Separate Strands, Connected Signals:Bio 3D Printing Shows Vascular Cells
Separate Strands, Connected Signals:
Bio 3D Printing Shows Vascular Cells
Can Form Networks Without Crowding into One Fiber
Scientific Reports 2026 · University of Bergen / King’s College London et al.
Growing a usable tissue in the lab most often fails not on the cells themselves but on the vasculature — without a vascular network delivering oxygen and nutrients, engineered tissue rarely survives the first weeks after implantation. In 2026, a team at the University of Bergen published a study in Scientific Reports using a custom fibrin-gelatin “vascular ink” to run a controlled co-culture experiment in bio 3D printing: two cell types, three spatial arrangements, 14 days of tracking — all to answer one question: to grow a vascular network, do the cells have to share the same printed fiber? The answer is more forgiving than expected.
Highlights
- Why vasculature is the lifeline of engineered tissue: the 200 μm oxygen diffusion limit, and the 100 μm nozzle that cannot print capillaries
- A custom “vascular ink”: fibrin builds the scaffold, gelatin serves as a temporary placeholder, and viscosity rises sharply below 22 °C
- Three cell arrangements go head-to-head: mixed in one fiber / adjacent separate fibers / paracrine across a membrane — 14 days decide
- Counterintuitive conclusion: cells do not need to be squeezed into the same fiber; within migration distance, the vascular network still forms
*All values quoted from the original paper: 200 μm from the introduction (oxygen diffusion limit, literature data); 6.8× from the results quantitative analysis; ≤10 μm from CAM histological observation.
PART 01 Why Engineered Tissue Always Falls Short: Vasculature Is Non-Negotiable
The demand gap for tissue and organ transplantation continues to widen, and lab-grown engineered tissue is held up as the great hope. But the paper states it plainly from the opening: whether an implant survives depends first on blood supply. Vasculature supports nearly every tissue — delivering oxygen and nutrients, carrying away metabolic waste. In most tissues, cells should be no more than about 200 μm from the nearest blood vessel, the oxygen diffusion limit; once that distance is exceeded or supply is interrupted, cell proliferation, differentiation, and the survival of the entire implant are compromised.
The most obvious way to give engineered tissue vessels is to directly print hollow channels. But the paper runs three calculations:
First, extrusion-based bio 3D printing nozzles are typically on the order of 100 μm — far thicker than capillaries at 4–10 μm;
Second, synthetic vascular grafts have succeeded at large diameters (arteries, large veins) but generally fail at small diameters — surfaces lacking an endothelial lining readily trigger thrombosis;
Third, “top-down” printing of hollow channels is inherently limited by fabrication precision, whereas truly physiological microvessels are actually self-assembled by endothelial cells “bottom-up.”
Attention thus shifted to the self-assembly route: embed vascular endothelial cells (HUVECs) in a printed 3D matrix and let them cluster into tubes on their own. Substantial evidence shows that pairing them with supporting cells in the “pericyte” role — bone marrow mesenchymal stromal cells (BMSCs) in this study — yields more mature and stable vascular networks. But a rival view has persisted: perhaps paracrine signaling alone from supporting cells (factors released across a distance) is sufficient. Does it take physical proximity, or is distance enough? This is precisely the question the paper sets out to answer.
PART 02 A Custom “Vascular Ink”: Fibrin Builds the Scaffold, Gelatin Runs Alongside
The team did not use a commercial ink but formulated their own specifically for vascularization, naming it directly — vascu-ink. The core component is fibrin: a natural polymer that is itself pro-angiogenic and involved in wound healing, serving as a temporary extracellular matrix scaffold in the ink. The paired gelatin has a thermoresponsive “temper” — it improves printability without crosslinking, and under 37 °C culture conditions it simply dissolves and gradually diffuses away, a standard “sacrificial component.” What actually holds the structure together is enzymatically crosslinked fibrin.
This “enzymatic crosslinking” property also determined the process selection: fibrinogen requires thrombin to convert into fibrin, and extrusion printing conveniently allows the ink to be deposited first and then converted, while also enabling multiple inks within the same construct — exactly what the later comparison experiments require. Rheological testing gave the formulation its “health report”: at low shear rates the ink first shows shear-thickening, transitioning to shear-thinning at high shear rates; when temperature drops below 22 °C, viscosity rises sharply (gelatin starts to “set”), while at printing temperature viscosity is only 0.173±0.055 Pa·s — quite thin, leaving room for high cell viability. Testing across nozzle diameters from 0.15 mm to 0.5 mm all extruded smoothly; printed four-layer cylinders maintained structural stability during culture.
vascu-ink formulation: fibrinogen 2% (w/v, Sigma F8630, dissolved in DPBS); gelatin stock 32% (w/v) prepared at 75 °C with 50 rpm stirring, added to a final concentration of 4% (w/v); stored at 37 °C with gentle stirring throughout; both powders UV-sterilized for 1 h
Rheological testing: TA Discovery HR-2 rotational rheometer; parallel plate 20 mm diameter, 1500 μm gap; shear rate 0.1–300 s⁻¹ (22 °C) plus temperature sweep 37–20 °C (1 s⁻¹); n=4 independent replicates
Key values: viscosity at printing temperature 0.173±0.055 Pa·s; viscosity rises sharply below 22 °C; extrudable nozzle range 0.15–0.5 mm (no qualitative difference in HUVEC network orientation across diameters)
Cell culture: HUVECs in EGM-2 (kit’s 2% FBS replaced with 2% human serum); BMSCs in α-MEM + 5% human serum + 5 ng/mL fibroblast growth factor; both at passages 4–5; one donor each for this study
PART 03 Three Arrangements: Same Fiber, Separate Fibers, or Calling Across the Gap
The cleverness of the experimental design lies here: same ink, same cell ratio, with the only variable being the spatial relationship between the two cell types.
Group 1 — Direct mixed: HUVECs and BMSCs mixed in proportion into the same ink and extruded together, at zero physical distance;
Group 2 — Direct separated: each cell type in its own ink, printed as two adjacent fibers, to see whether cells migrate across on their own;
Group 3 — Indirect: HUVECs printed in the ink, BMSCs in a hanging culture insert, exchanging only soluble signals across a membrane with no physical contact.
All three groups maintained a HUVEC:BMSC ratio of 5:1 (following the team’s prior work). Note that to preserve this ratio, the total cell number per sample was not identical across groups — the separated group used 10 million/mL HUVEC ink and 2 million/mL BMSC ink, giving a lower initial total than the mixed group; the paracrine group received 80,000 BMSCs per insert well. This detail resurfaces when interpreting results. Printing itself used a 0.25 mm inner-diameter nozzle and a four-layer 0/90° crosshatch structure, followed by immersion in thrombin + calcium solution for 10 min to crosslink and set, then 14 days of culture with “health checks” at days 1, 3, 7, and 14.
Cell loading: ink pre-warmed to 37 °C then combined with cells (cell suspension 50 μL/mL); Direct mixed total density 10 million cells/mL (HUVEC 8.33M + BMSC 1.67M); Direct separated two inks at 10M HUVEC/mL and 2M BMSC/mL respectively; Indirect ink HUVEC 10M/mL, BMSC 80,000 per insert (0.3 cm²) ≈ 267,000 cells/cm²
Printing parameters: nozzle inner diameter 0.25 mm; four-layer 0/90° crosshatch structure; printing temperature below 22 °C
Crosslinking and culture: thrombin 2.5 U/mL + 25 mM CaCl₂ for 10 min, DPBS wash 2× 5 min; EGM-2 + 2% human serum + 1% antibiotics + 100 μg/mL Normocin, 1.5 mL per well, medium changed three times per week; 14 days of culture, assays at days 1/3/7/14
Assays: Live/Dead staining, CCK-8 metabolic activity (n=6), immunofluorescence (CD31, vWF, α-SMA, Col-IV), confocal 3D imaging + IMARIS surface area quantification (n=4–8 ROIs per condition), chick chorioallantoic membrane (CAM) in vivo assay
PART 04 The 14-Day Test: All Alive, but Only “Close-Proximity” Groups Formed Networks
First the baseline: all three groups survived the “abuse” of printing — day-1 viability was high and maintained through day 14, with only sporadic dead cells. Metabolic activity rose over time in every group, with the separated group rising steepest; the paper’s explanation is candid: this group had the fewest initial cells, and the lower initial density actually provided more abundant nutrients; high density may cause local hypoxia, and hypoxia is associated with upregulation of pro-angiogenic factors such as VEGF. The HUVEC-only control without BMSCs had acceptable early viability but showed increasing dead cells over time and visibly lower cell density at two weeks.
The real watershed appeared at the tubular network level. Using type IV collagen (a basement membrane component) staining to track: by day 3 HUVECs in all groups had only mildly clustered, by day 7 all had begun to show ordered arrangement, but by day 14 only the two Direct groups containing BMSCs had grown longer, clearer tubular structures. CD31 (endothelial junction protein) 3D rendering similarly showed larger networks in the Direct groups. Quantitative data were even more direct: at days 3 and 7 differences among groups were modest; by day 14, both Direct groups showed multiple large tubular surface regions exceeding 250,000 μm², while the largest in the paracrine group was only about 60,000 μm² — the largest structures differed 6.8-fold between the separated and paracrine groups. The HUVEC-only control never formed a recognizable network.
Cell viability: all three groups showed high viability from day 1 through day 14; only the HUVEC-only control showed increasing dead cells and decreasing density over time
Metabolic activity: rising in all groups; steepest in the separated group (compensation for lower initial cell number); significant difference between Direct and paracrine groups at day 14 (**p<0.01, two-way ANOVA)
Tubular surface area: by day 14 both Direct groups showed structures >250,000 μm²; paracrine group maximum about 60,000 μm²; 6.8-fold difference in largest structures between separated and paracrine groups (Kruskal-Wallis + Dunn test)
Paracrine baseline: assembly in the paracrine group at day 14 still exceeded the HUVEC-only control — paracrine signaling helps, but cannot support networks of comparable scale and stability
PART 05 Cells Do “Visit”: Migration Evidence in the Separated Group
How did the separated group match the mixed group? The paper answers with a set of dual-labeling stains. α-SMA is a typical pericyte marker, usually expressed by migrating pericytes. In 3D confocal images, α-SMA-positive signal appeared between the CD31-positive HUVEC networks in both mixed and separated groups — in the separated group the two cell types originally lived in separate fibers, so α-SMA appearing within the same imaging plane can only mean BMSCs migrated over and took on a pericyte role in situ.
Whether migration can occur depends on whether the matrix “allows” it. The paper specifically cites a key reference: fibrinogen concentration affects the compressive modulus of the crosslinked matrix, and above 2.5 mg/mL it can restrict vascular sprouting and tubular network organization — this study used exactly 2.0 mg/mL, leaving ample space for cell migration and rapid sprouting. Additionally, nozzle diameters (150/250/500 μm) produced no qualitative differences in HUVEC network orientation; the true guide is likely chemokine gradients rather than printing traces. Endothelial cells also showed sprouting behavior toward chemokine gradients — cells have their own sense of direction; the ink only needs to stay out of the way.
PART 06 Evidence in the Chick Embryo: Lumen-Bearing Microvessels Formed
No matter how elegant the in vitro network, it must pass the “living environment” test. The team chose the classic chick chorioallantoic membrane (CAM) assay: printed samples were pre-cultured for 5 days, placed on the CAM surface at embryonic day 7, harvested for histology at day 12 — both blank ink and cell-laden ink were tested. Results showed that regardless of cell content, vascu-ink integrated well with CAM tissue without surrounding inflammation, clearing the biocompatibility exam first.
The more critical finding hides in the details: inside the tubular structures of cell-laden samples, microvessel lumens of ≤10 μm were observed — meaning HUVECs self-assembled not just “solid ropes” but hollow structures resembling vessels; immunostaining also captured a larger slit-like CD31-positive structure closely resembling a small-caliber vessel; HUVECs near the CAM even showed sprouting behavior toward the CAM tissue. The paper also draws the line clearly: the CAM assay does not confirm blood perfusion or functional anastomosis — it provides a proof of concept that “the ink can create an early vascular network-generation microenvironment.”
PART 07 One Practical Design Principle, and Four Questions the Paper Leaves Open
Condensing the three groups into one sentence: the physical presence of BMSCs is a necessary condition for growing strong, stable tubular networks; paracrine signaling alone is not enough; and “physical presence” does not require crowding into the same fiber — only that cells are within migration distance of one another.
For multi-material bio 3D printing this is genuinely good news: vascular cells and supporting cells need not be forced into the same nozzle and the same ink; designers can confidently assign different cell types to different inks and positions, leaving freedom to the tissue-specific architecture itself. The paper also looks forward to combining vascu-ink with tissue-specific inks — allowing vascular networks to “grow in place” inside complex engineered tissue.
The paper is equally direct about its own limitations, leaving four questions for future research:
First, vascular function and perfusion capacity were not tested (the CAM assay itself does not confirm perfusion);
Second, this study used tubular structure surface area for inter-group comparison — the method is direct and reproducible, but future work needs finer measurements of branch points, junctions, and vessel diameter;
Third, matching the ink with the host and in vivo functional validation are unavoidable steps toward translation;
Fourth, long-term maturation, pericyte recruitment, and perfusion bioreactor culture are all homework for the next stage.
Closing Thoughts
but a design principle pinned down by controlled experiments:
in bio 3D printing, giving vascular cells “social distance” is allowed
— provided the matrix is made soft enough and the distance kept within migration range.
For teams building multi-material, multi-cell constructs,
this amounts to another validated arrangement option;
and for the paracrine camp, the cold water is well measured:
paracrine signaling is useful — just not sufficient.
📄 Paper Information
Title: Johansen Å, Korkeamäki JT, Yamada S, Rashad A, Miettinen S, Mustafa K. Effect of 3D-printed co-culture design of mesenchymal stem cells and human umbilical vein endothelial cells on tubular formation. Scientific Reports, 2026; 16: 26067.
DOI: 10.1038/s41598-026-55959-y
Timeline: Received 5 March 2026, accepted 27 May 2026, open access
Author affiliations: University of Bergen (Department of Clinical Dentistry, Center for Translational Oral Research), King’s College London, Middle East Technical University (Turkey), Tampere University (Finland)
This article is a research literature interpretation; all data and conclusions are quoted from the original paper without extension. The bio 3D printing technologies and cell constructs described are still at preclinical research stage and have not entered clinical application. This article does not constitute any diagnostic or treatment recommendation, nor any health intervention reference. Please refer to professional physician advice for individual treatment plans.
· END ·
Separate Strands, Connected Signals:
Bio 3D Printing Shows Vascular Cells
Can Form Networks Without Crowding into One Fiber
Scientific Reports 2026 · University of Bergen / King’s College London et al.
Growing a usable tissue in the lab most often fails not on the cells themselves but on the vasculature — without a vascular network delivering oxygen and nutrients, engineered tissue rarely survives the first weeks after implantation. In 2026, a team at the University of Bergen published a study in Scientific Reports using a custom fibrin-gelatin “vascular ink” to run a controlled co-culture experiment in bio 3D printing: two cell types, three spatial arrangements, 14 days of tracking — all to answer one question: to grow a vascular network, do the cells have to share the same printed fiber? The answer is more forgiving than expected.
Highlights
- Why vasculature is the lifeline of engineered tissue: the 200 μm oxygen diffusion limit, and the 100 μm nozzle that cannot print capillaries
- A custom “vascular ink”: fibrin builds the scaffold, gelatin serves as a temporary placeholder, and viscosity rises sharply below 22 °C
- Three cell arrangements go head-to-head: mixed in one fiber / adjacent separate fibers / paracrine across a membrane — 14 days decide
- Counterintuitive conclusion: cells do not need to be squeezed into the same fiber; within migration distance, the vascular network still forms
*All values quoted from the original paper: 200 μm from the introduction (oxygen diffusion limit, literature data); 6.8× from the results quantitative analysis; ≤10 μm from CAM histological observation.
PART 01 Why Engineered Tissue Always Falls Short: Vasculature Is Non-Negotiable
The demand gap for tissue and organ transplantation continues to widen, and lab-grown engineered tissue is held up as the great hope. But the paper states it plainly from the opening: whether an implant survives depends first on blood supply. Vasculature supports nearly every tissue — delivering oxygen and nutrients, carrying away metabolic waste. In most tissues, cells should be no more than about 200 μm from the nearest blood vessel, the oxygen diffusion limit; once that distance is exceeded or supply is interrupted, cell proliferation, differentiation, and the survival of the entire implant are compromised.
The most obvious way to give engineered tissue vessels is to directly print hollow channels. But the paper runs three calculations:
First, extrusion-based bio 3D printing nozzles are typically on the order of 100 μm — far thicker than capillaries at 4–10 μm;
Second, synthetic vascular grafts have succeeded at large diameters (arteries, large veins) but generally fail at small diameters — surfaces lacking an endothelial lining readily trigger thrombosis;
Third, “top-down” printing of hollow channels is inherently limited by fabrication precision, whereas truly physiological microvessels are actually self-assembled by endothelial cells “bottom-up.”
Attention thus shifted to the self-assembly route: embed vascular endothelial cells (HUVECs) in a printed 3D matrix and let them cluster into tubes on their own. Substantial evidence shows that pairing them with supporting cells in the “pericyte” role — bone marrow mesenchymal stromal cells (BMSCs) in this study — yields more mature and stable vascular networks. But a rival view has persisted: perhaps paracrine signaling alone from supporting cells (factors released across a distance) is sufficient. Does it take physical proximity, or is distance enough? This is precisely the question the paper sets out to answer.
PART 02 A Custom “Vascular Ink”: Fibrin Builds the Scaffold, Gelatin Runs Alongside
The team did not use a commercial ink but formulated their own specifically for vascularization, naming it directly — vascu-ink. The core component is fibrin: a natural polymer that is itself pro-angiogenic and involved in wound healing, serving as a temporary extracellular matrix scaffold in the ink. The paired gelatin has a thermoresponsive “temper” — it improves printability without crosslinking, and under 37 °C culture conditions it simply dissolves and gradually diffuses away, a standard “sacrificial component.” What actually holds the structure together is enzymatically crosslinked fibrin.
This “enzymatic crosslinking” property also determined the process selection: fibrinogen requires thrombin to convert into fibrin, and extrusion printing conveniently allows the ink to be deposited first and then converted, while also enabling multiple inks within the same construct — exactly what the later comparison experiments require. Rheological testing gave the formulation its “health report”: at low shear rates the ink first shows shear-thickening, transitioning to shear-thinning at high shear rates; when temperature drops below 22 °C, viscosity rises sharply (gelatin starts to “set”), while at printing temperature viscosity is only 0.173±0.055 Pa·s — quite thin, leaving room for high cell viability. Testing across nozzle diameters from 0.15 mm to 0.5 mm all extruded smoothly; printed four-layer cylinders maintained structural stability during culture.
vascu-ink formulation: fibrinogen 2% (w/v, Sigma F8630, dissolved in DPBS); gelatin stock 32% (w/v) prepared at 75 °C with 50 rpm stirring, added to a final concentration of 4% (w/v); stored at 37 °C with gentle stirring throughout; both powders UV-sterilized for 1 h
Rheological testing: TA Discovery HR-2 rotational rheometer; parallel plate 20 mm diameter, 1500 μm gap; shear rate 0.1–300 s⁻¹ (22 °C) plus temperature sweep 37–20 °C (1 s⁻¹); n=4 independent replicates
Key values: viscosity at printing temperature 0.173±0.055 Pa·s; viscosity rises sharply below 22 °C; extrudable nozzle range 0.15–0.5 mm (no qualitative difference in HUVEC network orientation across diameters)
Cell culture: HUVECs in EGM-2 (kit’s 2% FBS replaced with 2% human serum); BMSCs in α-MEM + 5% human serum + 5 ng/mL fibroblast growth factor; both at passages 4–5; one donor each for this study
PART 03 Three Arrangements: Same Fiber, Separate Fibers, or Calling Across the Gap
The cleverness of the experimental design lies here: same ink, same cell ratio, with the only variable being the spatial relationship between the two cell types.
Group 1 — Direct mixed: HUVECs and BMSCs mixed in proportion into the same ink and extruded together, at zero physical distance;
Group 2 — Direct separated: each cell type in its own ink, printed as two adjacent fibers, to see whether cells migrate across on their own;
Group 3 — Indirect: HUVECs printed in the ink, BMSCs in a hanging culture insert, exchanging only soluble signals across a membrane with no physical contact.
All three groups maintained a HUVEC:BMSC ratio of 5:1 (following the team’s prior work). Note that to preserve this ratio, the total cell number per sample was not identical across groups — the separated group used 10 million/mL HUVEC ink and 2 million/mL BMSC ink, giving a lower initial total than the mixed group; the paracrine group received 80,000 BMSCs per insert well. This detail resurfaces when interpreting results. Printing itself used a 0.25 mm inner-diameter nozzle and a four-layer 0/90° crosshatch structure, followed by immersion in thrombin + calcium solution for 10 min to crosslink and set, then 14 days of culture with “health checks” at days 1, 3, 7, and 14.
Cell loading: ink pre-warmed to 37 °C then combined with cells (cell suspension 50 μL/mL); Direct mixed total density 10 million cells/mL (HUVEC 8.33M + BMSC 1.67M); Direct separated two inks at 10M HUVEC/mL and 2M BMSC/mL respectively; Indirect ink HUVEC 10M/mL, BMSC 80,000 per insert (0.3 cm²) ≈ 267,000 cells/cm²
Printing parameters: nozzle inner diameter 0.25 mm; four-layer 0/90° crosshatch structure; printing temperature below 22 °C
Crosslinking and culture: thrombin 2.5 U/mL + 25 mM CaCl₂ for 10 min, DPBS wash 2× 5 min; EGM-2 + 2% human serum + 1% antibiotics + 100 μg/mL Normocin, 1.5 mL per well, medium changed three times per week; 14 days of culture, assays at days 1/3/7/14
Assays: Live/Dead staining, CCK-8 metabolic activity (n=6), immunofluorescence (CD31, vWF, α-SMA, Col-IV), confocal 3D imaging + IMARIS surface area quantification (n=4–8 ROIs per condition), chick chorioallantoic membrane (CAM) in vivo assay
PART 04 The 14-Day Test: All Alive, but Only “Close-Proximity” Groups Formed Networks
First the baseline: all three groups survived the “abuse” of printing — day-1 viability was high and maintained through day 14, with only sporadic dead cells. Metabolic activity rose over time in every group, with the separated group rising steepest; the paper’s explanation is candid: this group had the fewest initial cells, and the lower initial density actually provided more abundant nutrients; high density may cause local hypoxia, and hypoxia is associated with upregulation of pro-angiogenic factors such as VEGF. The HUVEC-only control without BMSCs had acceptable early viability but showed increasing dead cells over time and visibly lower cell density at two weeks.
The real watershed appeared at the tubular network level. Using type IV collagen (a basement membrane component) staining to track: by day 3 HUVECs in all groups had only mildly clustered, by day 7 all had begun to show ordered arrangement, but by day 14 only the two Direct groups containing BMSCs had grown longer, clearer tubular structures. CD31 (endothelial junction protein) 3D rendering similarly showed larger networks in the Direct groups. Quantitative data were even more direct: at days 3 and 7 differences among groups were modest; by day 14, both Direct groups showed multiple large tubular surface regions exceeding 250,000 μm², while the largest in the paracrine group was only about 60,000 μm² — the largest structures differed 6.8-fold between the separated and paracrine groups. The HUVEC-only control never formed a recognizable network.
Cell viability: all three groups showed high viability from day 1 through day 14; only the HUVEC-only control showed increasing dead cells and decreasing density over time
Metabolic activity: rising in all groups; steepest in the separated group (compensation for lower initial cell number); significant difference between Direct and paracrine groups at day 14 (**p<0.01, two-way ANOVA)
Tubular surface area: by day 14 both Direct groups showed structures >250,000 μm²; paracrine group maximum about 60,000 μm²; 6.8-fold difference in largest structures between separated and paracrine groups (Kruskal-Wallis + Dunn test)
Paracrine baseline: assembly in the paracrine group at day 14 still exceeded the HUVEC-only control — paracrine signaling helps, but cannot support networks of comparable scale and stability
PART 05 Cells Do “Visit”: Migration Evidence in the Separated Group
How did the separated group match the mixed group? The paper answers with a set of dual-labeling stains. α-SMA is a typical pericyte marker, usually expressed by migrating pericytes. In 3D confocal images, α-SMA-positive signal appeared between the CD31-positive HUVEC networks in both mixed and separated groups — in the separated group the two cell types originally lived in separate fibers, so α-SMA appearing within the same imaging plane can only mean BMSCs migrated over and took on a pericyte role in situ.
Whether migration can occur depends on whether the matrix “allows” it. The paper specifically cites a key reference: fibrinogen concentration affects the compressive modulus of the crosslinked matrix, and above 2.5 mg/mL it can restrict vascular sprouting and tubular network organization — this study used exactly 2.0 mg/mL, leaving ample space for cell migration and rapid sprouting. Additionally, nozzle diameters (150/250/500 μm) produced no qualitative differences in HUVEC network orientation; the true guide is likely chemokine gradients rather than printing traces. Endothelial cells also showed sprouting behavior toward chemokine gradients — cells have their own sense of direction; the ink only needs to stay out of the way.
PART 06 Evidence in the Chick Embryo: Lumen-Bearing Microvessels Formed
No matter how elegant the in vitro network, it must pass the “living environment” test. The team chose the classic chick chorioallantoic membrane (CAM) assay: printed samples were pre-cultured for 5 days, placed on the CAM surface at embryonic day 7, harvested for histology at day 12 — both blank ink and cell-laden ink were tested. Results showed that regardless of cell content, vascu-ink integrated well with CAM tissue without surrounding inflammation, clearing the biocompatibility exam first.
The more critical finding hides in the details: inside the tubular structures of cell-laden samples, microvessel lumens of ≤10 μm were observed — meaning HUVECs self-assembled not just “solid ropes” but hollow structures resembling vessels; immunostaining also captured a larger slit-like CD31-positive structure closely resembling a small-caliber vessel; HUVECs near the CAM even showed sprouting behavior toward the CAM tissue. The paper also draws the line clearly: the CAM assay does not confirm blood perfusion or functional anastomosis — it provides a proof of concept that “the ink can create an early vascular network-generation microenvironment.”
PART 07 One Practical Design Principle, and Four Questions the Paper Leaves Open
Condensing the three groups into one sentence: the physical presence of BMSCs is a necessary condition for growing strong, stable tubular networks; paracrine signaling alone is not enough; and “physical presence” does not require crowding into the same fiber — only that cells are within migration distance of one another.
For multi-material bio 3D printing this is genuinely good news: vascular cells and supporting cells need not be forced into the same nozzle and the same ink; designers can confidently assign different cell types to different inks and positions, leaving freedom to the tissue-specific architecture itself. The paper also looks forward to combining vascu-ink with tissue-specific inks — allowing vascular networks to “grow in place” inside complex engineered tissue.
The paper is equally direct about its own limitations, leaving four questions for future research:
First, vascular function and perfusion capacity were not tested (the CAM assay itself does not confirm perfusion);
Second, this study used tubular structure surface area for inter-group comparison — the method is direct and reproducible, but future work needs finer measurements of branch points, junctions, and vessel diameter;
Third, matching the ink with the host and in vivo functional validation are unavoidable steps toward translation;
Fourth, long-term maturation, pericyte recruitment, and perfusion bioreactor culture are all homework for the next stage.
Closing Thoughts
but a design principle pinned down by controlled experiments:
in bio 3D printing, giving vascular cells “social distance” is allowed
— provided the matrix is made soft enough and the distance kept within migration range.
For teams building multi-material, multi-cell constructs,
this amounts to another validated arrangement option;
and for the paracrine camp, the cold water is well measured:
paracrine signaling is useful — just not sufficient.
📄 Paper Information
Title: Johansen Å, Korkeamäki JT, Yamada S, Rashad A, Miettinen S, Mustafa K. Effect of 3D-printed co-culture design of mesenchymal stem cells and human umbilical vein endothelial cells on tubular formation. Scientific Reports, 2026; 16: 26067.
DOI: 10.1038/s41598-026-55959-y
Timeline: Received 5 March 2026, accepted 27 May 2026, open access
Author affiliations: University of Bergen (Department of Clinical Dentistry, Center for Translational Oral Research), King’s College London, Middle East Technical University (Turkey), Tampere University (Finland)
This article is a research literature interpretation; all data and conclusions are quoted from the original paper without extension. The bio 3D printing technologies and cell constructs described are still at preclinical research stage and have not entered clinical application. This article does not constitute any diagnostic or treatment recommendation, nor any health intervention reference. Please refer to professional physician advice for individual treatment plans.
· END ·
Separate Strands, Connected Signals:
Bio 3D Printing Shows Vascular Cells
Can Form Networks Without Crowding into One Fiber
Scientific Reports 2026 · University of Bergen / King’s College London et al.
Growing a usable tissue in the lab most often fails not on the cells themselves but on the vasculature — without a vascular network delivering oxygen and nutrients, engineered tissue rarely survives the first weeks after implantation. In 2026, a team at the University of Bergen published a study in Scientific Reports using a custom fibrin-gelatin “vascular ink” to run a controlled co-culture experiment in bio 3D printing: two cell types, three spatial arrangements, 14 days of tracking — all to answer one question: to grow a vascular network, do the cells have to share the same printed fiber? The answer is more forgiving than expected.
Highlights
- Why vasculature is the lifeline of engineered tissue: the 200 μm oxygen diffusion limit, and the 100 μm nozzle that cannot print capillaries
- A custom “vascular ink”: fibrin builds the scaffold, gelatin serves as a temporary placeholder, and viscosity rises sharply below 22 °C
- Three cell arrangements go head-to-head: mixed in one fiber / adjacent separate fibers / paracrine across a membrane — 14 days decide
- Counterintuitive conclusion: cells do not need to be squeezed into the same fiber; within migration distance, the vascular network still forms
*All values quoted from the original paper: 200 μm from the introduction (oxygen diffusion limit, literature data); 6.8× from the results quantitative analysis; ≤10 μm from CAM histological observation.
PART 01 Why Engineered Tissue Always Falls Short: Vasculature Is Non-Negotiable
The demand gap for tissue and organ transplantation continues to widen, and lab-grown engineered tissue is held up as the great hope. But the paper states it plainly from the opening: whether an implant survives depends first on blood supply. Vasculature supports nearly every tissue — delivering oxygen and nutrients, carrying away metabolic waste. In most tissues, cells should be no more than about 200 μm from the nearest blood vessel, the oxygen diffusion limit; once that distance is exceeded or supply is interrupted, cell proliferation, differentiation, and the survival of the entire implant are compromised.
The most obvious way to give engineered tissue vessels is to directly print hollow channels. But the paper runs three calculations:
First, extrusion-based bio 3D printing nozzles are typically on the order of 100 μm — far thicker than capillaries at 4–10 μm;
Second, synthetic vascular grafts have succeeded at large diameters (arteries, large veins) but generally fail at small diameters — surfaces lacking an endothelial lining readily trigger thrombosis;
Third, “top-down” printing of hollow channels is inherently limited by fabrication precision, whereas truly physiological microvessels are actually self-assembled by endothelial cells “bottom-up.”
Attention thus shifted to the self-assembly route: embed vascular endothelial cells (HUVECs) in a printed 3D matrix and let them cluster into tubes on their own. Substantial evidence shows that pairing them with supporting cells in the “pericyte” role — bone marrow mesenchymal stromal cells (BMSCs) in this study — yields more mature and stable vascular networks. But a rival view has persisted: perhaps paracrine signaling alone from supporting cells (factors released across a distance) is sufficient. Does it take physical proximity, or is distance enough? This is precisely the question the paper sets out to answer.
PART 02 A Custom “Vascular Ink”: Fibrin Builds the Scaffold, Gelatin Runs Alongside
The team did not use a commercial ink but formulated their own specifically for vascularization, naming it directly — vascu-ink. The core component is fibrin: a natural polymer that is itself pro-angiogenic and involved in wound healing, serving as a temporary extracellular matrix scaffold in the ink. The paired gelatin has a thermoresponsive “temper” — it improves printability without crosslinking, and under 37 °C culture conditions it simply dissolves and gradually diffuses away, a standard “sacrificial component.” What actually holds the structure together is enzymatically crosslinked fibrin.
This “enzymatic crosslinking” property also determined the process selection: fibrinogen requires thrombin to convert into fibrin, and extrusion printing conveniently allows the ink to be deposited first and then converted, while also enabling multiple inks within the same construct — exactly what the later comparison experiments require. Rheological testing gave the formulation its “health report”: at low shear rates the ink first shows shear-thickening, transitioning to shear-thinning at high shear rates; when temperature drops below 22 °C, viscosity rises sharply (gelatin starts to “set”), while at printing temperature viscosity is only 0.173±0.055 Pa·s — quite thin, leaving room for high cell viability. Testing across nozzle diameters from 0.15 mm to 0.5 mm all extruded smoothly; printed four-layer cylinders maintained structural stability during culture.
vascu-ink formulation: fibrinogen 2% (w/v, Sigma F8630, dissolved in DPBS); gelatin stock 32% (w/v) prepared at 75 °C with 50 rpm stirring, added to a final concentration of 4% (w/v); stored at 37 °C with gentle stirring throughout; both powders UV-sterilized for 1 h
Rheological testing: TA Discovery HR-2 rotational rheometer; parallel plate 20 mm diameter, 1500 μm gap; shear rate 0.1–300 s⁻¹ (22 °C) plus temperature sweep 37–20 °C (1 s⁻¹); n=4 independent replicates
Key values: viscosity at printing temperature 0.173±0.055 Pa·s; viscosity rises sharply below 22 °C; extrudable nozzle range 0.15–0.5 mm (no qualitative difference in HUVEC network orientation across diameters)
Cell culture: HUVECs in EGM-2 (kit’s 2% FBS replaced with 2% human serum); BMSCs in α-MEM + 5% human serum + 5 ng/mL fibroblast growth factor; both at passages 4–5; one donor each for this study
PART 03 Three Arrangements: Same Fiber, Separate Fibers, or Calling Across the Gap
The cleverness of the experimental design lies here: same ink, same cell ratio, with the only variable being the spatial relationship between the two cell types.
Group 1 — Direct mixed: HUVECs and BMSCs mixed in proportion into the same ink and extruded together, at zero physical distance;
Group 2 — Direct separated: each cell type in its own ink, printed as two adjacent fibers, to see whether cells migrate across on their own;
Group 3 — Indirect: HUVECs printed in the ink, BMSCs in a hanging culture insert, exchanging only soluble signals across a membrane with no physical contact.
All three groups maintained a HUVEC:BMSC ratio of 5:1 (following the team’s prior work). Note that to preserve this ratio, the total cell number per sample was not identical across groups — the separated group used 10 million/mL HUVEC ink and 2 million/mL BMSC ink, giving a lower initial total than the mixed group; the paracrine group received 80,000 BMSCs per insert well. This detail resurfaces when interpreting results. Printing itself used a 0.25 mm inner-diameter nozzle and a four-layer 0/90° crosshatch structure, followed by immersion in thrombin + calcium solution for 10 min to crosslink and set, then 14 days of culture with “health checks” at days 1, 3, 7, and 14.
Cell loading: ink pre-warmed to 37 °C then combined with cells (cell suspension 50 μL/mL); Direct mixed total density 10 million cells/mL (HUVEC 8.33M + BMSC 1.67M); Direct separated two inks at 10M HUVEC/mL and 2M BMSC/mL respectively; Indirect ink HUVEC 10M/mL, BMSC 80,000 per insert (0.3 cm²) ≈ 267,000 cells/cm²
Printing parameters: nozzle inner diameter 0.25 mm; four-layer 0/90° crosshatch structure; printing temperature below 22 °C
Crosslinking and culture: thrombin 2.5 U/mL + 25 mM CaCl₂ for 10 min, DPBS wash 2× 5 min; EGM-2 + 2% human serum + 1% antibiotics + 100 μg/mL Normocin, 1.5 mL per well, medium changed three times per week; 14 days of culture, assays at days 1/3/7/14
Assays: Live/Dead staining, CCK-8 metabolic activity (n=6), immunofluorescence (CD31, vWF, α-SMA, Col-IV), confocal 3D imaging + IMARIS surface area quantification (n=4–8 ROIs per condition), chick chorioallantoic membrane (CAM) in vivo assay
PART 04 The 14-Day Test: All Alive, but Only “Close-Proximity” Groups Formed Networks
First the baseline: all three groups survived the “abuse” of printing — day-1 viability was high and maintained through day 14, with only sporadic dead cells. Metabolic activity rose over time in every group, with the separated group rising steepest; the paper’s explanation is candid: this group had the fewest initial cells, and the lower initial density actually provided more abundant nutrients; high density may cause local hypoxia, and hypoxia is associated with upregulation of pro-angiogenic factors such as VEGF. The HUVEC-only control without BMSCs had acceptable early viability but showed increasing dead cells over time and visibly lower cell density at two weeks.
The real watershed appeared at the tubular network level. Using type IV collagen (a basement membrane component) staining to track: by day 3 HUVECs in all groups had only mildly clustered, by day 7 all had begun to show ordered arrangement, but by day 14 only the two Direct groups containing BMSCs had grown longer, clearer tubular structures. CD31 (endothelial junction protein) 3D rendering similarly showed larger networks in the Direct groups. Quantitative data were even more direct: at days 3 and 7 differences among groups were modest; by day 14, both Direct groups showed multiple large tubular surface regions exceeding 250,000 μm², while the largest in the paracrine group was only about 60,000 μm² — the largest structures differed 6.8-fold between the separated and paracrine groups. The HUVEC-only control never formed a recognizable network.
Cell viability: all three groups showed high viability from day 1 through day 14; only the HUVEC-only control showed increasing dead cells and decreasing density over time
Metabolic activity: rising in all groups; steepest in the separated group (compensation for lower initial cell number); significant difference between Direct and paracrine groups at day 14 (**p<0.01, two-way ANOVA)
Tubular surface area: by day 14 both Direct groups showed structures >250,000 μm²; paracrine group maximum about 60,000 μm²; 6.8-fold difference in largest structures between separated and paracrine groups (Kruskal-Wallis + Dunn test)
Paracrine baseline: assembly in the paracrine group at day 14 still exceeded the HUVEC-only control — paracrine signaling helps, but cannot support networks of comparable scale and stability
PART 05 Cells Do “Visit”: Migration Evidence in the Separated Group
How did the separated group match the mixed group? The paper answers with a set of dual-labeling stains. α-SMA is a typical pericyte marker, usually expressed by migrating pericytes. In 3D confocal images, α-SMA-positive signal appeared between the CD31-positive HUVEC networks in both mixed and separated groups — in the separated group the two cell types originally lived in separate fibers, so α-SMA appearing within the same imaging plane can only mean BMSCs migrated over and took on a pericyte role in situ.
Whether migration can occur depends on whether the matrix “allows” it. The paper specifically cites a key reference: fibrinogen concentration affects the compressive modulus of the crosslinked matrix, and above 2.5 mg/mL it can restrict vascular sprouting and tubular network organization — this study used exactly 2.0 mg/mL, leaving ample space for cell migration and rapid sprouting. Additionally, nozzle diameters (150/250/500 μm) produced no qualitative differences in HUVEC network orientation; the true guide is likely chemokine gradients rather than printing traces. Endothelial cells also showed sprouting behavior toward chemokine gradients — cells have their own sense of direction; the ink only needs to stay out of the way.
PART 06 Evidence in the Chick Embryo: Lumen-Bearing Microvessels Formed
No matter how elegant the in vitro network, it must pass the “living environment” test. The team chose the classic chick chorioallantoic membrane (CAM) assay: printed samples were pre-cultured for 5 days, placed on the CAM surface at embryonic day 7, harvested for histology at day 12 — both blank ink and cell-laden ink were tested. Results showed that regardless of cell content, vascu-ink integrated well with CAM tissue without surrounding inflammation, clearing the biocompatibility exam first.
The more critical finding hides in the details: inside the tubular structures of cell-laden samples, microvessel lumens of ≤10 μm were observed — meaning HUVECs self-assembled not just “solid ropes” but hollow structures resembling vessels; immunostaining also captured a larger slit-like CD31-positive structure closely resembling a small-caliber vessel; HUVECs near the CAM even showed sprouting behavior toward the CAM tissue. The paper also draws the line clearly: the CAM assay does not confirm blood perfusion or functional anastomosis — it provides a proof of concept that “the ink can create an early vascular network-generation microenvironment.”
PART 07 One Practical Design Principle, and Four Questions the Paper Leaves Open
Condensing the three groups into one sentence: the physical presence of BMSCs is a necessary condition for growing strong, stable tubular networks; paracrine signaling alone is not enough; and “physical presence” does not require crowding into the same fiber — only that cells are within migration distance of one another.
For multi-material bio 3D printing this is genuinely good news: vascular cells and supporting cells need not be forced into the same nozzle and the same ink; designers can confidently assign different cell types to different inks and positions, leaving freedom to the tissue-specific architecture itself. The paper also looks forward to combining vascu-ink with tissue-specific inks — allowing vascular networks to “grow in place” inside complex engineered tissue.
The paper is equally direct about its own limitations, leaving four questions for future research:
First, vascular function and perfusion capacity were not tested (the CAM assay itself does not confirm perfusion);
Second, this study used tubular structure surface area for inter-group comparison — the method is direct and reproducible, but future work needs finer measurements of branch points, junctions, and vessel diameter;
Third, matching the ink with the host and in vivo functional validation are unavoidable steps toward translation;
Fourth, long-term maturation, pericyte recruitment, and perfusion bioreactor culture are all homework for the next stage.
Closing Thoughts
but a design principle pinned down by controlled experiments:
in bio 3D printing, giving vascular cells “social distance” is allowed
— provided the matrix is made soft enough and the distance kept within migration range.
For teams building multi-material, multi-cell constructs,
this amounts to another validated arrangement option;
and for the paracrine camp, the cold water is well measured:
paracrine signaling is useful — just not sufficient.
📄 Paper Information
Title: Johansen Å, Korkeamäki JT, Yamada S, Rashad A, Miettinen S, Mustafa K. Effect of 3D-printed co-culture design of mesenchymal stem cells and human umbilical vein endothelial cells on tubular formation. Scientific Reports, 2026; 16: 26067.
DOI: 10.1038/s41598-026-55959-y
Timeline: Received 5 March 2026, accepted 27 May 2026, open access
Author affiliations: University of Bergen (Department of Clinical Dentistry, Center for Translational Oral Research), King’s College London, Middle East Technical University (Turkey), Tampere University (Finland)
This article is a research literature interpretation; all data and conclusions are quoted from the original paper without extension. The bio 3D printing technologies and cell constructs described are still at preclinical research stage and have not entered clinical application. This article does not constitute any diagnostic or treatment recommendation, nor any health intervention reference. Please refer to professional physician advice for individual treatment plans.
· END ·
Separate Strands, Connected Signals:
Bio 3D Printing Shows Vascular Cells
Can Form Networks Without Crowding into One Fiber
Scientific Reports 2026 · University of Bergen / King’s College London et al.
Growing a usable tissue in the lab most often fails not on the cells themselves but on the vasculature — without a vascular network delivering oxygen and nutrients, engineered tissue rarely survives the first weeks after implantation. In 2026, a team at the University of Bergen published a study in Scientific Reports using a custom fibrin-gelatin “vascular ink” to run a controlled co-culture experiment in bio 3D printing: two cell types, three spatial arrangements, 14 days of tracking — all to answer one question: to grow a vascular network, do the cells have to share the same printed fiber? The answer is more forgiving than expected.
Highlights
- Why vasculature is the lifeline of engineered tissue: the 200 μm oxygen diffusion limit, and the 100 μm nozzle that cannot print capillaries
- A custom “vascular ink”: fibrin builds the scaffold, gelatin serves as a temporary placeholder, and viscosity rises sharply below 22 °C
- Three cell arrangements go head-to-head: mixed in one fiber / adjacent separate fibers / paracrine across a membrane — 14 days decide
- Counterintuitive conclusion: cells do not need to be squeezed into the same fiber; within migration distance, the vascular network still forms
*All values quoted from the original paper: 200 μm from the introduction (oxygen diffusion limit, literature data); 6.8× from the results quantitative analysis; ≤10 μm from CAM histological observation.
PART 01 Why Engineered Tissue Always Falls Short: Vasculature Is Non-Negotiable
The demand gap for tissue and organ transplantation continues to widen, and lab-grown engineered tissue is held up as the great hope. But the paper states it plainly from the opening: whether an implant survives depends first on blood supply. Vasculature supports nearly every tissue — delivering oxygen and nutrients, carrying away metabolic waste. In most tissues, cells should be no more than about 200 μm from the nearest blood vessel, the oxygen diffusion limit; once that distance is exceeded or supply is interrupted, cell proliferation, differentiation, and the survival of the entire implant are compromised.
The most obvious way to give engineered tissue vessels is to directly print hollow channels. But the paper runs three calculations:
First, extrusion-based bio 3D printing nozzles are typically on the order of 100 μm — far thicker than capillaries at 4–10 μm;
Second, synthetic vascular grafts have succeeded at large diameters (arteries, large veins) but generally fail at small diameters — surfaces lacking an endothelial lining readily trigger thrombosis;
Third, “top-down” printing of hollow channels is inherently limited by fabrication precision, whereas truly physiological microvessels are actually self-assembled by endothelial cells “bottom-up.”
Attention thus shifted to the self-assembly route: embed vascular endothelial cells (HUVECs) in a printed 3D matrix and let them cluster into tubes on their own. Substantial evidence shows that pairing them with supporting cells in the “pericyte” role — bone marrow mesenchymal stromal cells (BMSCs) in this study — yields more mature and stable vascular networks. But a rival view has persisted: perhaps paracrine signaling alone from supporting cells (factors released across a distance) is sufficient. Does it take physical proximity, or is distance enough? This is precisely the question the paper sets out to answer.
PART 02 A Custom “Vascular Ink”: Fibrin Builds the Scaffold, Gelatin Runs Alongside
The team did not use a commercial ink but formulated their own specifically for vascularization, naming it directly — vascu-ink. The core component is fibrin: a natural polymer that is itself pro-angiogenic and involved in wound healing, serving as a temporary extracellular matrix scaffold in the ink. The paired gelatin has a thermoresponsive “temper” — it improves printability without crosslinking, and under 37 °C culture conditions it simply dissolves and gradually diffuses away, a standard “sacrificial component.” What actually holds the structure together is enzymatically crosslinked fibrin.
This “enzymatic crosslinking” property also determined the process selection: fibrinogen requires thrombin to convert into fibrin, and extrusion printing conveniently allows the ink to be deposited first and then converted, while also enabling multiple inks within the same construct — exactly what the later comparison experiments require. Rheological testing gave the formulation its “health report”: at low shear rates the ink first shows shear-thickening, transitioning to shear-thinning at high shear rates; when temperature drops below 22 °C, viscosity rises sharply (gelatin starts to “set”), while at printing temperature viscosity is only 0.173±0.055 Pa·s — quite thin, leaving room for high cell viability. Testing across nozzle diameters from 0.15 mm to 0.5 mm all extruded smoothly; printed four-layer cylinders maintained structural stability during culture.
vascu-ink formulation: fibrinogen 2% (w/v, Sigma F8630, dissolved in DPBS); gelatin stock 32% (w/v) prepared at 75 °C with 50 rpm stirring, added to a final concentration of 4% (w/v); stored at 37 °C with gentle stirring throughout; both powders UV-sterilized for 1 h
Rheological testing: TA Discovery HR-2 rotational rheometer; parallel plate 20 mm diameter, 1500 μm gap; shear rate 0.1–300 s⁻¹ (22 °C) plus temperature sweep 37–20 °C (1 s⁻¹); n=4 independent replicates
Key values: viscosity at printing temperature 0.173±0.055 Pa·s; viscosity rises sharply below 22 °C; extrudable nozzle range 0.15–0.5 mm (no qualitative difference in HUVEC network orientation across diameters)
Cell culture: HUVECs in EGM-2 (kit’s 2% FBS replaced with 2% human serum); BMSCs in α-MEM + 5% human serum + 5 ng/mL fibroblast growth factor; both at passages 4–5; one donor each for this study
PART 03 Three Arrangements: Same Fiber, Separate Fibers, or Calling Across the Gap
The cleverness of the experimental design lies here: same ink, same cell ratio, with the only variable being the spatial relationship between the two cell types.
Group 1 — Direct mixed: HUVECs and BMSCs mixed in proportion into the same ink and extruded together, at zero physical distance;
Group 2 — Direct separated: each cell type in its own ink, printed as two adjacent fibers, to see whether cells migrate across on their own;
Group 3 — Indirect: HUVECs printed in the ink, BMSCs in a hanging culture insert, exchanging only soluble signals across a membrane with no physical contact.
All three groups maintained a HUVEC:BMSC ratio of 5:1 (following the team’s prior work). Note that to preserve this ratio, the total cell number per sample was not identical across groups — the separated group used 10 million/mL HUVEC ink and 2 million/mL BMSC ink, giving a lower initial total than the mixed group; the paracrine group received 80,000 BMSCs per insert well. This detail resurfaces when interpreting results. Printing itself used a 0.25 mm inner-diameter nozzle and a four-layer 0/90° crosshatch structure, followed by immersion in thrombin + calcium solution for 10 min to crosslink and set, then 14 days of culture with “health checks” at days 1, 3, 7, and 14.
Cell loading: ink pre-warmed to 37 °C then combined with cells (cell suspension 50 μL/mL); Direct mixed total density 10 million cells/mL (HUVEC 8.33M + BMSC 1.67M); Direct separated two inks at 10M HUVEC/mL and 2M BMSC/mL respectively; Indirect ink HUVEC 10M/mL, BMSC 80,000 per insert (0.3 cm²) ≈ 267,000 cells/cm²
Printing parameters: nozzle inner diameter 0.25 mm; four-layer 0/90° crosshatch structure; printing temperature below 22 °C
Crosslinking and culture: thrombin 2.5 U/mL + 25 mM CaCl₂ for 10 min, DPBS wash 2× 5 min; EGM-2 + 2% human serum + 1% antibiotics + 100 μg/mL Normocin, 1.5 mL per well, medium changed three times per week; 14 days of culture, assays at days 1/3/7/14
Assays: Live/Dead staining, CCK-8 metabolic activity (n=6), immunofluorescence (CD31, vWF, α-SMA, Col-IV), confocal 3D imaging + IMARIS surface area quantification (n=4–8 ROIs per condition), chick chorioallantoic membrane (CAM) in vivo assay
PART 04 The 14-Day Test: All Alive, but Only “Close-Proximity” Groups Formed Networks
First the baseline: all three groups survived the “abuse” of printing — day-1 viability was high and maintained through day 14, with only sporadic dead cells. Metabolic activity rose over time in every group, with the separated group rising steepest; the paper’s explanation is candid: this group had the fewest initial cells, and the lower initial density actually provided more abundant nutrients; high density may cause local hypoxia, and hypoxia is associated with upregulation of pro-angiogenic factors such as VEGF. The HUVEC-only control without BMSCs had acceptable early viability but showed increasing dead cells over time and visibly lower cell density at two weeks.
The real watershed appeared at the tubular network level. Using type IV collagen (a basement membrane component) staining to track: by day 3 HUVECs in all groups had only mildly clustered, by day 7 all had begun to show ordered arrangement, but by day 14 only the two Direct groups containing BMSCs had grown longer, clearer tubular structures. CD31 (endothelial junction protein) 3D rendering similarly showed larger networks in the Direct groups. Quantitative data were even more direct: at days 3 and 7 differences among groups were modest; by day 14, both Direct groups showed multiple large tubular surface regions exceeding 250,000 μm², while the largest in the paracrine group was only about 60,000 μm² — the largest structures differed 6.8-fold between the separated and paracrine groups. The HUVEC-only control never formed a recognizable network.
Cell viability: all three groups showed high viability from day 1 through day 14; only the HUVEC-only control showed increasing dead cells and decreasing density over time
Metabolic activity: rising in all groups; steepest in the separated group (compensation for lower initial cell number); significant difference between Direct and paracrine groups at day 14 (**p<0.01, two-way ANOVA)
Tubular surface area: by day 14 both Direct groups showed structures >250,000 μm²; paracrine group maximum about 60,000 μm²; 6.8-fold difference in largest structures between separated and paracrine groups (Kruskal-Wallis + Dunn test)
Paracrine baseline: assembly in the paracrine group at day 14 still exceeded the HUVEC-only control — paracrine signaling helps, but cannot support networks of comparable scale and stability
PART 05 Cells Do “Visit”: Migration Evidence in the Separated Group
How did the separated group match the mixed group? The paper answers with a set of dual-labeling stains. α-SMA is a typical pericyte marker, usually expressed by migrating pericytes. In 3D confocal images, α-SMA-positive signal appeared between the CD31-positive HUVEC networks in both mixed and separated groups — in the separated group the two cell types originally lived in separate fibers, so α-SMA appearing within the same imaging plane can only mean BMSCs migrated over and took on a pericyte role in situ.
Whether migration can occur depends on whether the matrix “allows” it. The paper specifically cites a key reference: fibrinogen concentration affects the compressive modulus of the crosslinked matrix, and above 2.5 mg/mL it can restrict vascular sprouting and tubular network organization — this study used exactly 2.0 mg/mL, leaving ample space for cell migration and rapid sprouting. Additionally, nozzle diameters (150/250/500 μm) produced no qualitative differences in HUVEC network orientation; the true guide is likely chemokine gradients rather than printing traces. Endothelial cells also showed sprouting behavior toward chemokine gradients — cells have their own sense of direction; the ink only needs to stay out of the way.
PART 06 Evidence in the Chick Embryo: Lumen-Bearing Microvessels Formed
No matter how elegant the in vitro network, it must pass the “living environment” test. The team chose the classic chick chorioallantoic membrane (CAM) assay: printed samples were pre-cultured for 5 days, placed on the CAM surface at embryonic day 7, harvested for histology at day 12 — both blank ink and cell-laden ink were tested. Results showed that regardless of cell content, vascu-ink integrated well with CAM tissue without surrounding inflammation, clearing the biocompatibility exam first.
The more critical finding hides in the details: inside the tubular structures of cell-laden samples, microvessel lumens of ≤10 μm were observed — meaning HUVECs self-assembled not just “solid ropes” but hollow structures resembling vessels; immunostaining also captured a larger slit-like CD31-positive structure closely resembling a small-caliber vessel; HUVECs near the CAM even showed sprouting behavior toward the CAM tissue. The paper also draws the line clearly: the CAM assay does not confirm blood perfusion or functional anastomosis — it provides a proof of concept that “the ink can create an early vascular network-generation microenvironment.”
PART 07 One Practical Design Principle, and Four Questions the Paper Leaves Open
Condensing the three groups into one sentence: the physical presence of BMSCs is a necessary condition for growing strong, stable tubular networks; paracrine signaling alone is not enough; and “physical presence” does not require crowding into the same fiber — only that cells are within migration distance of one another.
For multi-material bio 3D printing this is genuinely good news: vascular cells and supporting cells need not be forced into the same nozzle and the same ink; designers can confidently assign different cell types to different inks and positions, leaving freedom to the tissue-specific architecture itself. The paper also looks forward to combining vascu-ink with tissue-specific inks — allowing vascular networks to “grow in place” inside complex engineered tissue.
The paper is equally direct about its own limitations, leaving four questions for future research:
First, vascular function and perfusion capacity were not tested (the CAM assay itself does not confirm perfusion);
Second, this study used tubular structure surface area for inter-group comparison — the method is direct and reproducible, but future work needs finer measurements of branch points, junctions, and vessel diameter;
Third, matching the ink with the host and in vivo functional validation are unavoidable steps toward translation;
Fourth, long-term maturation, pericyte recruitment, and perfusion bioreactor culture are all homework for the next stage.
Closing Thoughts
but a design principle pinned down by controlled experiments:
in bio 3D printing, giving vascular cells “social distance” is allowed
— provided the matrix is made soft enough and the distance kept within migration range.
For teams building multi-material, multi-cell constructs,
this amounts to another validated arrangement option;
and for the paracrine camp, the cold water is well measured:
paracrine signaling is useful — just not sufficient.
📄 Paper Information
Title: Johansen Å, Korkeamäki JT, Yamada S, Rashad A, Miettinen S, Mustafa K. Effect of 3D-printed co-culture design of mesenchymal stem cells and human umbilical vein endothelial cells on tubular formation. Scientific Reports, 2026; 16: 26067.
DOI: 10.1038/s41598-026-55959-y
Timeline: Received 5 March 2026, accepted 27 May 2026, open access
Author affiliations: University of Bergen (Department of Clinical Dentistry, Center for Translational Oral Research), King’s College London, Middle East Technical University (Turkey), Tampere University (Finland)
This article is a research literature interpretation; all data and conclusions are quoted from the original paper without extension. The bio 3D printing technologies and cell constructs described are still at preclinical research stage and have not entered clinical application. This article does not constitute any diagnostic or treatment recommendation, nor any health intervention reference. Please refer to professional physician advice for individual treatment plans.
· END ·