The following materials were purchased from Sigma-Aldrich (United States) and used without further purification: sodium hyaluronate (M_n = 100 kDa), glycidyl methacrylate (GMA), tetraethylammonium bromide (TEAB), poly(ethylene glycol) diacrylate (PEG-DA, M_n = 575), acetone, N,N-dimethylformamide (DMF), potassium permanganate (KMnO₄), sodium tetraborate decahydrate (Na₂B₄O₇×10H₂O), concentrated hydrochloric acid (36%), glacial acetic acid, 4-(dimethylamino)benzaldehyde (DMAB), N-acetyl-D-glucosamine (NAG), and hyaluronidase from bovine testes. Flavin mononucleotide (FMN) was obtained from Pharmstandard (Russia). Triethanolamine (TEOHA) was purchased from Merck (United States). 2,2-Diphenyl-1-picrylhydrazyl (DPPH), 97% purity, was obtained from Hangzhou Dingyan Chem Co., Ltd (China). Phosphate buffered saline (PBS), pH 7.4, was prepared by dissolving biotechnology grade PBS tablet (VWR Life Science, Canada) in 100 ml of deionized water. Penicillin–streptomycin (5,000 U/ml and 5,000 µg/ml, respectively) was purchased from PanEco (Russia). Amphotericine B (5,000 µg/ml) was purchased from JSC “Sintez” (Russia).

The rheological testing routine was performed using viscometer Brookfield DV2T (Ametek Brookfield, United States) equipped with low volume cone-plate measuring system CP-51. Prior to the measurements, samples (0.5–0.7 ml) were thermostated in measurement chamber of the rheometer for 15 min. A temperature 25 ± 0.1°C was maintained using a Thermo Fisher Scientific Accel 500LC thermostat. The measurements were run at 0.1 rpm. Depending on the cone speed and torque the viscosity of the sample was calculated by Reocalc T software. To provide a constant shear stress the rotation was performed for 300 s until the shear rate/shear stress ratio reached an equilibrium.

The diffusion-limited photofabrication process was based on extrusion of photoinitiator-free bioink into aqueous solution containing free radicals. Bioink was loaded into a sterile syringe with a removable needle that was immersed into 0.066 mM FMN dissolved in PBS. To facilitate a cross-linking process, 33.5-mM TEOHA was added as the co-initiator. The photoinitiator solution was excited by a semiconductor laser at a wavelength of 450 nm and 900 mW power. The excitation uniformity was realized by irradiating the photoinitiator solution, while scanning a 16-mm laser beam across the solution. The external diameter of a fiber was controlled by a nozzle, while the inner diameter was controlled by the non–cross-linked bioink that was adjusted by an exposure dose. Removing non–cross-linked ink was performed in PBS by squashing the fiber from one end.

The compression test machine EZ-Test EZ-SX (Shimadzu, Japan) equipped with 500-N load cell was employed for mechanical testing of the samples with cylindrical shape. Young’s modulus E is directly proportional to the compressive stress to compressive strain ratio: E = S l o p e × L 0 A 0 , (2) where L ₀ is the gauge length, A ₀ is the cross-sectional area of the sample under test, and Slope is the ratio of the compressive force to compressive strain evaluated by the method of least squares: S l o p e = F Δ L , (3) where F is compressive force and Δ L is compressive strain.

The efficiency of the free radical generation by FMN and TEOHA was evaluated using DPPH. To carry out this assay, 0.25-mM solution of DPPH in ethanol was mixed with aqueous solutions of photoinitiators (FMN/TEOHA of ratio ranging from 1/1,000 to 1/127) in equal volumes and placed into a 2-mm quartz cuvette. Then the mixture was exposed to 450-nm laser light (intensity, 450 mW/cm²), thoroughly stirred and analyzed using an UV/VIS spectrophotometer Cary 50 (United States) at 200–700 nm range. The reaction of DPPH stable radicals with radicals produced by photoinitiators led to disappearance of DPPH absorption peaks at 340 and 520 nm resulting in color change from deep purple to pale yellow. The time required for DPPH radical depletion defined the efficacy of free radical formation by the FMN/TEOHA photoinitiating complex.

Hyaluronic acid was converted to photocurable form by chemical modification in aqueous medium with glycidyl methacrylate to introduce moieties with the vinyl group (Figure 1). The number of vinyl groups is the key parameter affecting cross-linking density of the hydrogel. Therefore, this parameter was precisely controlled in the implemented reaction proceeding as a reversible trans-esterification through the primary hydroxyl group and the irreversible epoxide ring-opening conjugation with the carboxylic and hydroxylic groups (van Dijk-Wolthuis et al., 1997; Li et al., 2003; Reis et al., 2009). The reaction parameters, such as the concentration of GMA and reaction time, governed the degree of the modification and enabled HA preparation with percentage of vinyl groups on demand for required cellular responses. We synthesized a number of HAGM modifications with various degrees of substitution (DS) from 11.0 to 69.3%. In order to achieve the lowest DS (11.0%) the reaction was carried out for 1 day with the lowest content of GMA (HA:GMA = 1:5). Increase of the reaction time and/or HA:GMA ratio results in expected magnification of DS up to 69.3% at 5 days of reaction and HA:GMA = 1:35). More detailed dependences of DS on reaction parameters are described in the study by Sochilina et al. (2021).

HAGM aqueous solution was used as the hydrogel ink for extrusion. The photoinitiation was realized by using the FMN/TEOHA complex characterized by an absorption band in the spectral range ∼450 nm (Figures 2A,B). We emphasized the advantage of blue light (>450 nm) as non-toxic for living cells (Opländer et al., 2011), contrary to cytotoxic UV irradiation (Clydesdale et al., 2001) that is necessary for excitation of conventionally employed photoinitiators (BASF, 2021). Implementation of TEOHA as the co-initiator provided homolytic decomposition of vinyl groups, followed by radical cross-linking due to the generation of long-lived radicals. As it is known (Orellana et al., 1999), FMN excitation results in FMN short-lived triplet states (∼1 µs), which can activate a double bond of photopolymer. Amine groups are capable to quench the FMN excited states through an electron transfer from the amine to a flavin molecule, converting them to neutral long-lived (∼100 µs) radical species. These radicals enhance the efficiency of initiation of the vinyl monomer polymerization. Thus, evaluation of an appropriate ratio of FMN/TEOHA is crucial for the photo-cross-linking kinetics of HA containing vinyl bond moieties.

We studied various ratios of FMN/TEOHA in terms of the free radical production using the 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical assay (Magalhães et al., 2008; Alamed et al., 2009). The DPPH radical is a long-lived organic nitrogen radical of deep purple color. DPPH turns yellow, when reducing agents, including radicals, are added. Monitoring the decrease of the DPPH absorbance at 528 nm allowed evaluation of the free radical generation in the mixture of FMN solution with TEOHA at various ratios (Figure 2C). It was found that decreasing the FMN/TEOHA ratio at a constant FMN concentration resulted in a decrease of the photobleaching time. We inferred that an increase of the TEOHA concentration yielded maximum concentration of free radicals, which was capable to promote the effective radical photo-cross-linking. Nevertheless, the FMN/TEOHA ratio lower than 1/300 did not lead to a further rapid decrease of the photobleaching time.

The schematic diagram in Figure 3 illustrates key stages of the fiber formation process and radical species formation in the FMN/TEOHA complex under 450-nm laser light irradiation. After photoactivation, radicals permeated through the bioink hydrogel being extruded and triggered cross-linking of the structure from the surface inwards. The penetration depth of the photoexcited radicals determined the extruded fiber cross-linking cross-section profile. Unlinked ink in the central part of structure was removed by dissolving in water. As a result hollow-core structures were produced.

It should be noted that the viscosity of extruded ink is the key parameter limiting the fabrication process. Permeation of the photoexcited radicals in the depth of the extruded volume followed by hydrogel cross-linking must occur prior to complete dissolution of the bioink in the aqueous solution of the FMN/TEOHA complex. To satisfy this condition, the viscosity of the bioink must be higher than 40,000 mPa•s and can be controlled by adjusting the concentration of the polymers, changing HAGM molecular weight or employing HAGM with variable DS (Figure 2D). The degree of substitution (DS) in HAGM used in inks (HAGM 20 wt%, PEG-DA 5 wt%) rising up to 60% led to a viscosity decrease, which is likely to associate with the changes in swelling caused by implementation of hydrophobic moieties in the polymer chain. As a result, the more hydrophobic polymer adsorbs less water, leading to mobility of polymer chains and the shear stress decrease, which translates into a viscosity decrease. Further DS increase probably does not alter the polymer hydrophobicity and, consequently, the viscosity. Therefore, we prepared bioinks containing HAGM (molecular weight ∼100 kDa, DS 46%) 20 wt% and PEG-DA (molecular weight 575 Da) 5 wt%, which resulted in the dynamic viscosity of ∼90,000 mPa•s. This composition ensured uniform extrusion of bioinks and controllable cross-linking in diffusion-limited photofabrication process.

Figure 4 illustrates images of hydrogel structures extruded in aqueous solution of 0.066-mM FMN and 33.5-mM TEOHA under excitation with light at 450 nm. Fibers were extruded through 1.3 mm nozzle at a rate of 0.15 ml/min. Aiming to form a hollow-core fiber, the samples after 3.5 min exposition were placed in PBS for developing to allow uncross-linked hydrogel clear from the fiber core volume by dissolution. As a result, a hollow-core fiber was formed with 3-mm outer diameter and 350-µm wall thickness. Figure 4A illustrates structure with the shape of “question mark sign” printed on the substrate. In order to demonstrate possibilities of retention and motion of the fluid the hollow-core fiber was connected to nozzles. Fiber ends were treated with pressurized air to partially dry and tightening of the hydrogel to fit to the nozzles. Figures 4B,C demonstrate the colorized liquid flux flowing through the fiber with and without bubbles. End facet view of the fiber is presented in Figure 4D. To estimate the cross-linking rate, we formed the fibers with extrusion velocity of 0.15 ml/min through a wide 2.4 mm nozzle. It was found that an increase of the cross-linking process time from 3 to 4.5 min increased the average fiber wall thickness from 330 to 380 µm. Increasing the cross-linking time up to 15 min resulted in solid-core fiber formation. The minimal wall thickness of hollow-core fiber was obtained at extrusion through the 600 µm nozzle. Figures 4E,F illustrate the cross-section of a hollow-core fiber with wall thickness ∼300 μm. Note: HAGM is the material that is subjected to swelling after formation. The extrusion and subsequent cross-linking occur in the water medium, so we avoided the negative factors associated with swelling of the final structure, and water loss by samples is excluded in our approach.

To demonstrate the cell-friendly gelation process, we loaded the bioink with HaCaT cells, immortalized human keratinocytes, which are often used in bioprinting due to their high viability and survivability (Koch et al., 2012; Jeffries et al., 2020; Boonyagul et al., 2021). Free radicals penetrating from the solution to the interior of a bioink hydrogel structure were spent for the activation of the HAGM double bonds. Although using FMN as a photosensitizer for cell damage is well known (Baier et al., 2006; Akasov et al., 2019), we speculate that the exhaustion of potentially cytotoxic free radicals in the hydrogel volume allowed cells embedded in the hydrogel structure volume to escape negative influence. In other words, reactive oxygen species (ROS) generated at the photoactivation FMN/TEOHA complex cannot cause cells oxidative stress because the free radicals quenching took place predominantly by HAGM molecules. Figure 5 represents microscopy images of cell-laden fibers extruded through 2.4-mm nozzle on 3 h and 1, 3, and 6 days of structure incubation. Cell-permeant dyes calcein AM and Hoechst 33342 were used to determine cell viability and nuclei staining, respectively. Fluorescence imaging indicated the cell viability for 6 days and confirmed cell-amiable hydrogel gelation. Since fluorescence of calcein AM could be used to quantify cell viability, we normalized the calcein AM fluorescent signal to Hoechst 33342 one. We found that calcein AM mean fluorescence significantly increased from 34.5 ± 4.7 on day 3–43.1 ± 7.1 (p < 0.05), while the calcein AM/Hoechst 33342 ratio increased from 1.09 on day 3 to 1.35 on day 6 (Supplementary Section S1). This observation could be explained with the increase of viability in cell population. It should be noted that the total cell amount calculated using Hoechst 33342 fluorescence increased insignificantly (from 31.8 ± 5.0 on day 3–32.0 ± 2.3 on day 6) and this could be explained with the slowdown in cell growth in the dense hydrogel microenvironment. We found that 95–100% of cells stained with Hoechst 33342 had also been labeled with calcein AM that confirmed a high survival rate of cells during extrusion, photopolymerization, and followed incubation. This survival rate did not concede the viability of cells in previously published researches devoted to 3D printing of cell-laden scaffolds (Sharma et al., 2020; Hull et al., 2021).

Additionally, hollow-core structures were tested on hemocompatibility. No toxicity was found for as-produced hydrogel structures containing HAGM 20 wt% and PEG-DA 5 wt% (see Supplementary Section S2). Since the concentration of PEG-DA in the bioink appeared to be an instrumental to control the mechanical properties of the produced structures (adding PEG-DA in HAGM:PEG-DA concentration as 2:1 leads to 3-fold increase of hydrogel Young’s modulus) (Savelyev et al., 2017), the hydrogel with up to 15 wt% concentration of PEG-DA was also studied and no detectable toxicity changes were revealed.

Biodegradation is an important requirement for hydrogel samples to control the longevity of an artificial implant in vivo. To evaluate the biodegradation property of as-produced extruded bioink hydrogels, we compared the stability of HAGM before and after the photo-cross-linking in vitro using bovine testicular hyaluronidase, which is known to cleave HA within the living organism (see Supplementary Section S3). Morgan–Elson colorimetric reaction experiments with HAGM 4% solution revealed a decrease of DD from 56 to 27% which corresponded to an increase of the GMA substitution in HAGM with DS from 0 to 63%. We hypothesized that the observed biodegradation was caused by formation of steric hindrance for enzymes during the substrate hydrolysis. A similar pattern of the DD change of the HAGM hydrogels depending on DS was confirmed. In the case of hydrogels, an increase in the degree of the cross-linking can lead to the formation of even greater steric hindrance for the enzyme. The hindered enzymatic degradation of HAGM hydrogels is associated with the prolonged lifetime of such hydrogels in the living body. Thus, variation of DS in initial HAGM appeared to be a promising parameter to control our hydrogel structure lifetime in vivo.

We performed measurements of the Young’s modulus of hydrogel samples. It should be noted, that drying of the samples was excluded during the measurements. The hydrogel containing HAGM 20 wt% and PEG-DA 5 wt% indicated Young’s modulus ∼1.6 ± 0.4 MPa. This value of Young’s modulus falls in the range of Young’s moduli of soft tissues under physiologic conditions including blood vessels (0.701–10.7 MPa) (Gao et al., 2007; Faturechi et al., 2019).

Poor processability and uncontrollable mechanical properties of native polymers limit their application scope in tissue engineering. A number of synthetic polymers, such as polylactic acid (Savioli Lopes et al., 2012), poly(lactic-co-glycolic acid (Castillo-Dalí et al., 2015), and modified polyvinyl alcohol (Velutheril Thomas and Nair, 2019) have been proposed to realize tunable mechanical and biodegradability properties. However, poor biocompatibility of these synthetic materials remains their main drawback. Hybrid composite material that comprises native and synthetic polymers has potential to overcome the limitations inherent to natural polymer while retaining the advantages of synthetic ones. Natural polymers (hydrogels) are derived from alginate, collagen, gelatin, chitosan, agarose, and hyaluronic acid. Among these, we chose endogenous hyaluronic acid due to its properties such as hydrophilicity and biocompatibility. The polymer can be modified with various moieties allowing for a multitude of cross-linking chemistries (Day et al., 2018), for example, glycidyl methacrylate.

Implementation of HAGM as the key component of the photocurable composition improved its toxicity and processability features. For instance, high viscosity of the bioink on the base of HAGM (dynamic viscosity >90,000 mPa•s) improved its printability (Chimene et al., 2020). Moreover, using HAGM in a combination with PEG-DA opens a possibility to produce hydrogel with controllable mechanical properties using the cell-friendly cross-linking. Taking into account our previous study of hydrogels on the base of methacrylated hyaluronic acid (Savelyev et al., 2017), we believe that stiffness of the hydrogel can increase >10-fold by simply changing the concentration of macromolecules. It either can be adjusted by changing the molar weight of the initial polymers or by varying DS of HA functional groups by GMA.

It should be noted that Xu et al. (2020) have proposed and demonstrated an extrusion method for fabrication of biomimetic vessels using conjugated polymers. Formation of small diameter and heterogeneous bilayer blood vessel-like constructs was carried out by using gelatin methacryloyl and the photoinitiator Irgacure 2959 by employing a layer-by-layer photo-cross-linking process. Every four layers produced by an integrated tissue–organ printing (ITOP) system were cross-linked by using UV (365-nm) radiation. We believe that the reported layer-by-layer photo-cross-linking using UV-cured Irgacure 2959 limits an application scope of this approach by exerting an excessive stress on cells. Coaxial technology represents another method of hollow-core tube fabrication. This method suffers from the requirement of machining coaxial nozzles specific for each hollow-core fiber configuration and makes control over the thickness of tube walls difficult. Our approach based on direct single-step extrusion of a hydrogel structure into aqueous solutions containing free radicals was demonstrated to produce controllable cell-amiable hollow-core tubular structures. Moreover, the presented approach has a potential to be implemented in a combination with coaxial nozzle due to advantages in combining several cell types in a highly ordered pattern making the fiber fabrication more versatile and demanded.

In our study, the minimal and maximal diameters of hollow-core fibers were limited by the diameters of nozzles, while the width was adjusted by exposure time. We assumed that the maximum diameter of fiber is unlimited, while the minimal diameter of fiber is affected by the technical limitations. Extrusion of fibers through nozzles 1 mm in diameter is challenging because the fiber starts curling, especially when small gas bubbles remain in the depth of bioinks, so the additional degassing procedure is required. Therefore, it becomes difficult to produce uniform samples. The minimal wall thickness should be chosen enough to provide adequate mechanics of the tube in order to avoid its collapse. The maximal wall thickness is limited by the exposure time. In the experiment for the solid-core fiber with diameter ∼3 mm the exposure time exceeded 15 min. In other words, for translating our approach to 3D printing technology, we speculated that it is maximally acceptable time for photo-cross-linking of the structure. Nevertheless, we believe that our technology is well applicable for fabrication of fibers in a wide range of outer/inner diameters suitable for mimicking blood vessels or bile ducts. Combination of the extrusion process with automatic removal of the fiber from the solution containing free radicals to a developing tank will enable fabrication of samples with virtually unlimited length. Furthermore, changing the time of the cross-linking process in the solution containing free radicals enabled accurate control over the inner diameter ranged from 0% (solid core) to 88% of the outer diameter. As-produced hydrogel hollow-core fibers are easily integrated in more complicated structures or used as a part of tissue engineering construction to facilitate the vascularization process. We believe our method represents a versatile fabrication platform for diverse tissue engineering applications. Thus, our cell-laden bioink diffusion-limited photofabrication is superior in comparison with the existing tissue engineering approaches, where the scaffold printing is followed by cell seeding, which made the process time-consuming. Our technique is facile, single-step, and more affordable.

However, an uptake of our technology in the life sciences and clinical applications hinges on several improvements. For example, permeability and porosity control of the hydrogel warrants further studies in the future. The problem of interconnection of native and artificial vessels is an inherent problem of biomimetic vessels, including our reported technology. Moreover, the biodegradation of grafts based on hollow-core fibers must be experimentally estimated in the full context of a living organism. Adaptation of diffusion-limited photofabrication process for implementation in 3D printing is also required (Supplementary Section S4).