Make your paper interactiveCreate

Text from the interactive explanation

Combined TIRF and 3D Super-Resolution Microscopy for Nanoscopic Characterization of Adhesion Molecules on Microvilli

Abdullah Alghamdi, Mansour M Aldehaiman, Maged F Serag, Shuho Nozue, Ioana-Andreea Ciocanaru, Karmen AbuZineh, Jasmeen S Merzaban, Satoshi Habuchi

Original paper

0The question

1Rolling

2The glow

3Landscape

4One molecule

5Overlay

6Experiment

7Back to the wall

•Sources

Analytical Chemistry · 2026

A blood-stem-cell transplant only works if the injected cells find their own way to the bone marrow. They brake by rolling along blood-vessel walls, catching on with thousands of tiny membrane fingers called microvilli. But which part of each finger does the catching? This study combined two light microscopes to map the fingers and the grip molecules on them in 3D, to about 30 nanometres.

Measured You are looking up at the underside of a real leukemia cell (KG1a) on glass: the TIRF image and single-molecule positions come from the paper’s example dataset. Height was computed from brightness with the paper’s equation and exaggerated about ×20, so the fingers look taller than they are. The flashes are molecule positions shown at their measured XY on the reconstructed surface.

Illustration The 3D view zooms into the patch where a rolling cell meets the floor (box). The patch is about 8 µm across; the whole cell is about 10–15 µm.

Rolling, not sliding

How a fast-moving cell in the bloodstream manages to stop.

Stem cells and leukemia cells reach the bone marrow by a series of steps. The first one is the most dramatic: a cell racing along in the blood brushes the vessel wall, snags on sticky proteins called selectins , and starts to roll, like a ball dragging its fingers along a carpet. The fingers are microvilli, and the molecules on them that grab selectins include PSGL-1 and CD44.

1 Tethering & rolling on selectins (this story)

2 Activation

3 Arrest

4 Crossing the vessel wall

5 Directed migration into the marrow

Try: slide the flow from zero to the top and watch what the fingertips do.

Rolling — tethers forming and breaking

Rolling is a balance: bonds at the front form while bonds at the back snap. Too little flow and the cell just sits; too much and the bonds can’t hold. (Speeds and bond behaviour here are illustrative; the experiment used 2 dyn/cm².)

A glow only 120 nanometres deep

The first trick: light that never really leaves the glass.

Ordinary light can’t sharpen anything smaller than its own blurry spot, about 130 nm in this microscope, and a microvillus is only a few hundred nanometres wide. So the team used a sideways trick. Shine a laser up through the glass at a steep enough angle and it bounces back completely (total internal reflection). Yet a thin evanescent glow leaks just above the glass, fading fast with height. Anything in it lights up; anything higher stays dark.

Try: tilt the laser past the critical angle. Watch the beam move to the edge of the lens, the glow appear and thin out, and the flashes of light gather at the fingertip touching the glass.

Glow depth d = 121 nm

d(θ) = λ4π (n1²sin²θ − n2²)−1/2 λ = 640 nm, glass n1 = 1.52, water n2 = 1.33

The angle is set by where the laser enters the objective: the team shifted the focused beam from the centre of a high-aperture oil lens (NA 1.49) towards its edge, so it leaves the lens steeply. Below 61° the light shoots straight into the cell and every dye molecule glows, so brightness says nothing about height. Above it, only a sliver next to the glass is lit, and dye there sends far more light back down to the camera. At the paper’s 66.6° that sliver is about 120 nm deep.

Predict first: a fingertip 120 nm above the glass glows what fraction as bright as one touching it?

At 60 nm: 61% as bright as at the glass

Brightness falls off as e−z/d, the same way at every height, so it works as a ruler : measure how bright a bit of membrane glows and you know how far it hangs above the glass.

Turn brightness into a landscape

From one TIRF photo to a 3D map of the cell’s underside.

The team stained the cell membrane with a red dye, set the laser to 66.6° and averaged 10 camera frames. Below is a real TIRF image from the paper’s example data: the underside of one KG1a cell, 11.7 µm across. Step through what their pipeline does with it.

Try: step through the four stages, then hover or tap a bright spot to read its height.

The raw picture: bright blobs are fingertips touching the glow, sitting on a soft haze from the cell body.

Raw brightness at the marked pixel: 100% of the brightest value

δz = d · ln(I0 / Iz), with d = 120 nm I0 = brightest pixel (taken as touching the glass)

Measured image; heights computed with the paper’s equation from the measured image, using a simplified haze filter.

Does the map catch every finger?

No, about half. Laterally, TIRF is still limited by blur, so neighbouring fingers can merge into one blob, and short fingers that never reach down into the glow stay invisible. Even so, the team showed that more than 50% of the microvilli are reliably captured, which is enough to study them one by one.

Lay a finger down

The brightness-to-height trick assumes that above every pixel there is one membrane. That holds for a finger pointing straight down. But what if a finger lies on its side?

Try: tilt the finger from upright to lying flat and compare the two sides.

One membrane per pixel — the reconstruction matches the real shape

A lying finger puts two membranes, its bottom and its top, above the same pixel. Their glows add up, the equation reads one brighter-than-real surface, and the top membrane disappears: the tube collapses into a flat ridge.

The paper measured the effect

They compared each TIRF finger map with the surface traced by single molecules on the same finger, using a similarity score (SSIM, 1 = identical). For all three molecules, upright fingers matched better than lying ones.

So the team analysed only upright fingers, those whose outline is nearly round (aspect ratio ≤ 1.2). In the example image above, 14 of 36 outlines pass that test.

Catch one molecule at a time

The second trick: make molecules take turns.

The finger map shows shape. To see the grip molecules, the team tagged them with antibodies carrying a green dye that can be made to blink . If all the dyes glow together, neighbours melt into one blur. If only a few glow in each camera frame, each blur belongs to one molecule, and its centre can be found far more precisely than the blur is wide.

Try: squeeze the two molecules closer than the blur, then let them blink.

One blur — you can’t tell it’s two molecules

Blinking turns a resolution problem into a counting problem: one molecule per blur, one centre per fit. The paper recorded 10,000 frames of 30 ms each per cell.

Depth from a squashed spot

Finding x and y is the easy part. For height, the team put a weak cylindrical lens in the camera path. It focuses sideways and up-down at slightly different depths, so a molecule’s spot stretches one way above focus and the other way below. Calibrated with tiny beads every 10 nm from −300 to +300 nm, the spot’s shape tells its depth to about 30 nm .

Try: move the molecule up and down and watch the spot change shape.

Round spot → molecule in the focal plane, z ≈ 0 nm

Mini-game: which molecule sits highest?

Tip: “above focus” means stretched sideways.

Watch a real acquisition build up

Watch: 55,746 measured localizations from the example dataset, appearing frame by frame. Pause any time.

Frame 0 / 10,000

Colour = measured depth z (same scale throughout). Positions are measured; the order of appearance is randomised, because the original frame numbers are not part of the released file.

One frame shows almost nothing. Ten thousand frames, about five minutes of recording, add up to a map: the dense patch in the middle is the cell, and the scattered speckle all around is dye that stuck to the bare glass.

Lay the molecules on the landscape

Two maps of the same cell, filtered and stacked.

Now the two pictures are combined, using the same cell imaged both ways. First, keep only the molecules inside the thin slab that the glow can see. Then keep only those over a finger outline. Then drape them on the height map. Here is that pipeline, run live on the example data.

Try: drag the green window along the histogram, then apply the finger mask.

55,746 localizations in the whole field

Every measured localization, coloured by depth. Molecules far from the cell can only be dye stuck on the bare glass.

Why the slab?

The glow only maps the bottom 120 nm of the cell, so only molecules in that slab can be compared with the map. The histogram also has a second population: molecules stuck on the glass itself, found everywhere, including far from the cell. The paper’s analysis script for this example puts the coverslip plane at 85 nm on this axis, so its slab runs from 85 to 205 nm.

Sanity check: does actin land in the core?

Before trusting the overlay with unknown molecules, the team tested it on one with a known address. F-actin filaments form the stiff column inside every microvillus. Overlaid on the finger maps, they did sit inside, as a column up the core. The TIRF map and the actin-traced surface agreed with a mean SSIM of 0.50 (0.56 for upright fingers, 0.41 for lying ones). The two surfaces sat less than 50 nm apart, consistent with actin lying just under the membrane, tethered by linker proteins.

The experiment: before and after rolling

Where do the grip molecules sit on a single finger, and does rolling move them?

Here is one upright microvillus as the paper draws it: a funnel whose tip touches the glass (0 nm) and whose top merges into the cell body 120 nm up. Choose a molecule and a condition. Resting cells sat still on glass; rolled cells had rolled on E-selectin at 2 dyn/cm² and were fixed after 1 minute.

The histogram looks at the finger side-on and counts localizations in each 10 nm height band, the same view the paper uses in Figs. 5–7. Bright squares near the bottom mean molecules crowd the tip. Funnel colour = height: red tip at the glass (0 nm) to pale top (120 nm).

Predict first: after rolling on E-selectin, what will PSGL-1 do?

Tip zone (bottom 30 nm): 25% of localizations

F-actin runs up the core of the finger like a column, the scaffold that holds a microvillus up. Rolling leaves it unchanged.

Simulated point clouds that reproduce the patterns the paper reports (about 30 nm depth scatter). Drag the 3D view or use arrow keys to orbit. The measured data are below.

Back to the vessel wall

So, where does a homing cell hold on?

At the fingertips, at least for PSGL-1.

In cells that had been rolling, PSGL-1 gathered at the tips of the microvilli, the very first part of the cell to meet the selectins on the wall. CD44 stayed spread over the whole finger, before and after rolling.

What the authors make of it

In the authors’ words, the tip enrichment of PSGL-1 versus uniform CD44 indicates that selectin–ligand interactions are “structurally regulated at the nanoscale level.” They suggest this arrangement “could enhance ligand exposure to selectins, increasing binding efficiency and stabilizing rolling adhesion.” The measurements show where the molecules sit; how much extra grip that brings is the next question.

The cells were fixed, not filmed live, so each map is a snapshot taken about a minute into rolling.

Why it matters

Earlier work from the same group showed that fine-tuning homing-specific adhesion can improve how transplanted stem cells migrate and engraft. Knowing exactly where the grip molecules sit gives that kind of work a nanoscale target.

The same TIRF + 3D single-molecule approach could be pointed at immune cells moving through tissue, cancer cells spreading, or microbes latching onto host cells. The authors also see a next step: AI-assisted super-resolution methods (such as Deep-STORM or ANNA-PALM) might one day make it fast enough to follow microvilli and selectin clusters in real time.

Brightness to height (eq. 1): δz = d(θ)·ln[I0/Iz]. I0 is the brightest pixel, taken as the point closest to the glass; Iz is the pixel’s brightness. A pixel at 37% of I0 sits one glow-depth (about 120 nm) up.

Glow depth (eq. 2): d(θ) = (λ/4π)·(n1²sin²θ − n2²)−1/2, with λ = 640 nm, n1 = 1.52 and n2 = 1.33. The critical angle is 61.04°; the objective (NA 1.49) allows up to 78.6°. The team checked the angle–depth relation with dye solutions and used angles above 65°.

Stain KG1a cells’ membranes (MemBrite Fix 640); roll them over recombinant human E-selectin at 2 dyn/cm² in a microfluidic channel (or let them rest on glass).

After about 1 minute of rolling, fix in place under flow (4% paraformaldehyde + 0.2% glutaraldehyde); label the target with Alexa Fluor 488 antibodies, or phalloidin for F-actin.

TIRF: 640 nm laser at 66.6° (120 nm glow), 10 frames averaged; pixel size 83.33 nm.

Filter (Laplacian-of-Gaussian plus smoothing) to separate fingers from cell-body haze; convert brightness to height with eq. 1; threshold (≥ 200 counts) to outline each finger.

3D-SMLM on the same cell: 488 nm laser, 10,000 frames of 30 ms; astigmatic spot shape gives depth (about 30 nm precision); spots wider than 200 nm rejected.

Keep localizations within 120 nm of the coverslip, keep those inside finger outlines, overlay on the topography.

Analyse upright fingers (outline aspect ratio ≤ 1.2); check overlay quality with SSIM; build side-on (XZ, YZ) histograms with 10 nm height bins.

Sources & credits

Alghamdi A, Aldehaiman MM, Serag MF, Nozue S, Ciocanaru I-A, AbuZineh K, Merzaban JS, Habuchi S. “Combined TIRF and 3D Super-Resolution Microscopy for Nanoscopic Characterization of Adhesion Molecules on Microvilli.” Analytical Chemistry 2026; 98(23): 16898–16911. doi:10.1021/acs.analchem.5c08159 · Full text (PMC13276846). Published under CC BY 4.0.

Figures 3–7 are reproduced from the article under CC BY 4.0 (resized).

Example TIRF image and single-molecule localizations: from the article’s Supporting Information (analysis-scripts archive), CC BY 4.0. The height map was recomputed with a simplified version of the paper’s pipeline (Gaussian haze subtraction in place of the authors’ Laplacian-of-Gaussian filter), then eq. 1 with d = 120 nm. The supplement does not state which molecule was labelled in this example, so it is shown simply as “single-molecule localizations”.

Rolling-cell scenes, the laser cross-section, the tilted-finger model, blinking and astigmatism spots, and the single-microvillus point clouds are simulations or illustrations based on the paper’s description and reported values.

Topics in this iPaper

Biology
We show distinct nanoscopic localizations of adhesion molecules on the microvilli of HSPCs/leukemic cells and their reorganization under shear stress during cell rolling

Abstract

Chemistry
Here, we develop a microfluidics-based super-resolution (SR) imaging platform that enables the three-dimensional (3D) mapping of the cell surface morphology

Abstract; journal Analytical Chemistry

Physics
Since the intensity of the evanescent field decays exponentially from the interface (i.e., glass–buffer interface), the nanoscopic depth profile of the cell surface is determined by fluorescence intensity variation

Results and Discussion, 3D Mapping of Cell Surface Morphology Using TIRFM

super resolution microscopy
by integrating total internal reflection fluorescence microscopy (TIRFM) with single-molecule localization microscopy (SMLM)

Abstract

single molecule localization microscopy
the 3D coordinates of the adhesion molecules characterized by the astigmatism-based 3D-SMLM imaging

Results and Discussion, 3D Localization of Adhesion Molecules Using SMLM

total internal reflection fluorescence microscopy
3D cell surface topography was determined by depth-dependent fluorescence intensity under the TIR illumination

Results and Discussion, 3D Mapping of Cell Surface Morphology Using TIRFM

microfluidics
we integrated the TIRFM and 3D-SMLM imaging setup into a microfluidics-based cell rolling assay platform

Results and Discussion, Characterization of the Effect of Cell Rolling

cell adhesion
Notably, PSGL-1 exhibited tip enrichment, whereas CD44 displayed a more uniform distribution, indicating that selectin/ligand interactions are structurally regulated at the nanoscale level.

Conclusions

microvilli
Microvilli, finger-like projections enriched with adhesion molecules, play a crucial role in HSPC homing by serving as anchoring and tethering points.

Conclusions

hematopoietic stem cells
The homing of hematopoietic stem/progenitor cells (HSPCs) and leukemic cells is a multistep process governed by complex spatiotemporal interactions between adhesion molecules under shear stress.

Abstract

Combined TIRF and 3D Super-Resolution Microscopy for Nanoscopic Characterization of Adhesion Molecules on Microvilli — Interactive explanation | iPaper.ai