Isolation of Mouse Retinal Capillaries and Subendothelial Matrix for Stiffness Measurement Using Atomic Force Microscopy

Abstract

Retinal capillary degeneration is a clinical hallmark of the early stages of diabetic retinopathy (DR). Our recent studies have revealed that diabetes-induced retinal capillary stiffening plays a crucial and previously unrecognized causal role in inflammation-mediated degeneration of retinal capillaries. The increase in retinal capillary stiffness results from the overexpression of lysyl oxidase, an enzyme that crosslinks and stiffens the subendothelial matrix. Since tackling DR at the early stage is expected to prevent or slow down DR progression and associated vision loss, subendothelial matrix, and capillary stiffness represent relevant and novel therapeutic targets for early DR management. Further, direct measurement of retinal capillary stiffness can serve as a crucial preclinical validation step for the development of new imaging techniques for non-invasive assessment of retinal capillary stiffness in animal and human subjects. With this view in mind, we here provide a detailed protocol for the isolation and stiffness measurement of mouse retinal capillaries and subendothelial matrix using atomic force microscopy.

Introduction

Retinal capillaries are essential for maintaining retinal homeostasis and visual function. Indeed, their degeneration in early diabetes is strongly implicated in the development of vision-threatening complications of diabetic retinopathy (DR), a microvascular condition that affects nearly 40% of all individuals with diabetes¹. Vascular inflammation contributes significantly to retinal capillary degeneration in DR. Past studies have demonstrated an important role for aberrant molecular and biochemical cues in diabetes-induced retinal vascular inflammation^2,3. However, recent work has introduced a new paradigm for DR pathogenesis that identifies retinal capillary stiffening as a crucial yet previously unrecognized determinant of retinal vascular inflammation and degeneration^4,5,6.

Specifically, the diabetes-induced increase in retinal capillary stiffness is caused by the upregulation of collagen crosslinking enzyme lysyl oxidase (LOX) in retinal endothelial cells (ECs), which stiffens the subendothelial matrix (basement membrane)^4,5,6. Matrix stiffening, in turn, stiffens the overlying retinal ECs (due to mechanical reciprocity), thus leading to the overall increase in retinal capillary stiffness⁴. Crucially, this diabetes-induced retinal capillary stiffening alone can promote retinal EC activation and inflammation-mediated EC death. This mechanical regulation of retinal EC defects can be attributed to altered endothelial mechanotransduction, the process by which mechanical cues are converted into biochemical signals to produce a biological response^7,8,9. Importantly, altered EC mechanical cues and subendothelial matrix structure have also been implicated in choroidal vascular degeneration associated with early age-related macular degeneration (AMD)^10,11,12, which attests to the broader implications of vascular mechanobiology in degenerative retinal diseases.

Notably, retinal capillary stiffening occurs early on in diabetes, which coincides with the onset of retinal inflammation. Thus, the increase in retinal capillary stiffness may serve as both a therapeutic target and an early diagnostic marker for DR. To this end, it is important to obtain reliable and direct stiffness measurements of retinal capillaries and subendothelial matrix. This can be achieved by using an atomic force microscope (AFM), which offers a unique, sensitive, accurate, and reliable technique to directly measure the stiffness of cells, extracellular matrix, and tissues¹³. An AFM applies minute (nanoNewton-level) indentation force on the sample whose stiffness determines the extent to which the indenting AFM cantilever bends- the stiffer the sample, the more the cantilever bends, and vice versa. We have used AFM extensively to measure the stiffness of cultured endothelial cells, subendothelial matrices, and isolated mouse retinal capillaries^4,5,6,11,12. These AFM stiffness measurements have helped identify endothelial mechanobiology as a key determinant of DR and AMD pathogenesis. To help broaden the scope of mechanobiology in vision research, here we provide a step-by-step guide on the use of AFM for stiffness measurements of isolated mouse retinal capillaries and subendothelial matrix.

Protocol

All animal procedures were performed in accordance with the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research and approved by the Institutional Animal and Care Use Committees (IACUC; protocol number ARC-2020-030) at the University of California, Los Angeles (OLAW institution animal welfare assurance number A3196-01). The following protocol has been performed using retinal capillaries isolated from adult (20-week-old) male C57BL/6J mice weighing ~25 g (diabetic mice) and ~32 g (nondiabetic mice; Jackson Laboratory).

1. Isolation of mouse retinal capillaries for AFM stiffness measurement (Days 1-4)

NOTE: This protocol, reported in a recent study⁴, details the enucleation and mild fixation of the mouse eye, retinal isolation, and trypsin digestion, and subsequent mounting of the resultant retinal vasculature on microscopy slides for AFM stiffness measurement.

1. Enucleation and mild fixation (Day 1)
   1. After euthanizing the mouse with carbon dioxide exposure followed by cervical dislocation, insert micro forceps behind the eyeball, hold onto the muscular attachment, and carefully pull out the eye without pinching the micro forceps too tightly, which might sever the optic nerve.
   2. Place the enucleated eye in 5% (v/v) formalin in PBS for 24 h at 4 °C for mild fixation.
      NOTE: Based on experience⁴, unfixed or more mildly fixed eyes yield fragile capillaries that become fragmented during AFM sample preparation and stiffness measurement.
2. Retinal isolation (Day 2)
   1. Place the fixed eye on a piece of wax paper under a dissecting microscope. Next, using micro forceps, hold the remnants of the muscle and optic nerve attached to the outside of the posterior eye and orient the eye so that the cornea faces one side (Figure 1).
   2. Using a surgical blade, make an incision 1-2 mm behind and parallel to the limbus (cornea-sclera junction). Next, while holding the eye in place with the micro forceps, apply a downward force with the blade and continue pressing until the anterior segment of the eye is totally separated from the posterior end (Figure 1). Discard the anterior segment and lens. Do not saw back and forth as that may cause retinal damage.
   3. Transfer the posterior segment (sclera, choroid, and retina) into a 6 cm dish filled with PBS (pH 7.4).
   4. Using a micro spatula and simultaneously gently holding the optic nerve with micro forceps, scoop out the retina. Store the retina in a 2 mL microcentrifuge tube containing PBS at 4 °C until capillary isolation.
       
3. Retinal rinses
   1. To rinse one retina, fill up six wells of a 12-well plate with 1 mL of double distilled H₂O (ddH₂O). Next, using an inverted Pasteur pipette, transfer the retina from the 2 mL microcentrifuge tube into the first well.
   2. Rinse the retina on the orbital shaker at 120 RPM for 30 min at room temperature (RT).
   3. Using a P1000 pipette, carefully pipette water up and down adjacent to the retina, blowing water at the retina to cause gentle agitation.
   4. Using the inverted glass pipette, transfer the retina to the second well and repeat steps 1.3.2 and 1.3.3 4 times, with each rinse done in a fresh (ddH₂O-containing) well.
   5. Finally, transfer the retina to the sixth well and rinse it overnight (O/N) at RT on the orbital shaker set at 100 RPM to facilitate the separation of retinal neuroglia from blood vessels during trypsin digestion.
4. Retinal trypsin digestion (Day 3)
   1. Prepare a 10% (w/v) trypsin solution by dissolving trypsin 1:250 powder in Tris buffer (pH 8; 0.1 M). Mix gently by inverting the conical tube several times to minimize bubble formation, which makes it harder to confirm trypsin solubility.
   2. Equilibrate the 10% trypsin solution in a 37 °C water bath for 10-15 min. Once the trypsin powder has completely dissolved, filter the solution using a 0.2 µm syringe filter.
   3. In a 12-well plate, add 2 mL/well/retina of the filtered 10% trypsin solution. Add some extra trypsin solution in the neighboring well (to equilibrate the glass pipette walls for subsequent steps).
   4. Add 3 mL of ddH₂O in three wells of a 6-well plate (dedicate these three wells for one retina).
   5. In a 15 mL conical tube, insert a P200 tip before adding 4 mL of the filtered 10% trypsin solution. Transfer this conical tube to the 37 °C bath to allow the tip to be soaked in trypsin for 2-3 min, as that helps prevent the retina from sticking to tip walls in the latter steps.
   6. After O/N rinsing of the retina (from step 1.3.5), use a P1000 pipette to carefully pipette water up and down adjacent to the retina, squirting water on the retina to cause gentle agitation.
   7. Using an inverted glass pipette, transfer the rinsed retina to trypsin-containing-well of the 12-well plate (from step 1.4.3). Ensure that the retina is transferred with a minimal amount of residual ddH₂O so as not to dilute the trypsin solution significantly. Incubate for 3 h at 37 °C.
   8. Centering the trypsin-soaked P200 tip (from step 1.4.5) on the optic nerve area, gently pipette the entire vascular network up and down to dissociate the non-vascular tissue¹⁴.
   9. Using an inverted glass pipette pre-rinsed in trypsin (by pipetting trypsin up and down 5 times), transfer the retinal vasculature to the first ddH₂O-containing well of the 6-well plate (from step 1.4.4).
   10. Swirl the plate and pipette the vasculature up and down using an inverted trypsin-rinsed glass pipette to remove any residual non-vascular tissue.
   11. Using the glass pipette, transfer the retinal vasculature to the second ddH2O-containing well of the 6-well plate and store at 4 °C until mounting on the slide for AFM stiffness measurement.
       NOTE: Hydrated retinal vasculature (in PBS) is structurally intact at 4 °C for up to 2 weeks. However, routine stiffness measurements should be made from freshly isolated capillaries.
5. Mounting the retinal vasculature for AFM stiffness measurement (Day 4)
   1. Using an inverted trypsin-rinsed glass pipette, transfer the retinal vasculature to the third ddH₂O-containing well of the 6-well plate to remove any residual trypsin solution.
   2. Thoroughly clean (using ddH₂O and wiper tissue) a charged surface microscopy slide that will be used for mounting the vascular network for AFM stiffness measurement.
      NOTE: The charged surface of the slide provides strong binding for the vascular network during AFM measurement.
   3. Label the slide and draw a 4-5 cm circular region around the center with a hydrophobic pen to avoid water spillage during the subsequent steps.
   4. Using an inverted trypsin-rinsed glass pipette, carefully transfer the retinal vasculature to the center of the marked region.
   5. Using a P200 pipette, carefully remove as much excess water as possible from one edge without accidentally aspirating the vasculature into the tip.
   6. Place the slide in a biosafety cabinet (BSL-2 hood) close to the front side (filtered mesh) and let the residual water evaporate.
   7. Once the vasculature is almost dry, under a phase contrast microscope (4x magnification), carefully rehydrate the vasculature by slowly adding ddH₂O with a P200 pipette on to one side of the marked region. Adding ddH₂O directly on the vasculature causes it to detach.
   8. Take phase contrast images of the rehydrated vasculature at 4x and 20x magnifications to ensure it has good structural integrity and is properly spread out (not folding on itself) and attached to glass slide, which are essential criteria for reliable AFM stiffness measurement.

2. Obtaining subendothelial matrix from retinal microvascular endothelial cell (REC) cultures (Days 1-17)

NOTE: This protocol, adapted from Beacham et al.¹⁵ and reported in recent studies^4,5,6, describes REC culture on modified glass coverslips, followed by decellularization to obtain subendothelial matrix for subsequent AFM stiffness measurement.

1. Preparation of gelatin-coated glass coverslips and cell plating (Day 1)
   NOTE: Perform the gelatin coating and subsequent PBS⁺⁺ rinsing steps in a tissue culture hood before moving to a chemical fume hood for subsequent glutaraldehyde and ethanolamine treatment steps.
   1. Place one autoclaved 12 mm diameter circular glass coverslip per well of a sterile 24-well plate and add 500 µL of pre-warmed sterile 0.2% (w/v) gelatin (diluted in PBS containing calcium and Magnesium; PBS⁺⁺) to each well containing coverslip. Incubate for 1 h at 37 °C.
   2. Aspirate the gelatin solution, rinse the coverslips once with sterile PBS⁺⁺, and crosslink the coated gelatin by adding 500 µL of 1% (v/v) glutaraldehyde for 30 min at RT.
   3. Collect and properly dispose (as per institutional guidelines) the glutaraldehyde solution before rinsing the coverslips with PBS⁺⁺ for 5 min on an orbital shaker at 100 RPM in RT. Repeat this rinsing step 5 times. During the first three rinses, lift the coverslips using a sterile tweezer to allow thorough rinsing of the glutaraldehyde. Any residual amount can reduce cell viability.
   4. Add 800 µL of 1 M ethanolamine to the gelatin-crosslinked coverslip for 30 min at RT to quench any remaining traces of glutaraldehyde in the well.
   5. Collect and properly dispose (as per institutional guidelines) of the ethanolamine solution and rinse the coverslips 5 times with PBS⁺⁺ for 5 min on an orbital shaker at 100 rpm in RT. During the first three rinses, lift the coverslips using a sterile tweezer to allow thorough rinsing of the ethanolamine. Any residual amount can reduce cell viability.
   6. The crosslinked coverslips are now ready for cell plating.
      NOTE: Although we routinely plate cells immediately after coverslip preparation, it may be worthwhile to compare cell plating on crosslinked gelatin coverslips stored in PBS⁺⁺ at 4 °C for 1-2 days.
   7. Under sterile conditions in a tissue culture hood, plate retinal microvascular endothelial cells (RECs) in 500 µL medium/well at a density that achieves 100% confluence within 24 h, as confirmed by phase contrast microscopy.
      NOTE: Reaching confluence within 24 h is important as the resulting REC quiescence will increase the ability of cells to secrete matrix (refer to the following step). In our experience with human RECs, a plating density of 4 x 10⁴ cells/cm² is sufficient to reach confluence within 24 h. However, as the REC size varies with species (e.g., mouse RECs are significantly smaller), the initial plating density may need to be optimized.
2. Subendothelial matrix production by REC culture (Day 2-16)
   1. After cells reach confluence (~24 h), replace the culture medium with fresh medium supplemented with sterile-filtered ascorbic acid (final concentration of 200 µg/mL).
      NOTE: Any REC treatment with disease risk factors (e.g., high glucose, advanced glycation end products, etc.) or pharmacological agents can begin now, along with ascorbic acid treatment.
   2. Change the ascorbic acid-supplemented culture medium every other day for 15 days.
      NOTE: Ascorbic acid is unstable in solution. Prepare fresh 100X stock solution every other day for medium supplementation.
3. Decellularization of REC cultures to obtain the subendothelial matrix (Day 16)
   1. After 15 days of ascorbic acid treatment, remove the medium and rinse the cells with calcium/magnesium-free PBS.
   2. Decellularize the REC cultures by adding 250 µL/per well of warm decellularization buffer (20 mM ammonium hydroxide and 0.5% Triton X-100 in PBS) for ~2-3 min. Confirm the removal of cells under a phase contrast microscope (10x magnification).
   3. Gently remove the decellularization buffer without disturbing the REC-secreted subendothelial matrix and add 0.5 mL of PBS to each well.
      NOTE: To ensure the subendothelial matrix does not detach during this and the subsequent rinsing steps, perform gentle rinsing only with a P1000 pipette. Do not perform vacuum aspiration.
   4. Store the plates at 4 °C overnight to stabilize the REC-secreted matrix.
4. DNase treatment of subendothelial matrix to remove cellular debris (Day 17)
   1. Rinse the subendothelial matrix from step 2.3.4 with 800 µL of PBS/well.
   2. Gently remove the PBS and incubate the matrix with 200 µL of RNase-free DNase I (30 K units) at 37 °C for 30 min to remove all traces of cellular debris. Check the efficiency of DNase treatment by carefully looking for cell debris under a phase contrast microscope.
   3. Gently remove the DNase I solution and rinse the subendothelial matrix twice with 800 µL of PBS/well.
   4. Check that the macroscale fibrous subendothelial matrix is visible under a phase contrast microscope (10x magnification). Visualization of the finer nanoscale matrix fibers requires AFM topographical scanning or high-resolution confocal imaging of immunolabeled matrix proteins at 100x magnification.
   5. Immediately use the fresh (unfixed) subendothelial matrix for AFM stiffness measurement.
   6. Following AFM measurement, fix matrix samples with 1% (v/v) paraformaldehyde (15 min at room temperature) and stored at 4 °C for immunolabeling with antibodies against subendothelial matrix proteins and/or crosslinking enzymes.

3. AFM stiffness measurement

NOTE: This protocol, adapted from a standard AFM user manual and reported in recent studies^4,5, details the acquisition and analysis of stiffness data from retinal capillaries and subendothelial matrix using an AFM and data analysis software. Although the steps outlined below are based on a specific model of AFM (see Table of Materials), the underlying principles are generally applicable to all AFM models.

1. Cantilever probe selection
   ​NOTE: Soft biological samples require soft cantilevers that can bend upon sample indentation while the probe dimensions are selected to match the dimensions of the sample.
   1. For stiffness measurement of 5-8 µm diameter mouse retinal vessels, use a 1 µm radius cantilever probe with a spring constant (k) of 0.2-0.3 N/m. For stiffness measurement of a subendothelial matrix composed of nanoscale fibers, use a 70 nm radius cantilever probe with a spring constant (k) of 0.06-0.1 N/m.
2. Cantilever mounting
   1. On the AFM computer, open the software that controls the AFM unit and the camera attached to a phase contrast microscope that is used to visualize the cantilever and sample.
   2. Fix the cantilever holder onto the cantilever changing stand and mount the selected cantilever on the holder using watchmaker forceps under a stereoscope without physically damaging the cantilever. Tighten the screw on the holder to secure the cantilever.
   3. Place and lock the cantilever holder on the AFM head that is resting on the stand.
   4. Using the step motor function in the AFM software, withdraw the AFM head to the highest point and set that position as point zero in the software. This step prevents the AFM cantilever from accidentally hitting the sample stage while mounting the AFM head (next step).
   5. Carefully lift the AFM head from its stand and mount it on the AFM sample stage by placing the legs in their respective slots.
3. Laser alignment
   ​NOTE: The initial laser alignment is performed without any sample (i.e., in Air mode) as it ensures a more precise alignment of the laser beam on the cantilever tip (by preventing laser refraction in liquid). However, as biological samples are immersed in a medium that has a different refractive index than air, it is important to realign the laser in the medium before stiffness measurement.
   1. For laser alignment in air, place the AFM head on the sample stage and use the 10x objective and attached camera to visualize the cantilever on the monitor in live mode. The infrared laser is hidden from the user's view but is visible with a CCD camera.
   2. Select the Contact Mode Force Spectroscopy function on the software and open the window entitled Laser alignment.
   3. Using the two screws marked laser lateral on the AFM head, focus the laser beam on the cantilever such that the SUM value in the laser alignment window is the highest. To achieve this, focus the laser at or around the center of the cantilever.
   4. Using the Detector adjustment screws, adjust the photodetector such that the laser spot is at the center of the laser alignment window.
   5. Using the Mirror adjustment screw, adjust the mirror to ensure that the SUM value is at the highest possible value.
      NOTE: A high SUM value ensures a sensitive and accurate assessment of sample stiffness. Thus, both the laser alignment and mirror adjustment steps should be performed to obtain the highest SUM value.
   6. Next, for laser alignment in liquid, add the desired culture medium in a dish and mount it firmly on stage to prevent any vibration or drift, which creates measurement artifacts during sample indentation.
   7. While looking at the live camera feed focused on the cantilever, lower it into the liquid using the step motor function.
   8. Repeat steps 3.3.4 and 3.3.5 to ensure that the SUM value remains the highest and the laser spot is at the center of the laser alignment window.
4. Calibration of cantilever spring constant
   NOTE: Although cantilever probes come with a manufacturer-calibrated spring constant (k), it is good practice to independently verify its value (in liquid) before the start of a measurement. Use a clean glass surface that minimizes interactions between the cantilever probe and the glass surface.
   1. The setpoint force applied by the cantilever sets the maximum laser deflection from the cantilever that is allowed for a force indentation. Set it initially at 1.5 V. If the cantilever fails to approach at this given setpoint force, increase it incrementally until it indents the sample and generates a force indentation curve.
   2. Z Length signifies the maximum distance by which the z-piezo withdraws after the cantilever has reached the setpoint during sample indentation. Z length should be at least enough to ensure the cantilever probe separates cleanly from the sample. For calibration of spring constant on a glass surface, set Z Length at 1 µm.
   3. Z speed refers to the speed at which the z-piezo moves the cantilever vertically down towards the sample during force indentation. It should be optimized because a very low speed will primarily capture the viscous behavior of the sample, while a very high speed will primarily capture the elastic behavior. The optimal speed is expected to capture the true viscoelastic nature of biological samples. Set the Z speed to 2 µm/s for the viscoelastic biological samples.
   4. The Target height on the Z range indicates the approximate distance from the sample at which the z-piezo rests after measuring a force curve and before moving to a different location on the sample. The Z Range target height should be greater than the height of the sample features. Set target height on the Z range at 7.5 µm.
   5. Using the Step Motor function, bring the cantilever fairly close to the sample surface (judging by the focus in phase contrast image at 10x magnification) before selecting the Approach function in the contact mode force spectroscopy window to bring down the cantilever in smaller increments of 15 µm.
   6. Click Acquire to capture a force curve.
   7. Select the Force Curve, open it in the calibration manager window, and select Contact-based Mode in the method section.
   8. Using the Select-fit Range function, select the Cantilever Retraction Curve for a linear curve fit. Next, check the Sensitivity Check box to convert the force unit from V to N.
   9. Lift the cantilever by 100-200 µm in the liquid and select the Thermal Noise function. Again, using Select Fit Range, fit the thermal noise bell curve with a Lorentz curve. After curve fitting, select the Spring Constant (k) box. Confirm that spring constant (k) is close to the manufacturer's value and note it down for future reference. After calibration, the unit for setpoint will change from mV to nN.
      ​NOTE: Thermal noise is the natural frequency of the cantilever at a particular temperature. Correction for thermal noise is required for accurate spring constant measurement.
5. Acquisition of force-distance curves
   1. Place the slide-mounted sample (retinal vessels or subendothelial matrix) on the AFM stage and select Contact Mode Force Spectroscopy in the software's experiment section.
   2. Set the setpoint force at 0.5 nN and click Approach, which is significantly higher than the thermal deflection of both SAA-SPH 1 µm and PFQNM-LC-A 70 nm cantilever probes, ensuring a smooth cantilever approach towards the sample.
   3. After the cantilever has detected the surface and returned to its resting target height, set the setpoint at 0.2 nN for the stiffer SAA-SPH 1 µm cantilever probe or 0.1 nN for the softer PFQNM-LC-A 70 nm cantilever probe. This setpoint adjustment ensures that both stiff and soft cantilevers bend readily during sample indentation (refer to section 3.1).
   4. Set Z Length at ~2.5 µm to ensure clean separation of the cantilever probe from biological samples during retraction (refer to step 3.4.2) and Z speed at 2 µm/s (refer to step 3.4.3).
   5. Using the stage screws on the AFM stage, carefully position the cantilever probe at a desired location on the sample.
      NOTE: Since capillaries do not always spread flat on the slide, it is safe to withdraw the cantilever a further 10-20 µm from its resting target height before moving the stage.
   6. Click Approach to first bring the cantilever closer to the sample and then click Acquire to capture the force curves for the desired locations. Save all force curves for analysis.
6. Data analysis
   1. Open the force curve in the data processing software.
   2. Select the Retraction Force Curve for data analysis (similar to step 3.4.8).
   3. Using the Data smoothing function, smoothen the force curve with a Gaussian filter to remove the unwanted noise in the acquired data.
   4. Using the Baseline subtraction function, adjust the value of the slope and magnitude of the (non-contact) baseline portion of the force curve to zero.
   5. Click on the Contact Point function to automatically bring the contact point of the force curve to the (0,0) coordinates on X- and Y-axes.
   6. Using the Vertical Tip Position function, calculate the actual vertical position of the cantilever on the Y-axis by correcting for any sample indentation.
   7. On the processed force curve, apply the Elasticity Fit function by first selecting the tip shape and radius (based on the selected cantilever probe), followed by force curve fitting using the Hertz/Sneddon model.
      1. If matrix indentation by the 70 nm radius probe exceeds 70 nm, which is typically the case, select the Paraboloid tip shape. For capillary stiffness measurement, sample indentation by the 1 µm radius probe never exceeds 1 µm, so select the Sphere tip shape. Further, if the contact point of the fitted curve does not coincide with the contact point of the actual retraction curve, select the Shift Curve check box.
   8. Note and save the value of Young's modulus (stiffness).

Representative Results

Mouse retinal capillaries
AFM stiffness measurement of isolated retinal capillaries involves sample handling steps that could potentially damage their mechanostructural integrity. To prevent this and thereby ensure the feasibility, reliability, and reproducibility of AFM measurements, the enucleated eyes are fixed in 5% formalin overnight at 4 °C prior to vessel isolation. This mild fixation protocol with reduced formalin concentration, low fixation temperature, limited fixation time, and lack of corneal puncture was developed to minimize any potential crosslinking/stiffening artifacts caused by chemical fixation. As shown in Figure 2B,C, this relatively mild fixation ensures that the isolated retinal vasculature is structurally robust and sufficiently durable for the AFM measurement. In contrast, retinal vessels isolated from unfixed eyes (using the hypotonic method)⁴ or briefly fixed eyes (for 8 h) become fragmented or collapse, thereby making them unsuitable for AFM measurement (Figure 2B).

Retinal subendothelial matrix
Vascular stiffness reflects the combined stiffness of vascular cells and the basement membrane (subendothelial matrix)⁴. Since cells adapt to matrix stiffness by undergoing a similar change in their own stiffness, a process termed mechanical reciprocity⁹, subendothelial matrix stiffness, becomes an important determinant of the overall vascular stiffness. For matrix stiffness measurement, it is important to obtain a homogeneously dense subendothelial matrix. For human retinal ECs grown in ascorbic acid-supplemented culture medium, this usually takes 10-15 days (Figure 3A)^4,5,6. This difference in culture period may arise from lot-to-lot differences in commercially available primary retinal ECs. Further, we generally find that commercially available retinal ECs from C57BL/6 mice deposit a denser matrix when compared with primary human retinal EC culture, thus indicating species-specific differences. As shown in Figure 3A, phase contrast images only provide a gross view of the matrix at a macro scale. However, the finer nano-to-micro-scale fibrillar structure becomes apparent in high-resolution confocal fluorescence images of the matrix immunolabeled with antibodies against matrix structural proteins collagen IV and fibronectin (Figure 3B). It should be noted that these matrix proteins also provide instructive cues to endothelial cells by binding to specific integrin receptors⁹.

AFM stiffness measurement
The selection of appropriate cantilever stiffness (spring constant, k) and probe dimension is essential for reliable and sensitive measurements. These parameters should be chosen to match the stiffness and dimensions of the biological sample. After the selected cantilever has been mounted on the AFM, the infrared laser beam must be focused on the tip of the cantilever, and the reflected laser spot must be centered on the photodetector. This laser alignment ensures precise and sensitive detection of cantilever deflection and, consequently, stiffness measurement. An AFM stiffness measurement begins with the z-piezo moving the cantilever vertically down toward the sample. There is no cantilever deflection at this time, which produces a flat baseline of the approach curve (Figure 4A). As the cantilever probe contacts and indents the sample, the cantilever bends, causing laser deflection on the photodetector, which is depicted by the vertical deflection of the approach curve. After applying a preset indentation force (setpoint force) on the sample, the cantilever retracts to the starting position (Z target height) away from the sample. The deflected retraction curve is then fitted to the Hertz/Sneddon model to calculate the sample's Young's modulus (stiffness).

From the representative force indentation measurement shown in Figure 4, it is clear that the approach and retraction curves obtained from a retinal capillary isolated from a diabetic mouse (Figure 4B) are substantially steeper than those obtained from a nondiabetic mouse (Figure 4A). The steeper slope of force indentation curves indicates greater cantilever deflection caused by higher sample resistance to force indentation, which reflects higher sample stiffness¹³. Indeed, subsequent data analysis of multiple force indentation curves revealed that mouse retinal capillaries become significantly stiffer in diabetes⁴. It should also be noted that contact between the cantilever probe and biological samples often causes nonspecific surface adhesion, which leads to negative cantilever deflection during retraction, as seen from the extension of the retraction curve beyond the baseline (Figure 4B). Further, biological samples like cells, matrix, and blood vessels are viscoelastic by nature and thus may undergo some permanent deformation and/or change in apparent stiffness following force indentation. If so, this will be reflected in the misalignment of approach and retraction curves (hysteresis). Indeed, comparing Figure 4A and Figure 4B confirms the expected trend where force indentation of softer capillaries in nondiabetic mice produces greater hysteresis than that seen in their stiffer counterparts. As previously reported^5,6, stiffness (Young's modulus) of subendothelial matrices obtained from retinal EC cultures is also calculated in the aforementioned manner.

Figure 1: Schematic illustration of the incision cut made on a mouse eye for retinal isolation. Using tweezers to hold the optic nerve, a scalpel is used to make a full incision posterior to the limbus to separate the mouse retina from the lens and anterior chamber. The purple dashed line shows the location of the vertical incision. This schematic was drawn using a scientific image and illustration software. Please click here to view a larger version of this figure.

Figure 2: Protocol optimization for isolation of intact mouse retinal vessels for AFM measurement. (A) The stereoscope image shows an intact retina isolated from a mildly fixed mouse eye prior to the capillary isolation steps. Scale bar: 2 mm. (B) Representative phase contrast images at 4x magnification show mouse retinal vessels obtained using the different isolation methods. Comparing the structural integrity and durability of the isolated vessels, trypsin digestion of retinas from enucleated eyes fixed in 5% formalin for 24 h at 4 °C (red box) was found to yield the most suitable retinal vasculature for AFM stiffness measurement. Retinal vessels isolated using this method exhibited a clear vascular network that spread uniformly along the glass surface. Scale bar: 500 µm. (C) High magnification (20x) view of an intact retinal capillary network, similar to that shown in (B), confirms the high structural integrity of capillaries obtained from mildly fixed eyes. Scale bar: 100 µm. This figure has been modified from⁴. Please click here to view a larger version of this figure.

Figure 3: Decellularized matrices obtained from primary human and mouse REC cultures. (A) Representative phase contrast images at 20x magnification showing subendothelial matrix aggregates on glass coverslips following decellularization of 10-day (10 d) or 15-day (15 d) cultures of human or mouse RECs. Scale bar: 100 µm. (B) Representative high magnification (100x) confocal fluorescence images of decellularized matrices obtained from 15 d human REC cultures and labeled with anti-collagen IV and anti-fibronectin antibodies reveal a dense nanofibrillar collagen IV and fibronectin matrix. Scale bar: 20 µm. Please click here to view a larger version of this figure.

Figure 4: Force distance curves from AFM stiffness measurement of mouse retinal capillaries. Line graphs indicate a representative approach (light color) and retraction (dark color) curve from a single force indentation measurement at one location of a mouse retinal capillary isolated from a (A) nondiabetic or (B) diabetic mouse. The force curves obtained using an SAA-SPH 1 µm radius hemispherical cantilever probe, plot the relationship between cantilever-sample distance (controlled by the z-piezo) and the applied vertical force that causes cantilever deflection. (A) The yellow arrow indicates the z-piezo-driven cantilever approach towards the sample, the white arrow indicates the contact point where the cantilever probe makes contact with the sample, and the green arrow indicates the cantilever deflection up to a setpoint (peak) indentation force (*). (B) Both approach and retraction curves obtained from a retinal capillary isolated from diabetic mice exhibit a markedly steeper slope than those from their nondiabetic counterparts (shown in A), which indicates higher capillary stiffness in diabetic mice. The arrowhead indicates a dip in the retraction curve below the baseline, which reflects the negative deflection of the cantilever probe caused by adhesion between the probe and sample during indentation. Please click here to view a larger version of this figure.

Discussion

AFM has been widely used to measure disease-associated changes in the stiffness of larger vessels, such as the aorta and arteries¹⁶. These findings have helped establish the role of endothelial mechanobiology in cardiovascular complications such as atherosclerosis¹⁷. Based on these findings, we have begun to investigate the previously unrecognized role of endothelial mechanobiology in the development of retinal microvascular lesions in early DR. Success in this pursuit, however, relies on the accurate measurement of diabetes-induced changes in retinal capillary stiffness. Recent studies have revealed that similar to large vessels, stiffness of retinal capillaries and retinal EC-secreted subendothelial matrix can also be directly, accurately, and reliably measured using AFM^4,5,6.

This approach involves the isolation of mouse retinal capillaries from mildly fixed eyes using trypsin digestion. Based on experience, this mild fixation step is necessary as capillaries isolated from unfixed eyes (using the hypotonic method) are fragile and become fragmented, rendering them unsuitable for AFM stiffness measurement⁴. Conversely, stronger fixation and the ensuing need for extensive trypsin digestion, which has been reported in other retinal trypsin digestion protocol¹⁸, may also compromise capillary structural integrity, as judged by the disruption of tight junctions. Thus, the mild capillary fixation and gentle trypsin digestion employed here help reduce stiffness artifacts arising from excessive formalin-induced crosslinking and ensure the structural integrity of the capillaries for reliable AFM stiffness measurement.

As an alternative approach, a recent study reported AFM stiffness measurement of retinal capillaries from lightly fixed retinal flat mounts¹⁹. By performing AFM force indentations on retinal capillaries within the intact retina, this approach enables in situ stiffness measurement. However, these capillary stiffness measurements likely include the stiffness of the inner limiting membrane. Further, this approach is restricted to the measurement of only superficial capillaries that are accessible on a retinal flat mount while leaving out the deeper-lying capillary plexus that is specifically affected in early DR²⁰. These issues can be addressed using this approach, which extracts vessels cleanly (devoid of residual retinal tissue) from all retinal layers. We are currently optimizing this protocol to isolate choroidal vessels from animal models of early AMD. That said, we realize that the mild formalin fixation employed in the protocol may cause a crosslinking-associated stiffening artifact in AFM measurements. Therefore, it would be prudent to interpret the vascular stiffness values obtained using this approach more in terms of relative changes in stiffness (between different experimental groups) rather than absolute stiffness.

In contrast to retinal capillaries that require mild fixation, the subendothelial matrix obtained from retinal EC cultures can be assessed without any modification. This is largely due to the robustness of the deposited matrix, which results from a combination of several factors, including the addition of ascorbic acid in culture medium (which enhances collagen synthesis¹⁵), chemical modification of glass coverslips that prevents matrix detachment during decellularization, and prolonged duration of culture (10-15 days). Importantly, we have shown that the hyperglycemia-induced increase in subendothelial matrix stiffness in vitro is consistent with the diabetes-induced increase in retinal capillary stiffness seen in vivo⁴. This is not surprising as an increase in matrix stiffness is expected to increase the stiffness of overlying retinal ECs (due to mechanical reciprocity) that, together, increase the overall capillary stiffness. Indeed, we recently showed that retinal ECs become stiffer when grown under diabetic conditions⁴, likely via an increase in Rho/ROCK-dependent actin cytoskeletal tension²¹. These findings also imply that the choroidal EC stiffening seen in the rhesus monkey model of early AMD reflects the increased stiffness of choroidal subendothelial matrix and intact vessels¹², an idea that is being tested in ongoing studies. Overall, the fact that stiffness alterations in the unfixed retinal subendothelial matrix and ECs mirror the trend seen in mildly fixed intact retinal vessels testifies to the validity of the mild fixation approach.

Direct stiffness measurement of retinal capillaries from animal models of DR not only helps establish endothelial mechanobiology as a novel therapeutic target for DR management, but it also provides a strong rationale to develop new sensitive imaging techniques for non-invasive assessment of retinal capillary stiffness in DR patients in the future. Such noninvasive techniques could potentially detect subtle changes in blood flow that are expected to arise from capillary stiffening, similar to the changes in arterial pulse wave velocity that are commonly detected in diabetes-induced cardiovascular diseases. AFM stiffness measurements of retinal capillaries isolated from post-mortem human eyes from diabetic donors will provide important proof-of-concept in this regard. Given that retinal capillary stiffness increases early on in the (streptozotocin) mouse model of type 1 diabetes⁴, successful use of AFM in validating stiffness-measuring imaging modalities may lead to the identification of retinal capillary stiffness as a clinical biomarker for early DR pathogenesis.

Materials

Name	Company	Catalog Number	Comments
Retinal Capillary Isolation
0.22 µm PVDF syringe filter	Merck Millipore	SLGVM33RS	Low Protein Binding Durapore
10X Dulbecco's Phosphate Buffered Saline without calcium % magnesium	Corning	20-031-CV	Final concentration 1X, pH 7.4
12-well plate	Falcon Corning	353043
15 mL centrifuge tube	Corning	430791	Rnase-Dnase-free, Nonpyrogenic
20 mL Luer-Lok TIP syringe	BD	302830
5 3/4 inch Disposable Borosilicate Glass/Non-sterile Pasteur pipette	FisherBrand	13-678-20A
60x15 mm Tissue Culture Dish	Falcon Corning	353002
6-well plate	Falcon Corning	353046
Aqua-Hold 2 Pap - 13 mL Pen	Scientific Device Laboratory	9804-02
Blade holder	X-ACTO
Carbon Steel Surgical Blade #10	Bard-Parker	371110
Dental Wax	Electron Microscopy Sciences	50-949-027
Dissecting microscope	Am-scope
Formalin solution, neutral buffered, 10%	Millipore Sigma	HT501128-4L	Final concentration 5% (v/v)
Kimwipes - wiper tissue	Kimtech Science	34133
Micro spatula	Fine Science Tools	10089-11
Orbital Shaker	Lab Genius	SK-O180
PELCO Economy #7 Stainless Steel 115mm  Tweezer	Ted Pella, Inc.	5667
Phase contrast microscope	Nikon TS2
Purifier Logic+ Class II, Type A2 Biosafety Cabinet	Labconco	302380001
Safe-Lock microcentrifuge tubes 2 mL	Eppendorf	22363352
Stereoscope	AmScope	SM-3 Series Zoom Trinocular Stereomicroscope 3.5X-90X
Superfrost Plus microscopy slide - White tab - Pre-cleaned - 25x75x1.0 mm	FisherBrand	1255015
Tris Buffer, 0.1M solution, pH 7.4 - Biotechnology Grade	VWR	E553-500ML	pH 8 for trypsin solution
Trypsin 1:250 powder Tissue Culture Grade	VWR	VWRV0458-25G	10 % (w/v) trypsin solution
Water Molecular Biology Grade	Corning	46-000-CM
Subendothelial Matrix
10X PBS	Corning	20-031-CV
1X PBS with calcium and magnesium	Thermo Fisher Scientific	14040-117	pH 7.4
Ammonium hydroxide	Sigma-Aldrich	338818
Ascorbic Acid	Sigma-Aldrich	A4034
Collagen IV antibody	Novus Biologicals	NBP1-26549
DNase I	Qiagen	79254
Ethanolamine	Sigma-Aldrich	398136
Fibronectin antibody	Sigma-Aldrich	F6140
Fluoromount	Invitrogen-Thermo Fisher Scientific	00-4958-02
Gelatin	Sigma-Aldrich	G1890
Glass coverslips (12mm)	Fisher	12-541-000
Glutaraldehyde	Electron microscopy Sciences	16220
Human retinal endothelial cells (HREC)	Cell Systems Corp	ACBRI 181
MCDB131 medium	Corning	15-100-CV
Mouse retinal endothelial cells (mREC)	Cell Biologics	C57-6065
Triton X-100	Thermo Fisher Scientific	BP151-100
Trypsin	Gibco-Thermo Fisher Scientific	25200-056
AFM Measurement
1 µm Probe	Bruker	SAA-SPH-1UM	A 19 micron tall hemispherical probe with 1 micron end radius, Spring constant 0.25N/m
70 nm LC probe	Bruker	PFQNM-LC-V2	A 19 micron tall hemispherical probe with 70nm end radius,  Spring constant 0.1N/m
camera	XCAM family	Toupcam	1080P HDMI
Desktop to run the camera	Asus	Asus desktop	Intel i5-6600 CPU , 8GB RAM
Dish holder for coverslip	Cellvis	D29-14-1.5-N	29mm glass bottom dish with  14 mm micro-well
Nanowizard 4	Bruker	Nanowizard 4	Bioscience atomic force microscope mounted on an optical microscope for sensitive measurement of the mechanostructural properties (stiffness and topography) of soft biological samples
Phase contrast micrscope	Zeiss	Axiovert 200	Inverted microscope with 10X objective