In Situ Fabrication of Constraints for Multicellular Micro‐Spheroids Using Two‐Photon Lithography

2‐photon polymerization is a promising technology for creating complex, microscale 3D matrices for biomedical and also bioprinting applications. Cancer research provides compelling uses for this strategy, in particular, for generating a 3D constraint around multicellular spheroids. Because these spheroids are inhomogeneous in size and shape, the ability to target a spheroid composed of a few living cells requires geometrical control of the printing shape in situ. In this study, it is presented that two‐photon lithography can be used to study complex phenomena involved in cancer progression, such as collective 3D cell migration in situ in vitro. This method allows the spatial and temporal control of cancer cell migration from single spheroids, using dome‐shaped confinements with micrometer‐sized openings. The confinement of the spheroids leads to a decreased migration speed and affects actin dynamics. Furthermore, this methodology provides a novel way of analyzing the behavior of specific regions of multicellular structures, by enabling the separation of multicellular structures, while keeping them alive. Ultimately, this study demonstrates a new way to use two‐photon lithography for controlling the growth, migration and morphological cues of live cells, thus opening new avenues toward the dynamic in situ control of living 3D structures.


Introduction
3D additive manufacturing strategies are among the most prominent technologies for microfabricating polymer-based scaffolds, with high accuracy and defined morphologies. [1]Among these strategies, 2-photon polymerization (2PP) utilizes femtosecond lasers to polymerize liquid polymers, enabling DOI: 10.1002/adfm.202302356 the creation of precise structures at microand even nano-scale. [2]As many cell types interact with three-dimensional environments, 2PP is a promising technique for creating complex 3D matrices to control the cell growth, migration and spreading in 3D milieus, with unprecedented precision. [1]ne of the recent advances in the application of 2PP is the use of proteins as photoresists. [3]This allows the generation of tissue-like matrices, resembling most relevant features found in the extracellularmatrix (ECM). [4]Due to its broad spectrum of applications, including bioprinting, a widely used protein-based photoresist is gelatin. [5]Since 2PP enables the performance of special features at temporal and spatial scale, this can be used to control and direct several biophysical parameters characterizing the cell behavior, such as their collective migration within and into/out of confined spaces. [1,6]Conventional techniques to study collective migration of cells in 3D milieus have been carried out by the entrapment of 3D cellular aggregates, known as spheroids, in soft hydrogels. [7]owever, these matrices are characterized by their isotropy, and therefore, they are incapable of directing the migration and growth of cells in a selected plane.Alternative strategies to guide the collective migration of cells include the use of microfluidic devices [8] and stimulation by laser ablation. [9]These methods, however, do not permit to accurately control the behavior of cellular aggregates embedded in a 3D hydrogel-based matrix, or even within a specific section of the microspheroid, guided by mechanical or metabolic restrictions (i.e., the presence of a physical barrier or mediated by micro-gradients of oxygen or nutrients respectively).
Since multicellular aggregates are widely analyzed in the field of cancer research, we seek the ability to develop new tools for controlling the morpho-mechanical fate of cells found in 3D milieus during the different stages of the malignant progression (i.e., invasion, extravasation, and early metastasis).This can be addressed by constraining multicellular structures in defined spaces, designed and performed in situ, and on-demand.To progress toward this goal, we present here a strategy to print a 3D polymer-based cage around single microspheroids, as well as isolating parts of it, by 2PP.The purpose of using this strategy is to control cell migration from the confined milieu, while A femtosecond laser with center wavelength at 780 nm selectively polymerizes a photosensitive material to create intricate structures with sub-micron resolution.B-D) 3D CAD model of a designed dome structure, having a total diameter of 260 μm, a wall thickness of 50 μm, four 80 μm wide openings on the side and a 80 μm wide opening on the top.For visualization purposes half of the structure is presented semi-transparent D).Brightfield image of a spheroid E) before printing and F) directly after printing.Scale bar 50 μm.G) Schematic representation showing how to constrain a spheroid with 2PP in situ.The time typically needed to constrain the spheroid is 6 min.
simultaneously studying the impact of 3D confinement on cell movement.The presented method enables us to engineer complex 3D scaffolds around multicellular populations constituted of a very low amount of cells (i.e., having less than 200 cells), while maintaining their morphological organization and internal cellcell communication.

Results and Discussion
Initially, a dome-like 3D structure was printed around one single cancer micro-spheroid, to demonstrate the viability of 2PP for in situ printing around a living multicellular system.Such microspheroids are made up of few cells (normally less than 10.000 cells).They are generated by culturing cells on non-adherent surfaces, where the cell-cell binding is enhanced due to the reduction of cell-matrix anchorages.Since they exhibit morphomechanical and metabolic similarities to the environment found in primary tumors, these 3D cellular constructs have been widely used in cancer research.In fact, these tumor models have gained relevance for studying the role of the cellular morphology in drug sensitivity, as well as upregulation of hallmarks of cancer pathogenicity.However, we, among others, have shown that cancer cells need the existence of confinement and mechanical stress to upregulate the markers associated with invasion and metastasis. [10]Therefore, in this work we decided to constrain single micro-spheroids within a dome-like structure.These 3D structures contain openings for permitting cell migration in desired directions, thus providing a miniaturized model for cell invasion.The printed structure consists of a hemisphere with a wall thickness of 50 μm and an outer diameter of 260 μm.These scaffolds constrain spheroids consisting of less than 200 cells, and can be adapted according to the spheroid size.The dome has four openings on its periphery (80 μm in diameter) and one on its top (Figure 1A-D).These apertures allow the analysis of cell migration in 3D.The dimensions of the side openings have been chosen to generate a gradient of nutrients and oxygen, while keeping the three-dimensional morphology and the cell density of the multicellular aggregate constant.These conditions may resemble the micro-environmental conditions sensed by entrapped cancer cells in vivo, in which a dense and packed group of cells are surrounded by an external envelope of crosslinked and low degradable extracellular matrix proteins. [11]o entrap the cell aggregates, MCF7 breast cancer spheroids were placed on a glass coverslip and incubated overnight to induce their attachment (Figure 1E).It is important to mention that these spheroids are constituted by less than 200 cells (data not shown), having a diameter of ≈150-200 μm.This enables their anchorage on the glass surface, facilitating the printing process (Figure 1G).A biocompatible and biodegradable gelatinbased photoresist was utilized in this study.The cured photoresist has a storage modulus ranging from 4 to 26 kPa according to the manufacturer's product information sheet (Bioinx, 2023), in which the stiffness can be varied depending on print parameters, e.g., laser power, print speed, and hatching and slicing distance.This procedure, as well as cell monitoring, can be done via bright field imaging.A print process in this case requires ≈6 min, as demonstrated in a timelapse video available in Supporting Information (Video S2, Supporting Information).Using a 25x objective lens, the print field covers a circular area of ≈200 μm radius.This limitation affects the available printing coverage and the number of structures that can be printed together.Despite this, each printing job takes only 6 min, allowing for efficient printing of multiple cages, without compromising cell aggregate integrity.By careful planning and optimization, it is possible to print a few cages in a single run, while maintaining the viability and activity of the multicellular spheroid.This gives the potential to directly adjust the printing design toward the spheroid of interest, which is relevant when working with inhomogeneous 3D cell constructs constituted by cells with different proliferation rates, morphologies, and even cell sizes. [12]Figure 1E,F show the spheroid before and directly after the printing process, demonstrating the accuracy of the process.It is worth noting that 2PP can enable the encapsulation of floating organoids. [13]However, this approach reduces the probability of finding a particular spheroid in a desired location, so a larger number of spheroids are required before printing.As a result, it becomes challenging to control this process.The difficulty of precisely positioning a freely floating spheroid during printing has also been discussed in the literature. [13]n important prerequisite for the use of 2PP in the presence of living cells, is their viability.Figure 2A shows that the MCF7 located on the surface of the spheroid and close to the printed dome are highly viable, exhibiting neither damage by the printing process, nor due to the presence of the photoinitiator (Bioinx) in the medium.Although dead cells can be detected in the center of the multicellular aggregate, we assume that these so-called necrotic zones were not caused by the 3D printing process itself, but rather by factors such as limited diffusion of oxygen and nutrients, as shown in Figure S1 (Supporting Information), and as previously observed by other research groups. [14]oth the constrained aggregates and the control spheroids were monitored daily, up to five days.Figures 2B-G demonstrate that constrained cells spreaded out from the confined space through the openings of the dome structure.This phenomenon was already detected after one day post-printing (Please check the video showing 3D reconstruction of day 0 and day 5 in the Video S1 (Supporting Information).Additionally, Figure S2 (Supporting Information) includes cross-sections and reconstructed 3D images.It is important to mention that no penetration of migrating cancer cells into the bulk of the hydrogel-based structure was observed during the experimental time, indicating that cells were incapable to enzymatically or mechanically penetrate the polymer-based matrix, as already reported. [15]To improve our understanding of the effect of the confinement on the spheroid behavior, the convexity ratio was analyzed.Convexity of an object is defined as the ratio of its convex closure to the perimeter of the object itself, and denotes how it differs from a convex shape (i.e., the more irregular the boundary of an object is, the lower its convexity).This parameter gives information about the morphological changes detected on the migrating front of a specific multicellular tissue.This is assessed by measuring the perimeter obtained from the fluorescent images of the cells (i.e.stained with Calcein AM).
When performing this analysis on control experiments (i.e., the unconstrained condition), no changes in the convexity ratio was detected during the experimental time, showing that unconfined cells grow and migrate without any directional preference, as they maintain their convexity.In contrast, we observe that there is a significant decrease of the convexity ratio in the spheroids confined by the dome structures (Figure 2I).This can be explained by the fact that they have no pathway to grow out of the confinement other than conforming to the longitudinal shape to spread out of the opening.Hence the multicellular aggregate took on a geometrically irregular shape.The formation of this non-convex tissue shape is also evident in the data showing the perimeter to area ratio, which increases significantly in the samples constricted by the printed structures, suggesting that the boundaries of the caged spheroids are growing at a faster rate than the area they occupy.However, no significant changes can be seen in the control experiment.This phenomenon resembles fluids, in which the propagation is defined by surface minimization. [16]These models are in agreement with our data, as the control does not show a decrease in convexity ratio, whereas the cell spheroid growing in the dome structure had to overcome the tendency of surface minimization, to grow through the dome openings.
To further explore the potential application of the 2PP method for fabricating constraints for spheroids, larger spheroids were anchored onto a glass coverslip, so that a cage could be printed on the aggregate's edge (See Figure 3).This strategy enables isolation of specific cell clusters localized inside a morphologically heterogeneous spheroid.It allows an accurate control of populations exhibiting a specific hallmark, such as increasing proliferation rates, cell-surface markers, or expression of relevant proteins involved in the matrix degradation, to name a few.In these assays, it was observed that cells located within the printed hydrogel matrix died (Figure 3A,B).This observation suggests that the photolithographic process, namely the combination of ink and laser exposure, might be toxic for cells located in exposed positions.In order to probe this hypothesis, experiments were carried out to show that neither the ink nor the laser alone induced cell mortality (Figure S1A-D, Supporting Information), (Figure S3, Supporting Information, presents the relative quantitative data).These observations were also confirmed by the detection of caspase proteins (i.e., Caspase 3/7), involved in programmed cell death by apoptosis.As shown in Figure 3, Caspase 3/7 are mostly localized in the core of the spheroids, rather than on their surface (Figure S1E,F, Supporting Information), or even on the zone where the printing process is taking place.Since the expression of Caspase 3/7 requires several minutes, and it is mostly mediated by metabolic factors (lack of oxygen and/or nutrients exchange between the spheroid surface and its core), our results suggest that the cells die immediately after contacting the laser and the photoresist, permitting no time to upregulate the expression of programmed cell death markers in printed areas, in comparison to the zones the were not directly impacted by the 2PP process.
The cells within the encaged area and outside of the structure are viable, and continue to grow.though the shape of the spheroid is deformed.Hence, the 2PP method can actively intervene in the growth of the spheroid, opening up an avenue toward interactively studying the impact of well-defined physical barriers on the function of multicellular systems.Furthermore, it also allows the isolation of sections of a multicellular system, e.g. for detailed studies on specific cells from the edge of a spheroid, or potentially also of organoids.
Despite the fact that 2PP allows for the creation of highresolution structures, the geometry of the hydrogel-based scaffolds may be altered by their ability to absorb a significant amount of water. [17]Understanding this effect is crucial for designing polymer-based matrices with optimal dimensions, to better tune the cell migration.In addition, the dimension of channels plays a very important role in cell migration,due to the restrictions imposed by the size and the compactability of the nuclei. [18]Therefore, after establishing the 2PP method, the swelling behavior of the printed hydrogel was investigated.To this purpose, several dome structures having apertures with different diameters (80, 40, and 20 μm) (see Figure 4A inset) were printed.The pillar structure was chosen for the swelling test because of its flat top.This shape allows observation of the increase in projected area as a result of swelling, by using a stereomicroscope.We observed a 1.6-fold increase of the projected area of the printed scaffold after immersion in cell culture media (Figure 4A).Surprisingly, the hydrogel then gradually shrunk and stabilized after 1 day, re-sulting in a projected area increase of ≈1.28-fold, compared to the print design (i.e., CAD model).Our results agree with previous publications in the field, that measure the increase of a structure's projected area as a result of the uptake of water after printing. [17]Swelling behavior can also be influenced by changes in laser power and speed. [19]o see the effect of swelling on the dimensions of the openings of the domes, printed shapes were incubated in cell culture medium at 37 °C for one day, to reach equilibrium.These assays were carried out in absence of cells.To characterize the structures of the printed material after swelling, the medium was fluorescently labeled with FITC-dextran (2000 kDa).This strategy was required, because the autofluorescent signal of the photoresist is very weak, making detailed measurements of the geometrical changes difficult to detect in the hydrogel-based scaffold.As Figure 4B shows, the cross-section area of 12 measured openings decreased significantly, compared to the theoretical area of the CAD file (2513 μm 2 ).Regarding the design of smaller openings (20 μm in diameter), the swelling led to a complete closure of the designed openings (data not shown).This is also evident in experiments where spheroids were confined in domes with 40 μm-diameter openings, showing that the cells were unable to escape from their confinement (Figure 4D).18a] Furthermore, the deformation of the hydrogels allowed the cells to expand the channels, similar to how tumor cells might invade dense tissues. [20]In this work, it was not observed that the cells migrated through the openings closed by swelling, suggesting that the cells were unable to deform the hydrogel, or to compromise their nuclei to the extent necessary for migration.These results can be related to the nature of the cells utilized in this experiment, which generally exhibit more collective behavior in comparison to more aggressive cell types (i.e., MDA-MB-231 breast cancer cells) that tend to act and migrate as individuals, instead of as clusters. [10]t is also relevant to remark that the swelling and shrinking of the structure also affects the shape of the openings, as it can be seen in Figure 4C.As already reported, the channel geometry affects cell behavior, like cell migration, orientation or differentiation, [21] and therefore, it is essential to monitor the changes in the shape of the apertures that result from the hydrogel swelling.As Figure 4B,C show, the cross-sectional profiles appear dissimilar to the modeled shape, exhibiting uneven swelling patterns along the aperture.This may be attributed to localized and anisotropic properties of the hydrogel, induced by varying laser exposure during the printing process. [22]o show the potential of the presented method, we further investigated how the printed structures influence the cellular orientation, as well as the intracellular arrangement of actin stress fibers.To achieve this, we observed the actin dynamics of constrained and unconstrained spheroids by live cell imaging over a timeframe of 6 h using SiR-actin and SPY505-DNA (Figure 5).Here we observe relatively high actin alignment at the interface of the spheroid and the printed structure.On the other hand, the actin alignment in the openings seemed to be much more dynamic, demonstrating sudden changes, and implying a stronger role of biomechanical cues, such as surface tension or tensile forces on reorganizing stress fibers over time.In the control, the dynamic rearrangement of actin is occurring at a lower and more constant rate.It is highly interesting that our findings disagree with some studies investigating cell alignment in tubular confinement.According to these experiments, actin fibers orient themselves along the long axis of channels, during the cell migration in tubes of diameters ranging from 25 to 100 μm, which is a comparable length scale to the presented structure. [23]The reason for this difference could be that, in the presented dome shape, the cells are attached on one side to the flat glass surface, experiencing the curvature only from the top side by the openings.In the work by Xi. et al. the cells are localized in a tube, allowing the cells to sense from all sides a concave curvature in the circumferential direction of the tube and zero curvature only in the longitudinal direction. [23]The ability of the cells to sense the curvature could explain this discrepancy.As demonstrated by this analysis, the method we provide to control 3D multicellular systems in situ using 3D structures, provides novel opportunities for studies on multicellular migration. [24]n this work, we have used 2PP to generate 3D domes in situ, and in this way confined 3D multicellular cancer aggregates .This approach allows for the systematic analysis of cell invasion , from miniaturized spheroids into their surroundings, which is highly relevant to understand the metastatic potential of solid tumors. [25]Due to the tunability of printing procedure, in principle also parameters can be controlled that are highly relevant in tumors, such as mechanical stress, or hypoxia. [10]Additionally, our 3D printing strategy offers flexibility, allowing for easy adjustment of various morpho-mechanical parameters of the scaffolds, at the single-organoid scale.This represents a significant advancement in the field, as it addresses a previously unexplored aspect of biofabrication techniques.Finally, the versatility of this method is highlighted by its ability to accommodate various types and quantities of cells for embedding in a3D matrix.This suggests numerous potential biomedical applications, such as employing primary cancer cells or co-culturing them with immune cells, as well as confining organoids, among other possibilities.Additionally, our 3D printing strategy offers flexibility, allowing for easy adjustment of the geometry and the mechanical properties of the confinements.Other methods have so far employed bioprinters to construct 3D tumor models, by integrating various cell types, substrates, and bioprinting techniques for drug screening and therapy purposes. [26]Nevertheless, bioprinters lack the level of precision provided by 2PP printing.Only the sub-micrometer precision of the 2PP process allows to create realistic extracellular environments in situ, after the cells have already grown within 3D structures.
With our method it is also possible to separate and isolate parts of a cellular 3D aggregate, which is clearly superior to other bioprinting approaches, and can in the future give excellent opportunities to study parts of a 3D cellular aggregate.Hence, our strategy of encapsulating multicellular systems in situ with 2PP provides a significant advancement in the field, paving the way for many studies where the extrusion of cells from multicellular structures is decisive.This is particularly relevant for studies on metastasis from solid tumors, where the formation of secondary tumors is often driven by cells escaping from the original malignant tissue . [27]Similarly, also in tissue engineering and regenerative medicine, where shaping multicellular structures in situ would be highly valuable, our method can be applied complementarily to, e.g., optogenetics. [28]

Conclusion
In this work, we present a refined technique for in situ twophoton lithography on living spheroids constituted by less than 200 cells.As we demonstrate in this work, living cancer spheroids can be either totally or partly encaged, yielding new tools for future applications in cancer research, where the use of very small spheroids and/or multicellular organoids are currently replacing traditional 2D studies. [29]Furthermore, the method could also provide novel technical strategies for assessing cell migration in a three-dimensional environment, instead of traditional 2D scratch assays.Finally, since the method described in this work has the advantage to constrain a very low number of cells in very short time, we can envisage its utilization as a high throughput platform for personalized medicine, where microsections of a dissected tumor can be exposed to several anticancer drugs, or where metastatic activity through micro-cavities of a confining sphere can be tested.The ability to manipulate living cellular systems in situ with two-photon lithography promises great potential in applications where dynamic intervention of cellular systems at the micrometer scale is envisioned.

Experimental Section
Printing Material: A commercial natural gelatin-based hydrogel resin (HYDROBIO INX N100, Bioinx) was used as Bioink for this study.Briefly, the resin was prepared by mixing photoinitiator (10 μL) and of resin (90 μL) at 40 °C.Prior to print, the spheroids were placed in the middle of coverslips (see section "Spheroid culture").The media inside the PDMS ring were gently replaced with the prepared resin and covered with a coverslip to avoid dehydration of the resin.
Printing Process: Computer-aided designs of a collection of microdomes and micro-wells were achieved using the CAD software Autodesk Inventor 2021.The dome models were adjusted to the average size of the spheroids so that the small spheroids were enclosed inside the dome.In addition, pores with diameters of 20, 40, and 80 μm were created in the domes, allowing for cancer cells to migrate out.To create a relatively rigid barrier for the cells, the wall thickness of the dome was set at 50 μm.Once the designs were complete, the models were hatched and sliced using DeScribe (Nanoscribe) software in preparation for the print jobs.
A commercial Direct Laser Writing setup (Photonic Professional GT2, Nanoscribe GmbH) with a 25X, NA = 0.8 oil immersion objective was used to print the hydrogel-based domes around the spheroids.A laser power of 40 mW and a scanning speed of 60 mm.s −1 were used as print parameters.The structures were developed with a complete DMEM medium at 37 °C for 10 min.
Swelling Test: To measure the swelling behavior of the hydrogel, micropillar (50 μm in diameter) and dome shapes were printed on coverslips.The printed well structures were imaged at different time points after developing with the 37 °C warm buffer belonging to the Bioink kit (1, 3, 5 h, and 1, 2, 3, 4, 5 days) using a stereomicroscope (SZX10, Olympus).The print area for each time point was measured using ImageJ and normalized to the area of the model.The dome shapes were incubated after printing in complete cell culture medium at 37 °C for one day.To visualize the openings, the medium was exchanged to cell culture medium containing 0.5 mg mL −1 2000 kDa Fluoresceinisothiocyanat-dextran (FITC dextran) (Sigma-Aldrich, Germany).The imaging was performed using a Nikon Ti2 AX confocal light scanning microscope located at the Nikon Imaging Center in Heidelberg equipped with a Plan Apo  20x objective.FITC was excited at 488 nm and detected at 499-551 nm.The resulting images were processed by using Fiji ImageJ v1.53u and a custom made python3 code.Briefly, the image stacks were preprocessed by applying a threshold using Fiji followed by reslicing along the opening axis.The cross-sectional area of the openings was evaluated in 2 μm steps through the apertures using PyCharm 2022.3.2 (Community Edition).
Spheroids Preparation: MCF7 cells were grown as suspended spheroids using non-adhesive agarose coated 24-well plates.The plates were prepared using a sterile 1% w/v agarose (Sigma-Aldrich, Germany) solution in PBS.Prior to seeding, the cells were resuspended in collagen type I solution from rat tail (Sigma-Aldrich, Germany) diluted in cell culture medium at the final concentration of 25 μg mL −1 .The collagen solution was incubated on ice for 1 h for preassembly.The cells were then resuspended in the collagen type I solution and seeded in the agarosecoated well plate with a final concentration of 5000 cells/spheroid in a volume of 250 μL well −1 .The cells were incubated for 1 h at 37 °C and 5% CO 2 , and finally, 250 μL of cell growth medium was gently added to each well, before returning the plate to the incubator.Spheroids were 2-5 days in culture before the experiment.
Spheroid Culture: For the printing process, the spheroids were broken apart by pipetting and were placed in the center of a PDMS ring on a glass coverslip immersed in medium.After overnight incubation at 37 °C with 5% CO 2 , the spheroids adhered to the glass surface.
Control Culture: For the monolayer control experiments, ≈5600 cells cm −2 MCF7 cells were seeded on a coverslide and incubated over night at 37 °C and 5% CO2 before printing.
Fluorescent Staining: For all stainings of the spheroids, a CO 2 independent media (Gibco, USA) supplemented with 10% v/v fetal bovine serum, 1.0% v/v penicillin-streptomycin, 1x GlutaMax and 50 μg mL −1 gentamicin was used.For the live/dead assay, the cells were incubated 15 min prior imaging in media with 1 μM Calcein AM (invitrogen, USA) and 1 μg mL −1 propidium iodide (invitrogen, USA) at 37 °C.After imaging the media was exchanged to supplemented DMEM for further cell culture.The spheroids tested for apoptosis marker (Caspase 3/7) were incubated for 30 min in media containing 80 μL mL −1 CellEvent Caspase 3/7 Green ReadyProbes Reagent (invitrogen) at 37 °C.Subsequently, the media was changed to media containing Image-iT LIVE Plasma Membrane and Nuclear Labeling Kit (invitrogen, USA) with a final concentration of 1 μg mL −1 AlexaFluor 594 wheat germ agglutinin and 1.0 μM Hoechst 33 342.The samples were incubated for 10 min at 37 °C before imaging.After imaging, the media was exchanged to DMEM with supplements for further cell culture.
Imaging: For the live/dead and the apoptosis assay, a Nikon Ti2 Ax confocal light scanning microscope located at the Nikon Imaging Center in Heidelberg was used.The excitation wavelength for Hoechst 33 342 was at 405 nm and the emission range at 430-475 nm.Calcein AM and Caspase 3/7 were excited at 488 nm and detected at 499-551 nm.Propidiumiodide, AlexaFluor 594 and the print were excited at 561 nm and detected at 571-625 nm.The images were acquired using a Plan Apo  10x objective.
To visualize actin dynamics and cell motility, a spheroid constrained by a dome and a spheroid without confinement were stained two days after printing with live cell staining media containing 200 nM SiR actin (Spirochrome, USA), 10 μM Verapamil (Spirochrome, USA) and 0.1% v/v SPY-505 DNA (Spirochrome, USA) in phenol red free DMEM with 4.5 g L −1 glucose (PAN-biotech, Germany) supplemented with 10% v/v fetal bovine serum (PAN-biotech, Germany), 1% v/v penicillin-streptomycin, 1x Gluta-MAX and 50 μg mL −1 gentamicin.The spheroids were incubated in staining media for 1 h at 37 °C and 5% CO 2 prior imaging.Imaging was performed using a spinning disk microscope Olympus IX81 with an Olympus UPlansApo 20X objective.SiR actin was excited at 640 nm and recorded at 677 nm.SPY-505 DNA was excited at 488 nm and recorded at 523 nm.The microscope was equipped with a stage top incubation system by ibidi to provide optimal growth conditions (37 °C, 5% CO2 and 90% humidity).
Data Analysis: The shape of different spheroids has been extracted from fluorescence images in Fiji employing a macro by Nowell et al. [30] The statistical analysis of the collected data was carried out utilizing Numpy, Scipy, Pandas and Statannotation libraries. [31]ctin alignment characterization has been performed based on the method explained in Marcotti et al. [32] utilizing fast Fourier transform (FFT) of square windows of 33 by 33 pixels chosen from the maximum intensity projection of stacked fluorescence image stacks in the original space.The windows were overlapped by 70%, and an intensity threshold of 850 was considered to filter the blank spaces.By calculating the central image moments of FFT at each position and consequently constructing the covariance matrix of the image and computing its eigenvalues, the orientation of the filaments and the eccentricity, as a measure of how oblong the FFT was at a given position, was quantified.These pieces of information were used to generate orientation maps (considering that the orientation of features in the real image was orthogonal to the orientation of FFT, hence applying a 90-degree rotation) and eccentricity maps.Finally, to calculate the order parameter as an alignment indicator at each time point of flattened 2D images, the following Equation (1) was used: Here,  ij represents the angle between the central reference vector and its neighbors, and 〈 • • • 〉denotes the average over all filaments.The value of S ranges from −1 to 1, where S = 1 indicates perfect alignment of the actin filaments in a single preferred direction, S = −1 represents opposite alignment, and S = 0 signifies a random or isotropic distribution of orientations.

Figure 1 .
Figure 1.Overview of the 2PP method used to fabricate constraints for micro-spheroids.A) Illustration of the principle of the 2PP 3D printing method.A femtosecond laser with center wavelength at 780 nm selectively polymerizes a photosensitive material to create intricate structures with sub-micron resolution.B-D) 3D CAD model of a designed dome structure, having a total diameter of 260 μm, a wall thickness of 50 μm, four 80 μm wide openings on the side and a 80 μm wide opening on the top.For visualization purposes half of the structure is presented semi-transparent D).Brightfield image of a spheroid E) before printing and F) directly after printing.Scale bar 50 μm.G) Schematic representation showing how to constrain a spheroid with 2PP in situ.The time typically needed to constrain the spheroid is 6 min.

Figure 2 .
Figure 2. Cell migration through the openings of the dome structure is shown in Figure 1D.A) Live-dead staining and cross-section of spheroid and print ˜6 h after the printing (living cells (green), dead cells (red) and autofluorescence of printed structure (light red background)).B-G) Time-lapse images of cells spreading through the openings of the dome structure.Images show the print and the live-dead staining at B) day 0, C) day 1, D) day 2, E) day 3, F) day 4, and G) day 5.The presence of necrotic cells at the core of a spheroid is commonly observed as a consequence of restricted access to oxygen and nutrients caused by border cells.H) Control: Confocal image of spheroids without confinement stained with Calcein AM (green) and propidium iodide (red) at day 0 to day 2. Convexity ratio I) and perimeter to area ratio J) of the spheroid without constriction (control) and constricted by the dome shape (Print) at day 1 and day 2 after printing.Number of samples in control = 3, Number of samples in print = 3. Statistical analysis of dispersion: student's t-test on each pair of units (*: 1 × 10 −2 < p ≤ 5 × 10 −2 ).In Figure 2A,B, ˜100 cells were confined in each case.

Figure 3 .
Figure 3.In situ printing used to isolate cells from an mm-sized spheroid.A) 3D representation of the spheroid and the print on day 0. B-D) Time series of growing tissue surrounding the print on B) day 0, C) day 1, D) day 2. Living cells (green), Dead cells (red) and autofluorescence of printed structure (red).

Figure 4 .
Figure 4. A) Observed swelling behavior of 3D printed wells over the course of 5 days at 37 °C (number of measured prints 9).B,C) Effect of hydrogel swelling on opening size and shape in dome structure with 80 μm pores was visualized by imaging FITC-dextran (2000 kDa) in medium one day after printing.B) Histogram based on the cross-sectional area of the dome's apertures with 80 μm opening size.The cross-sections were derived by reslicing confocal images of domes along the axis of the openings.The images were taken one day after printing and incubation in media at 37 °C.(Number of positions in pores to analyze the cross section area in total = 1153).C) Examples of the cross-sections of measured openings (top) compared with the theoretical cross-section of the .stlfile (bottom).The theoretical cross-section: 2513 μm 2 .D) Confining cells with 40 μm opening prints.Living cells are stained with Calcein AM (green).Dead cells are stained with PI (red).Autofluorescence of print is visible in red.In Figure4Dan approximate number of 20 cells were confined by 2PP.The cells in the left corner of the Day 4 image are pre-existing cells that were outside the printed constraints before the printing process, and they continued to grow throughout the experimental period.

Figure 5 .
Figure 5. A) Maximum fluorescence intensity projection of z-stack images showing the actin filaments and nuclei of cells in a spheroid spreading outside of a channel at intervals of 1 h.The actin filaments were stained by SiR-Actin (magenta) and the nuclei by SPY505-DNA (green).B) The actin alignment at two different regions of the image.Region 1 represents the cells around the printed structure and region 2 displays the cells in the channel.An overlay of alignment vectors in yellow on the images is shown in (I).The FFT eccentricity map of the image ranging from 1 (circular) to 0 (elongated) is represented in (II).The orientation angles of actin fibers are displayed in 2D orientation map (III).C) The order parameter for 6 different regions of constricted spheroids and non-constrained spheroids at time points of 30 min for 6 h is shown.The order parameter value of 1 indicates complete alignment, 0 random alignment and −1 shows the orthogonal alignment.It is evident that the actin fibers around the print stayed more aligned compared to the cells in the channels and in non-constrained spheroids for a longer time.