This article presents a protocol for seeding scarce population of cells using pipette-tips to droplet microfluidic devices in order to provide higher encapsulation efficiency of cells in droplets.
Amongst various microfluidic platform designs frequently used for cellular analysis, droplet-microfluidics provides a robust tool for isolating and analyzing cells at the single-cell level by eliminating the influence of external factors on the cellular microenvironment. Encapsulation of cells in droplets is dictated by the Poisson distribution as a function of the number of cells present in each droplet and the average number of cells per volume of droplet. Primary cells, especially immune cells, or clinical specimens can be scarce and loss-less encapsulation of cells remains challenging. In this paper, we present a new methodology that uses pipette-tips to load cells to droplet-based microfluidic devices without the significant loss of cells. With various cell types , we demonstrate efficient cell encapsulation in droplets that closely corresponds to the encapsulation efficiency predicted by the Poisson distribution. Our method ensures loss-less loading of cells to microfluidic platforms and can be easily adapted for downstream single cell analysis, e.g., to decode cellular interactions between different cell types.
In recent years, the use of microfluidics as a robust and versatile platform for cellular analysis at the single cell level has rapidly increased1. These platforms provide high-throughput screening of single cells and biological molecules with high precision and sensitivity using very small sample volume2,3,4. Among different types of microfluidic designs, droplet-based platforms enable high-throughput analysis of single cells by isolating them in an aqueous phase droplet surrounded by an immiscible phase that allows precise and accurate control over the cellular microenvironment5,6. Droplet-based microfluidics gives the flexibility to isolate single or multiple-cells in, both, aqueous and hydrogel droplets and is valuable in probing complex cellular behavior, such as protein secretion or cellular interactions7,8,9. Signaling and cross-talk amongst immune cells can be influenced by interactions with other cells in the microenvironment10. Isolation of single cells in droplets provides an effective noise-free analytical laboratory, free from the influence of external environmental factors for more efficient and accurate results11,12. Modifying the design of a droplet-microfluidic platform with multiple inlets allows the encapsulation of multiple cell types to study cellular interactions via cell-pairing12,13.
The process of encapsulation of cells in droplets is random and the rate of encapsulation of cells can be statistically determined using the formula for the Poisson distribution14,15. This rate of encapsulation can be estimated by considering the average rate of the arrival of cells at the droplet junction and assuming that the arrival of each cell is independent from the arrival of other cells16. Even though independent cell arrival cannot be guaranteed, in cases of sparsely distributed cells, the assumption of independence can be considered and the probability of a droplet containing one or more cells can be predicted as a function of the number of cells present in each droplet and the average number of cells per droplet16,17. Since this estimation of cellular encapsulation in droplets is dependent on the number of cells present in each droplet, one can suggest that increasing the concentration of the cells at the inlet will increase the average number of cells present in each droplet16. Therefore, to ensure single cell encapsulation, the cell concentrations must be reduced but this often leads to a large number of empty droplets18.
Loss of cells during loading by either attachment, sedimentation, and/or clumping in the syringe, tubing, or production device is a common drawback responsible for the deviation of actual encapsulation values from the predicted encapsulation values19. This problem gets further exaggerated when seeding rare immune cells or clinical specimens as they are already scarce in population and the encapsulation of only a few cells, much lower than expected, does not provide sufficient data for experimental analysis. Plasmacytoid dendritic cells (pDCs) are a rare subset of immune cells that only constitutes approximately 0.2 – 0.6 percent of the entire white blood cell population20. These cells secrete massive amounts of type I interferons upon activation and thereby play a critical role in immune responses21. When studying the cellular behavior of such rare cells in droplets, it is imperative to prevent cell loss during cell seeding and encapsulation22. There are several design related developments that have ensured the encapsulation of single cells in droplets using active encapsulation methods that utilize different physical forces such as acoustic or electrical forces for generation of droplets containing single-cells23,24. However, these methods have their own limitations in terms of droplet production16.
In this study, we established a robust and straightforward method that circumvents the shortcomings of traditional methods for loading single or multiple cells to microfluidic devices. Our method, inspired by Rho et al., utilizes differently-sized pipette tips for seeding small volumes of rare immune cells to droplet microfluidic platforms without significant sample loss and yielded results that are coherent with theoretical predictions25. This methodology can be easily and successfully adapted for several applications involving droplet-based microfluidics and applied for a wide variety of cell types or even microparticles.
1. 3-Inlet Polydimethylsiloxane (PDMS) Device Fabrication
2. Loss-Less Cell Encapsulation
3. Cell Pairing
For our experiments, we used a 3-inlet PDMS based microfluidic device with the height of 25 microns (Figure 1). In this device setup, we used the outer inlet for flushing the oil with surfactant and the two inner inlets for flushing the aqueous phases with cell suspensions or media. After generation and collection, the droplets are incubated for a couple of hours off chip before downstream analysis using flow cytometry. During the incubation period, serum components present in the media can interact with the surfactant and cause droplets to become unstable and disintegrate. It is therefore important to add an optimized concentration of surfactant to the fluorinated oil. We tested the stability of the monodispersed droplets containing hematopoietic serum-free culture media supplemented with 2% human serum with different concentrations of surfactant in fluorinated oil. It can be inferred from Figure 2 that these monodispersed droplets are highly stable for up to 24 hours when at least 3% surfactant is added to the oil phase. Similar results were obtained with RPMI media with and without the addition of 10% FCS (data not shown). Therefore, droplet stability is highly dependent on optimal surfactant concentrations when working with different sources of culture media and serum components.
To demonstrate the encapsulation efficiency of our approach we first seeded the cells using tubing connected to syringes, which is the most conventional approach for seeding cells (Figure 3A). We harvested Jurkat T cells at different concentrations of 1.0×106 cells/mL, 2.0×106 cells/mL, and 4.0×106 cells/mL and obtained an encapsulation efficiency that was lower than predicted values (Figure 3B). At 1.0×106 cells/mL, the fraction of droplets that contained a single cell was 2.5%, which did not increase even upon using higher cell concentrations. To increase the cell-loading efficiency, we modified our previous approach and mounted the tubing at half the length to an elevated tripod and loaded the cell suspension in the half that was attached to the PDMS device (Figure 4A). Using this approach, we encapsulated Jurkat T cells at different concentrations of 1.0×106 cells/mL, 2.0×106 cells/mL, and 4.0×106 cells/mL, and also rare pDCs at different concentrations of 1.0×106 cells/mL, 2.0×106 cells/mL, and 12.0×106 cells/mL. We expected improved encapsulation rates by preventing cell sedimentation with this method. However, at all the concentrations tested, the experimental results were much lower that the predicted Poisson values (Figure 4B and Figure 4C).
Using our tip-loading approach we optimized our cell encapsulation rates to obtain experimental results coherent with the statistically predicted values (Figure 5A). For different concentrations of Jurkat T cells, the obtained encapsulation efficiency matched our calculated values at all concentrations (Figure 5B). Remarkably, even with adherent cells like A549 tumor cells, which tend to clump, we observed a slightly improved encapsulation efficiency at a cellular concentration of 1.0×106 cells/mL (Figure 5C). We also assessed the efficacy of our system to encapsulate less available and scarce pDCs at different cell concentrations of 1.0×106 cells/mL, 3.0×106 cells/mL, and 13.0×106 cells/mL (Figure 5D). To facilitate the loading of possibly larger volumes exceeding 200 µL, e.g., when working with cell lines or more abundant primary immune cells, we also investigated the cell encapsulation efficiency by using 1000 µL tips (blue). We demonstrated that these 1,000 µL tips gave a similar encapsulation efficiency in comparison to the 200 µL tips (yellow) (Figure 5E).
Dependent on the chip design and research question at hand, our tip loading technique can be used to load cells through either one inlet, for probing into cellular heterogeneity, or multiple inlets in parallel, for decoding cellular interactions. We compared the loading of Jurkat T cells (at a concentration of 10.0×106 cells/mL) from one inlet to two differently labelled populations of Jurkat T cells (at a combined concentration of 10.0 x106 cells/mL) from two inlets (Figure 6A and Figure 6B). During encapsulation, the droplets were generated using ultra-low gelling temperature agarose and gelled after production to form agarose hydrogel beads that allowed subsequent downstream analysis via microscopy and flow cytometry (Figure 6C and Figure 6D). Microscopic analysis revealed that cell pairing was achieved at different combinations indicating for high throughput cell pairing (Figure 6C). Furthermore, analysis of the same population of hydrogel beads by flow cytometry revealed that beads without cells could be separated from beads with cells based on the distinct forward (FSC, size) and sideward (SSC, granularity) scatter pattern (Figure 6D). Gating on the population of beads without cells confirmed a lack of cell encapsulation by the absence of fluorescent signals. Additionally, gating on the bead population with cells revealed the existence of multiple sub-populations indicative for the encapsulation of differently labeled Jurkat T cells. Our results demonstrate that efficient cell pairing can be achieved, based on both microscopic and flow cytometric analysis, and showed a slightly increased encapsulation efficiency compared to the Poisson prediction (Figure 6E).
Figure 1. PDMS based droplet microfluidic device with three inlets and one outlet. The device consists of three inlets for continuous oil phase, cell culture media, and cell suspension, respectively. The generated droplets are collected at the outlet. The samples flow laminarly to the flow-focusing junction where they are encapsulated in droplets. At the inlets, filter structures hold large particles like protein or cell aggregates back. The diameter of the gaps in the filter structure are indicated by blue lines. The diameter of the channels at the production nozzle are indicated by red lines. The channel height on the entire chip was 25 µm. Please click here to view a larger version of this figure.
Figure 2. Droplet stability over 24 hours. The graphs show the area of droplets containing hematopoietic serum-free culture media + 2% human serum, over time for three different concentrations of surfactant A) 0.5% B) 3% C) 5%. Please click here to view a larger version of this figure.
Figure 3. Tubing based cell loading approach. Jurkat-T cells are loaded at different concentrations to the device using a syringe connected to tubing. A) The illustration shows the experimental setup B) The cell encapsulation rate as determined by light microscopy. Dots: experimentally determined values; Closed lines: Poisson distribution. Error bar represents standard error of mean. Please click here to view a larger version of this figure.
Figure 4. Encapsulation of various cell types at different concentrations using a vertical tube loading approach. Jurkat T cells and pDCs (of different concentrations) were encapsulated to determine the efficiency of cell encapsulation. A) The illustration shows the experimental setup for the vertical tube loading approach. B) The graph shows the encapsulation efficiency of Jurkat T cells. C) The graph shows the encapsulation efficiency of pDCs. Dots: experimentally determined values; Closed lines: Poisson distribution. Error bar represents standard error of mean. Please click here to view a larger version of this figure.
Figure 5. Tip loading approach to encapsulate different cells types. A) Schematic illustration of the tip loading technique. B) The graph shows the Encapsulation efficiency of Jurkat cells. C) The graph shows the encapsulation efficiency of A549 cells. D) The graph shows the encapsulation efficiency of pDCs. E) The graph shows the encapsulation efficiency of Jurkat T cells using 200 µL pipette tips (yellow) and 1,000 µL pipette tips (blue). Dots: experimentally determined values; Closed lines: Poisson distribution. Error bar represents standard error of mean. Please click here to view a larger version of this figure.
Figure 6. Cell pairing in droplets. A) Schematic illustration of the tip loading approach for pairing distinct cells from 2 inlets in droplets. B) The graph shows the encapsulation of Jurkat T cells using one inlet or two inlets in parallel. The cell concentration for one tip is 2.0×106 cells/mL and the cell concentration for two tips both 1.0×106 cells/mL. Dots: experimentally determined values; Closed lines: Poisson distribution. C) The Fluorescence microscopic overlays of hydrogel beads and labelled Jurkat T cells. D)The graph shows the flow cytometric analysis of paired cells in agarose hydrogel beads. The plot demonstrates both forward scatter and sideward scatter. E) Comparison of cell numbers in agarose hydrogel beads as detected by fluorescence microscopy and flow cytometry. Bars: mean value; Whiskers: standard error of mean, n ≥ 4. Error bar represents standard error of mean. Please click here to view a larger version of this figure.
In this protocol, we have demonstrated an efficient and straightforward technique to load and encapsulate cells in droplets for high-throughput, single-cell analysis and to perform controlled cell pairing for cellular interaction studies. Furthermore, we have compared several conventional approaches to load cells to microfluidic devices and showed that our tip loading approach is a more efficient technique in comparison to other methods.
Studying clinical specimens or rare cell types scarce in number by droplet-based microfluidics possess some inherent challenges. Like we have also demonstrated, cells tend to sediment in syringes and surface of the tubing, thereby, preventing cellular encapsulation to conform to the predicted values. To evade this problem, some groups use stirring bars in the syringes. However, when using rare and limited cell populations, the total cell volume is also limited, thereby, limiting the use of large syringes and stirring bars. Furthermore, we also replaced more commonly used tubing with Teflon coated tubing to prevent cell attachment but this method did not improve the results and if the tubing is too long, the problem of cell attachment aggravates (data not shown). Alternatively, we used a vertical tube loading approach where the cells were loaded in the tubing and not in the syringe to prevent the loss of cells in large syringe volumes. Using this technique, cells with small sample volume can be loaded, e.g., pDCs which are rare and limited. Also, the sample from the tubing is loaded to the device vertically to prevent cell sedimentation. The tubing used for cell seeding has small dimensions and can be compared to microchannels. The flow in the tubing is pressure driven and follows a parabolic velocity profile26. This implies that the maximum flow velocity is at the center of the tubing and minimum velocity is at the edges of the tubing27. When flushing a population of cells through the tubing, the velocity gradient causes the cells to be pushed towards the edges where they settle down because the velocity at the boundary is close to zero. The sedimentation or settlement of cells in the tubing, thereby, reduces the encapsulation efficiency as shown in the representative results where the experimental data did not match with the predicted model.
Another commonly adapted solution used by scientists, working with droplet microfluidics, is to increase the density of the cell culture media by addition of density matching reagents such as Iodinaxol to prevent cell sedimentation in syringes19. However, density matching reagents can influence cellular behavior and adversely affect the cytokine secretion by cells (data not shown)28.
Even though several small and big modifications in conventional cell loading techniques showed slight improvements in encapsulation efficiencies, the obtained experimental results still did not match the theoretical calculations. However, with the tip loading approach we could the overcome the limitations of previous methods and encapsulation efficiency governed by Poisson statistics. This technique is not only advantageous for loading suspension cells but can also be applied for loading adherent cells, such as primary keratinocytes and A549 to microfluidic chips. When using abundant cell lines, for example A549, K562, etc., larger sample volume can be used. Therefore, depending on the volume of the sample, different sized pipette-tips can also be used and this simple technique can be adapted for both single-cell encapsulation and multiple cell encapsulation.
While low cell concentration is required to ensure the encapsulation of single cells in droplets, higher cells concentrations are desired to increase the average number of cells encapsulated in each droplet for studies related to cell pairing. There are several single-cell methods that have been previously described to pair immune cells on microfluidic chips or microfabricated nanowells29,30,31. In droplet microfluidics, Poisson statistics dictates that 1:1 cell pairing for two different cell types can be achieved at optimal cell concentrations. Based on the Poisson prediction, there is also a probability that droplets might contain other combinations. While 1:1 cell pairing can be desirable to study cellular interactions at single cell level and results in increased cellular understanding, multiple cell pairing also has major advantages. It allows to comprehend the influence of multiple cells of one cell type on the other cell type. Cross talk between different immune cells help to generate an effective immune response against several infections and pathogens and also adds robustness to our immune system32. As such, cellular communication can be interrogated with high precision in distinct contexts, e.g., 1:1, 2:1, 1:2, 2:2, 3:1, etc. yielding increased understanding on how single or pairs of cells control the induction of immune responses. This is particularly interesting to study for example the capacity of natural killer cells or cytotoxic T cells to serially kill their respective target cells.
As discussed, for multiple encapsulation of cells in droplets, higher cell concentrations are desired. However, when loading cells from one inlet for cell encapsulation, higher concentrations of cell sample can cause cells to aggregate at the inlet. This results in lower encapsulation rates and higher deviation from the theoretical values. To evade this problem, the cells can be loaded from two separate inlets as well. Theoretically, it would be possible to develop other microfluidic devices with multiple inlets to achieve even higher levels of cell encapsulation where an average on x number of cells is warranted. In this study we investigated the encapsulation efficiency of Jurkat T cells when loaded from both one inlet and two inlets using the same total concentration and obtained similar encapsulation efficiency. This modification allows researchers to pair different cell types on chip.
While this method aids in loading cells to microfluidic devices without significant loss of cells, there are certain precautions that need to be kept in mind. When filling the syringes with mineral oil and aspirating the cell sample in pipette tips, incorporation of air bubbles should be avoided and the entire system should be air-free. It is also important to keep in mind that the mineral oil should not mix with the sample. Pipette tips, containing samples, should be inserted firmly in the inlets of the microfluidic device, with utmost precaution, to prevent leakage and further incorporation of air bubbles. To summarize, tip-loading is a straightforward, yet, robust technique that allows for high-throughput analysis of cellular behavior through cell encapsulation without significant loss of cells in a cost-effective manner. When used with optimal sample concentrations at the inlet, this approach of loading cells with pipette-tips is very flexible and can be adapted for different cell types, especially for rare primary immune cells, to obtain higher encapsulation efficiency, close to predicted models.
The authors have nothing to disclose.
We thank the Eindhoven University of Technology for generous support.
1H,1H,2H,2H-Perfluoro-1-octanol | Sigma-Aldrich | 171468-5G | |
1H,1H,2H,2H-Perfluorooctyltriethoxysilane | Fluorochem/UK | S13150 | Silane (toxic) |
Agarose (Ultra-low Gelling Temperature) | Sigma-Aldrich | 9012-36-6 | |
BD Wegwerpspuiten met Luer-Lok-punten | Fisher Scientific | 10630694 | Syringe |
Biopsy Punch 1.2 mm | Harris Uni-Core | ||
Cell Proliferation Dye eFluro 670 | eBioscience | 65-0840-85 | |
CellTrace CFSE | Invitrogen | C34554 | |
CellTrace Far Red Cell | Invitrogen | C34564 | |
Eppendorf Tubes | Eppendrof Tubes | Safe-Lok tubes 1 mL and 2 mL | |
Glass Slide | Sigma Aldrich | CLS294775X38-72EA | Corning microscope slides, plain L × W 75 mm × 38 mm |
Harvard Pumps | Harvard Apparatus | C-400750; C-400727 | Syringe pumps |
HFE-7500 3M Novec Engineered fluid | Fluorochem/UK | 51243 | Flourinated oil |
Kai Biopsy Punch 5 mm | Amstel Medical | 1980130 | |
Luer stub | Instechlabs/USA | LS20S | Luer stub, 20ga (pink) x 0.5in (12mm), non-sterile |
Mineral oil (Light) | Sigma Aldrich | M8410-1L | |
Phosphate buffered saline | Sigma-Aldrich | P4417-50TAB | Tablets |
Pico-Surf 1 (5%in Novec 7500) | Sphere Fluidics | 020317-09 | Surfactant |
Plasma Asher | Emitech | K1050X | Plasma asher |
RPMI 1640 Medium | Gibco | 11875093 | |
Silicone Elastomer Base 184 | Sylgard | 9355218 | PDMS base |
Silicone Elastomer Curing Agent | Sylgard | 9355218 | Curing Agent |
Stainless steel catheter coupler | Instechlab/USA | SC20/15 | 20ga x 15mm, non-sterile |
TFE Teflon Tubing | Sigma-Aldrich | 58696-U | PTFE Tubing L × O.D. × I.D. 50 ft × 1/16 in. × 0.031 in. |
Thinky mixer ARE-250 | EX-4025F | Conditioning mixture |