一种用于自动化和标准化制造基于水凝胶的三维培养模型的开源技术平台

摘要

本方案为使用一种新型开源自动化技术进行黏性材料标准化且可重复混合的全面教程。文中详细介绍了新开发的开源工作站的操作方法、开源方案设计软件的使用，以及验证和确认可重复混合物的方法。

摘要

目前，粘性材料的混合步骤依赖于重复且耗时的操作，主要以手动方式进行，通量较低。这些问题在工作流程中构成了缺陷，最终可能导致研究结果无法重现。基于手动操作的工作流程进一步限制了粘性材料的发展及其广泛应用，例如用于生物医学领域的水凝胶。通过采用具有标准化混合流程的自动化工作流程，可以克服这些挑战，从而提高实验的可重复性。本研究提供了逐步操作指南，介绍如何使用开源实验方案设计工具、操作开源工作站，并确定可重复的混合物。具体而言，该开源实验方案设计工具可引导用户完成实验参数的选择，并生成可直接用于操作工作站的协议代码。该工作站针对粘性材料的移液操作进行了优化，通过集成温度控制模块（适用于温敏性材料）、正置换式移液器（适用于粘性液体）以及可选的吸头触碰模块（用于去除移液头表面多余材料），实现自动化且高度可靠的样品处理。混合物的验证通过快速且低成本的橙黄G（Orange G）吸光度测定完成。本方案展示了制备80% (v/v) 甘油混合液、明胶甲基丙烯酰基（GelMA）的系列稀释液，以及5% (w/v) GelMA与2% (w/v) 海藻酸钠双网络水凝胶的结果。文中还包含故障排除指南，以帮助用户顺利采用该协议。所述工作流程可广泛应用于多种粘性材料，实现用户自定义浓度的自动化配制。

引言

Reproducibility and replicability are of paramount importance in scientific work^1,2,3,4. However, recent evidence has highlighted significant challenges in repeating high-impact biomedical studies in fundamental science as well as translational research^4,5,6,7. Factors contributing to irreproducible results are complex and manifold, such as poor or biased study design^6,8, insufficient statistical power^3,9, missing compliance with reporting standards^7,10,11, pressure to publish⁶, or unavailable methods or software code^6,9. Amongst them, subtle changes in the protocol and human errors in the execution of experiments have been identified as further elements accounting for irreproducibility⁴. For instance, manual pipetting tasks introduce intra- and inter-individual imprecision^12,13 and increase the probability of human errors¹⁴. While commercial liquid handling robots are able to overcome these drawbacks and have demonstrated increased reliability for liquids^15,16,17, automated handling of materials with significant viscous properties is still challenging.

Commercial liquid handling robots commonly use air cushion pipettes, also known as air piston or air displacement pipettes. The reagent and the piston are separated by an air cushion which shrinks during dispensing steps and expands during aspirating steps. Using air cushion pipettes, viscous materials ‘flow’ only slowly into and out of the tip, and early withdrawal of the pipette from the reservoir may result in the aspiration of air bubbles. During dispensing tasks, the viscous material leaves a film on the inner tip wall which ‘flows’ only slowly or not at all when being forced by air. To overcome these issues, positive displacement pipettes were introduced commercially to actively extrude the viscous material out of the tip using a solid piston. Although these positive displacement pipettes enable accurate and reliable handling of viscous materials, automated solutions with positive displacement pipettes are still too expensive for academic laboratory settings, and, therefore, most workflows with viscous materials rely solely on manual pipetting tasks¹⁸.

In general, viscosity is defined as the resistance of a fluid to flow, and viscous materials are further being defined as materials with a greater viscosity of water (0.89 mPa·s s at 25 °C). In the field of biomedical applications, experimental setups often contain multiple materials with a greater viscosity than water, such as dimethyl sulfoxide (DMSO; 1.99 mPa·s at 25 °C), glycerol (208.1 mPa·s at 25 °C for 90% glycerol [v/v]), Triton X-100 (240 mPa·s at 25 °C), and water-swollen polymers, referred to as hydrogels^19,20. Hydrogels are hydrophilic polymer networks arranged in a physical or/and chemical mode used for various applications, including cell encapsulation, drug delivery, and soft actuators^19,20,21,22. The viscosity of hydrogels depends on the polymer concentration and molecular weight¹⁹. Routinely used hydrogels for biomedical applications exhibit viscosity values between 1 and 1000 mPa·s, while specific hydrogel systems have been reported with values of up to 6 x 10⁷ mPa·s^19,23,24. However, viscosity measurements of hydrogels are not standardized in terms of measurement protocol and sample preparation, and, therefore, viscosity values between different studies are difficult to compare.

Since commercially available automated solutions specifically designed for hydrogels are either missing or too expensive, current workflows for hydrogel depend on manual handling¹⁸. To understand the limitations of the current manual-based workflow for pipetting of hydrogels, it is important to comprehend essential handling tasks¹⁸. For example, once a novel hydrogel material has been synthetized, a desired concentration or a dilution series with varying concentrations is generated to identify reliable synthesis protocols and crosslinking characteristics with subsequent analysis of the mechanical properties^25,26,27,28. In general, a stock solution is prepared or purchased, and subsequently mixed with a diluent and/or other reagents to obtain a mixture. The mixing tasks are mostly not performed directly in a well plate (or any output format), and are rather performed in a separate reaction tube, which is commonly referred to as master mix. During these preparation tasks, various aspirating and dispensing steps are required to transfer the viscous material(s), mix the reagents, and transfer the mixture to an output format (e.g., a 96 well plate). These tasks require a high amount of human labor¹⁸, long experimental hours, and increase the probability of human errors which could potentially manifest as inaccurate results. Moreover, manual handling prevents efficient preparation of high sample numbers to screen various parameter combinations for detailed characterization. The manual processing also impedes the usage of hydrogels for high-throughput screening applications, such as the identification of promising compounds during drug development. The current manual-based preparation steps are not feasible to screen drug libraries consisting of thousands of drugs. For these reasons, automated solutions are required to provide an efficient development process and enable the successful translation of hydrogels for drug screening applications.

To move from manual-based workflows to automated processes, we have optimized a commercial open source pipetting robot for the handling of viscous materials by the integration of temperature docks for thermoresponsive materials, the usage of off-the-shelf positive displacement pipettes using capillary piston tips, and an optional tip touch dock for pipette tip cleaning. This pipetting robot has been further integrated as a pipetting module into a newly developed open source workstation, which consists of ready-to-install and customizable modules^18,29. Detailed assembly instructions for the developed workstation including hardware and software files are freely accessible from the GitHub (https://github.com/SebastianEggert/OpenWorkstation) and the Zenodo repository (https://doi.org/10.5281/zenodo.3612757). In addition to the hardware development, an open source protocol design application has been programmed and released to guide the user through the parameter selection process and generate a ready-to-use protocol code (https://github.com/SebastianEggert/ProtocolDesignApp). This code runs on the commercial open source pipetting robot as well as on the developed open source workstation.

Herein, a comprehensive tutorial is provided on the operation of the open source workstation to automate mixing tasks for viscous materials (Figure 1). The tutorial-specific protocol steps can be carried out with the developed open source workstation as well as the commercial open source pipetting robot. Supported by an in-house developed open source protocol design application, automated mixing and preparation of required concentrations for glycerol, gelatin methacryloyl (GelMA) and alginate is demonstrated. Glycerol has been selected in this tutorial, since it is well characterized^30,31, it is inexpensive and readily available, and, therefore, it is commonly used as viscous reference material for automated pipetting tasks. As examples for hydrogels used in biomedical applications, GelMA and alginate hydrogel precursor solutions have been applied for automated mixing experiments. GelMA presents one of the most commonly used hydrogels for cell encapsulation studies^32,33, and alginate was selected in this study to demonstrate the ability to manufacture double network hydrogels^34,35. Using Orange G as a dye, a fast and inexpensive procedure was implemented to validate and verify the mixing results with a spectrophotometer¹⁶.

A commercial open source pipetting robot has been integrated as a pipetting module into the developed open source workstation (Figure 2a), and therefore, the name ‘pipetting module’ is further used to describe the pipetting robot. A detailed description of the installed hardware is beyond the scope of this protocol and is available via the provided repositories which also include step-by-step instructions for the general assembly of the open source platform. The pipetting module can be equipped with two pipettes (single- or 8-channel pipette) which are installed on axis A (right) and axis B (left) (Figure 2b). The pipetting module offers a 10-deck capacity according to American National Standards Institute/Society for Laboratory Automation and Screening (ANSI/SLAS) standards, and the following location positions are defined on the deck: A1, A2, B1, B2, C1, C2, D1, D2, E1, E2 (Figure 2c). To initiate photo-induced polymerization of hydrogel solutions, a separate crosslinker module is required and has been added to the workstation. The crosslinker module is equipped with LEDs with a wavelength of 400 nm and, therefore, substances that excite at a visible light wavelength can be used with the current systems, such as lithium phenyl-2,4,6 trimethylbenzoylphosphinate (LAP)^36,37. The intensity (in mW/cm²) of the LEDs can be addressed by the user in the protocol design application to study the crosslinking behavior³⁸. The workstation includes also a storage module to enable increased throughput studies; however, this module is not used within this study and, therefore, not further described. In general, it is recommended to operate the pipetting module in a biological safety cabinet to avoid sample contamination. The main power circuit to operate the pipetting module is a 12 V circuit, which is considered as a low-voltage application in most countries. All electrical components are based in a dedicated control box preventing users from coming into contact with the source of an electrical hazard.

By following these standardized mixing protocols, researchers are able to achieve reliable mixtures for viscous as well as non-viscous materials in an automated fashion. The open source approach allows users to optimize mixing sequences and share newly developed protocols with the community. Ultimately, this approach will facilitate the screening of multiple parameter combinations to investigate the interdependencies between different factors and, thereby, accelerate the reliable application and development of viscous materials for biomedical applications.

方案

NOTE: The protocol starts with an introduction to (1) the software and (2) the hardware setup to familiarize the user with required installations and the workstation. Following a section on (3) material preparation and (4) the usage of the protocol designer application, (5) the calibration of the pipetting module and (6) the execution of the automated protocol is highlighted in detail. Finally, (7) validation and verification procedures are described, including absorbance reading and data analysis. A general protocol workflow with individual tasks is displayed in Figure 1.

1. Software setup

NOTE: This section includes a detailed instruction to install the application programming interface (API) as well as the required protocol designer application and the calibration terminal. The following instructions are written for a Raspberry Pi (RPi) single-board computer; however also Windows 8, 10 and macOS 10.13+ have been successfully used with the API and the applications.

1. Set up the computer environment.
   NOTE: Be familiar with the basics of Python³⁹, how to set up and use a Raspberry Pi^40,41, and how to connect to the internet⁴². The following tutorial steps focus on protocol-specific steps and additional information on the usage of a Raspberry Pi is available online⁴⁰.
   1. Open a terminal window from the taskbar or application menu.
   2. Update the system’s package list:
      sudo apt-get update
   3. Upgrade all the installed packages:
      sudo apt-get dist-upgrade
   4. Restart the Raspberry Pi:
      sudo reboot
   5. Check installed Python version:
      python3 --version
      Make sure that at least Python 3.5 is installed; if not, install the latest version⁴³.
   6. Install python pip, which publishes Python packages with the Python Package Index⁴⁴:
      sudo apt-get install python3-pip
   7. Install dependencies:
      pip install numpy
      pip install python-resize-image
      NOTE: If you are using Windows, you need to install the windows-curses package via: python -m pip install windows-curses
2. Install the application programming interface (API).
   NOTE: The API provides a simple Python framework designed to write experimental protocols script and operate the workstation. The following two APIs are required to successfully execute the generated protocol code.
   1. Install workstation API:
      pip install openworkstation
   2. Install Opentrons API to operate the pipetting module:
      pip install opentrons==2.5.2
   3. Verify, if API is installed successfully:
      python3
      >>> import openworkstation
      >>> import opentrons
      NOTE: The size of the API and the protocol design application is 2.2 MB and 1.2 MB, respectively. No issues were experienced during the installation when used with limited disk space (200 MB). However, the disk space requirements depend on the operating system.
3. Select a directory for file download (calibration terminal, protocol design application, etc.).
   NOTE: Files can be copied and pasted elsewhere afterwards.
4. Clone files from GitHub repository:
   git clone https://github.com/SebastianEggert/OpenWorkstation
   NOTE: The ‘git clone’ command clones and subsequently saves all files into the directory, which is open in the terminal at this time. Since the repository also includes the hardware files for the assembly, not the entire repository is required to execute the presented protocols. All required files to replicate the experiments are available as Supplemental File and in the GitHub repository under “/examples/publication-JoVE”.
5. Open the downloaded folder. If the entire repository was downloaded, navigate to the ‘publication-JoVE’ folder via
   cd openworkstation/examples/publication-JoVE
   NOTE: This folder includes files required for the operation of the workstation and the usage of the protocol designer application and the calibration terminal.

2. Hardware setup

1. Place the workstation in a biological safety cabinet to avoid sample contamination.
2. Install pipettes on the workstation.
   1. Select the pipette size based on the experimental setup. In general, take a pipette size which volume to be aspirated is on the higher end of the range. If mixing tasks with volumes greater than 1 mL are required for a specific setup (e.g., aspirating/dispensing of 2 mL), choose the M1000E with a maximal aspirating/dispensing volume of 1,000 µL to minimize pipetting steps and save time.
      NOTE: A detailed instruction for air-displacement pipettes is available online⁴⁵. The developed pipetting module is able to integrate the following off-the-shelf positive displacement pipettes: M10E (1–10 µL), M25E (3–25 µL), M50E (20–50 µL), M100E (10–100 µL), M250E (50–250 µL), M1000E (100–1,000 µL).
   2. Use an M4 Allen key to loosen and tighten screws.
   3. Attach the two pipette fixation plates (white acrylic plates) to the aluminum rail and tighten M5 screws loosely.
   4. Insert the pipette into the two pipette fixation plates and ensure that the ergonomic tail of the pipette is resting on the opposite side of the acrylic mounting plate.
   5. Tighten the four screws of the two pipette fixation plates firmly.
   6. Slide the two square fastening nuts, which are attached to the acrylic mounting plate, into the extrusion slot of the z-axis and tighten screws.
      NOTE: Fasten pipette tightly to avoid any movement during operation.

3. Material preparation

NOTE: The viscous materials (glycerol, GelMA, alginate) are used for the experiments presented in this study, and, therefore, prepared volumes and handling tasks (e.g., add 5 mL of stock solution in 5 mL reaction tubes) are specifically for this experimental setup.

1. Gelatin methacryloyl (GelMA)
   NOTE: GelMA functionalization, dialysis, and lyophilization are not the scope of this paper, and a step-by-step protocol is available in Loessner et al.³³. The protocol starts using lyophilized GelMA, which can be prepared in-house or purchased commercially.
   1. Calculate required mass of GelMA (m_GelMA) based on the desired final stock concentration (c_GelMA) and volume (V_GelMA) using the equation:
      m_GelMA = c_GelMA x V_GelMA
      NOTE: VGelMA depends on the experimental setup and it is recommended to prepare 20−30% excess material. The presented protocols start with 5 mL of 20% (w/v) GelMA as a stock solution.
   2. Weigh the required amount of lyophilized GelMA, add it into a 50 mL reaction tube and add the required amount of phosphate buffered saline (PBS).
   3. Mix GelMA either by soaking into the solvent at 4 °C overnight or by heating to 60 °C for 6 h in a water bath.
      NOTE: Sterile GelMA solutions can be stored protected from light at 4 °C for at least six months.
   4. Fill 5 mL of GelMA into 5 mL reaction tubes.
2. Photoinitiator: Lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP)
   NOTE: Avoid additional exposure to room light, since LAP is light sensitive.
   1. Calculate required mass of LAP (m_LAP) based on the desired final stock concentration (c_LAP) and required volume (V_LAP) using the equation:
      m_LAP = c_LAP x V_LAP
      NOTE: It is recommended to prepare a 3% (w/v) stock solution.
   2. Weigh the required amount of LAP, add it into a 15 mL reaction tube and add PBS.
   3. Wrap the tube in aluminum foil to prevent photo-induced decomposition.
   4. Dissolve LAP by placing the reaction tube in a water bath at 37 °C for 2 h or until fully dissolved.
   5. Fill 1 mL of LAP stock solution in 5 mL tubes.
3. Alginate
   1. Calculate required amount of alginate (m_alginate) based on the desired final stock concentration (c_alginate) and volume (V_alginate) using the equation:
      m_alginate = c_alginate x V_alginate
      NOTE: Valginate depends on the experimental setup and it is recommended to prepare 20−30% excess material. The presented protocols start with 5 mL of 4% (w/v) alginate as a stock solution.
   2. Weigh the required mass of alginate, add it into a 50 mL reaction tubes, and add PBS.
   3. Place the alginate mix into a water bath at 37 °C for 4 h.
      NOTE: The usage of a vortex mixer will accelerate the dissolution process, but also generate air bubbles. Dissolved alginate can be stored at 4 °C for at least six months.
   4. Fill 5 mL of alginate into 5 mL reaction tubes.
4. Fill 5 mL of glycerol in 5 mL reaction tubes.
5. Orange G solution
   1. Prepare a 10 mg/mL stock solution of Orange G in a 50 mL reaction tube.
      NOTE: The volume depends on the number of experiments. Depending on the diluent type, prepare stock solution either in ultrapure water, PBS or a suitable diluent reagent. In the presented experiments, ultrapure water was used for diluting glycerol and PBS for diluting GelMA and alginate. PBS was used as a diluent for GelMA and alginate, and can be either prepared using tablets or purchased off-the-shelf.
   2. Mix for 10 s by vortexing.
   3. Wrap the tube in aluminum foil to prevent photo-induced decomposition.
      NOTE: Stock solution can be used after 24 h to ensure proper dissolution of Orange G.
   4. Dilute stock solution to a 1 mg/mL working solution in a 50 mL reaction tube.
   5. Transfer working solution to appropriate flasks/tubes for the experimental setup.
      NOTE: For the presented experiments, the working solution was filled into 5 mL tubes. Orange G stock and working solution can be stored at 4 °C and used within three months upon preparation.
   6. Fill 5 mL of the 1 mg/mL Orange G working solution in 5 mL reaction tubes.

4. Generate protocol code with the protocol designer application

NOTE: The specified parameters in steps 4.2−4.7 are the same for all conducted experiments, except for the material’s stock concentration and the final output concentration. These parameters are summarized in Table 1 and, in the following, parameters are used to prepare double network hydrogels with 5% (w/v) GelMA, 2% (w/v) alginate, 0.15% (w/v) LAP, and PBS as a diluent.

1. Open the protocol designer application by running ‘ProtocolDesignApp.html’.
   NOTE: The app “ProtocolDesignApp.html” guides the user through the parameter selection process and automatically generates the ready-to-use protocol to operate the workstation. The user interface runs on every commonly used internet browser (i.e. Chrome, Firefox, Safari, edge, Internet Explorer).
2. Enter protocol name (e.g., double-network-hydrogels) on setup page.
3. Click ‘Continue’ to confirm protocol name and proceed to the next step.
4. Define input tray by selecting ‘3x4 Heating Block’ from drop-down menu and the following input parameters:
   1. Select ‘Gel 1’ from drop-down menu, enter name ‘GelMA’, enter stock-concentration ‘20%’, set ‘Number of Samples’ to ‘3’ to fill one column. Click ‘+add’ to save entries.
   2. Select ‘Gel 2’ from drop-down menu, enter name: ‘Alginate’, enter stock-concentration ‘4%’, set ‘Number of Samples’ to ‘3’ to fill one column. Click ‘+add’ to save entries.
   3. Select ‘Photoinitiator’ from drop-down menu, enter name: ‘LAP’, enter stock-concentration ‘3%’, set ‘Number of Samples’ to ‘3’ to fill one column. Click ‘+add’ to save entries.
   4. Select ‘Diluent 1’ from drop-down menu, enter name: ‘PBS’, set ‘Number of Samples’ to ‘3’ to fill one column. Click ‘+add’ to save entries.
      NOTE: The visualization of the input tray automatically updates, once ‘+add’ is clicked. If more inputs are added than the tray’s capacity, the warning ‘Too many samples for this tray’ will be shown to the user.
5. Define crosslinking parameters by checking ‘Photo crosslinking’ and typing the time in seconds, ‘30’, and the intensity W/m² with ‘2’.
6. Finish the input setup by clicking ‘CONTINUE’.
7. Define output tray setup by selecting ’96 well plate’ in the drop-down menu for well plate type.
8. Click on ‘Group1’ to define outputs by creating a group of samples.
   1. Specify output composition by entering desired concentrations and sample volume into the fields for each input: GelMA = ‘5’, Alginate = ‘2’, LAP = ‘0.15’, Total Volume = ‘60’.
   2. Check box to apply advanced mixing protocol.
9. Specify number of samples by entering the number of samples into the field ‘Number of Samples’: ‘96’.
   NOTE: The visualization of the sample tray automatically updates, once ‘+add group’ is clicked. If more samples are added than the tray’s capacity, the warning ‘Too many samples for this tray’ will be shown to the user.
10. Finish the output setup by clicking ‘CONTINUE’.
    1. Select tray position on the deck layout and prepare the platform accordingly:
    2. Check checkmark in field ‘SLOT A1’ and select ‘Empty_Cell’ from drop-down menu.
    3. Check checkmark in field ‘SLOT A2’ and select ‘Trash_Cell’ from drop-down menu.
    4. Check checkmark in field ‘SLOT B1’ and select ‘Tips_Cell_100µL’ from drop-down menu.
    5. Check checkmark in field ‘SLOT B2’ and select ‘Tips_Cell_1000µL’ from drop-down menu.
    6. Check checkmark in field ‘SLOT C1’ and select ‘Input_Cell’ from drop-down menu.
    7. Check checkmark in field ‘SLOT C2’ and select ‘Empty_Cell’ from drop-down menu.
    8. Check checkmark in field ‘SLOT D1’ and select ‘Mixing_Cell’ from drop-down menu.
    9. Check checkmark in field ‘SLOT D2’ and select ‘Output_Cell’ from drop-down menu.
    10. Define type and characteristics of first pipette (M100E) by checking ’Pipette Left’, selecting ‘10-100µL positive displacement’ from drop-down menu and setting aspirating speed = ‘600’, dispensing speed = ‘800’.
    11. Define type and characteristics of second pipette (M1000E) by checking ’Pipette Right’, selecting ‘100-1000µL positive displacement’ from drop-down menu and setting aspirating speed = ‘800’, dispensing speed = ‘1200’.
11. Click on ‘GENERATE PROTOCOL’ to confirm the setup and generate the protocol script.
    NOTE: The developed protocol designer app generates automatically a new folder whenever a new protocol is generated. All files which are required for this experiment and to operate the workstation are saved in this folder which is named after the protocol name. The folder can be copied into different directories without causing issues.
12. Do not close the interface, as it will be used to execute the protocol (see step 6.6.).

5. Calibration of the pipetting module

NOTE: Containers (e.g., well plates, tip rack, trash) and pipettes (e.g., M1000E) must be calibrated initially. If a container and/or a pipette position are modified/changed, the new position must be calibrated.

1. Navigate to the protocol folder and open the calibration terminal by executing the file ‘calibrate.py’ in a terminal windox (see step 1.1.1):
   phython.calibrate.py
   NOTE: The 'calibrate.py' interface guides the user through the calibration of the deck setup and pipettes. Make sure that the file is in the same folder as the protocol file and the module files. It is automatically generated in step 4.10.
2. Select movement increments for plungerx,y,z movement with the numeric keypad (1−8): ‘1’ for 0.1 mm, ‘2’ for 0.5 mm, ‘3’ for 1 mm, ‘4’ for 5 mm, ‘5’ for 10, ‘6’ for 20 mm, ‘7’ for 40 mm, and ‘8’ for 80 mm.
3. Calibrate the pipette.
   1. Press keyboard shortcut P to select pipette size.
   2. Press keyboard shortcut V to enter the plunger calibration mode.
      NOTE: It is recommended to start with small increments (2, 5, and 10 mm) to get familiar with the increment size and movement action of the pipette head.
   3. Calibrate the following plunger positions for a positive displacement pipette: T–Top = rest position; B–Bottom = plunger is pushed until resistance is met; P–Pick-up = plunger is pushed to a position where a piston tip can be attached; E–Eject = plunger is pushed until an attached tip is ejected. Vary plunger positions using the upwards and downwards arrows on the keyboard, and save the final position using S on the keyboard.
   4. Leave the pipette plunger calibration mode by pressing keyboard shortcut V.
4. Calibrate container position relative to the pipette tip.
   1. Press keyboard shortcut P to select pipette type. Make sure that a tip is connected to the selected pipetted.
   2. Press keyboard shortcut C to select container type.
   3. Select an appropriate movement increment and move pipette tip to the following positions. For well plates, calibrate to the ‘A1’ well position at the bottom; For tip rack, calibrate to the ‘A1’ position; For trash, choose a position (defined as a point) where the tip can be ejected into the trash.
   4. Press keyboard shortcut S to save position.
   5. Repeat steps 5.3.1−5.3.3 for all containers listed under ‘C’ for the selected pipette type.
   6. Repeat 5.3.1−5.3.5 for the second pipette type.
   7. Close the calibration script.

6. Protocol execution with the workstation

NOTE: Protocol files are accessible via the repository and are also available as Supplemental File.

1. Position trash container, tip racks, input tray, mixing tray, and output on the deck (defined in step 4.3).
2. Calibrate pipettes and instruments as defined in section 5.
3. If required, switch the temperature dock ON and select the temperature for input and mixing tray.
   NOTE: The experiments in this tutorial were conducted without temperature control and at 40 °C for glycerol, and 37 °C for GelMA and alginate pipetting.
4. Position tubes with input reagents in the aluminum blocks on the temperature docks according to the selected setup.
5. Wait until input reagents have reached the desired temperature.
   NOTE: To ensure proper temperature distribution, an incubation time of 30 min is recommended for GelMA and alginate.
6. Execute the protocol file by clicking on ’RUN PYTHON SCRIPT’
   NOTE: The selected protocol is now executed by the workstation. The accompanying video highlights automated mixing of GelMA and the distribution of 60 µL into a 96 well plate.
7. Run is completed, when ‘Finished’ is displayed.

7. Validation and verification process

1. Remove the well plate from workstation and transport the well plate with the samples to a spectrophotometer.
2. Read absorbance with a spectrophotometer at 450 nm. Read each plate 2x to compare results and ensure consistent results.
3. Export and save absorbance readings.
4. Data analysis.
   NOTE: Experimental data can be processed individually or copied and pasted into the provided template to evaluate the mean, standard deviation, and coefficient of variance (CV) value using spreadsheet software.
   1. Open the Supplemental File ‘supplementary_template-analysis.xlsx”, which is also vailable within the GitHub repository under ‘openworkstation/examples/publication-JoVE.
   2. Copy absorbance readings into the ‘raw data’ sheet, ensure that all cell references are correctly defined in all tables, and click on the ‘analysis’ sheet for information on the average, standard deviation, and coefficient of variance (CV) values.
      NOTE: Depending on the sample distribution on a well plate, the following preset evaluation types are available with the template: the ‘Uniform’ type is used when all samples have the same composition, the ‘By rows’ type is used when samples in different rows have different compositions, the ‘By columns’ type is used when samples in different columns have different compositions, and the ‘Customized’ type is used when the sample positions are user specific.

结果

本教程展示了甘油（图3）以及 GelMA 与 LAP 和海藻酸钠（图4）实验的结果。

研究了在三种不同条件下制备80%（v/v）甘油溶液的过程：无温度控制（室温，22 °C）且不进行吸头触碰（定义为设置1）、有温度控制（40 °C）但无吸头触碰（设置2），以及有温度控制（40 °C）并进行吸头触碰（设置3）（图3a-i）。选择这两个温度条件是为了评估操作差异，因为当甘油从22 °C（139.5 mPa·s）加热至40 °C（46.6 mPa·s）时，其黏度几乎降低至原来的三分之一³⁰。将85%（v/v）的甘油储备液稀释至终浓度为80%，并均匀分配至96孔板中（每种设置n = 96）。实验总耗时为30分42秒，包括将各组分加入混合管、相应的混匀操作以及样品分配至96孔板的过程。为了识别稀释混合物之间的差异，使用含1 mg/mL橙黄G（Orange G）的超纯水作为甘油的稀释剂。吸光度读数显示，温度控制与吸头触碰功能的引入对混合效果具有显著影响（p < 0.0001）。除进行双因素方差分析（ANOVA）外，还计算了变异系数（CV）以评估相对标准偏差。变异系数是衡量偏差程度相对于均值的标准化指标，以百分比表示。当关注点并非样本均值本身，而是测量值之间的变异性时，变异系数可提供额外信息以识别可重复的混合效果⁴⁶。在本实验的三种不同设置中，设置1、设置2和设置3的吸光度CV值分别为5.6%、4.2%和2.0%，依次降低，表明温控模块和吸头触碰功能对获得可靠结果具有显著影响（图3a-ii）。绘制设置3的样品吸光度值（96孔板中样品编号#1至#96）显示，整个实验过程中未出现系统性上升或下降趋势，因此表明样品位置对吸光度值无显著影响（图3a-iii）。通过热图可视化每块检测孔板的数据，可进一步识别特定行或列是否存在非均一性，或在分配过程中吸光度值的变化情况。三种设置的热图显示，从设置1到设置3，整块孔板上的非均一性逐渐减少（图3b）。最后，在八次独立运行中评估了所进行混合操作的可重复性（图3c-i,ii），每次运行耗时6分57秒。单次混合运行的CV值较低，介于1.1%至2.6%之间，表明各次运行中的混合与分配操作高度可靠。八次运行的吸光度值整体CV值为3.3%，证明所建立的混合方案具有良好的可重复性。

通过将20% (w/v) 的 GelMA 储备液用 PBS 稀释至14%、12%、10%、8%、6%、4%、2% 和 0% (w/v)，并加入 LAP 至终浓度恒定为 0.15% (w/v)，制备 GelMA 梯度稀释系列（图4a-i），总耗时为 55 分 12 秒。根据实验方案脚本的要求，水凝胶在 400 nm 波长、2.0 mW/cm² 光强下交联 30 秒。为评估不同混合物之间的差异，将作为 GelMA 和海藻酸钠稀释剂的 PBS 配制为含 1 mg/mL 橙黄 G（Orange G）的溶液。因此，可通过分光光度计检测同一种混合物内各样品之间以及连续稀释梯度之间的吸光度差异。各浓度梯度的吸光度测量值之间差异显著（p < 0.0001），且各浓度梯度间的变异系数（CV）极低，介于 1.2% 至 3.4% 之间（每浓度梯度 n = 12）。线性回归分析显示拟合度高，R² 值达 0.9869（图4a-ii），热图分析进一步证实了每种浓度分布的均一性以及不同浓度之间的差异（图4a-iii）。采用自动化混合方式制备含 5% (w/v) GelMA、2% (w/v) 海藻酸钠、0.15% (w/v) LAP 及 PBS 作为稀释剂的混合液，分别在不接触孔壁（setup 2）和接触孔壁（setup 3）两种条件下进行（每种条件 n = 96），交联参数相同（30 秒，2.0 mW/cm²，400 nm）。四种试剂的分配、混合及分装至 96 孔板共耗时 32 分 22 秒。所有涉及 GelMA 和海藻酸钠的实验均在 37 °C 下进行，以防止发生热致凝胶化，从而避免 GelMA 无法移液。使用触碰孔壁（touch tip）功能后，CV 值从 5.2% 降低至 3.4%，特别是通过去除吸头末端多余液体，有效避免了低值区域的离群值（图4b-i）。尽管 setup 2 和 setup 3 的平均值分别为 1.927 和 1.944，非常接近，但变异系数表明相对于均值的偏差显著减小。可通过热图可视化方式比较 96 孔板中单个行之间的差异，以检测行和/或列方向上的不均一性（图4b-ii）。

图1：包含各项具体任务的实验流程图。 本研究所述工作流程分为七个任务，归入设置、准备、执行和分析四个阶段。首先，需完成软件（任务1）和硬件（任务2）的设置。在完成材料准备（任务3）和实验程序脚本生成（任务4）后，通过定义移液器和容器的位置对移液模块进行校准（任务5）。随后，在工作站上运行程序脚本（任务6），并对混合物进行验证与确认（任务7），以评估混合物的质量。请点击此处查看此图的放大版本。

图 2：移液模块的开源工作站及工作台布局。（a）所开发的工作站借鉴了流水线式设计思路，样品可在不同模块之间传送，整个系统由以下模块组成：移液模块、交联模块、存储模块、传输模块和计算模块。（b）移液模块的工作台布局根据实验设计（例如孔板类型、管体积等）进行配置。图中所示的工作台布局用于本研究中的实验，包括量程为 10−100 µL（M100E）和 100−1,000 µL（M1000E）的正位移移液器、配备毛细活塞（CP）的吸头架（100 µL 使用 CP1000，1,000 µL 使用 CP1000）、废料容器、混匀托盘以及用于输入试剂的进样托盘。（c）可用的工作台位置由图中标注的数字定义。请点击此处查看该图的放大版本。

图3：甘油混合物自动移液结果。（a）灵活的自动化工作站设计支持评估三种不同设置（i），以确定可重复结果的最佳参数。（ii）增加吸头触碰（tip touch）步骤并对材料加热后，体系3的变异系数（CV）显著降低，所得混合物高度可重复。每次实验均使用96个样本。（iii）单个样本值的绘图显示移液顺序无显著影响。（b）各体系的实验结果通过热图可视化，以评估行/列差异、边缘效应或母液混合情况的影响。（c）体系3的可重复性在八次独立运行中进行了分析，包括（i）中位数、标准差、CV值，以及（ii）热图。图中面板 a-ii（n = 96）和 b-i（n = 12）的数据以均值及单个数据点表示。统计学显著性定义为 ****p < 0.0001，采用双因素方差分析（ANOVA）。请点击此处查看该图的放大版本。

图 4：水凝胶混合实验结果。 （a）从 20% (w/v) 的明胶甲基丙烯酰 (GelMA) 储备溶液出发，在一个实验流程中使用 96 孔板制备了 14、12、10、8、6、4、2 和 0% (w/v) 的系列稀释液（每种浓度 n = 12）。 （i）所配制各浓度的变异系数 (CV) 值介于 1.2% 至 3.4% 之间，（ii）线性回归显示高度拟合，R² 值为 0.9869。（iii）通过生成的热图可直观确认稀释的均一性。（b）采用 5% (w/v) GelMA、2% (w/v) 海藻酸钠和 0.15% (w/v) LAP 制备双网络水凝胶，（i）分别在有和无触碰吸头操作的条件下进行（每种设置 n = 96），并在 400 nm 波长、2.0 mW/cm² 光强下交联 30 秒。引入触碰吸头操作后，CV 值从 5.2% 降低至 3.4%。（ii,iii）热图证实，使用触碰吸头去除吸头多余材料可减少偏差。图中 a-i 和 b-i 面板的数据以均值及单个数据点表示。统计学显著性定义为 *p < 0.05，***p < 0.001，以及 ****p < 0.0001，采用单因素方差分析 (ANOVA)。 请点击此处查看该图的放大版本。

图5：移液器类型差异及粘性生物材料操作中的问题总结。（a）试剂与活塞之间由空气垫隔开，分配过程中空气垫收缩，吸液过程中空气垫膨胀。在吸液和分配粘性物质时，缓慢的“流动”会引发气泡和移液行为不规则等问题。（b）正置换移液器通过在吸头内部使用活塞，实现对粘性材料可靠地吸液和分配。（c）移取高粘性材料（例如4% (w/v) 海藻酸钠）可能导致吸头表面残留过量材料，从而在整个实验过程中造成体积误差。（d）使用简单的吸头触碰托盘可有效去除吸头表面的过量材料，从而实现精确的吸液和分配体积。该方法可通过将孔板盖的内侧放置于吸头架容器上来实现。请点击此处查看此图的放大版本。

材料 #1（储备浓度）	材料 #1 的终浓度	材料 #2（储备浓度）	材料 #2 的终浓度	材料 #3（储备浓度）	材料 #3 的终浓度	稀释液（Orange G 工作液）	混合物中 Orange G 的终浓度	图中显示
甘油（85% (w/v)）	80% (w/v)					水（1 mg/mL Orange G）	0.059 mg/mL	图 3a−c
GelMA（20% (w/v)）	0% (w/v)	LAP（3% (w/v)）	0.15% (w/v)			PBS（1 mg/mL Orange G）	1 mg/mL	图 4a
GelMA（20% (w/v)）	2% (w/v)	LAP（3% (w/v)）	0.15% (w/v)			PBS（1 mg/mL Orange G）	0.85 mg/mL	图 4a
GelMA（20% (w/v)）	4% (w/v)	LAP（3% (w/v)）	0.15% (w/v)			PBS（1 mg/mL Orange G）	0.75 mg/mL	图 4a
GelMA（20% (w/v)）	6% (w/v)	LAP（3% (w/v)）	0.15% (w/v)			PBS（1 mg/mL Orange G）	0.65 mg/mL	图 4a
GelMA（20% (w/v)）	8% (w/v)	LAP（3% (w/v)）	0.15% (w/v)			PBS（1 mg/mL Orange G）	0.55 mg/mL	图 4a
GelMA（20% (w/v)）	10% (w/v)	LAP（3% (w/v)）	0.15% (w/v)			PBS（1 mg/mL Orange G）	0.45 mg/mL	图 4a
GelMA（20% (w/v)）	12% (w/v)	LAP（3% (w/v)）	0.15% (w/v)			PBS（1 mg/mL Orange G）	0.35 mg/mL	图 4a
GelMA（20% (w/v)）	14% (w/v)	LAP（3% (w/v)）	0.15% (w/v)			PBS（1 mg/mL Orange G）	0.25 mg/mL	图 4a
GelMA（20% (w/v)）	5% (w/v)	海藻酸钠（4% (w/v)）	2% (w/v)	LAP（3% (w/v)）	0.15% (w/v)	PBS（1 mg/mL Orange G）	0.2 mg/mL	图 4b

表1：所进行实验的参数概述。

实验步骤	问题	可能原因	解决方案
1.1	软件无法安装或更新	SD卡磁盘空间不足	检查SD卡的磁盘空间。如有需要，删除不必要的文件、清空回收站，或使用容量合适的SD卡
1.2	API无法安装	用户安装权限受限（无root用户权限）	在指定命令前使用“sudo”命令以获取管理员权限。在Linux系统中，此类访问权限称为超级用户（superuser）
3.1	GelMA相关问题	功能化、透析或冷冻干燥过程问题	详细分步操作流程及故障排除列表参见Loessner等人的文献33
5.1 和 6.2	工作站对指令无响应	连接问题	关闭所有设备并关闭计算机，切断电源10秒后，重新启动计算机和工作站
5.1 和 6.2	工作站对指令无响应	连接问题	检查计算机是否识别USB连接，并确认USB端口设置正确。确保防火墙未阻止连接过程（参见表2下方链接）
5.1 和 6.2	无法打开文件	目录错误	检查目录（文件夹路径），确保使用了正确的路径。若无法找到某个文件（例如interface.py），可能是路径设置错误
6.6.2	吸头未正确安装或在移动过程中脱落	校准问题	重复移液器的校准步骤，并确保毛细管活塞与移液器正确连接
6.6.2	吸头未正确安装或在移动过程中脱落	安装问题	移液器未正确连接至移液轴，导致在移动过程中发生位移。请拧紧螺丝以防止此类情况
6.6.2	吸头在材料上方吸液	校准问题	重新校准该类型样品架，以正确定义高度
6.6.2	吸头在材料上方吸液	校准问题	检查管内液体体积，并确保其与实验方案设计软件中设定的体积一致
6.6.2	材料在移动过程中发生浸入	吸头携带过多多余材料	启用吸头触碰（tip touch）停靠选项；可选地，也可延长触碰时间
6.6.2	材料呈固态或过于黏稠，无法移液	材料的热响应特性	检查材料的热响应特性表征结果，并相应调整温控停靠位的加热/冷却温度
https://support.opentrons.com/en/articles/2687601-c-having-trouble-connecting-try-this-basic-troubleshooting

表2：问题排查表，列出了已识别的问题、可能的原因以及解决问题的方案。

补充文件。请点击此处下载该文件。

讨论

移取黏性材料，尤其是用于生物医学应用的水凝胶^{19,20,21,33,47}，是许多研究实验室中的常规操作，用于制备用户自定义浓度或不同浓度的稀释系列。尽管该操作具有重复性且执行过程相对简单，但目前大多仍依赖手动完成，样品通量较低¹⁸。本教程介绍一种开源工作站的操作方法，该工作站专为处理黏性材料设计，可实现黏性材料的自动化混合，从而可重复地生成目标浓度。该工作站针对水凝胶的移液操作进行了优化，通过集成温度控制模块（适用于温敏性材料）、正置换移液器（适用于黏性材料）以及可选的吸头触碰模块（用于去除吸头表面多余材料），实现自动化且高度可靠的样品处理。移液模块经过专门优化，可对黏性材料进行标准化和自动化处理。与空气垫移液器（图5a）相比，正置换移液器（图5b）在 dispensing 黏性材料时不会在吸头内残留材料，从而确保吸液和 dispensing 体积的准确性。可选的吸头触碰模块可去除吸头表面多余的样品材料（图5c,d），对于黏稠材料（例如 4% (w/v) 海藻酸钠）尤为有用。

该实验方案设计应用程序专为水凝胶实验而编程，可实现最多四种试剂（具有不同浓度）和最多两种稀释液的稀释操作。在此应用程序中，用户仅需选择目标浓度或连续稀释步骤，即可避免最终稀释比例计算过程中出现错误。所需的吸液和分液体积将自动计算，并保存至一个独立的文档文本文件中，随后填入实验方案脚本。该实验方案设计应用程序使用户能够完全控制所有实验参数（例如移液速度），并确保对关键参数进行内部记录。该应用程序会考虑储液区（例如孔）的液位高度，并相应调整吸液/分液高度，以防止吸头过度插入黏稠材料中。这一集成功能可避免液体在吸头外壁积聚，从而在整个实验过程中确保吸液和分液操作的可靠性。尽管该实验方案设计应用程序是为水凝胶的稀释步骤开发的，但它也可用于非黏稠液体的稀释，例如橙G染料。可通过仓库路径“/examples/publication-JoVE”访问的实验方案设计应用程序版本，即为本实验方案部分所解释并在视频中重点展示的版本。此版本将不再更新。然而，更新版本的实验方案设计应用程序可通过主仓库页面获取。校准终端最初由Sanderson⁴⁸ 开发，现已针对正向置换移液器的校准进行了优化。

如方案第4节所述，移液器以及容器必须在初始阶段进行校准。该校准过程对于定义并保存位置至关重要，这些位置随后用于计算移动增量。因此，成功执行方案依赖于明确定义的校准位置，因为错误的校准点可能导致吸头撞击容器。由于移液器的活塞位置必须手动校准，移液的准确性和精密度在很大程度上取决于所执行的校准质量。这些校准程序高度依赖用户对移液模块的操作经验，因此建议在初期由经验丰富的人员进行培训，以确保正确的校准流程。除了在移液模块上进行手动校准外，移液器本身也必须进行校准，以确保移液的准确性。建议至少每12个月校准一次移液器，以满足ISO 8655中规定的验收标准。如Stangegaard等¹⁶所述，可采用验证和确认方法对移液器校准进行内部评估。

为了获得可靠的数据集，必须使用高质量的试剂作为起始材料。这一点在水凝胶处理过程中尤为重要，因为试剂批次间的差异可能会影响本实验方案中所得的结果。除了批次间的变异外，小体积溶液配制过程中的微小变化也可能导致性能差异。为避免此类问题，建议配制较大体积的试剂溶液，并将其用于整个实验过程。

验证和确认过程依赖于使用染料来识别可靠的混合物。本方案描述了橙G的应用，但该通用方案和分析流程也可适用于荧光染料^49,50。使用橙G可降低对分光光度计的技术要求，并避免荧光染料在光照后发生漂白所需采取的防护措施。在实验过程中，所用材料未观察到染料溶解行为异常或形成聚集体的问题，但在其他材料中可能出现此类情况。潜在的聚集体形成以及染料与材料之间的相互作用可通过显微镜轻松检测。

本教程介绍的流程和技术为当前粘性材料的操作流程增加了自动化能力，可实现高度可靠的实验任务，同时最大限度地减少人工操作。所提供的故障排除表（表2）列出了已识别的问题，并给出了可能的原因及相应的解决方案。该工作站已成功应用于多种天然（如明胶、结冷胶、基质胶）和合成（如聚乙二醇[PEG]、Pluronic F127、Lutrol F127）聚合物材料的自动移液操作。特别是，该开放源代码工作站与专为粘性材料设计的开放源代码实验方案设计应用程序相结合，将对从事生物医学工程、材料科学和微生物学领域研究的科研人员具有重要帮助。

材料

本文使用的材料清单

姓名	公司	目录编号	评论
15 个反应管	Fisher Scientific， Inc. （美国）	14-959-53A
5 mL 管	Pacific Laboratory Products Australia Pty. Ltd. （澳大利亚）	SCT-5ML	大小取决于实验方案;也可以使用 Eppies（0.5、1、1.5 mL）或 Falcon 管（15、50 mL）;产品由 Axygen， Inc. 生产。https://www.pacificlab.com.au/shop/tubes-plastic/sct-5ml-tubewith-screwcap-blue-unassembled-5ml-self-standing/1/name
50 mL 反应管	Fisher Scientific， Inc.（美国）	14-432-22
70% w/w 乙醇	LabChem， Inc.（美国）	aja726-5Lpl
96 孔板	Thermo Fisher Scientific， Inc.（美国）		https://www.thermofisher.com/order/catalog/product/168055
藻酸盐	NovaMatrix		4200001 https://www.novamatrix.biz/store/pronova-up-lvg/
脱盐或超纯 （MilliQ） 水
明胶甲基丙烯酰 （GelMA）	合成		详细方案（包括材料和参考资料）可在 Loessner 等人（2016 年）的《自然方案》中找到。https://www.nature.com/articles/nprot.2016.037
苯基-2,4,6-三甲基苯甲酰膦酸锂 （LAP）	Sigma-Aldrich， Inc. （美国）	900889
M4 和 M5 内六角扳手	OpenBuilds， inc. （美国）	179， 190	也可在每家五金店买到。https://openbuildspartstore.com/allen-wrench/
OrangeG	Fisher Scientific （美国）	O267-25	https://www.fishersci.com/shop/products/orange-g-certified-biological-stain-fisher-chemical/O26725
磷酸盐缓冲盐水 （PBS）	Thermo Fisher Scientific， Inc. （美国）	14190-144	交替使用：PBS 片剂：18912014 （Thermo Fisher Scientific）
设备
温度对接的铝块	Ratek Instruments Pty. Ltd.（澳大利亚）	可提供适用于不同试管尺寸的	SB16
分析天平	Sartorius AG （德国）	ED224S
开源液体处理机器人：商业产品	Opentrons Laboratories， Inc. （美国）	OT-One S Pro	https://shop.opentrons.com/products/ot-one-pro
开源液体处理机器人：开放式源硬件	按照开源方法		硬件和软件文件可在GitHub和Zenodo上免费访问（提供链接;提供构建说明。https://github.com/SebastianEggert/OpenWorkstation. https://zenodo.org/record/3612757#.XipEjBV7F24
外置活塞式移液器：MicromanE	Gilson， Inc. （美国FD10006		取决于所需的大小。https://www.gilson.com/default/shop-products/pipettes/positive-displacement.html
分光光度计	BMG LABTECH GmbH （德国）	CLARIOstar
小贴士：毛细管活塞	Gilson， Inc. （美国）	F148180	取决于所需的尺寸。https://www.gilson.com/default/shop-products/pipette-tips.html?technique_en_ww_lk=191