Step‐Ladder Bioprinting to Align Collagen Fibers for Anisotropic Tissue Fabrication
Ilayda Namli, Yogendra Pratap Singh, Deepak Gupta, Syed Hasan Askari Rizvi, Yasar Ozer Yilmaz, Joao Vitor Silva Robazzi, Medine Dogan Sarikaya, Mehmet Baykara, Ibrahim T. Ozbolat
TL;DR
A new bioprinting method aligns collagen fibers to create tissues with better structure and function, such as corneas and cartilage.
Contribution
A novel step-ladder printing approach that enhances collagen fiber alignment and cellular guidance in 3D bioprinting.
Findings
SLP 3D-printed constructs showed improved collagen fiber anisotropy and narrower angle distributions.
SLP guided seeded cells to align with collagen fibers, enhancing tissue organization.
Corneal and cartilage constructs demonstrated high transparency, shape fidelity, and native-like mechanical properties.
Abstract
Aligned collagen microstructure is essential for the mechanical and biological function of anisotropic tissues. However, conventional engineering methods often fail to achieve consistent and tunable fiber alignment within complex geometries. In this study, we developed a step‐ladder printing (SLP) approach by incorporating successive segments of channels of variable widths into a custom barrel design, combining controlled extensional flows with 3D bioprinting to enhance collagen fiber alignment. The results revealed that constructs 3D‐printed via SLP demonstrated improved anisotropy of collagen fibers and narrower fiber angle distributions compared to both extrusion‐based bioprinting with a conventional straight nozzle and drop casting methods. Furthermore, SLP effectively guided the directionality of seeded cells, aligning them consistently with underlying collagen fibers. To exemplify…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6
FIGURE 7Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
Topics3D Printing in Biomedical Research · Collagen: Extraction and Characterization · Electrospun Nanofibers in Biomedical Applications
Introduction
1
Aligned collagen networks represent a fundamental structural component of tissue extracellular matrix (ECM), critically influencing tissue morphogenesis, coordinated cell migration, wound healing, and cancer metastasis [1]. The continuous remodeling of ECM, driven by reciprocal biochemical and biophysical interactions, creates a viscoelastic environment with a wide range of structural and functional properties [2, 3]. The anisotropic organization of fibrous collagen plays a pivotal role in cell signaling and development, orchestrating cell orientation in vitro and directing cell proliferation and migration during tissue regeneration in vivo [4, 5, 6].
Given the physiological significance of collagen alignment, several strategies have been explored to engineer biomimetic scaffolds that replicate the aligned architecture of collagen fibers, mimicking native ECM, such as wet‐spinning, magnetic field induced alignment, mechanical stretching, electro‐compaction, electrospinning, surface patterning, cell‐mediated remodeling, and microfluidic systems. For instance, magnetic field‐induced alignment has demonstrated success in orienting collagen fibers to mimic the anisotropic properties of native tissues. However, early implementations required high intensity magnetic fields (4T–12T), limiting their practicality for scalable tissue engineering [7]. Recent advancements have introduced the use of magnetic nanoparticles, enabling collagen alignment under lower magnetic fields (<0.25T), improving feasibility for biomedical applications [8]. Mechanical stretching is another technique utilized, where uniaxial strain is applied to collagen gels to reorganize fibers, enhancing mechanical properties and cellular orientation [9, 10]. This approach has been particularly effective in guiding cell infiltration and ECM remodeling, making it a valuable tool in tissue engineering [11]. However, its application is often limited to 2D constructs, restricting its use in complex, 3D tissues architectures. A biologically inspired approach involving cell‐mediated remodeling, wherein cells actively organize collagen fibers through contractile forces, is another strategy to induce collagen alignment [12, 13]. While this method offers biological relevance, it lacks precision and scalability, making it unsuitable for fabrication of large‐scale biomimetic tissues.
Microfluidic systems have emerged as a promising platform to achieve collagen fiber alignment in 3D through controlled extensional flows [14]. These flows involve changes in fluid velocity that stretch materials along the flow direction, promoting the self‐assembly of collagen monomers into aligned fibrillar structures [15, 16]. By designing microfluidic channels with variable cross‐sectional areas, researchers can generate gradients of extensional force that drive long‐range fiber alignment within 3D collagen hydrogels [17]. Haynl et al. demonstrated that collagen microfibers produced using microfluidic techniques exhibited superior mechanical properties compared to those fabricated through traditional wet‐spinning methods [18]. Microfluidic approaches have been used to create aligned constructs for corneal tissue engineering, offering a promising alternative to conventional alignment methods [19, 20, 21].
Although techniques including magnetic field guidance, electrospinning, mechanical stretching, and microfluidics effectively induce collagen fiber alignment as discussed above, they are inherently limited in their ability to fabricate structurally complex 3D constructs. For instance, electrospun 2D or thin‐layered scaffolds require additional assembly steps to achieve volumetric tissues. Mechanical stretching aids in alignment but does not contribute to the formation of 3D architecture. Similarly, microfluidic techniques often require additional molding or casting steps to form anatomically relevant geometries. As a result, the fabrication of full‐thickness, complex‐shaped 3D tissue constructs with spatially controlled collagen architecture remains a significant challenge due to current limitations in 3D biofabrication technologies [22, 23].
To overcome the challenges associated with fabricating complex tissue architectures, 3D bioprinting has emerged as a transformative approach, enabling the precise spatial deposition of bioinks to construct multi‐layered, biomimetic tissue constructs with defined structural and functional organization. Despite its advantages in spatial control and construct fidelity, current 3D bioprinting strategies alone lack the intrinsic capability to induce microscale collagen alignment [24]. Indeed, integrating microscale collagen alignment with the fabrication of architecturally complex 3D constructs is essential for replicating the hierarchical organization of native ECM. In this context, the integration of microfluidic alignment strategies with 3D bioprinting presents a novel and highly advantageous approach, enabling in situ induction of collagen fiber alignment during bioprinting.
Herein, we introduce a new bioprinting approach termed ‘step‐ladder printing (SLP),’ featuring a custom‐designed extrusion barrel incorporating microfluidic segments with progressively decreasing cross‐sectional width. This design is specifically engineered to generate localized extensional flows during extrusion and synergistically integrate microfluidic flow dynamics with layer‐by‐layer additive biofabrication, enabling in situ alignment of collagen fibers during bioprinting. SLP facilitates the fabrication of structurally anisotropic constructs, such as cornea and cartilage, mimicking the hierarchical organization of the native ECM. The resulting aligned collagen architecture promotes spatial patterning of cells, leading to defined cellular organization within bioprinted tissues. The successful application of this method in the context of corneal and cartilage tissue engineering highlights its broader relevance to regenerative medicine and opens new avenues for developing implantable, biomimetic tissues, where microscale control over the ECM architecture is critical for guiding cell behavior and restoring tissue function.
Results and Discussion
2
To facilitate controlled collagen fiber alignment via extensional flow, a custom‐designed, microfluidics‐inspired step‐ladder printhead was developed and integrated into an extrusion‐based bioprinting platform (Figure S1) (Movie S1). The printhead was fabricated using a sacrificial 3D printing strategy, creating a staircase‐like internal geometry, engineered to induce step‐wise contractions along the flow path. These geometric constrictions generated localized extensional strain fields during extrusion (Figure 1A), promoting uniaxial alignment of collagen fibrils. Bioprinting and alignment performance were systematically characterized through quantitative assessments of filament morphology and fiber orientation (Figure 1B). The functional utility of SLP was demonstrated through applications in anisotropic tissues, including cartilage and cornea (Figure 1C,D).
(A) Development of the step‐ladder printhead (SLP). A photocrosslinkable silicon composite (PhSC) was loaded into a barrel and a sacrificial ink was 3D printed within PhSC. After crosslinking PhSC, the sacrificial ink was removed to form a hollow channel inside the barrel. (B) Characterization of alignment in bioprinting, including, filament fusion test, collapse test, extrusion test, and cellular directionality assessment. (C,D) Applications of the method: horizontal and vertical alignment of fibers for cartilage tissue engineering, and stromal fiber alignment for corneal tissue engineering.
Collagen Alignment via SLP‐Induced Extensional Flow
2.1
To assess how many fibrils were formed throughout the printing process, we examined their presence before printing, after printing, and after incubation. Confocal imaging revealed progressive fibril development in the collagen bioink across three stages. Prior to printing, initial fibril formation was observed, likely due to spontaneous self‐assembly during neutralization. Fibril density was further enhanced during printing, and post incubation, which promoted additional fibrillogenesis. This stepwise increase in fibrillar content suggests that pre‐existing fibers may align during printing, while newly formed fibrils post‐incubation could adopt the organized structure of the printed hydrogel, supporting overall alignment (Figure S2). The fibril orientation was guided during printing via SLP‐induced extensional flow within stepwise narrowing segments. Each segment transitioned into the next through stepwise contractions, creating periodic constrictions that induced accelerated fluid flow in the narrower regions. As the channel width decreased, fluid velocity increased due to the conservation of mass, giving rise to extensional flow localized near the constrictions. These regions of high extensional strain were hypothesized to play a critical role in directing collagen fibril alignment along the principal flow axis. The interplay between extensional and shear forces in these regions produced significant extensional strain gradients, effectively stretching and aligning collagen fibers during extrusion [16] (Table S1). To evaluate the proposed approach, collagen was printed through the SLP nozzle at a nozzle speed of 600 mm min^−1^. The calculated extensional strain rate was 9.45 ± 0.28 min^−^ ^1^ in constriction A‐B, 12.82 ± 0.38 min^−^ ^1^ in B‐C, and increased to 38.41 ± 1.05 min^−^ ^1^ in constriction C‐D. The strain rate increased progressively along the channel, reaching a peak of 1045.6 ± 20.1 min^−^ ^1^ at the constriction between regions D and E as enumerated in Table S1. Previous studies have demonstrated that strain rates in this range are sufficient to induce alignment of collagen fibers during flow‐based processing conditions [16, 25]. In these high‐strain regions, collagen molecules tend to orient along the principal deformation axis, validating the potential of SLP in guiding fibril organization. In this study, the reported experimental velocities represent average values over a cross‐section, and therefore the calculated strain rate is expected to be lower than the actual local maximum extensional strain rate. COMSOL simulations further predicted higher velocity values compared to experiments. This discrepancy can be attributed to several factors present in real systems but not accounted for in the simulation. Experimentally, the presence of complex particulate nature of collagen, which may cause the fibrils to interact with each other and the nozzle walls and instigate a complicated yield stress behavior. It may also result in partial clogging inside the nozzle, which can cause flow irregularities and reduce flow velocity. Moreover, viscous resistance and energy dissipation within the bioink contribute to additional flow reduction. Since the computational model does not fully capture these rheological and microstructural effects, it overestimates the velocities compared to experimental measurements.
Collagen, as a non‐Newtonian Herschel‐Bulkley (HB) fluid, exhibits shear‐thinning behavior and requires a yield stress to initiate flow. Within the step‐ladder microchannel architecture, the sequential constrictions generate localized regions of elevated strain rate, which reduces the apparent viscosity of the collagen bioink and promotes extensional flow. Unlike shear flow, which predominantly affects fluid layers near boundaries, extensional flow exerts uniform deformation across the entire cross‐section, making it particularly effective for aligning collagen fibrils in confined geometries [16]. The segmented architecture enables repeated and progressive fiber reorganization as the bioink traverses each constriction, avoiding abrupt transitions and preserving structural integrity. Its tunable constriction geometry allows adaptation to diverse hydrogel viscosities and supports consistent fiber orientation across different bioink formulations. Previous studies have also shown that modulating flow and geometry can critically influence fiber alignment and mechanical strength [16, 26]. However, insufficient strain rates or overly short segments may lead to partial or random fiber alignment, highlighting the need to maintain optimal flow conditions and segment lengths.
Printability and Rheological Evaluation of Collagen Bioinks
2.2
Optimizing collagen concentration is critical for achieving both fiber alignment and acceptable printability. Higher concentrations (e.g., 16 mg/mL) offer improved mechanical robustness and structural fidelity, while lower concentrations (e.g., 12 mg/mL) may support better molecular mobility and alignment but compromise print quality due to lower viscosity and gel strength [27]. Rheological assessments revealed the mechanical basis for these observations. As shown in Figure 2A‐(i), viscosity of the 16 mg/mL gel was significantly higher than that of the 12 mg/mL gel at all shear rates, contributing to lower spreading and improved filament stability. Storage modulus was also markedly elevated in the 16 mg/mL gel (Figure 2A‐(ii)), indicating enhanced elastic behavior and better resistance to deformation during extrusion. These rheological characteristics are essential for maintaining shape fidelity and facilitating layer‐by‐layer deposition in bioprinting. To assess these trade‐offs, comprehensive printability and rheological evaluations were conducted.
Printability characterization. (A) Rheological assessment of collagen bioinks (n = 5). (i) Viscosity profiles of SLP12 and SLP16 bioinks as a function of shear rate, measured under steady shear conditions. (ii) Storage (G′) and loss (G″) modulus of SLP12 and SLP16 bioinks obtained from oscillatory shear strain sweeps. (B) Grid printing for bioinks with concentration of 12 and 16 mg/mL (scale bar: 3 mm). (C) Printability quantification (n = 4). (D) Percentage diffusion (n = 5). (E) Filament collapse gap distance (mm). (F) Hanging filament test (n = 6). Data were presented as mean ± SD, where * p < 0.05, ** p < 0.01 and *** p < 0.001.
Figure 2B shows representative bilayer grid structures printed using 12 and 16 mg/mL collagen bioinks. Constructs built with 12 mg/mL collagen exhibited lower resolution, rounded corners, and significant filament spreading, whereas constructs built with 16 mg/mL collagen maintained sharper edges and well‐defined geometries. These visual differences were confirmed by quantitative printability assessments (Figure 2C), where the printability index of the 12 mg/mL gel decreased below the threshold (1.0), particularly for smaller grid sizes (e.g., 2×2), indicating impaired shape retention. In contrast, the 16 mg/mL gel consistently maintained values within the optimal range [28], demonstrating its structural reliability during extrusion. Print diffusion analysis further confirmed the superior performance of the 16 mg/mL gel (Figure 2D). The lower viscosity of the 12 mg/mL gel led to increased spreading upon deposition, compromising spatial resolution and line integrity. Filament collapse test reinforced this result (Figure 2E), where filaments printed with the 12 mg/mL gel visibly sagged and deformed under gravitational force, while those printed with the 16 mg/mL gel maintained their form with minimal deflection. Similarly, in the hanging filament test (Figure 2F), the 12 mg/mL gel failed to support consistent filament formation, showing drooping and irregular lengths. In contrast, the 16 mg/mL gel produced straight, uniform filaments of lengths, reflecting better mechanical resilience and structural retention during unsupported deposition.
The effect of printing speed on filament quality was also analyzed (Figure S3). At low speeds (200 mm/min), over‐deposition and filament merging were observed, which reduced print resolution. At higher speeds (800 mm/min), filament thinning and detachment occurred due to inadequate gel extrusion. An intermediate speed of 600 mm/min was identified as optimal, producing continuous filaments with consistent width and minimal distortion. Analysis of filament corners revealed that the printing speed of 800 mm/min closely replicated the original design, achieving the highest corner fidelity. In contrast, filaments printed at 200 and 400 mm/min exhibited reduced corner printability. Additionally, edge straightness analysis based on pixel distribution indicated that increasing nozzle speed resulted in smoother and more linear filament edges. For these reasons, an intermediate speed of 600 mm/min was identified as the optimal to produce continuous filaments with consistent width and minimal distortion and used for alignment experiments as expounded below.
The flow regime was further characterized by calculating the Weissenberg (Wi) and Deborah (De) numbers to quantitatively link extensional flow behavior with collagen fiber alignment (Figure S4). The computed dimensionless parameters (Wi ≈ 3.3 and De ≈ 5.49) delineate an elastic‐dominated flow regime (Wi > 1, De > 1), wherein elastic stresses are sustained over the characteristic deformation timescale. This rheological state favors persistent extensional deformation of the bioink, promoting alignment and consequently yielding the experimentally observed anisotropic collagen fiber organization within the SLP‐printed constructs.
Fiber Orientation and Anisotropy Quantification
2.3
The alignment of collagen fibers was systematically analyzed in constructs fabricated using drop‐cast (DC) (16 mg/mL), straight‐channel (SC) (16 mg/mL), and SLP configurations at two different collagen concentrations: 12 mg/mL (SLP12) and 16 mg/mL (SLP16). A key objective was to determine whether the extensional flow generated within the SLP geometry could promote collagen alignment beyond the limitations typically observed in strain‐based systems (SC). The presence of initial collagen fibers in the bioink may contribute to alignment under flow, thereby influencing the microstructural organization in final constructs (Figure S2). Additionally, fibers that are already present in the bioink may be aligned during printing through SLP, while newly forming fibers are likely to align concurrently as they assemble within the flow‐oriented hydrogel environment. Confocal reflectance microscopy across z‐planes revealed pronounced fiber alignment in SLP12 constructs compared to the disordered networks observed in DC and SC samples (Figure 3A,B). Hue‐coded orientation maps indicated that SLP12 achieved more consistent angular alignment across depth, whereas deeper fibers in DC and SC appeared largely isotropic. SLP16 also exhibited fiber alignment, though slightly less pronounced than SLP12, suggesting concentration‐dependent effects. Quantitative evaluation using the coefficient of alignment (CoA) and anisotropy metrics provided further insights (Figure 3C‐(i,ii)). The DC group had the lowest CoA (0.17), followed by SC (0.24), both below the threshold (∼0.50), generally considered indicative of significant alignment [29]. In contrast, SLP12 and SLP16 achieved a CoA value of 0.55 and 0.50, respectively, confirming that extensional flow in the segmented channel facilitated robust fiber alignment. Although SLP16 constructs demonstrated better structural fidelity, the marginally lower CoA compared to that for SLP12 may reflect the impact of increased matrix viscosity, which restricted molecular reorientation during the flow. Quantitative anisotropy analysis revealed that the SLP method significantly improved directional organization compared to control groups (Figure 3C‐(ii)). DC and SC constructs exhibited low anisotropy values, consistent with isotropic fiber networks. In contrast, SLP12 and SLP16 yielded anisotropy values of 0.154 and 0.076, respectively. The decline in anisotropy at higher collagen concentration aligns with previous observations that increased matrix density and faster gelation kinetics can inhibit flow‐induced fiber alignment [30]. Although the measured anisotropy values are lower than the anisotropy observed in the native corneal stroma (typically 0.5–0.7) [31, 32], the SLP approach offers superior depth uniformity and scalability not achievable with conventional alignment methods. Importantly, the SLP approach enables the generation of anisotropically‐aligned collagen architectures in constructs with several millimeters in thickness, surpassing the limitations of prior microfluidic or surface‐constrained systems, which are generally restricted to thin‐film formats [33].
(A) Confocal reflectance microscopy (CRM) images showing collagen fibril organization (n = 4) in: (i) drop‐cast (DC), (ii) straight‐channel (SC) at 16 mg/mL, (iii) SLP‐printed at 12 mg/mL (SLP12), and (iv) SLP‐printed at 16 mg/mL (SLP16). (B) Corresponding color maps indicating fibril orientation. (C) Quantitative anisotropy analysis for all collagen groups (n = 4). (D) Orientation distribution plots of collagen fibrils. (E) Compass plots displaying dominant alignment direction for: (i) DC, (ii) SC at 16 mg/mL, (iii) SLP12, and (iv) SLP16. (F) COMSOL simulation results showing: (i) surface velocity magnitude (m/s), (ii) surface shear rate (1/s), (iii) surface pressure distribution (kPa), and (iv) predicted fiber orientation near the nozzle tip. Fibril orientation profiles were shown at two time points: (v) t = 0.1 s and (vi) t = 0.7 s. Data were presented as mean ± SD. Statistical significance: * p < 0.05, ** p < 0.01, *** p < 0.001.
Orientation histograms and compass plots provided complementary insights into the degree and consistency of fiber alignment across different groups (Figure 3D,E). As shown in Figure 3D, DC collagen constructs exhibited broad orientation distributions with no dominant angular direction, suggesting minimal structural organization, a finding consistent with prior reports indicating that DC collagen gels form largely isotropic networks due to the absence of directional cues during fibrillogenesis [26]. In contrast, SC printed constructs demonstrated partial fiber alignment, with a moderate peak near 90°, indicating that even basic extrusion introduces some degree of fiber orientation, although limited by uniform flow fields. SLP printed constructs exhibited markedly improved alignment. SLP12 showed sharp, narrow peaks at 90°, reflecting efficient and uniform fiber orientation under extensional flow. SLP16 also displayed strong alignment with peaks at similar angles, though the peaks were slightly broader, indicating a slight decrease in alignment uniformity likely due to higher viscosity and resistance to flow‐induced strain at elevated collagen concentration. Also, one possible reason could be related to assembly kinetics, where faster fibrillogenesis may occur at higher concentrations, which could lock in a less aligned structure [30]. Compass plots in Figure 3E further support these observations, where DC and SC constructs showed widely dispersed angular distributions, while SLP constructs exhibited tightly clustered orientation vectors around 90°, with SLP12 showing the most concentrated alignment. These results validate that the SLP approach enabled controlled and reproducible alignment of fibers and offers scalability in the bioprinting process itself, allowing for the fabrication of customized 3D constructs with microscale aligned fibers. Moreover, long‐chain polymers subjected to extensional flows can exhibit conformational hysteresis, becoming kinetically trapped in coiled or stretched states depending on flow history [34]. Collagen alignment observed in our constructs might be partially driven by a similar hysteretic mechanism at the nanoscale, where repeated extensional deformation favors chain extension and stabilization, ultimately contributing to microscale fiber organization.
Furthermore, a comprehensive understanding of how flow parameters influence collagen fiber alignment is essential for rationally designing nozzle geometries capable of directing fiber orientation during bioprinting. Toward this, a computational model was developed using COMSOL Multiphysics to simulate flow behavior within an SLP nozzle and predict fiber response to local shear and extensional stress fields. The simulation study revealed that geometric constrictions caused the velocity magnitude to increase most markedly from the nozzle entrance and within the last two to three ladder segments (Figure 3F‐(i)). This acceleration generated extensional flow, enhancing fluid deformation across successive segments. The corresponding surface shear rate map showed elevated values at the interface between segments (Figure 3F‐(ii)), identifying zones of concentrated shear stress that coincide with constriction points. Pressure distribution analysis further supported these findings, revealing pressure drops across ladder steps (Figure 3F‐(iii)). The particle‐based fiber orientation model incorporated into the simulation demonstrated how flow‐induced stress reorients fibers during extrusion (Figure 3F‐(iv)). Fibers introduced at the inlet with random orientation gradually aligned with the principal flow axis as they moved through the channel (Figure 3F‐(v)). By the nozzle outlet, fibers were predominantly oriented at around 90° with the horizontal, demonstrating the effectiveness of the step‐ladder geometry inducing uniaxial alignment (Figure 3F‐(vi)) (Movie S2). This behavior aligns with theoretical models indicating that once extensional strain rates surpass a critical threshold, fiber‐like structures undergo ordering along the deformation axis [34, 35]. Moreover, COMSOL simulations were performed to evaluate the shear and extensional shear rate distributions within different nozzle configurations, including (i) a 4‐step SLP, (ii) a 2‐step SLP 1, (iii) a 2‐step SLP 2 and (iv) a single‐step SLP and (v) a conical nozzle (see Table S2 for geometric specifications of nozzle configurations). As shown in Figure S5, all configurations reached comparable maximum shear rates near 105 s^−1^ at the nozzle outlet, indicating that the local peak was mainly governed by the terminal orifice diameter. However, the internal shear distribution varied significantly among designs. The 4‐step SLP exhibited the highest average shear rate (873 s^−^ ^1^), followed by the 2‐step SLP 1 (605 s^−^ ^1^), 2‐step SLP 2 (645 s^−^ ^1^), and single‐step SLP (458 s^−^ ^1^) configurations. In multi‐step designs (4‐step and 2‐step SLP), the flow repeatedly exposed to shear at the narrow and wide regions, creating ongoing areas of strong deformation that extended several hundred micrometers backward. This progressive shear buildup distributed the stress more evenly throughout the channel rather than concentrating on it near the outlet. In contrast, the 2‐step and single‐step configurations showed broader low‐shear regions (10^2^–10^3^ s^−^ ^1^) interrupted by short, isolated high‐shear zones while the 4‐step configuration provided broader high‐shear regions (10^3^–105 s^−1^). Moreover, as shown in Figure S5, additional simulation tests were run to compare the SLP channel extended with a straight nozzle with respect to a conventional tapered (conical) nozzle. The results revealed that the tapered nozzle exhibited a lower shear rate and a smaller region of influence compared to SLP, confirming the effectiveness of the SLP geometry in inducing controlled shear and fibril alignment.
A similar pattern was observed for the extensional flow fields. While all configurations reached comparable peak extensional rates (around 10^4^ s^−^ ^1^) at the outlet, the average extensional shear rate revealed clear differences among configurations. The 4‐step SLP again generated the highest average extensional rate (406 s^−^ ^1^), followed by the 2‐step SLP 1 (278 s^−^ ^1^), 2‐step SLP 2 (300 s^−^ ^1^), and single‐step SLP (211 s^−^ ^1^) configurations. In the 4‐step SLP, distributed elongational zones were maintained throughout the stepped region, with alternating moderate‐to‐high extension regions (from 10^2^–10^3^ s^−^ ^1^ to 10^4^ s^−^ ^1^), indicating continuous strain accumulation along the flow path. In contrast, the 2‐step and single‐step channels exhibited localized elongation primarily at constriction points, leading to shorter and less uniform deformation exposure. These findings demonstrate that while the terminal orifice determines the maximum achievable shear and extensional rates, the internal geometry plays an essential role in shaping the total deformation area. This extended and evenly distributed flow field is expected to facilitate gradual molecular orientation and collagen fiber alignment, providing a more controlled mechanical environment compared to the abrupt stress peak characteristic of conventional nozzle designs.
The analysis showed that the distribution and magnitude of high‐shear and high‐extensional zones are highly sensitive to the presence and number of step transitions. Increasing the step features expanded the regions of strong extensional deformation and elevated both the average shear and extensional rates. Increasing the number of steps from one to four enhanced the average shear rate by approximately 90% and nearly doubled the average extensional rate, confirming strong geometric sensitivity to the number of steps. These results indicate that the alignment efficiency is governed primarily by step‐induced extensional flow rather than simple geometric constrictions.
Taken together, these simulations confirm that the combination of stepwise extensional flow, localized shear gradients, and pressure transitions drives collagen fiber alignment corroborating the experimental results. Following the quantification of fiber orientation and anisotropy within the SLP system, the detailed morphology of the aligned collagen fibers was examined and discussed in the subsequent section.
Quantitative Analysis of Collagen Fiber Morphology
2.4
To systematically evaluate the morphological effects of different fabrication methods, Figure 4 presents a comparative analysis of collagen fiber width, length, straightness, and alignment in constructs produced via DC (16 mg/mL), SC (16 mg/mL), and SLP (SLP12 and SLP16). In DC samples (Figure 4A), fibers exhibited a narrow width distribution (3–5 pixels), high straightness (∼0.9), short lengths (predominantly 30–50 µm), and random orientation. The resulting network was dense but disorganized, reflecting the absence of directional cues during assembly. This morphology is consistent with isotropic nucleation and growth, typical of static self‐assembly [26] leading to thicker, less ordered fibers. The spatial heatmaps confirmed this phenomenon with broad, non‐directional clustering and heterogeneous fiber distribution. SC printed constructs exhibited similar width (∼4 pixels) and straightness values (0.8–1.0) (Figure 4B), but with a broader distribution of fiber lengths. While most fibers were below 50 µm, outliers extended to nearly 180 µm. Despite increased variability in length, orientation remained poor, indicating that shear alone, without extensional flow, provides limited control over fiber alignment. The vector plots confirmed a lack of coherent directionality across the field of view. In contrast, SLP‐printed constructs demonstrated improved organization. SLP12 constructs exhibited fiber width (3–5 pixels), lengths extending up to 200 µm, and high straightness values (∼0.98), indicative of long, tightly aligned fibers (Figure 4C).
Width, length, and straightness analysis of collagen fibrils. (A) DC fibrils with 16 mg/mL collagen (B) SC with 16 mg/mL collagen. (C) SLP12 and SLP16: (i) color map, (ii) vector field, and (iii) heat map.
All groups exhibited comparable straightness values; however, straightness alone does not fully capture anisotropy. Directionality is also essential, as structures can appear straight without exhibiting a consistent orientation. Therefore, despite similar straightness, some groups (DC, SC) demonstrated reduced directional alignment, highlighting the need to evaluate both parameters for accurate assessment of anisotropy. These observations suggest that the unique stepwise contractions in the SLP geometry generated an extensional‐shear that supports directional fibrillogenesis. Constructs exposed to low shear stress or no shear stress showed shorter collagen fibers with prior studies indicating that collagen fiber elongation follows a non‐linear response. Specifically, maximal fiber extension occurs within a moderate shear window, while both very low and excessively high shear conditions inhibit fibril growth [32]. This suggests that in our low‐shear group, the shear stress may have fallen below the threshold necessary to induce significant collagen extension, explaining the reduced fiber length. SLP12 appears to introduce controlled extensional strain to reinforce alignment and promote elongation. SLP16 constructs showed similar trends, with consistent fiber widths, length and high straightness values (Figure 4D). The alignment of fibers also remained high, with spatial clustering and directional vector fields indicating coherent, uniaxial organization across the constructs.
Taken together, these findings validate SLP as a robust and tunable approach for guiding collagen fiber organization, significantly enhancing fiber alignment and morphology compared to conventional methods, and inspiring future nozzle designs for biofabrication of anisotropic tissues. Simulations and experimental data collectively demonstrate that the synergistic effects of stepwise extensional flow, localized shear gradients, and pressure transitions within the SLP system drive the formation of straighter, more uniformly aligned collagen fibers. While flow dynamics play a central role in directing fiber orientation, additional parameters, including precursor concentration, crosslinking conditions, pH, and buffer composition, also modulate collagen self‐assembly [36, 37]. Among the tested formulations, the 12 mg/mL gel exhibited superior alignment due to lower resistance to flow‐induced strain but lacked print fidelity and mechanical robustness. In contrast, the 16 mg/mL formulation achieved an optimal balance, offering sufficient viscosity and elasticity for high‐resolution printing while maintaining effective fiber alignment under optimized flow conditions. These insights underscore the importance of co‐optimizing bioink composition and flow parameters to engineer hierarchically‐organized collagen constructs that closely mimic the anisotropic architecture of the native ECM.
Cellular Activity on Aligned Collagen Constructs
2.5
To assess the biological relevance of the aligned collagen constructs generated via SLP, a comparative study was conducted using Normal human lung fibroblasts (NHLFs)‐seeded constructs. The aim was to evaluate cell orientation and cell viability in relation to fiber architecture. First, the spatial distribution and density of collagen fibers were visualized with Sirius Red staining (Figure 5A). The stained images showed dense and disorganized collagen deposition in the DC group, whereas the SC group displayed partially‐organized fiber matrix. In contrast, the SLP group supported densely packed, linearly arranged collagen fibers, indicative of a well‐organized matrix architecture. These observations were further corroborated by Scanning Electron Microscopy (SEM) analysis, which showed randomly oriented fiber network in DC, partially aligned fibers in SC, and highly aligned submicron‐scale fiber alignment in the SLP group, suggesting aligned construct topology (Figure 5B).
(A) Sirius Red staining of collagen (scale bar:100 µm). (B) SEM images of collagen fibrils (scale bar: 5 µm). Anisotropy of NHLF orientation. (C) (i) Immunofluorescent images of NHLFs on DC, SC, and SLP constructs (scale bar:100 µm), (ii) color map of NHLF orientation distribution. (D) Bar graphs representing anisotropy values calculated from stained NHLFs. (E) Proliferation of NHLFs at Days 1 and 7. (F–H) AI‐based tracking of MDA‐MB‐231 cells depicting the analysis of cell alignment, directionality, and migration patterns. (F) Orientation and distribution of cells at three time points (6, 18, and 36 h) during time‐lapse imaging. (G) Cell trajectories along the x‐ and y‐axes, with each color representing an individual tracked cell. (H) Directionality analysis of the tracked cells: (i) scatter plot of directionality persistence for each cell, illustrating how consistently cells migrate in the same direction in the x‐y plane, (ii) directionality persistence specifically along the y‐axis, and (iii) variance in mean migration angles of individual cells based on their trajectories, (iv) graph shows the displacement along the x‐axis for each tracked cell in DC, SC, and SLP. Data were presented as mean ± SD, where * p < 0.05, ** p < 0.01 and *** p < 0.001.
The fibroblasts morphology was then studied via immunofluorescent imaging of their cytoskeletons (Figure 5C‐(i,ii)), which revealed clear morphological differences across substrates. On DC constructs, cells displayed a random, spread‐out morphology, consistent with the disorganized underlying collagen network. SC constructs exhibited moderate alignment of cells, whereas cells on SLP constructs displayed elongated, spindle‐like morphology with pronounced uniaxial alignment along the direction of underlying aligned fibers. This was further supported by the orientation distribution analysis (Figure 5A‐(iii)), which showed broader angular dispersion in DC and SC samples, whereas SLP16 constructs demonstrated a narrow peak around −70°, indicative of highly aligned cell orientation. Quantification of cellular anisotropy (Figure 5D) confirmed this trend, with the SLP16 group exhibiting significantly higher alignment (p < 0.001) compared to both DC and SC. To assess cytocompatibility, cell proliferation assays were performed. As shown in Figure 5E, proliferation was enhanced in all groups on Day 7, indicating biocompatibility.
To further quantify alignment‐dependent cellular response, Artificial Intelligence (AI)‐based cell tracking analysis was conducted (Figure 5F–H). Time‐lapse imaging combined with automated tracking revealed distinct differences in cellular migration behavior across the DC, SC, and SLP constructs. The results revealed consistent directional migration of cells along the SLP axis, while cells on SC and DC constructs showed more randomized trajectories (Movie S3). The directional coherence in the SLP group confirms that topographical alignment can effectively guide cell movement, a key requirement in wound healing and tissue regeneration applications (Figure S6). Orientation distribution at the final time point exhibited broad angular spread in the DC group, indicating random and multidirectional migration patterns (Figure S7). In contrast, the SC group showed a moderately narrower distribution (Figure S8), while the SLP group exhibited the most confined angular spread, with minimal deviation below 30° or above 150°, suggesting highly restricted and aligned movement (Figure S9).
Trajectory plots illustrated individual cell paths, where each color represents a distinct cell plotted along the x‐ and y‐axes (Figure 5G). The results indicated disorganized movement patterns in the x‐y plane for the cells seeded on both DC and SC groups, whereas cells on SLP predominantly migrated along the y‐axis. Directionality persistence, calculated as the ratio of net displacement to the total path length, quantified the degree of directed migration, with values approaching 1 indicating highly persistent movement (Figure 5H). This metric is critical for evaluating alignment‐dependent behavior, as it directly reflects how consistently cells migrate in a preferred direction. Mean persistence values were approximately 0.05 for DC, 0.13 for SC, and 0.32 for SLP, with a subset of SLP cells exhibiting values closer to 1, indicating enhanced directional guidance in SLP that was significantly different from both DC and SC (Figure 5H‐(i)).
When directional persistence was analyzed specifically along the y‐axis, the results were in agreement with the overall directionality trends, with the SLP group exhibiting a significant higher persistence compared to DC and SC, confirming that cells preferentially migrated along the aligned fiber or 3D printing direction (Figure 5H‐(ii)). Furthermore, analysis of mean cell migration angles demonstrated reduced angular variance among cells on SLP, in contrast to the broader angular distribution observed for DC and SC groups (Figure 5H‐(iii)). Statistical testing revealed that the variance in SLP was significantly lower, highlighting enhanced alignment‐driven guidance. Conversely, evaluation of x‐directional movement showed significantly higher lateral movement for DC and SC, while SLP exhibited limited x‐axis migration, consistent with its anisotropic structural constraints (Figure 5H‐(iv)). These findings underscore the importance of precise cell tracking and alignment analysis for revealing the impact of substrate architecture on migration patterns and persistence.
Collectively, these results demonstrate that the SLP fabrication method yields an anisotropic microenvironment that promotes highly directional, vertically aligned cell migration with minimal lateral deviation. In contrast, DC and SC constructs permit more random and multidirectional migration patterns. These findings support the hypothesis that embedded structural cues within constructs can effectively direct cellular organization and behavior, offering a promising strategy for tissue engineering applications, where aligned ECM and directional cell orientation are critical.
Applications of SLP‐Fabricated 3D Constructs
2.6
We demonstrate the applicability of SLP‐fabricated 3D constructs, featuring extensional flow‐induced aligned collagen fibers, in the context of corneal and articular cartilage grafts. Despite their structural differences and functional roles, both the corneal stroma and articular cartilage highlight the importance of collagen fiber alignment in maintaining their respective functionality, tissue integrity, and guided cell behavior. These applications highlight the translational potential of SLP fabricated construct designs that incorporate anisotropic microarchitecture for tissue engineering applications requiring organized ECM and directional cellular responses.
3D Bioprinted Corneal Constructs with Aligned Collagen Architecture
2.6.1
The corneal stroma is a highly organized ECM, where the precise lamellar arrangement of type I collagen fibers is essential for maintaining transparency and biomechanical strength. The stroma houses keratocytes, quiescent cells that preserve ECM homeostasis under normal conditions [38]. However, upon corneal injury, keratocytes undergo apoptosis, triggering a wound healing cascade of events that leads to stromal cell differentiation into fibroblasts and myofibroblasts. These cells secrete disorganized and optically heterogeneous ECM, resulting in scar formation and opacity, which impairs vision [39, 40]. Collagen fiber alignment and diameter are key regulators of the quiescent keratocyte phenotype, suggesting that preservation of native ECM architecture may help prevent fibrotic transformation during corneal wound healing [41]. Engineering a corneal construct that replicates the native collagen orientation is therefore a key objective in regenerative ophthalmology.
To mimic the native corneal architecture, human corneal cells isolated from donor tissues were embedded within collagen matrices in this study. To replicate the anatomical dome‐shaped curvature and layered architecture of the native cornea, corneal constructs were 3D bioprinted in concentric circular paths. Layer by layer deposition was performed to approximate the hemispherical geometry, with each printed layer subjected to UV crosslinking to ensure structural stability and maintain the curvature. This radial printing strategy enabled controlled filament alignment along the corneal contour while preserving the overall dome‐shaped geometry of the construct. Optical transparency, an important requirement for corneal substitutes, was assessed using UV‐visible spectrophotometry across 400–700 nm. As shown in Figure 6A, the thin collagen (∼0.5 mm) exhibited high initial transparency, reaching ∼98% transmittance at 700 nm on Day 0, which slightly decreased to ∼90% by Day 7. In contrast, the thicker collagen (∼1 mm) construct showed reduced transparency, reaching ∼82.8% by Day 7, due to continued collagen fibrillogenesis and network densification. These values are comparable to >90% light transmission typically observed in the native cornea [42] and are better than other bioengineered systems, such as aligned electrospun collagen mats (63% transmittance under 500 nm) and non‐aligned mats (as low as 40% under 500 nm) [27, 35]. Similarly, hydrogel‑based systems often report transmittance values ∼80% at 400 nm and up to 95% at 700 nm, depending on the composition and crosslinking strategy [43].
(A) Transparency measurement of native cornea, thin crosslinked collagen (100 µL), thick crosslinked collagen (200 µL) and PBS (n = 3). (B) FACs analysis of isolated corneal fibroblasts. (C) Immunofluorescent images of bioprinted corneal constructs (scale bar: 100 µm). (D) Tail imaging of whole bioprinted corneal construct (scale bar: 500 µm, inset image's scale bar: 200 µm) (E) H&E staining of bioprinted corneal construct (scale bar: 100 µm). Data were presented as mean ± SD, where * p < 0.05, ** p < 0.01 and *** p < 0.001.
Flow cytometry analysis was performed to validate the phenotypic identity of isolated cells, confirming over 80% were CD166^+^ (Figure 6B) [44]. Immunofluorescent imaging of the printed constructs demonstrated 4',6‐diamidino‐2‐phenylindole (DAPI), phalloidin, and CD166 staining, revealing well‐organized and stratified cell layers aligned along the printing direction (Figure 6C). This alignment reflects the construct's anisotropic cues, which promoted actin cytoskeleton orientation and facilitated functional ECM deposition. Notably, phalloidin and CD166 signal co‐alignment across layers confirmed that the SLP‐printed architecture supported multi‐layered cellular orientation, critical for recreating the native corneal stroma. Anatomical fidelity was verified via whole‐construct imaging and histology. A macroscopic stitched image showed curvature mimicking native corneal geometry, achieved without the use of external molds (Figure 6D). Hematoxylin and eosin (H&E) staining revealed a layered architecture with regionally nuclei distribution (Figure 6E). These results suggest that the SLP‐based constructs not only achieved transparency but also promoted cellular and ECM alignment, replicating critical features of the native cornea.
3D Bioprinted Anisotropic Cartilage Constructs
2.6.2
To further validate the broader applicability of the SLP approach beyond corneal tissues, we applied it to fabricate structurally anisotropic cartilage constructs designed to mimic the native zonal architecture of articular cartilage. Articular cartilage consists of a highly organized ECM rich in type II collagen fibers and proteoglycans, which provides mechanical resilience and facilitates efficient distribution of mechanical loads across the joint [45]. It exhibits a zonally stratified structure, with each zone fulfilling distinct mechanical and biological functions. The superficial zone primarily resists shear stress and secretes lubricating molecules to maintain joint function, while the middle, deep, and calcified zones are specialized for bearing compressive loads [46]. This functional stratification is supported by zone‐specific collagen fiber orientations: in the superficial zone, collagen fibers are aligned parallel to the articular surface to counteract shear forces, whereas in the deep zone, fibers are oriented perpendicular to the surface, providing structural support against compressive stress. These distinct alignment patterns are critical for maintaining its mechanical integrity and load‐bearing capacity. Several studies have attempted to replicate collagen fiber alignment in engineered cartilage by applying mechanical stimulation, directional freeze‐casting, and scaffold patterning techniques [47, 48, 49]. However, these methods often rely on prolonged cell‐driven remodeling, provide limited control over zonal organization, or lack the ability to produce alignment during fabrication. In contrast, our approach enables the direct fabrication of multi‐zonal collagen alignment, reducing dependence on post‐fabrication cellular activity.
To replicate this hierarchical structure, we prepared µed bioink encapsulating adipose‐derived stem cells (ADSCs)‐derived chondrocytes. We then strategically bioprinted an array of vertical cylindrical pillars arranged circumferentially to form a hollow ring, mimicking the collagen orientation of the deep zone. This framework was then overlaid with horizontal collagen filaments to reproduce the superficial zone's tangential fiber alignment (Figure 7A, top two rows). Prior to bioprinting, chondrogenic differentiation of ADSC‐derived cells was validated in 2D culture. Immunofluorescence staining of these differentiated cells showed robust expression of collagen type II (COLII) and aggrecan (ACAN) (Figure S10, left), confirming commitment to the chondrogenic lineage. These pre‐characterized cells were subsequently used in the bioink formulation for 3D bioprinting, ensuring that the constructs were seeded with cells capable of producing cartilage‐specific matrix components. By guiding collagen alignment along the printing path, SLP enabled the fabrication of multi‐zonal constructs with spatially defined architecture, supporting both mechanical function and biomimicry, which is essential for cartilage regeneration. The constructs retained print fidelity and spatial geometry post bioprinting. H&E staining confirmed uniform nuclei distribution and structural preservation (Figure 7A, bottom two rows), validating that the SLP approach enabled uniform cell distribution within complex‐shaped 3D constructs. Moreover, Days 1 and 7 images showed time‐dependent changes in matrix staining (Figure 7B). Safranin O/Fast Green exhibited an increase in intensity by Day 7, indicating greater accumulation of cartilaginous matrix components. Alcian Blue/Nuclear Fast Red also showed an increase in GAG‐associated staining over the same period. In contrast, Masson's Trichrome demonstrated relatively similar collagen staining at both time points, suggesting limited detectable changes in collagen deposition within the 7‐day time course. A high‐magnification immunofluorescent image revealed ACAN localization, suggesting that construct anisotropy promoted organized ECM formation (Figure S10, right). Compression testing of the constructs revealed non‐linear stress‐strain behavior with strain‐stiffening, mimicking the viscoelastic profile of native cartilage (Figure 7C). The compressive modulus increased from ∼112 kPa on Day 1 to ∼250 kPa on Day 7, which fell within a range of 240–1000 kPa reported for the native articular cartilage [50, 51, 52]. This increase was likely due to matrix organization and/or collagen crosslinking occurring during the 7‐day incubation period. This mechanical robustness likely results from the interplay of bioprinting‐induced fiber alignment and the vertically/horizontally organized construct geometry.
(A) Zonal‐mimetic hydrogel constructs for cartilage tissue engineering (image created using BioRender). (A) The schematic shows cartilage tissue, which exhibits a distinct zonal architecture composed of a superficial zone, a middle (transitional) zone, and a deep zone. Schematic panels represent construct geometries (top row). Corresponding bioprinted constructs were shown below (second row, scale bar: 1 mm). H&E staining shows internal architecture and cell distribution (third row, scale bar: 500 µm). Close‐up views highlight organized, layered morphology within the constructs (bottom row, scale bar: 500 µm). (B) Histological staining of transverse sections were taken from the bottom region of the engineered cartilage constructs. Sections were processed with Masson's Trichrome, Safranin O/Fast Green, Alcian Blue and visualize collagen, matrix components and glycosaminoglycans. (scale bars:100 µm.) (C) (i) Bioprinted constructs before and during unconfined compression, showing sample deformation under loading. (ii) Representative stress–strain curves for constructs at Days 1 and 7, illustrating their elastic deformation behavior under unconfined compression. Arrows indicate the stress response at approximately 20% compressive strain. (D) (i) Quantification of the total DNA content on Days 1 and 7 and (ii) DNA content normalized to hydrogel mass (n = 3). (E) (i) The GAG content normalized to hydrogel mass (n = 5). (ii) The GAG content normalized to the DNA content on Days 1 and 7 (n = 3). (F) (i) Quantification of the total collagen content in hydrogel constructs on Days 1 and 7. (ii) The collagen content normalized to the DNA content on Days 1 and 7 (n = 4). Data were presented as mean ± SD, where * p < 0.05, ** p < 0.01 and *** p < 0.001.
DNA quantification indicated a 1.8‐fold increase in total cellularity from Day 1 to 7 (Figure 7D‐(i)); p < 0.001). When normalized to the hydrogel mass (Figure 7D‐(ii)), the DNA content remained relatively stable. These results are consistent with previous studies using ADSC‐derived chondrocytes, where the DNA content similarly increased from ∼189 ng on Day 1 to ∼357 ng by Day 7 [53]. Glycosaminoglycan (GAG) analysis revealed a substantial (∼8‐fold) increase in total GAG content by Day 7 (Figure 7E‐(i)); p < 0.01), indicating active ECM deposition. GAG normalized to DNA (Figure 7E‐(ii)) showed a ∼1.2‐fold increase, supporting enhanced matrix production per cell. These values align with previous publications [54, 55, 56] and confirm that the SLP‐based constructs promote functionally‐relevant matrix synthesis within a short culture duration. Interestingly, the total collagen content (Figure 7F‐(i)) showed no significant increase between Days 1 and 7. However, collagen normalized to DNA (Figure 7F‐(ii)) increased ∼3.5‐fold (p < 0.001), reflecting enhanced collagen deposition per cell and early matrix remodeling. These findings are consistent with prior studies, indicating active chondrocyte‐driven ECM formation even when total collagen changes were moderate [56]. Thus, the SLP approach enables the fabrication of biomimetic, anisotropic cartilage constructs with regionally organized cellular and structural architecture as evidenced by H&E staining. These results highlight the translational potential of aligned collagen deposition for cartilage reconstruction, especially in applications such as osteochondral grafting and joint resurfacing.
Together, the results demonstrated that the SLP approach enabled the fabrication of anatomically relevant, aligned, and structurally tunable constructs across multiple tissue types, including cornea and cartilage. The ability to guide both collagen fiber alignment and cellular orientation in 3D without the need for external magnetic, electrical, or mechanical fields represents a significant advancement in biofabrication. Constructs exhibited high structural fidelity, promoted lineage‐specific matrix deposition, and achieved performance benchmarks that either matched or exceeded those reported in the literature. While the SLP approach aligns collagen fibers through the extensional flow, several aspects remain to be optimized. First, the alignment depends strongly on collagen concentration. Lower concentrations (<12 mg/mL) generally promote higher degrees of alignment due to reduced shear resistance; however, maintaining adequate printability requires concentrations of approximately 16 mg/mL or higher, where the increased viscosity ensures filament stability during deposition. Second, prolonged printing times or nozzle residence can lead to partial fibril pre‐assembly before deposition, reducing reproducibility. Additionally, while the SLP geometry enhances flow control, it increases shear exposure at step transitions, which may influence embedded cell viability in highly sensitive cell types. Under cyclic loading, the crosslinked collagen network is expected to retain its alignment with minimal degradation in mechanical integrity. However, long‐term fatigue behavior remains to be characterized. Optical scatters could arise from microscale heterogeneities or minor fiber misalignments introduced during deposition. Finally, scale‐up of SLP constructs poses challenges related to maintaining uniform alignment across larger geometries, necessitating further optimization of the nozzle design and printing parameters. These future investigations will be vital to validate the clinical potential of this biofabrication strategy across diverse tissue engineering applications. Furthermore, the in‐vivo assessment would provide understanding of the translational potential of the SLP approach. Future work will explore multi‐material and multi‐cellular bioinks for reconstructing more complex tissue interfaces.
Conclusion
3
This study introduces an SLP approach to direct collagen fibrillogenesis via extensional flow. The modular constriction‐based design enabled tunable fiber alignment across varying concentrations of collagen bioinks, offering a significant advancement over conventional strain‐based approaches for collagen alignment. Structural and functional evaluations, including COMSOL simulations, fiber morphology, mechanical testing, and cellular assays, collectively confirmed SLP's capability to generate anisotropic tissue constructs with enhanced fidelity, mechanical integrity, and cell alignment. Demonstrations with engineered corneal and cartilage tissues highlight the translational relevance of SLP. Overall, the SLP approach offers a scalable and adaptable strategy for fabricating engineered tissues that require precise collagen fiber alignment and restoration of native biomechanical properties.
Materials and Methods
4
Materials
4.1
Acetic acid, xanthan gum, and sodium hydroxide (NaOH), and ascorbic acid were purchased from Sigma‐Aldrich (St. Louis, MO, USA). Riboflavin was obtained from Santa Cruz Biotechnology (Dallas, TX, USA) and dialysis tubing was acquired from Spectrum Labs (Rancho Dominguez, CA, USA). Dulbecco's phosphate‐buffered saline (DPBS), Opti‐MEM, phosphate buffered saline with tween‐20 (PBST), Alexa Fluor 568 (A12379), and) DAPI were purchased from Thermo Fisher Scientific (Waltham, MA, USA). The pH indicator was from EMD Millipore (Burlington, MA, USA). Trichloro(1H,1H,2H,2H‐perfluorooctyl) silane was obtained from MilliporeSigma (Darmstadt, Germany). Needles were sourced from Grainger (Lake Forest, IL, USA) and 3‐cc barrels were purchased from Nordson EFD (East Providence, RI, USA). NHLFs were obtained from (Lonza, Houston, Texas, USA), green fluorescent protein (GFP)‐tagged MDA‐MB‐231 (^GFP+^MDA‐MB‐231) cells, a triple‐negative breast cancer cell line, were obtained from Dr. Danny Welch, University of Kansas (Kansas City, KS) and ADSCs were purchased from Lonza (Walkersville, MD, USA). DMEM was purchased from Gibco (Thermo Fisher Scientific, Waltham, MA, USA), and fetal bovine serum (FBS) from Avantor (Radnor, PA, USA). TrypLE enzyme and penicillin‐streptomycin (PS) were obtained from Corning (New York, NY, USA) and Life Technologies (Carlsbad, CA, USA), respectively. Basic fibroblast growth factor (bFGF) was purchased from PeproTech (Cranbury, NJ, USA), and the chondrocyte differentiation medium was obtained from Cell Applications (San Diego, CA, USA). For immunostaining, the Aggrecan monoclonal antibody (BC‐3) was acquired from Invitrogen (Carlsbad, CA, USA) and anti‐collagen II antibody was purchased from Abcam (Cambridge, UK). The anti‐CD166/ALCAM (FITC) antibody was obtained from SinoBiological (Houston, TX, USA). The LIVE/DEAD viability assay kit and AlamarBlue proliferation reagent were both from Invitrogen. Additional reagents included paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA, USA), Triton X‐100 (Alfa Aesar, Ward Hill, MA, USA), and bovine serum albumin (BSA) from VWR (Radnor, PA, USA).
Fabrication of the SLP Nozzle
4.2
A 3‐cc barrel was filled with a silicone‐based material, which undergoes photocrosslinking upon exposure to 405 nm light, as described in our previous work [57]. This photocurable silicone matrix served as a support bath for embedded printing of a xanthan gum‐based sacrificial ink. The sacrificial ink consisted of 5% (w/v) xanthan gum dissolved in deionized water. A 23G, 2‐inch stainless steel nozzle was employed to print the sacrificial ink within the silicone‐filled barrel, to create a step‐ladder shaped channel structure using an extrusion‐based bioprinter (CELLINK, Sweden) at an applied pressure of 20–40 kPa. The printed structure consisted of four sequential segments of equal lengths but with increasing widths of 2, 4, 6, and 9.5 mm, spanning a total length of 20 mm, as shown in Figure S1. Following the printing process, the silicone matrix was photo‐crosslinked by exposing it to 405 nm light for 5 min. After photocrosslinking, the printed sacrificial ink was subsequently removed via gentle flushing with warm deionized water, resulting in formation of a well‐defined, hollow step‐ladder microfluidic channel embedded within the crosslinked silicone matrix inside a barrel (Figure 1).
Collagen Preparation and Neutralization
4.3
Rat tails were purchased from ROCKLAND (RT‐T297). Collagen type I was extracted from rat tail tendons following a published protocol [58]. Tendon fibers were carefully dissected from the tails and dissolved in 0.02 N acetic acid to obtain collagen solution. The resulting solution was then freeze‐dried to obtain dry collagen, which was weighed and re‐dissolved in 0.02 N acetic acid to achieve final concentrations of 12 and 16 mg/mL. The solution was centrifuged at 15 000 × g for 1 h at 4°C using a Sorvall Legend X1R centrifuge (Thermo Fischer) to remove insoluble residues. Sterilization was performed using Spectra/Por 1 dialysis Tubing (6–8 kDa MWCO), according to a previously described method [58]. Collagen neutralization was performed using NaOH to adjust the pH to 6.0–6.5, preventing premature fibrillogenesis. All neutralization steps and reagent handling were carried out on ice to maintain collagen solubility. For physical crosslinking, riboflavin (5 mg/mL in DPBS) was added to the collagen solution at a 2:100 (v/v) ratio for photocrosslinking post bioprinting.
Rheological Measurements
4.4
Rheological properties of collagen at concentrations of 12 and 16 mg/mL were characterized using a rheometer (MCR 302, Anton Paar, Austria). Each sample was tested in quintuplicate (n = 5). A 25‐mm parallel‐plate geometry was employed, with temperature maintained at 10°C using a Peltier‐controlled system. Shear‐thinning behavior was assessed through a flow sweep test, where the shear rate varied logarithmically from 0.1 to 100 s^−^ ^1^. Viscoelastic properties were evaluated via amplitude sweep tests at a constant frequency of 1 Hz, applying strain from 0.1% to 100% to identify the linear viscoelastic region. Frequency sweep tests were conducted within this region to measure the storage (G′) and loss (G″) modulus, using a fixed shear strain of 0.1% while varying angular frequency from 0.1 to 100 rad/s. To evaluate self‐healing behavior, alternating low (0.1 s^−^ ^1^ for 60 s) and high (100 s^−^ ^1^ for 10 s) shear rates were applied in repeated cycles. Viscosity recovery after each cycle was recorded to assess structural integrity and resilience of collagen gels under shear.
3D Printing with the SLP Nozzle
4.5
Collagen inks of different concentrations supplemented with riboflavin were prepared and loaded into the SLP nozzle. It was then centrifuged at 2000 × g at 4°C for 5 min to remove air bubbles before printing. Inner and outer diameter (ID/OD) for the nozzle was 0.337 and 0.641 mm, respectively. Bioprinting was performed on a substrate maintained at 10°C with relative humidity of 35%–60%. A 23G needle was used and collagen was extruded at an applied pressure of 20–50 kPa using an INKREDIBLE+ bioprinter (CELLINK, Sweden). During extrusion, 405‐nm light was applied directly at the nozzle tip to initiate in situ crosslinking, which stabilized the printed structures and maintained geometric fidelity throughout the printing process. UV crosslinking was performed using a Lumen Dynamics OmniCure S1500A light source equipped with a 200‐W mercury arc lamp and a 365‐nm band‐pass filter (model 019–01045R), which provided controlled and wavelength‐specific UV output to the samples. Samples were positioned 50 mm away from the light‐guide tip, where the surface irradiance was approximately 10 mW cm^−^ ^2^ at full intensity according to manufacturer specifications. Under these conditions, a 10‐min exposure corresponded to a total radiant dose of 6 J cm^−^ ^2^.
All constructs were printed directly onto glass coverslips, which were sterilized by sequential rinsing with 100% ethanol, air drying, and UV irradiation. Prior to printing, the coverslips were functionalized via vapor‐phase silanization using trichloro(1H,1H,2H,2H‐perfluorooctyl) silane vapor under vacuum for 24 h to enhance surface compatibility. Post 3D printing on coated coverslips, samples were incubated at 37°C for 30 min to facilitate gelation and crosslinking before confocal microscopy.
Hanging Filament Test
4.5.1
To assess the extrusion fidelity of collagen at 12 and 16 mg/mL, the maximum continuous filament length before filament break during extrusion was quantified. This length, defined as the point where the filament detached from the needle tip due to insufficient mechanical strength or cohesion, was captured on video using a Nikon D810 camera with a Nikon 105 mm Micro lens. The length of the continuous filament extruded from the needle tip was measured using ImageJ (National Institute of Health, Bethesda, MD).
Filament Fusion Test
4.5.2
A filament fusion test was performed to analyze the printability of collagen. Grids with square holes of varying sizes, ranging from 1 mm × 1 mm to 5 mm × 5 mm, were extruded onto a glass slide. The images were captured by a Canon camera and analyzed using ImageJ. The fusion characterization for each collagen concentration was performed by calculating the percentage ratio of the measured perimeter of extruded squares to the theoretical perimeter. A lower perimeter ratio indicates greater fusion or spreading between adjacent filaments, suggesting reduced print fidelity, while a percentage ratio close to 100% indicates minimal fusion and high fidelity.
Filament Collapse Test
4.5.3
A filament was extruded over a linear array of pillars spaced at distances of 1, 2, 3, 4, 5, and 6 mm, as previously described [59]. Photos of the continuous printed filament suspended over the gaps between pillars were taken to evaluate structural integrity. The collapse area factor (*C_f_ *), defined as the percentage of the area deflected or sagged between the pillars relative to the theoretical area that would be occupied if the filament remained perfectly straight, was calculated using Equation (1):
where 𝐴_𝑎_ and 𝐴_𝑡_ represent the actual and theoretical areas, respectively. 𝐴_𝑎_ was taken as zero for regions, where filament collapsed, and *C_f_
- was 100%.
Evaluation of Edge Straightness and Corner Fidelity of Printed Constructs
4.5.4
To investigate how the printing speed affects structural accuracy, such as edge straightness and corner fidelity, 2D square waves were printed at 200, 400, 600, and 800 mm/min. Edge straightness was evaluated by first applying image thresholding to convert filament images into binary format. The edge of each filament was determined by locating the topmost white pixel in each image column. Columns with aligned white pixels were considered indicative of straighter regions. A linear regression line was fitted to these edge points, and the standard deviation from the fitted line was calculated. Lower standard deviation values indicate smoother and more consistent filament edges. Corner fidelity was assessed by fitting a bounding rectangle around each printed corner region. The rectangle area was defined as 100% and the proportion of white pixels (representing extruded bioink) within this area was calculated. This percentage reflects how accurately the printed material filled the intended corner shape. Together, these metrics enable comparison of filament quality across different extrusion speeds. A detailed description of the analysis methods can be found in the Supporting Information.
Collagen Fibers Alignment
4.6
The alignment of collagen fibers achieved via the SLP nozzle was studied both qualitatively and quantitatively using the following microscopy techniques and compared against appropriate control groups.
Confocal Reflectance Microscopy (CRM)
4.6.1
Collagen fibers were visualized using a Zeiss LSM 880 confocal microscope equipped with a 40× water immersion objective in the reflection mode. CRM facilitated the assessment of collagen fiber orientation at 10‐µm intervals throughout the sample's depth. The maximum‐intensity z‐projection was generated and used for fiber alignment analysis.
Scanning Electron Microscopy (SEM)
4.6.2
SEM (Zeiss SIGMA VP‐FESEM) was also used to investigate collagen fiber alignment. 16 mg/mL collagen supplemented with riboflavin was line‐printed, crosslinked using 405 nm light and then incubated at 37°C for 30 min to facilitate gelation. To preserve the structure of the crosslinked filaments, samples were fixed using 2.5% glutaraldehyde prepared in PBS. Following fixation, the samples were dehydrated using a graded ethanol series (25%, 50%, 70%, 90%, and 100%, 10 min each). Upon complete dehydration, samples were sputter coated with iridium using a Leica sputter coater. SEM imaging was performed at an accelerating voltage of 3–5 keV.
Fiber Quantification Metrics
4.6.3
The extent of collagen fiber formation was evaluated at three key stages: before printing, immediately after extrusion, and following incubation. This assessment aimed to determine whether spontaneous fibrillogenesis occurred or if the printing process itself promoted or enhanced fiber assembly.
To quantify the average shear rate across different regions of the SLP nozzle, the average velocity of collagen at the nozzle tip was first determined. For this test, 1 mL of collagen was extruded at a pressure of 40 kPa for a predefined time. Using the known nozzle geometry, the average flow velocity at the nozzle tip was calculated using the equation:
where V is the extruded volume, A is the cross‐sectional area of the nozzle, and t is the extrusion time. Since the flow rate was constant across all nozzle segments, the average flow velocity in each segment was calculated by dividing the flow rate with its respective cross‐sectional area. To compute the average shear rate (γ) within the SLP nozzle, we employed a spatial gradient approach based on average velocity differences between adjacent segments. The average shear rate was calculated using Equation (3):
where Δ𝑣 represents the change in average velocity between two neighboring segments, and Δ𝑦 denotes the vertical distance (mm).
Anisotropy of collagen fibers was determined using FibrilTool, an ImageJ plug‐in. The analysis was performed within defined regions of interest (ROIs) as previously described [60]. The degree of anisotropy was reported on a scale from 0 to 1, where 0 indicates completely isotropic fibers (no alignment) and 1 represents perfectly anisotropic fibers (highly aligned) [60]. For each image, anisotropy was calculated from an average of six ROIs. All alignment measurements were conducted on maximum intensity z‐projections generated from 20 µm‐thick z‐stacks acquired using CRM. The angle range with the highest number of fibers (i.e., the mode) was identified, and fibers oriented within ±15° of this dominant angle were selected. The number of these aligned fibers was then divided by the total number of detected fibers to calculate the CoA. CoA value 1 indicated perfect alignment, whereas 0 represented complete randomness. Collagen fiber quantification was carried out using CT‐FIRE and CurveAlign, open‐source software tools designed for automated segmentation and analysis of individual collagen fibers [61]. Specifically, CurveAlign was used to evaluate distribution of fiber orientation within each image, while CT‐FIRE provided measurements of individual fiber properties, including length, width, angle, and straightness.
Computational Modeling and Simulation
4.7
Rheological Model Fitting (Herschel–Bulkley)
4.7.1
The shear stress and shear rate data obtained from the rheological modeling was used to fit HB model in MATLAB (MathWorks, Natick, MA, USA) using least square error minimization technique available with the curve‐fitting add‐on in MATLAB. The following is the expression for shear stress ‘τ’ of HB fluid as a function of shear rate (γ˙):
where ‘τ_ y ’ is the yield‐stress, ‘K’ is the consistency index and ‘n’ is the power law index of the HB fluid. From the curve fitting, we got 3.8164 [Pa], 27.5543 [Pa.s^n^], and 0.2591 for the τ y _, K, and n, respectively.
The apparent dynamic viscosity ‘μ’ of the HB fluid is expressed as:
COMSOL Simulation of Microchannel Flow
4.7.2
A 2D finite element model of the microchannel array was studied using COMSOL Multiphysics (COMSOL Inc., version 6.2). The geometries of the microchannel array were designed in COMSOL Multiphysics using the built‐in geometry module and the pressure boundary conditions (inlet = 40 kPa, outlet = 0 kPa) were applied on the inlet and outlet of the nozzle. We assumed a laminar and steady‐state operation of the bioprinting process for the gel; therefore, we used stationary laminar flow study condition. A regularized version of the HB fluid model was used for numerical stability [62]. The numerical study was performed by neglecting the influence of the collagen fiber on the bulk gel; thus, providing a pathway of considering a one‐way coupling, i.e., the transfer of load from the gel to collagen fibers. The following are the balance laws and the regularized constitutive model for collagen:
where ‘u−’ is the fluid velocity vector, ‘ρu−·∇u− ’ is the convective acceleration, ‘∇p’ is the pressure gradient, ‘ ∇u−’ is the velocity gradient, ‘∇ · ’ is the divergence operator, ‘m’ is the viscosity regularization parameter, ‘∇u−s’ is the symmetric part of the velocity gradient tensor, ‘: ’ is the tensor inner product and ‘eps’ is the machine epsilon added for avoiding numerical singularities.
Particle (Fiber) Orientation Simulation
4.7.3
To demonstrate the effect of channel geometry on local shear rate and particle alignment, fluid velocity data from COMSOL Multiphysics was imported into MATLAB. The simulation was focused on the two terminal channel regions (Ø = 2 mm and Ø = 0.34 mm) with high shear rates. A set of several hundred line‐particles, each 50 µm in length, was randomly initialized in position and orientation within the Ø = 2 mm region using a pseudo‐random number generator. The fluid velocity was applied to the endpoints of each particle and displacements were computed using Euler's time‐stepping method. To eliminate numerical errors due to length distortion during simulation, a constraint was imposed to preserve constant particle length. This was achieved by rescaling endpoint positions relative to the particle midpoint while maintaining the original orientation angle. At sufficiently small‐time steps, the results of the constrained and unconstrained models were equivalent. Particle positions and orientation angles were recorded at each timestep to evaluate alignment behavior under varying shear conditions.
Cell Alignment Studies
4.8
Cell lines used in this study include NHLFs, ^GFP+^MDA‐MB‐231 for alignment characterization within collagen constructs, morphology tracking via time lapse imaging, respectively. NHLFs and ^GFP+^MDA‐MB‐231 were cultured in DMEM supplemented with 10% FBS and 1% PS.
Cell Seeding, Viability, Proliferation, and Alignment Characterization
4.8.1
NHLFs were trypsinized using TrypLE for 3–5 min and centrifuged at 200 × g for 5 min and used for cell proliferation assay and alignment characterization. Trypsinized cells were mixed into the bioink at a concentration of 1 × 10^6^ cells/mL. To evaluate cell proliferation, an AlamarBlue dye reduction assay was performed, which indicates metabolically active cells. A 10% (v/v) AlamarBlue dye solution was prepared in culture medium, and cells were incubated in this medium for 3 h. After incubation, 100 µL of the solution was transferred to 96‐well flat‐bottom plates, and absorbance was measured using a microplate reader (Tecan Infinite 200 Pro, Switzerland) at 570/600 nm (excitation/emission). The results obtained were normalized with respect to those at Day 0. Post‐exposure cell viability was evaluated using the LIVE/DEAD Viability/Cytotoxicity Kit (Invitrogen, Thermo Fisher Scientific) at 24 h after bioprinting, which utilizes Calcein‐AM to label viable cells and Ethidium Homodimer‐1 (EthD‐1) to label dead cells. Samples were first washed with PBS and incubated with the staining solution prepared according to the manufacturer's protocol. Briefly, Calcein‐AM (final concentration 2 µM) and EthD‐1 (final concentration 4 µM) were diluted in PBS and applied to constructs to fully cover the surface. Samples were incubated at 37°C for 20–30 min in dark. After incubation, samples were gently rinsed with PBS and imaged immediately using a Zeiss Axio Observer microscope (Figure S11). For alignment characterization, NHLFs were seeded onto three types of collagen constructs: DC collagen, SC collagen fabricated using a traditional straight nozzle mounted on a 3‐cc straight barrel, and SLP collagen constructs. A total of 250 000 cells were seeded and cultured under standard conditions (37°C, 5% CO_2_) for 7 days. Following the culture period, ^GFP+^MDA‐MB‐231 cellular alignment was assessed using color‐mapped alignment detection and frequency distribution analysis. To quantitatively assess cell behavior, time‐lapse videos were acquired using the Zeiss Axio Observer microscope. Cell migration and orientation were analyzed using an AI‐based tracking pipeline. We used a fast YOLOv8 model to detect cells in each frame of time‐lapse videos, generating a bounding mask per cell. Temporal tracking was achieved by matching each detected cell with its nearest neighbor in the previous frame based on Euclidean distance, with tolerance for brief detection gaps to maintain trajectory continuity. To assess orientation, an ellipse was fitted to each cell mask, and the angle of its major axis was recorded per frame. The mean migration angle was computed for each cell to represent its average directional trajectory. Additionally, the initial and final orientation angles were calculated by averaging the angles from the first and last five frames, respectively, to quantify directional shifts over time. A cell's position (x and y coordinates) was obtained by tracking its centroid coordinates across all frames. Start and end coordinates were computed by averaging the first and last five centroids, respectively, and displacements in x and y axes were calculated. To minimize noise and improve tracking reliability, short‐duration tracks‐defined as trajectories formed by cells that appeared in only a few frames‐were excluded from analysis. The directionality persistence was calculated as the ratio of net displacement to the total path length, providing a measure of trajectory linearity along with x and y direction. Since fibers were oriented vertically, directionality persistence along the y‐axis was further determined to assess alignment‐driven migration. Furthermore, lateral displacement along the x‐axis (ΔX) was calculated to quantify the extent of lateral movement, providing an additional measure of cell deviation from the aligned y‐axis trajectory.
Corneal Tissue Engineering
4.9
Corneal Cell Isolation for Bioprinting
4.9.1
Corneal tissue was obtained from corneas deemed unsuitable for transplantation at the Gift of Life Eye Bank in Hershey, PA. The corneal cells were isolated according to a previously established protocol [63]. Briefly, corneas were washed with DPBS supplemented with 1% (v/v) PS. For the collection of cells, cornea was transferred to an expansion medium containing collagenase A and digested at 1 mg/mL for 2–4 h at 37°C. The sample was then centrifuged, and the cell pellet was resuspended in complete expansion medium as described before [63] (Figure S12). Cells were plated onto well precoated with FNC coating mix (fibronectin/collagen). Isolated human primary corneal cells used in corneal construct fabrication were maintained in Opti‐MEM supplemented with 8% FBS, 5 ng/mL bFGF, and 20 µg/mL ascorbic acid, and the medium was changed every 2–3 days. Once confluence was reached, the culture was switched to maturation medium for 7–28 days. The morphology and viability of all isolated cell types was checked under the microscope. Additionally, the isolation of pure cell population was confirmed using flow cytometry for the cell type (Figure S13). Primary human corneal cells were isolated and prepared as a single‐cell suspension in fluorescence‐activated cell sorting (FACS) buffer consisting of PBS, 25 mm 4‐(2‐hydroxyethyl)‐1‐piperazineethanesulfonic acid (HEPES) and 1–5 mm Ethylenediaminetetraacetic acid (EDTA). Cells were stained with CD166/ALCAM (FITC) monoclonal antibody (1:100 dilution), Anti‐EpCAM (Alexa Fluor 647) monoclonal antibody (1:100 dilution) for 30 min at 4°C, then washed and filtered through a 40‐µm cell strainer. CD166‐positive and CD326‐positive populations were sorted using a Bigfoot BSL2 cell sorter (Thermo Fisher Scientific), gated on single cells, and collected into tubes containing culture medium. The cells below passage 3 were used for all experiments. For bioprinting of corneal tissue construct, a bioink composed of 16 mg/mL collagen, riboflavin, and primary human corneal cells were prepared and bioprinted.
Transparency Measurement
4.9.2
Transparency of collagen for corneal constructs was assessed by measuring visible‐light transmittance (400–700 nm) on a spectrophotometer. Thin collagen gels (100 µL, pH 6.5) with riboflavin (2:100, v/v) and thick collagen gels (200 µL, pH 6.5) with the same riboflavin concentration were cast into 24‐well plates. Native corneal tissue was included as a reference, and PBS was used as background control to represent 100% transparency. Spectral absorbance was recorded immediately after gelation (Day 0) and again on Days 1, 3 and 7 post‐incubation to track changes in transparency over time.
Cartilage Tissue Engineering
4.10
For the cartilage tissue engineering application, ADSCs were used at passage 2–4 for chondrocyte differentiation. They were cultured in flasks until they reached 75% confluence in DMEM/F12 medium supplemented with 20% FBS and 1% PS. At the end of 3 weeks, chondrocyte differentiation was initiated using an all‐in‐one chondrocyte differentiation medium, which was refreshed every two days. To confirm differentiation, cells were immunostained with aggrecan monoclonal antibody (BC‐3) and anti‐collagen II antibody. Chondrocytes were then trypsinized and centrifuged at 200 × g for 5 min. Next, chondrocytes were suspended in a 16% collagen bioink. A bioink containing 16 mg/mL collagen, riboflavin, and chondrocytes were used to print multi‐zonal constructs. Following 3D bioprinting, all constructs were incubated at 37°C for 30 min to facilitate gelation and crosslinking, after which culture medium was added.
Histological Analysis
4.10.1
Constructs were fixed in 4% paraformaldehyde, cryoprotected in sucrose, embedded in OCT, and cryosectioned at 15 µm thickness using a cryostat (Leica CM1950, Leica Biosystems, Wetzlar, Germany). Transverse sections (collected from the bottom region of the constructs) were used for histological analyses. For morphological evaluation, H&E staining was performed using an automated slide stainer (Leica Autostainer ST5010 XL, Leica Biosystems, Wetzlar, Germany). For assessment of collagen distribution, Masson's Trichrome staining was performed using a commercial MTS kit (Sigma‐Aldrich, cat. No. HT15‐1KT) according to the manufacturer's protocol. To visualize GAGs and nuclei, sections were stained with Alcian Blue/Nuclear Fast Red by hydrating the cryosections, incubating them in 1% Alcian Blue (pH 2.5) for 30 min, rinsing thoroughly, and counterstaining with Nuclear Fast Red for 5 min. Cartilaginous matrix components were further evaluated using hematoxylin/Safranin O/Fast Green. After hydration, sections were stained with hematoxylin, incubated in Fast Green to label non‐cartilaginous proteins, differentiated in 1% acetic acid, and finally stained with Safranin O to highlight GAG‐rich regions. Following all staining procedures, sections were mounted using a xylene‐compatible mounting medium. 20× images were acquired using an Olympus BX61 microscope (Olympus Corporation, Tokyo, Japan).
DNA Quantification
4.10.2
For bioprinted cartilage constructs, the total DNA content was measured using the Quant‐iT PicoGreen dsDNA Assay Kit (Invitrogen, Thermo Fisher Scientific) following the manufacturer's instructions. The digested constructs were mixed with PicoGreen reagent and fluorescence was measured at an excitation/emission of 480/520 nm using a microplate reader. DNA concentrations were calculated against a standard curve generated with λ‐DNA and were normalized either to the total hydrogel mass or total DNA content of constructs.
Glycosaminoglycan (GAG) Quantification
4.10.3
The total sulfated GAG content was measured using a 1,9‐dimethylmethylene blue (DMMB) assay. Constructs were first digested using papain enzyme (e.g., 125 µg/mL papain in 0.1 m sodium acetate, 5 mm EDTA, 5 mm cysteine‐HCl, pH 6.0) at 60°C for 18 h. The resulting digests were reacted with DMMB dye, and absorbance was measured at 525 nm. GAG concentration was calculated using a chondroitin sulfate standard curve and values were normalized to the total DNA content and weight of the sample.
Collagen Quantification
4.10.4
Collagen content was assessed using the Sirius Red assay, which selectively stains hydroxyproline‐containing proteins. Prior to staining, samples were digested using pepsin in 0.5 m acetic acid at 4°C for 48 h to solubilize the collagen. Sirius Red dye solution was added to the pepsin digests and incubated for 1 h with gentle shaking. After washing with 0.01 N HCl to remove unbound dye, the bound dye was eluted using 0.1 m NaOH and absorbance was measured at 540 nm. Collagen concentration was calculated using a type I collagen standard curve and normalized to the total DNA content and weight.
Mechanical Testing
4.10.5
Compression tests were performed to study the mechanical properties of cartilage constructs after 7 days of culture in cell culture medium at 37°C. Unconfined compression tests were conducted using a rheometer (MCR 302, Anton Paar) equipped with a parallel‐plate geometry. For each sample, a small preload was applied until full contact between the top plate and the sample surface was achieved, followed by a brief relaxation period to eliminate pre‐stress. Samples were then compressed at a rate of 0.001 mm s^−^ ^1^ until reaching a maximum strain of 10%. The compressive modulus was determined from the slope of the initial linear region of the stress‐strain curve.
Immunofluorescence Staining and Imaging
4.11
For alignment analysis, NHLFs were fixed in 4% paraformaldehyde after 3 h of seeding, permeabilized using Triton X‐100 for 15 min, and blocked with 4% BSA in PBS. Cells were washed with PBST after each step. Cells were labeled with phalloidin conjugated to AlexaFluor568 for 30 min (1:400 dilution), followed by Hoechst 33342 (1:500 dilution) for 10 min to stain actin fibers and cell nuclei, respectively. Lastly, cells were washed with PBST for 15 min, followed by washing with 1× PBS for 5 min before imaging. Imaging was carried out using a Zeiss LSM 880 confocal microscope (Thornwood, NY, USA) using a 10× objective.
Bioprinted cartilage and corneal constructs were embedded in OCT (Fisher HealthCare) and cryosectioned at 10 µm thickness using a Leica CM1950 cryostat. Cartilage sections were stained with an ACAN antibody and counterstained with DAPI to visualize cell nuclei. Corneal sections were stained with phalloidin for F‐actin, DAPI for nuclei, and anti‐CD166 antibody to identify epithelial cells. Confocal imaging of all cryosections was performed using a Zeiss LSM 880 confocal microscope. Additionally, whole bioprinted corneal constructs were stained with phalloidin and DAPI and imaged using a Leica SP8 DIVE microscope (Leica Microsystems, Wetzlar, Germany). Maximum intensity projections were generated, and image channels were merged using ImageJ.
Statistical Analysis
4.12
One‐way analysis of variance (ANOVA) followed by Tukey's post hoc test was used to compare more than two experimental groups (anisotropy, proliferation analysis for DC, SC, and SLP). Unpaired Student's t‐test was applied for comparisons between two independent groups (printability analysis for 12 and 16 mg/mL and biochemical analysis for cartilage tissue fabrication for Days 1 and 7). Levene's test was used to evaluate the equality of variances among groups (AI analysis for cell alignment for DC, SC, SLP). A p‐value of <0.05 was considered statistically significant (^^), while p < 0.01 (^^) and p < 0.001 (^^) indicated increasing levels of significance.
Ethics Statement
Human corneal tissues were obtained from the Gift of Life Donor Program Eye Bank under Institutional Review Board (IRB) approval and in accordance with the Declaration of Helsinki and institutional guidelines for research involving human‐derived materials. The tissues were provided as de‐identified, non‐transplantable donor specimens for research purposes, and therefore qualified for IRB exemption at the Pennsylvania State University.
Conflicts of Interest
I.T.O. has an equity stake in Biolife4D and is a member of the scientific advisory board for Biolife 4D and Healshape. The remaining authors declare no competing interests.
Code Availability
The developed SLP nozzle in this study was fabricated using a custom 3D printing code. Cell alignment was characterized using an AI‐based YOLOv8 tracking framework. In addition, filament printing fidelity was evaluated using custom image analysis scripts for corner detection, white pixel quantification, and edge straightness detection. All the codes are publicly available at https://github.com/ilaydanamli/SLP.
Supporting information
Supporting File 1: smll72261‐sup‐0001‐SuppMat.docx
Supporting File 2: smll72261‐sup‐0002‐MovieS1.mp4
Supporting File 3: smll72261‐sup‐0003‐MovieS2.mov
Supporting File 4: smll72261‐sup‐0004‐MovieS3.mp4
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1R. M. Crossley , S. Johnson , E. Tsingos , et al., “Modeling the Extracellular Matrix in Cell Migration and Morphogenesis: A Guide for the Curious Biologist,” Frontiers in Cell and Developmental Biology 12 (2024): 1354132, 10.3389/fcell.2024.1354132.38495620 PMC 10940354 · doi ↗ · pubmed ↗
- 2O. Chaudhuri , J. Cooper‐White , P. A. Janmey , D. J. Mooney , and V. B. Shenoy , “Effects of Extracellular Matrix Viscoelasticity on Cellular Behaviour,” Nature 584 (2020): 535–546, 10.1038/s 41586-020-2612-2.32848221 PMC 7676152 · doi ↗ · pubmed ↗
- 3T. R. Cox and J. T. Erler , “Remodeling and Homeostasis of the Extracellular Matrix: Implications for Fibrotic Diseases and Cancer,” Disease Models and Mechanisms 4 (2011): 165–178, 10.1242/dmm.004077.21324931 PMC 3046088 · doi ↗ · pubmed ↗
- 4N. Matsumoto , S. Horibe , N. Nakamura , T. Senda , K. Shino , and T. Ochi , “Effect of Alignment of the Transplanted Graft Extracellular Matrix on Cellular Repopulation and Newly Synthesized Collagen,” Archives of Orthopaedic and Trauma Surgery 117 (1998): 215–221, 10.1007/s 004020050232.9581247 · doi ↗ · pubmed ↗
- 5S. Mc Dougall , J. Dallon , J. Sherratt , and P. Maini , “Fibroblast Migration and Collagen Deposition During Dermal Wound Healing: Mathematical Modelling and Clinical Implications,” Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 364 (2006): 1385–1405, 10.1098/rsta.2006.1773.16766351 · doi ↗ · pubmed ↗
- 6S. Guido and R. T. Tranquillo , “A Methodology for the Systematic and Quantitative Study of Cell Contact Guidance in Oriented Collagen Gels Correlation of Fibroblast Orientation and Gel Birefringence,” Journal of Cell Science 105 (1993): 317–331, 10.1242/jcs.105.2.317.8408268 · doi ↗ · pubmed ↗
- 7C. Guo and L. J. Kaufman , “Flow and Magnetic Field Induced Collagen Alignment,” Biomaterials 28 (2007): 1105–1114, 10.1016/j.biomaterials.2006.10.010.17112582 · doi ↗ · pubmed ↗
- 8N. Y. Patrawalla , R. Raj , V. Nazar , and V. Kishore , “Magnetic Alignment of Collagen: Principles, Methods, Applications, and Fiber Alignment Analyses,” Tissue Engineering Part B: Reviews 30 (2024): 405–422, 10.1089/ten.teb.2023.0222.38019048 PMC 11404687 · doi ↗ · pubmed ↗