Growing a piece of tissue in a dish, cells rarely survive beyond the “few millimeters thick” mark — not because nutrients are insufficient, but because there are no blood vessels to deliver oxygen and nutrients deep inside. To connect engineered tissue with “plumbing,” researchers have tried four routes: embedded printing, layer-by-layer stacking, coaxial nozzles, and sacrificial ink. All share a common frustration: extruded soft tubes cannot stand, and wall thickness is hard to control.

A protocol published in 2026 by the Heilshorn group at Stanford in BMC Methods flips the idea: what the nozzle extrudes is not the “wall” but a sacrificial ink loaded with a crosslinking initiator; the initiator diffuses outward, and wherever it reaches, the surrounding gel solidifies in place. The wall is “seeped” into existence, not extruded. This article breaks down the entire pipeline: modeling, formulation, printing, calibration, and endothelialization.

PART 01 Engineered Tissue Doesn’t Grow Large Because It Lacks Plumbers, Not Cells

The paper states it plainly from the opening: blood vessels transport oxygen, signaling molecules, nutrients, and metabolic waste. Without effective vascularization, engineered tissue lacks complexity, size, and long-term function — building complex, perfusable vascular networks is therefore listed as a key challenge in the field.

What makes it hard? Cells are not hard to culture — growing them “thick” is. Once tissue exceeds a thin layer, cells deep inside cannot get oxygen or nutrients, and waste cannot be removed — tissue relying solely on diffusion is destined to stay small. So the problem was never “making cells” but connecting every cell to plumbing.

For this reason, bio 3D printing has been highly anticipated over the past decades: it can place biomaterials, cells, and growth factors precisely where needed. But which route to take — the field has offered several answers.

PART 02 Four Old Routes: Each Stuck at the “Wall”

The paper’s background lays out the existing routes:

Sacrificial ink is the most “construction-like”: print a “wax core,” then remove it, leaving a lumen that becomes a vessel. But the paper points out its weak spot — conventional sacrificial ink extrusion struggles to create “self-supporting, vessel-like shells.” The reason is simple: once the wax core is removed, there is no load-bearing wall around the cavity; the soft matrix cannot hold, and the channel closes under compression.

In other words, every old route gets stuck at the “wall”: either thick and bulky, or no wall truly belonging to the channel at all. This wall is precisely what the protocol sets out to dismantle.

PART 03 Reverse It: Draw a Solid Line First, Let the Wall Grow Itself

Four steps:

1. Extrude a sacrificial ink containing a crosslinking initiator into a gel precursor support bath;
2. The initiator diffuses outward from the ink;
3. Where concentration exceeds the threshold, in-situ crosslinking occurs, forming a gel shell around the ink filament;
4. Finally, at 37 °C, melt and flush away the sacrificial ink, leaving a hollow, interconnected, perfusable channel.

The beauty is here: the channel wall is “drawn” by the diffusion front, wall thickness equals diffusion distance, controlled by time and concentration; channel inner diameter is the ink filament thickness, controlled by extrusion — two originally entangled design variables are split into two independent knobs. This is also the core difference from previous sacrificial ink extrusion methods: the channel has its own load-bearing shell, and the structure stands on its own when removed.

Compatibility is broad: photocrosslinkable GelMA, PEGDA, HAMA; enzymatically crosslinked gelatin, fibrin; ionically crosslinked sodium alginate — all usable systems. Channels can be seeded with cell suspension after printing to lay endothelium, or printed in one step with cell-laden ink.

PART 04 Raw Material Lesson: Why Cold-Water Fish Skin

The support bath base is GelMA (gelatin methacryloyl) — biocompatible, tunable mechanical properties, on-demand crosslinkable. But there is a picky detail: common porcine gelatin solidifies at room temperature and cannot serve as a support bath; the team switched to cold-water fish skin gelatin, which stays liquid at room temperature, perfectly matching GUIDE-3DP operating temperatures.

In-house GelMA synthesis: 20% w/v gelatin dissolved in carbonate buffer, pH adjusted to 10, heated to 70 °C, methacrylic anhydride added at 0.0833 mL per gram of gelatin, stirred for 2 hours. Two purification routes: ethanol precipitation is fast; dialysis is higher purity but time-consuming — dilute to 5 wt%, 10 kDa dialysis bag, >2 L ultrapure water for 5 days, then freeze-dry, store at −80 °C.

Quality control looks at two numbers: concentration by gravimetric method; methacryloyl degree by ¹H NMR — using aromatic amino acid signals as internal reference, measuring how much lysine signal dropped. Four batches measured 70.7 ± 5.2%: higher acylation means more crosslinking sites, denser gel, slower degradation.

PART 05 One Bath, One Ink: The “Jelly” Must Hold, the “Slush” Must Not Melt

The support bath has two clearly divided roles: the gel precursor handles “being crosslinked”; the thickener handles “keeping the ink filament from spreading.” Demonstration formulation: 40 wt% GelMA stock mixed 1:1 with 4 wt% thickener stock to give 20 wt% GelMA + 2 wt% thickener; when using Eosin Y, add 0.5 wt% TEA and 0.2 wt% NVP co-initiator; to help endothelium adhere, add 0.5–1 wt% fibrinogen. Plate preparation: about 8 g per well of a 6-well plate, seal and centrifuge at 4000 rpm for 5 min to remove bubbles and level the surface, store at 4 °C until use.

Sacrificial ink is gelatin microparticles resuspended at 100 mg/mL in cold buffer or medium — keep cold throughout: ice bath, hold only the tube rim, never the body. The troubleshooting table calls out this failure mode: hand warmth melts the ink, and what prints out is a pancake instead of a round filament. Initiator final concentration: LAP 1–5 mM or Eosin Y 0.025–0.1 mM; for alginate systems, CaCl₂ 50–100 mM. Mount a 27 G blunt needle and start printing.

PART 06 From a Vascular Map to a String of Coordinates: What Happens Before Printing

It all starts with a line drawing: in 3D modeling software, trace the vessels as curves following reference images (public vascular model libraries offer ready assets), deliberately overlapping adjacent paths — otherwise the printed lumen will disconnect. Then use the “divide” command to split each path into equally spaced points (example 0.5 mm), and export the coordinates. Filament diameter is determined jointly by point spacing and extrusion amount: at 0.5 mm spacing, extrusion amount is 0.005–0.02 mm.

Coordinates are converted into lines of G-code: each line “move to a point + extrude a certain amount,” starting with a relative extrusion mode declaration, setting the starting point, and writing the printing speed. Each segment must be individually reviewed in G-code simulation software — if direction is reversed, reorder in a spreadsheet.

The most masterful detail is “travel moves”: when two paths are not connected, the nozzle carrying ink must not cross the bath directly — it would drag ink where it should not go. The protocol requires retracing along the already printed path (extrusion set to 0), then lifting vertically to finish, without disturbing the freshly printed structure.

PART 07 Two Knobs: Inner Diameter Follows Extrusion, Outer Diameter Follows Diffusion

The nozzle plunges into the support bath, follows G-code paths, and pink ink filaments remain inch by inch in the bath. Stand at room temperature for 5–15 minutes — wait longer for thicker walls. Then crosslink: GelMA + LAP under UV for 5 min, Eosin Y under white light for 10 min, sodium alginate removed immediately. Wash away uncrosslinked bath material, melt sacrificial gelatin at 37 °C, cut both ends of the lumen, flush residual ink with warm medium — a self-supporting, perfusable empty tube stands up.

The inner diameter knob is composed of needle and flow rate. At a fixed printing speed of 180 mm/min, the paper compares 32 G + 0.6 µL/s, 32 G + 1 µL/s, and 27 G + 2 µL/s: inner diameter increases in that order, with significant differences between groups. Reading from the figure, the three settings span roughly 0.15 mm to 1.5 mm — about a 10× range within the same system. More practical is the orientation experiment: printed upward, downward, or horizontally, inner diameters are nearly identical (not significant), with only horizontal direction showing slightly larger variance due to filament spreading — when printing free-form surfaces, diameter will not fail due to orientation.

The outer diameter knob is entirely delegated to diffusion: wall thickness increases with diffusion time, and at any given time point, higher concentration groups are thicker. The paper compares 0.05 and 0.1 mM Eosin Y at five time points from 5 to 20 min, reading a wall thickness of about 0.65/0.76 mm at 20 min. The farther it seeps, the thicker the wall — even edge trimming is unnecessary.

PART 08 Dual Nozzles and Living Cells: Two Inks, Two Cell Types, Each in Place

How do cells get onto the wall? The paper offers two routes.

Fluorescence images tell it most clearly: ink-laden HUVECs immediately cover the channel after printing; 5 days later, the endothelial-specific junction protein VE-cadherin is expressed at cell junctions, forming a continuous monolayer. The two-color experiment is even more intuitive — two fluorescently labeled endothelial populations, each in its own ink, settle in designated zones after printing; in a zigzag interleaved structure, the two cell types follow their respective ink paths to assemble a red-and-green vascular network.

Closing Thoughts

For labs doing bio 3D printing, this protocol lays the entire pipeline out in the sun — even the “holding the tube body will melt the ink” detail is written into the troubleshooting table. The value of a methodology paper lies exactly here: it does not declare victory; it lowers the barrier. When later generations print vascular networks that look more and more real, they will probably thank this recipe for being so verbose about every step.

📄 Paper Information

· END ·