection paths. This challenge was addressed using an optical isolator composed of a Faraday rotator and two Glan-Laser calcite polarizers, which serve as the input polarizer (horizontally polarized) and the output polarizer (polarized at 45°), respectively, as shown in Fig. 4. The forward-travelling imaging illumination beam, initially horizontally polarized after passing through PBS1, is rotated 45° by the Faraday rotator, allowing it to pass through the output polarizer. It then returns to horizontal polarization after HWP3 and is subsequently merged with the vertically polarized treatment beam using a polarized beam splitter for MPP treatment. In contrast, the backward-travelling reflected confocal signals are rotated to 45° by HWP3, then undergo an additional 45° rotation in the same direction via the Faraday rotator, becoming vertically polarized. These signals are then reflected by the input polarizer, pass through a pinhole, and are focused onto the avalanche photodiode (APD) via lens L3, enabling real-time fsRCM imaging guidance.

Fig. 4.

Schematic single-laser configuration of the fsRCM imaging-guided MPP treatment system. An additional 785 nm CW laser was employed to facilitate comparison between CW laser treatment and MPP treatment. L1-L8: lens, HWP1-HWP4: half-wave plate, PBS1-PBS3: polarized beam splitter, QWP1: quarter wave plate, M1-M5: Ag mirror, M6: flip mirror, D2: dichroic mirror, APD: avalanche photodiode, GL1-2: Glan-Laser calcite polarizers

For the treatment comparison study, the 785 nm CW laser was integrated into the optical pathway via flip mirror M6, enabling convenient switching between the fs and the CW laser beams. After the comparison study, the CW laser could be removed, allowing the fs laser to perform both imaging guidance and treatment, further simplifying the system. The white light imaging channel is the same as the two-laser configuration shown in Fig. 2. In future studies, the fs laser could also be tuned to 830 nm for more efficient MPP treatment.

2.6. Treatment point calibration for the CW laser

Before conducting the treatment comparison study, a pre-calibration procedure was performed to determine the treatment point of the CW laser within the fsRCM imaging FOV. Unlike the fs laser, the CW laser could not burn the paper with black ink prints, even at its maximum power (150 mW). Instead, a silicon wafer was used as the sample. The fsRCM imaging laser power was reduced to its minimum level (∼1 mW), and the CW laser power was gradually increased until a tiny bright spot, representing the focal point of the CW laser, appeared near the center of the fsRCM image of the wafer surface (Fig. 5). This location was then marked with a software-generated white crosshair to indicate the CW treatment point.

Fig. 5.

Treatment point calibration for the CW laser. The focus position of the CW laser beam is shown in the FOV of the fsRCM imaging of the silicon wafer surface. The focal point of CW laser is indicated with the white arrow.

3. Results

3.1. RCM imaging for guiding treatment in the limbus of the normal mouse eye using the two-laser configuration

The limbus area in the normal mouse eye was imaged first with the white light imaging channel to obtain both mm-scale macro images and µm-scale micro images. The limbal plexus near the corneoscleral junction was clearly visualized with the large FOV white light imaging, which covers a circular area with a diameter of about 1.8 mm (Fig. 6(a)). The selected region was then imaged with µm-scale white light and cwRCM imaging channels. In this example, the FOV area contains two blood vessel branches: the smaller vessel is approximately 24 µm in diameter, and the larger one is 37 µm, as marked by a red rectangle in Fig. 6(a) taken before the MPP treatment. Post-laser treatment images (Fig. 6(b) and 6(c)) show the occluded vessel branch, indicated by the white arrow, appearing paler in the white-light image. Untreated vessels still appear bright red. White light imaging visualizes blood flow without depth perception, while cwRCM imaging’s optical sectioning capability allows us to see it at the exact depth. Real-time cwRCM imaging showed that the closed vessel had no blood flow, and the nearby untreated larger vessel had moving blood cells, indicated by dynamic reflectance signal changes of moving cells (see video Visualization 1 ^{(2.8MB, avi)} ). Figure 6(c) shows one captured cwRCM imaging frame from the video; the closed vessel and the untreated vessel are indicated by one white arrow and three white arrowheads, respectively. The cwRCM and high-magnification white-light imaging FOVs are set to 200 × 200 µm². The CW laser power on the tissue is 40 mW and the cwRCM imaging frame rate is 15 frames per second.

Fig. 6.

In vivo white light imaging and cwRCM imaging of the limbus in a mouse eye. (a) large FOV white light imaging of the limbus recorded before treatment as guidance for vessel targeting, (b) high-magnification white light imaging, and (c) cwRCM imaging of the vessel branches with a similar FOV as (b). Both images b and c were recorded after we had closed the vessels (indicated by one white arrow). The closed vessel shows a pale red colour in white light imaging and has no blood flow in cwRCM imaging, whereas the normal vessel branch (indicated by three white arrowheads) without receiving laser irradiation has a red colour, full of blood cells running through it (see Visualization 1 ^{(2.8MB, avi)} for playing the video).

3.2. MPP treatment of a single blood vessel in the mouse eye limbus using the two-laser configuration

The spatially selective vessel closure process was demonstrated in the normal mouse eye limbus using the MPP method under real-time RCM imaging guidance. The fs laser power was set to 130 mW, with a pulse width of 150 fs and the peak irradiance was estimated to be 3.7 × 10¹¹ W/cm² following the method described by T. Luo et al [24]. To prevent excessive heat accumulation and dispersion into surrounding tissues, a multiple-pulse packet mode was performed using a programmable shutter. Ten pulse packets were delivered to the target vessel, with each pulse packet having an on-time of 100 ms and an off-time of 100 ms. Figure 7 shows a sequence of extracted cwRCM video frames at 10 time points during MPP treatment in multiple-pulse-packet mode, illustrating gradual changes within the targeted vessel branch after each pulse packet. The white spot represents a strong cwRCM signal, likely due to blood cell coagulation, becoming visible after the first laser pulse packet and gradually diminishes after the laser beam is blocked. Blood flow cessation, indicating vessel closure, occurred after five pulse packets (Fig. 7(f)). The whole process is shown more clearly in video Visualization 2 ^{(8.8MB, avi)} .

Fig. 7.

The process of closing a normal limbal vessel with MPP treatment in multiple pulse packets mode. A total of ten pulse packets were delivered to the single vessel. (a) cwRCM image of the vessel branch before MPP treatment; (b)–(k) show the progressive coagulation of blood cells following each pulse packet; and (l) shows the cwRCM image acquired 30 s after the laser was stopped. Blood flow cessation, indicating vessel closure, occurred after five pulse packets (f). The whole process is shown more clearly in video Visualization 2 ^{(8.8MB, avi)} .

3.3. Demonstration of tissue damage mechanisms during MPP treatment using the one-laser configuration

To demonstrate that tissue damage from the fs laser is primarily due to two- rather than one-photon absorption of near-infrared light, a comparison study was conducted. A CW laser and a fs laser with the same power and wavelength were used to treat the same vessel branch under real-time fsRCM imaging guidance. The targeted vessel was first irradiated with the CW laser (785 nm, 130 mW) for 50 seconds. If no coagulation effects were observed, the fs laser was then used to close the same vessel. The interval between laser irradiations was set to over 1 minute to prevent heat accumulation. The focal points of both lasers were at the same depth but slightly different lateral positions to avoid the CW treatment effects.

fsRCM imaging was recorded during both CW and fs laser treatments. Figure 8(a)–(c) shows fsRCM images of the vessel branch before laser irradiation, after 3 seconds of CW laser exposure (the time required to close a vessel with the fs laser at 130 mW is about 1 to 3 seconds, depending on vessel size), and after 50 seconds of CW laser exposure. The selected vessel's diameter was approximately 30 µm. No changes in vessel morphology or blood flow velocity were observed after 3 seconds of CW laser exposure. One-photon absorption led to heat accumulation after prolonged exposure, but no shrinkage or damage to the blood vessel was noticed. During the 50-second exposure, blood flow velocity slowed slightly (See video Visualization 3 ^{(62.7MB, avi)} ), and two cells (presumably red blood cells, ∼7 µm in diameter) stuck to the vessel wall, as indicated by white arrows in Fig. 8(c).

Fig. 8.

Comparison of the treatment effect on normal limbal vessels using 785 nm CW laser and 785 nm fs laser. The first row (a-c) displays the fsRCM images of the blood vessel before, during (3-second exposure), and after CW laser irradiation (50-second exposure) (The whole procedure is shown in video Visualization 3 ^{(62.7MB, avi)} ). The two white arrows indicate two red blood cells sticking to the vessel wall. The second row (d-f) shows the same blood vessel before, during fs laser exposure, and after 1.1 seconds of fs laser irradiation. The focal points of the laser beams are indicated by white crosshairs. Bright spots occurred at the focal point during the treatment process. The vessel is closed after about 1.1 s exposure. The whole procedure is shown in video Visualization 4 ^{(9.1MB, avi)} .

Figure 8(d)-(f) presents fsRCM images of the same vessel before fs laser treatment (one minute after blocking the CW laser), during fs laser irradiation, and immediately after stopping the fs laser. A multiple pulse packet mode with an on-time of 100 ms and an off-time of 100 ms was applied to the same blood vessel. An intensive photocoagulation effect was observed during laser irradiation, as shown in Fig. 8(e). The vessel closed after 1.1 seconds of fs laser exposure, resulting in cessation of blood flow (See video Visualization 4 ^{(9.1MB, avi)} ). Compared to the unchanged vessel morphology after the first 3 seconds of CW laser irradiation, these results suggest that the fs laser treatment is dominated by two-photon absorption rather than one-photon absorption.

4. Discussion and conclusions

A mouse corneal vessel treatment system was successfully developed based on our previously reported antivascular therapeutic method - multiphoton photothermolysis (MPP). A customized stage with a mouse holder and eye stabilizer was incorporated to expose the corneal limbus for imaging and treatment. Additionally, a white-light imaging channel was integrated to facilitate macroscopic vessel localization. The combination of MPP treatment with real-time RCM imaging enables precise positioning of the laser beam focal spot, continuous treatment process monitoring, and ensures optimal therapeutic outcomes.

In this study, two MPP treatment systems were evaluated: a two-laser configuration previously employed in our work and a newly developed simplified single-laser configuration. Both systems demonstrated selective and precise closure of individual blood vessels in the cornea. For demonstration purposes, normal blood vessel branches at the corneal limbus in mice eye were selected as the target. Successful treatment was achieved using a laser power of 130 mW in pulse-packet mode (100 ms on-time and 100 ms off-time). Real-time observation of the treatment process revealed that MPP treatment induces vessel closure by initiating a central coagulative effect within the vessel. The coagulum gradually expands, while the blood vessel contracts inwardly, ultimately leading to vessel collapse and cessation of blood flow.

The single-laser configuration offers a more compact design and enables simultaneous imaging and treatment with a single laser. Adding another CW laser to the system enabled a direct comparison study between CW and fs laser treatment under the guidance of the same fsRCM imaging channel. This is the first comparison of its kind conducted in vivo, although in our earlier work [19] we performed a similar, albeit less rigorous, comparison using ex vivo mouse skin tissue samples. The results demonstrate that two-photon absorption is the inherent mechanism driving the treatment effect. The MPP approach achieves spatially selective two-photon photothermolysis by focusing the fs laser beam within the vessel, confining the two-photon absorption effect to a tiny focal volume of approximately 1 micrometer. The heat generated by the absorbed energy then diffuses within the blood vessel, inducing coagulation and ultimately leading to vessel collapse and cessation of blood flow. This precise localization within the targeted blood vessels significantly reduces the risk of collateral damage to surrounding tissues. As a result, typical side effects observed in conventional laser treatments, such as corneal and limbal epithelium thinning, limbal stem cell damage, and potential harm to the iris, are largely avoided.

Additionally, the equivalent wavelength of 415 nm falls within hemoglobin’s largest major absorption band, leveraging its high absorption coefficient. Compared to the 532 nm lasers commonly used in clinical treatments, which require higher power levels (120–700 mW) [15,16], the MPP method achieves effective treatment at a significantly lower power level (130 mW), enhancing safety while maintaining efficacy.

In summary, our MPP treatment system offers unique advantages for the treatment of corneal neovascularization (CNV), including enhanced microscopic spatial selectivity, reduced risk of collateral damage, and improved procedural control through real-time imaging guidance. The current experiments, performed on normal limbal vessels, were primarily intended to establish and validate the treatment methodology. As a next step, we will conduct a large-scale animal study using a suture-induced CNV mouse model to evaluate MPP performance on abnormal neovessels. We hope this approach can be translated into human clinical trials in the future for the effective treatment of CNV and other ocular diseases.

Funding

Canadian Institutes of Health Research https://ror.org/01gavpb45 ( PJT-166023); Canadian Dermatology Foundation https://ror.org/00env1778; BC Hydro Employees Community Services Fund; VGH and UBC Hospital Foundation.

Disclosures

The authors declare no conflicts of interest.

Data availability

The authors confirm that the data supporting the findings of this study are either available within the article or could be obtained from the authors upon reasonable request.