Recent advancements in tissue dissociation techniques for cell manufacturing single-cell analysis and downstream processing

Abstract

Tissue dissociation into single-cell suspensions is a critical technique for cell therapy manufacturing, single-cell analysis, and downstream processing. The process is traditionally carried out via enzymatic and mechanical dissociation of the tissue using standard laboratory techniques, but there have also been efforts made to translate these techniques onto microfluidic devices, as well as efforts into performing nonenzymatic digestion. Conventional methods face a number of challenges regarding viability, yield, long processing times, as well as the potential for the processing to create artifacts that can distort downstream analyses. In this review, we discuss the current state-of-the-art technology, go over advancements made in recent years to improve technologies and protocols related to tissue dissociation, and then consider the future of the technique, highlighting ways in which we envision it could be improved.

Graphical abstract

Graphical Abstract.

The graphical abstract was created with Biorender.

Significance Statement.

The current bottleneck in manufacturing of tissue-engineered and cell-based regenerative medicine therapies or single-cell isolation for downstream applications is the lack of rigorous, standardized, and validated systems that enable the reproducible dissociation of tissues into highly purified cell population(s) prior to initiating the subsequent processes. Here, we discuss the status quo of tissue dissociation technologies and protocols, highlight new technologies and improvements made since the start of the decade, and provide insights into what is needed to further optimize these technologies.

Introduction

Tissues are complex systems of cells that display a high degree of heterogeneity. Not only can they be composed of multiple cell types that behave very differently, but the stochiometric nature of interactions and behaviors of cells within the tissue mean that a high degree of heterogeneity can even be observed within populations of the same cell type within a tissue.^1-5 As such, in order to gain a more holistic understanding of the tissue, it is necessary to characterize this heterogeneity, which has led to the field of single-cell analysis. To perform such analyses, dissociating the tissue into a single-cell suspension is a necessary first step, which entails breaking down the extracellular matrix and cell–cell junctions holding the tissue together. Tissue dissociation is thus vital to several downstream applications, including single-cell sequencing,^6-12 flow cytometry analysis,^13-17 establishing cell lines,¹⁸ cultivating organoids,^19,20 isolating specific cell types,^21-24 and therapeutic cell replacement.^25,26 However, the process of preparing single-cell suspensions is one of the greatest sources of technical errors and variations in single-cell studies, and so, it is crucial that these techniques are well optimized and appropriate to the application.^5,9,27

In similarity, the current bottleneck in manufacturing of tissue-engineered and cell-based regenerative medicine therapies is the lack of rigorous, standardized, and validated systems that enable the reproducible dissociation of tissues into an optimal cell population(s) to a high degree of purity prior to initiating the manufacturing process.

Tissue dissociation is traditionally done through a mixture of mechanically mincing the tissue and then enzymatically breaking down the resultant tissue fragments, often in combination with mechanical disruption through agitation.^1,7,8,28-42 A number of different enzymes have been used for tissue dissociation, including collagenase,^{7,8,29-37,39,41,42} dispase,^{7,30,32,35,36} trypsin,^{29,31,33,38,40,43} papain,²⁵ and hyaluronidase.^{8,29,32-34,41,42} Additionally, the chelating agent ethylene diamine tetra-acetic acid (EDTA) has also been used to dissociate tissue in a similar manner to enzymatic methods.^29,40,44,45 While these traditional approaches have been used widely, there are several drawbacks associated with them. The enzymatic digestion can take a long time, often requiring hours of work—some protocols even call for overnight digestion—limiting the speed at which analyses can be done and increasing the time in which contamination and errors can occur.^{5,7,32,34,37,38,41,42} Furthermore, the enzymes used for dissociation can often damage the very cells the investigator is attempting to isolate for study, leading to reduced viability^1,7,8 and destruction of cell surface proteins,^1,5,44,46 jeopardizing the subsequent use of these cells for manufacturing, and increasing artifacts in downstream analysis.^5,28,47,48 These shortcomings can lead to tradeoffs, such as shortening digestion to avoid compromising the viability of cells for downstream processing, albeit with lower recovery. Additionally, the heterogeneous nature of samples/specimens both within a given type of tissue and between tissues with vastly different functions has led to these techniques being developed independently and separately for different types of tissues, creating a situation with very little in the way of a standardized protocol. As a result, much work has gone into optimizing the processes as well as developing novel methods such as incorporating microfluidics into the workflow^49-57 or exploring nonenzymatic alternatives for digesting the tissue.^{49,50,52,58-63}

This review will discuss the status quo of tissue dissociation technology and highlight the new technologies and improvements made since the start of the decade for the dissociation of complex tissue into single-cell suspensions (Table 1). While much work is also being done to further the dissociation of organoids,^25,26,64,65 as the challenges presented by these systems are not the same as with more complex native tissues, work in this related field will not be discussed in detail here. We will divide the review into three sections, highlighting advancements in traditional enzymatic dissociation methods, microfluidic adaptations of these methods, as well as the development of nonenzymatic alternatives. We will discuss the state of each of these approaches, the advantages, challenges, and drawbacks of each (Table 2) and finally present the overall state of tissue dissociation and where the technology can and should be improved in future research.

Table 1.

New tissue dissociation technologies since the start of the decade.

Technology	Dissociation type	Tissue type	Dissociation efficacy	Viability	Time	Source
Optimization of a Clinically Relevant Chemical-Mechanical Tissue Dissociation Workflow for Single-Cell Analysis	Enzymatic Mechanical	Bovine Liver Tissue MDA-MB-231 Breast Cancer cells	37%-42% (Enzymatic only) (Bovine Liver Tissue) 92% ± 8% (Enzymatic and Mechanical Dissociation) (Bovine Liver Tissue)	>90% (MDA-MB-231)	15 min	(29)
Optimized protocol for tissue dissociation for single-cell RNA sequencing	Mechanical Enzymatic	Triple-negative human breast cancer tissue	2.4 × 106 viable cells	83.5% ± 4.4%	>1 h	(8)
Optimized tissue dissociation protocol for single-cell RNA sequencing of skin biopsies	Mechanical Enzymatic	Human skin biopsy	∼24 000 cells/4 mm biopsy punch	92.75%	∼3 h	(7)
Automated Mechanical Tissue Dissociation Device	Mechanical Enzymatic	Mouse Lung Tissue Mouse Kidney Tissue Mouse Heart Tissue	1 × 105 to 6 × 105 cells (Lung) 1 × 106 to 1.5 × 106 cells (Lung) 1 × 105 to 5 × 105 cells (Heart)	60%-80% (Lung) 60%-80% (Kidney) 50%-60% (Heart)	∼1 h	(39)
Mixed modal microfluidic platform for tissue dissociation from minced tissue	Microfluidic Mechanical Enzymatic	Mouse Kidney Tissue Mouse Breast Tumor Tissue Mouse Liver Tissue Mouse Heart Tissue	∼20 000, 1700, 900 cells/mg tissue (epithelial, leukocyte, endothelial) (kidney) ∼9000, 900, 300 cells/mg tissue (epithelial, leukocyte, endothelial) (Breast tumor) ∼2400, 3500, 200 cells/mg tissue (hepatocyte, leukocyte, endothelial) (liver) ∼18 000, 1500, 1500 cells/mg tissue (Cardiomyocyte, leukocyte, endothelial)	∼95%, 60%-90%, 60%-90% (epithelial, leukocyte, endothelial) (Kidney) ∼70%-80%, 50%-60%, ∼80% (epithelial, leukocyte, endothelial) (Breast Tumor) ∼85%-90%, 70%-85%, 70%-85% (hepatocyte, leukocyte, endothelial) (liver) ∼70%-90%, 85%, 85% (cardiomyocyte, leukocyte, endothelial) (heart)	1-60 min	(55)
Optimized mixed modal platform for tissue dissociation from minced tissue	Microfluidic Mechanical Enzymatic	MCF-7 Breast cancer cell Mouse Kidney Tissue	∼170% of control (MCF-7) ∼400 000, ∼75 000, ∼3500, ∼10 000 cells/mg tissue (total, epithelial, leukocyte, endothelial) (kidney)	∼70% (MCF-7) ∼90%, ∼90%, ∼50%, ∼50% (total, epithelial, leukocyte, endothelial) (Kidney)	20-60 min	(56)
Microfluidic platform for dissociating clinical scale tissue samples into single cells	Microfluidic Enzymatic	Human Placental Tissue Human Endometrial Tissue Rat Liver Tissue Rat Adipose Tissue	2262 viable cells/mg tissue (average across all tissue types)	Not reported	45 min to 2 h	(57)
Electric Field Facilitated Rapid Dissociation	Electrical	Bovine liver tissue MDA-MB-231 triple-negative breast cancer cells Human Clinical Glioblastoma Tissue (GBM)	95% ± 4% (bovine liver tissue) >5× higher than traditional enzymatic mechanical method (GBM)	90% ± 8% (MDA-MB-231) ∼80% (GBM)	5 min	(58)
Ultrasound High Frequency Sonication Dissociation	Ultrasound Enzymatic	Bovine liver tissue MDA-MB-231 triple-negative breast cancer cells	53% ± 8% (sonication alone) 72% ± 10% (sonication plus enzymatic) (bovine liver tissue)	91%-98% (Sonication only, MDA-MB-231)	30 min	(59)
Enzyme-Free, Cold-Process Acoustic Method using Bulk Lateral Ultrasound for Tissue Dissociation	Ultrasound	Mouse heart tissue Mouse lung tissue Mouse brain tissue Mouse B16 melanoma tissue	3.6 × 104 live cells/mg (heart) 1.4 × 104 live cells/mg (lung) 1.4 × 104 live cells/mg (brain) 2.0 × 105 live cells/mg (lung)	36.7% (heart) 26.0% (lung) 51.5% (brain) 74.0% (tumor)	2 min	(60)
Novel Enzyme-Free approach to generate single cell suspensions for miRNA sequencing	Mechanical	Human Liver Human Spleen Human Cortex	4.80 × 105 cells/mL (liver) 2.16 × 106 cells/mL (spleen) 6.45 × 104 cells/mL (cortex)	45% (liver) 90% (spleen) 64% (cortex)	Not listed	(61)

Table 2.

Pros and cons of different approaches to dissociation.

Method		Pros	Cons
Enzymatic		Current standard Well established and documented with wide range of tissues Automated options and kits available No complex instrumentation required to perform or set up	Long processing times (>1 h, even overnight in some cases) Often induces stress response Poor viability for some tissue types
Microfluidic		Allows for modular design incorporating multiple functionalities Faster processing times (<1 h) High viability and yield from small sample volumes Can avoid stress response Can incorporate other methods easily	Limited work performed so far Tested with small range of complex tissues Devices are often prone to clogging and fouling Often requires some preprocessing Complex instrumentation needed to create new devices
Nonenzymatic	Mechanical	Well established and documented Easy to incorporate with other methods No complex instrumentation required to perform or set up Can preserve subpopulations of cells lost in enzymatic digestion	Typically provides low yields and viabilities when not paired with other methods Induces stress response and artifacts in downstream processes
	Electrical	Potential for higher dissociation efficiency than enzymatic methods High viability Very short processing times (<30 min) Low stress response	Very limited research performed at this time Tested with very small range of tissues so far Requires custom instrumentation set ups
	Ultrasound	Very short processing times (<30 min) Viability and yields at or above those from enzymatic methods Low stress response Relatively simple instrumentation	Very limited research performed at this time Tested with very small range of tissues so far

Traditional enzymatic and chemical approaches

Traditional approaches to tissue dissociation include finely mincing the tissues first using scalpels or scissors followed by enzymatic treatment to break down the cell–cell junctions and extracellular matrices. These enzyme treatments often need to be carried out at 37 °C, which can induce cellular stress responses, dramatically altering the transcriptome and thereby creating artifacts that appear in downstream analyses such as single-cell RNA sequencing (scRNA-seq), one of the most powerful and widely employed forms of single-cell analysis.^5,66,67 Another challenge these approaches face is that they can take hours to complete. This can lead to poor viability and/or recovery, as the aggressive form of dissociation these techniques employ can cause significant cell damage.⁸ In recent times, the potential of cold active protease as a method for dissociating at lower temperatures to avoid the stress response has started to be looked into; however, research in this area is still very early and has not yet been conducted with a wide range of tissues.⁶⁷ As such, much of the recent work in these approaches has sought to further optimize the processes for specific tissue types to best obtain a yield of high-quality cells for downstream processing or manufacturing.

Other areas where advancements are being made beyond optimization of protocols is the adoption of automated tissue dissociation devices using preset programs to automatically cycle through the dissociation protocol without human intervention. Of particular note are commercial devices which have been widely described in the literature to dissociate a broad range of tissues including human pancreatic and lung tumor tissue,^14,68,69 human gastric tissue,^6,70 human skin,⁷¹ mouse brain tissue,^21,72 mouse liver tissue,^73,74 and mouse muscle tissue⁷⁵ among many others. These platforms perform automated shaking and heating (within the maufacturer's proprietary tubes), using either the manufacturer's pre-existing programs or custom, user-designed programs.The investigator simply needs to load the desired tissue and appropriate enzymes to achieve dissociation into the tube and start the appropriate program. However, while these devices are useful, they can be quite expensive, they face the same issues as other enzyme-based digestion protocols, and they still need user-adjusted protocols when applied beyond the range of tissues for which the manufacturer has predefined kits. As such, it will not always be practical to adopt this platform for all users and use cases.

Welch et al.²⁹ have reported on developing an optimized workflow for chemical–mechanical isolation of cells from tissues. Using 10 mg of cryopreserved bovine liver tissue, they first tested and optimized the dissociation using several different proteolytic dissociation enzymes, as well as testing different digestion durations. They found an optimal enzymatic dissociation occurred at 15 min using collagenase-based digestion solutions, with the 1% collagenase and hyaluronidase solution achieving 42% dissociation efficiency and the 1% collagenase and pronase solution achieving 38% dissociation efficiency compared to the expected total cells calculated from a model. They then demonstrated that by combining the collagenase and pronase treatment with automated mechanical shaking on a heated incubator to further agitate and mechanically dissociate the tissue during the enzymatic digestion, they could achieve a 92% ± 8% dissociation efficiency. Next, they tested the impact this treatment had on cellular viability by using the protocol on cultured MDA-MB-231 triple-negative breast cancer cells and performing a viability assay, which showed over 90% of the cells were viable. While these results suggest this protocol yields a high recovery while preserving a high level of viability, it is important to note that the dissociation data was obtained with a primary tissue, while the viability data was derived with a cancer cell line. Furthermore, since dissociation efficiency was only assessed with cryopreserved tissue, results could differ markedly when performed on fresh tissue samples. Somewhat surprisingly, given that the goal of the study was to obtain single-cell suspensions to perform scRNA-seq, no actual testing of scRNA-seq with the dissociated tissues was conducted, so the effects of heat-stress and other artifacts that the protocol might have induced cannot be assessed from the work that was performed. The authors also noted that further investigation will be needed to ascertain how well this protocol translates to different types of tissues, especially fibrotic and cross-linked tissues.

Frolova et al.⁸ also performed a study to optimize a protocol for the dissociation of tissue for use in single-cell RNA sequencing. They collected tumor samples from 15 patients with triple-negative breast cancer and first mechanically minced the samples into 1-2 mm fragments and then compared enzymatic dissociation with three different commercial mixtures (Tumor Dissociation Kit, Liberase, and Collagenase/hyaluronidase) for 45 min followed by a 15 min DNase step in a BioSan TS-100 Thermo-Shaker at 37 °C and 800g. They found that the collagenase/hyaluronidase produced the highest yield of viable cells of the three kits, with an estimated viability of 84.9%, compared to 80.1% for the Tumor Dissociation kit and 30.2% for the Liberase. They then further optimized the collagenase/hyaluronidase digestion by comparing the effects of different culture media and digestion times and found that the optimal yield of about 2.4 million viable cells was achieved after 45 min of collagenase/hyaluronidase digestion in either DMEM or EpiCult Media. While a 60-min digestion yielded similar results, digestions of 30 min and 12 h both reduced the yield of viable cells. The authors demonstrated the reproducibility of these results by achieving an average viability of 83.5% ± 4.4% from 10 different biopsy samples, although they also discovered that the viability was reduced to below 80% for samples with dense fibrosis. Importantly, they also validated their dissociation method/protocol by performing a practical demonstration of running a scRNA-seq experiment with the resultant single-cell suspension. This work demonstrates the use of optimization to obtain a high yield of viable cells for scRNA sequencing from a small sample of tissue.

In another study, Burja et al.⁷ developed an optimized protocol for dissociating skin tissue into single-cell suspensions for scRNA-seq applications. A 4 mm biopsy punch of tissue was strained and then underwent multiple digestion steps with dispase, collagenase, and trypsin. This approach enabled the authors to obtain a cell yield of ∼24 000 cells per 4 mm biopsy punch with a viability of 92.75%, but the entire protocol required roughly 3 h and was fairly labor intensive, even compared to other enzymatic digestion methods, which may create more room for errors or contamination from poor handling. Like the study by Frolova et al.,⁸ Burja and colleagues also demonstrated the ability to use the cell suspensions generated by this protocol for scRNA-seq to capture the diversity of cells by both subcategorizing the cells into different cell types and examining the expression levels of key genes in specific cell types contained within the highly heterogeneous suspension. This work demonstrated the ability to dissociate skin tissue, which is harder to digest than many other tissue types, and obtain high viable cell yields to get relevant quantities for scRNA-seq. Comparing this work to the study by Frolova et al.⁸ also demonstrates how different tissues present different challenges in dissociation, since even with a larger biopsy sample and longer and more involved dissociation protocol, the viable cell yield from skin tissue was two orders of magnitudes lower than what that study obtained with breast cancer tissue.

Amosu et al.³⁹ took a different approach to improving tissue dissociation techniques as they developed an automated device that could reliably perform enzymatic dissociation with less direct work by a human operator. Their work describes the development of a motorized device that can digest up to 12 tissues at once by using 12 individual motors that help dissociate the tissue through mechanical agitation from controlled shaking during the enzymatic digestion. The dissociation of mouse lung, kidney, and heart tissue were tested on this device and compared to manual digestion using more conventional protocols. Interestingly, the viability and cell yield for these tissues did not differ significantly between the manual and automated device digestion, although the yield was more consistent for the automated dissociation and there was a notable increase in yield for CD4+ and CD8+ T cells from heart tissue. Moreover, despite the purpose of the device being to reduce the variability compared to manual digestion, standard deviations of results were not consistently improved compared to manual digestion. While viability was more consistent for device digestion compared to manual digestion for heart tissue, standard deviations were noticeably worse when comparing the viability of lung and kidney tissue digested with the automated device to results obtained with manual dissociation. It is possible that the differences observed would have been more significant if the manual techniques had been performed by less skilled operators, as the purpose of the device is to allow the task of tissue dissociation to be offloaded to less skilled individuals. It is also important to note that the authors did not directly compare their device to other automated tissue dissociation devices, which makes the comparative merits of this device unclear relative to what is already on the market. It is, however, notable that simultaneous processing of 12 samples would enable higher throughput than the 8 samples that the popular Miltenyi GentleMACS^TM Octo Dissociator can process at once.

Microfluidic approaches

The use of microfluidic approaches in tissue dissociation has gained interest in recent years due to the many advantages provided by such systems. Microfluidic systems are flexible and inherently lend themselves to modular design, which enables the integration of multiple functionalities, allowing for the incorporation of downstream analyses within a single device and increased automation.^1,76 Since microfluidics intrinsically involve working with small volumes, they can also reduce reagent use and dilution, making them useful for high-sensitivity and high-throughput analyses.^1,76 The earliest reported uses of microfluidics in sample dissociation tended to focus on the mechanical dissociation of cellular aggregates, in vitro cultured neurospheres, and similar structures by taking advantage of filters and the sheer forces applied when running the cells through the microfluidic devices.^49-53,77 Some of these devices did not use enzymatic digestion steps within the microfluidics, relying entirely on mechanical forces to achieve enzyme-free digestion.^49,50,52 However, as these early studies were working with simpler cultured cellular aggregates or tissues that had already been partially digested, they did not demonstrate the ability to dissociate the more complex structures inherent to native tissue. Nevertheless, early work on digesting bovine and mouse liver and kidney tissues demonstrated the potential of microfluidic devices to achieve high viability and yield in a fraction of the time required for traditional enzymatic approaches,⁵⁴ illustrating the promise of this avenue of research.

Lombardo et al.⁵⁵ combined aspects of their previous work to create a three-modal microfluidic device that greatly improved upon their prior endeavors. The device had two main components, a minced tissue digestion device and an integrated filtration/dissociation device (Figure 1A). First, minced tissue was inserted into the loading port and then moved into fluidic channels where the minced tissue fragments were broken down by shear force and contact with the collagenase solution that flowed through the device. This part of the device utilized a symmetrical design with two outlets on either side of the loading dock to reduce the risk of clogging that is often a hurdle for microfluidic designs. Once the tissue had been broken down sufficiently in the minced digestion device, the outlets were then connected by tubing to the dissociation/filtration device that further broke down the tissue by introducing shear forces from the branching channel array as well as by passing the tissue fragments/cellular aggregates through nylon mesh filters. After optimizing the device for each module individually, the investigators then tested dissociation of murine kidney, liver, heart, and mammary tumor samples, with the tissue spending 1-60 min (1-15 min for heart and liver tissues) in the digestion device before making a single pass through the filtration/dissociation device. The results were then compared to those obtained for each of these tissues using traditional enzymatic methods using cell-specific markers and flow cytometry. Outcomes calculated on a cell type by cell type basis for each tissue. For the kidney, the authors measured epithelial cells, endothelial cells, and leukocytes, and they observed that a 60-min dissociation in their device yielded over a 2-fold increase in epithelial cells and leukocytes and a 4-fold increase in endothelial cells compared to the control dissociation that consisted of 60 min of digestion with enzyme alone. Similar increases in cellular recovery were also demonstrated with the mammary tumor tissue. Studies dissociating liver tissue revealed a 4-fold enhancement in hepatocyte yield after processing for only 15 min in the device, when compared to a 60-min enzymatic digestion, but hepatocyte yield decreased with longer digestion and processing through the on-board filtration portion of the device, likely due to the large size and marked fragility of hepatocytes. Interestingly, however, yields of endothelial cells and leukocytes increased with longer processing times, leading the authors to propose that it may be possible to crudely select for a desired cell population by simply choosing the appropriate elution time. Remarkably, leukocyte yield from liver tissue after only 1 min of digestion in the device surpassed that of the 60-min enzyme digestion control. Processing cardiac tissue with the device yielded a 2-fold increase in cardiomyocyte recovery in just 15 min when compared to 60-min enzyme digestion controls; however, the recovery of endothelial and leukocyte cells on the digestion device was significantly decreased compared to yield obtained in the enzyme digestion controls. Importantly, irrespective of tissue type, cell viability was not significantly different when tissues were processed in the microfluidic device versus traditional enzyme digestion. To further qualify this new device’s utility for obtaining cells for subsequent downstream study, the resultant cell suspensions were analyzed by scRNA-seq, which demonstrated that this method of dissociation avoids the induction of the stress response normally seen with enzymatic digestion at 37 °C. Overall, this device shows the potential of microfluidics to greatly improve the yield of high-quality single cells while substantially reducing the processing time of dissociation.

Figure 1.

Microfluidic devices for tissue dissociation. (A) Fabricated minced tissue digestion device (top) and fabricated dissociation/filter device (bottom) from a previous study⁵⁵ under a Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/). Minced tissue digestion device is composed of a Luer port for loading tissue into the tissue chamber in between inlets and outlets for enzymatic solutions which are connected to the tissue chamber by fluidic channels designed to facilitate the dissociation with the application of hydrodynamic shear forces. The integrated dissociation/filter device composed of microfluidic channels designed to further break down tissues through the application of hydrodynamic shear forces and then by the use of two nylon mesh filters at the end of the device. (B) Schematic representation of Integrated Disaggregation and Filtration (IDF) device from a previous study⁵⁶ under a Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/). Large aggregates are broken down into single cells through exposure to stepwise increases in shear stress as they progress through the branching channel array which narrows from 1 mm to 125 µm. The nylon mesh filter module at the end can then further break aggregates down into single cells. Aggregates are not shown to scale with channel and pore sizes.

Further work was carried out on this platform to optimize the integrated dissociation and filtration device (without the minced tissue digestion device) in a subsequent study that worked with both cultured MCF-7 cells (a human breast cancer cell line that grows as loosely attached three-dimensional clusters) and murine kidney tissue⁵⁶ (Figure 1B). In this follow-up study, the authors tested different flow rates through the device, alongside different numbers of passthroughs through the filter and channels of the device. For the MCF-7 cells, 10 passages at flow rates of 40 mL/min through both the channel and filter modules yielded a viability of approximately 70% and an increase in live cell yield by roughly 150% when compared to enzyme-digested controls. However, this yield was actually below that obtained by just using the filtration 10 times without flowing the MCF-7 cell clusters through the channels. The authors speculated that these results were likely due to the simple nature of the MCF-7 aggregates, combined with the fact that the starting sizes of the clusters were already smaller than what the channels were designed to break down.

To begin delineating the parameters to achieve optimal dissociation of mouse kidney, the authors digested minced tissues with collagenase for either 20, 30, or 60 min before they loaded it into the microfluidic device and tested different numbers of passes through the filtration and channel modules similarly to what they did with the MCF-7 cells. In contrast to the results obtained with the MCF-7 cell aggregates, optimal dissociation of the kidney tissue was achieved with 10 passes through the channel and 1 pass through the filter (10C, 1F), which were the only conditions to show statistically significant improvements from the 20-min enzyme-digestion control. The total cell yield was consistent across all pre-digestion times for the 10C, 1F passes; however, some increases in yield of specific cell types like endothelial cells did seem to occur when the kidney tissue was pre-digested with collagenase for 60 min compared to 20 min. The authors then executed studies in an attempt to further optimize the dissociation process by altering the number of passes and pre-digestion time, and they found that performing a 60-min pre-digestion followed by 10 passes through the channels and 1 pass through the filter yielded the best overall results, liberating about 400 000 cells/mg tissue, with about 75 000, 3500, and 10 000 cells/mg tissue of epithelial, leukocyte, and endothelial cells, respectively. This represented a roughly 3-fold increase in total cell yield compared to the enzyme-digested control. The total cell yield and the epithelial cell yield remained consistent even after only 20 min of pre-digestion. However, reducing the digestion time significantly decreased leukocyte and endothelial cell yields, but especially the latter which exhibited a roughly 4-fold decrease in yield when the pre-digestion time was reduced from 60 to 20 min. The overall cell viability for epithelial, endothelial, and leukocytes was around 90%, 50%, and 50%, respectively. The ability of this device to obtain yields several fold higher than traditional enzymatic controls is impressive and demonstrates the ability to continue making improvements on an existing platform. However, the difference in response to channel passthroughs between the MCF-7 cells compared to the murine kidney tissue highlights that this device must be optimized with a specific application in mind and that it might not be suitable for all applications needing tissue dissociation.

A similar microfluidic device for tissue dissociation was also developed by Al-Mofty et al.,⁵⁷ who focused on using filters within the microfluidic design to remove all large aggregates and deliver a cell suspension ready for use in downstream applications without any additional processing. These investigators designed two prototypes: prototype-1, which was the primary one that was tested thoroughly, and prototype-2, which was designed to incorporate additional functionality. Prototype-1 featured a circular chamber in which tissue would be dissociated as well as branching channels incorporating 50 µm gap filter posts to trap larger aggregates and debris, leading into a simple outlet. Prototype-2 expanded on this design by including 100 µm gap retaining posts around the dissociation chamber to keep tissue in place when exposing it to flow, incorporating a cross-flow filter to prevent tissue debris from contaminating collected cells, and a final downstream filter to trap cells at the outlet and allow for cell staining. The authors tested the device with human placental and endometrial tissues as well as rat liver and adipose tissues. The placental and endometrial cells were dissociated with a Trypsin-EDTA solution, while the rat adipose and liver tissues were digested using a collagenase solution. Digestion with the device lasted 45 min for prototype-1 and 1-2 h for prototype-2. These studies confirmed that the basic functionality of both devices was as expected and demonstrated that placental cells could successfully be cultured from tissue dissociated with the device. The authors also showed that they were able to obtain roughly twice the yield of viable cells from the prototype-1 device compared to the control consisting of enzyme digestion alone, when examining yield as an average across the different tissue types. Unfortunately, however, the authors did not perform a similar comparison for prototype-2, nor did they report the overall viability of the cells. Moreover, the yields were reported only as an average across tissues, rather than from each of the tissues individually, making it hard to assess how performance might vary between different types of tissues with varying properties. As such, while these results seem promising, this device will clearly require further and more comprehensive testing to accurately evaluate its overall merits.

Nonenzymatic approaches

There have been efforts to investigate novel approaches to dissociate tissue that can work in the absence of enzymes or chelating agents. In addition to the decrease in cellular damage/stress, such approaches offer the benefit of cutting down on the number of reagents needed, which can lower costs and simplify workflows, and could potentially make them more accessible in lower resource settings. One early work in this field utilized mechanical disruption via pipetting to dissociate retinal tissue into single cells, but did not compare cell yield or viability to other methods, and, depending upon the application, introduced artifacts that could affect downstream analyses/applications.⁷⁸ While purely mechanical methods of dissociating such as scraping,⁷⁹ fine-needle aspiration,⁸⁰ or tissue grinding⁸¹ have existed for decades, they tend to produce lower recovery and higher cellular damage when not paired with enzymatic digestion.^37,82 Nevertheless, purely mechanical dissociation has shown some value in limited settings. For example, mechanical dissociation by passing tissue through a steel mesh within the Medimachine system yielded higher numbers of cells than enzymatic methods for testicular tissue. However, viabilities for the mechanically dissociated tissues were low, ranging from 8% with cryopreserved tissues to only 38% with fresh tissues.⁸³ Interestingly, one of the unique benefits of purely mechanical dissociation was showcased in studies dissociating breast cancer tissue, wherein the authors demonstrated that avoiding the use of enzymes enabled them to preserve subpopulations of cells from the tumor tissue that were absent in enzymatically digested samples of this same tissue due to enzymatic clipping of unique, defined, cell surface proteins.⁶² Most of the work performed with the goal of yielding high quantities of viable cells to date in this field has been limited to the use of microfluidic devices that employ sheer forces, meshes, and other physical objects to mechanically disrupt/break apart the tissue in question.^49,50,52 Further limiting their overall impact and broad applicability, the devices in many of these prior reports were only tested with simple cellular aggregates taken from culture or tissue that had already been partially digested through other means. As such, these studies often failed to completely demonstrate the ability of the devices being described to dissociate fully intact tissue. One notable exception to this generalization was a promising, nonmicrofluidic study by Garaud and colleagues,⁶³ who demonstrated that the benchtop tissue grinder instrument could be used to successfully dissociate murine liver tissue mechanically and produce comparable yields and viability to tissue that had been dissociated using standard enzymatic and mechanical methods.⁶³

Despite getting off to a rather slow start, several completely novel approaches to breaking down tissue without the need for enzymes have been reported in recent years. In one such study, Welch et al.^84-87 developed a novel platform using electric fields to dissociate tissue. Taking advantage of preexisting knowledge of electrical fields being used for electrophoresis, and the potential of these fields to induce cell movement and weaken adhesion, Welch and colleagues created a device designed to apply an electric field across a cuvette with parallel electrode plates, with the goal of using this electric field to dissociate tissues (Figure 2A). They first performed studies to determine the optimal media to be used in the device and found that nonionic solutions, such as ultrapure water or sucrose solutions, yielded the best results as they reduced bubbling and heating that could lead to cellular/tissue damage and reduced viability. Having determined the optimal dissociation solution, the authors next tested a range of applied DC voltages, comparing the dissociation of cryopreserved bovine liver tissue achieved with this device to that obtained using traditional collagenase digestion, with and without mechanical agitation, over a short time period (<5 min). The results demonstrated that increasing the applied voltage increased the degree of tissue dissociation up to an applied 90 V/cm, which yielded about 41% ± 3% dissociation in just 2 min, compared to less than 10% for collagenase digestion in the 5-min time period, irrespective of whether agitation was applied during the digestion. However, further increasing the voltage to 100 V/cm dramatically reduced the degree of dissociation achieved to less than that seen with the collagenase controls, which the authors concluded was likely due to the “bubbling” that occurred under these conditions, even with nonionic solutions. Surprisingly, however, the 100 V/cm sample exhibited the highest sample purity of 44% ± 12%, where sample purity represents the ratio of single cells to total particles, including cellular aggregates, fragments, and debris. Testing 10, 50, and 100 V/cm over longer time scales of up to 30 min revealed that longer treatments with constant electric fields were not an efficient means of improving tissue dissociation. In an effort to further increase the degree of tissue dissociation and reduce the “bubbling” at higher voltages, the authors next tried applying an oscillating square wave and testing the effect of varying the frequency up to 1 kHz for a 5-min period and revealed that, in general, dissociation efficiency increased with increasing frequency, such that the authors were able to achieve 95% ± 4% dissociation of the liver tissue in just 3 min with a 1 kHz electric field. To assess the impact that these electric field conditions were likely to exert on cellular viability, the authors exposed MDA-MB-231 cancer cells to the same 100 V/cm DC field and 100 V/cm 1 kHz square wave conditions for 5 min and observed no significant change in morphology or viability with the cells exposed to the 100 V/cm 1 kHz electric field exhibiting 90% ± 8% viability compared to 93% ± 2% viability for the control cells prior to electric field exposure. The investigators next tested the device with clinical glioblastoma multiforme (GBM) samples to establish the suitability of this device for dissociating other tissues. The degree of dissociation from exposure to a 100 V/cm 1 kHz electric field for 15 min was compared to enzymatic digestion using mincing, collagenase, and dispase for 60 min. Quite encouragingly, the electric field dissociation produced a 5-fold higher yield compared to the enzymatic treatment that took four times longer, while maintaining a similar viability of approximately 80%. Importantly, the authors confirmed that this new approach to tissue dissociation did not cause leakage of cell-free DNA, which could create problems in downstream single-cell sequencing. They also demonstrated that the RNA content and integrity in MDA-MB-231 cancer cells was unaffected by either the mechanical/enzymatic digestion or exposure to the electric field and also found that the yield of RNA was higher from the cells exposed to the electric field compared to those exposed to mechanical/enzymatic digestion. Finally, analysis of RNA for transcripts associated with the stress response and cellular adhesion demonstrated that the stress response was not induced in either the enzymatic/mechanical control or the samples dissociated in the electric field, except for an increase in SERPINE1 expression and decrease in PLAT expression. These analyses also revealed a much more pronounced downregulation of cell adhesion molecules in samples exposed to the electric field, perhaps providing at least a partial mechanistic explanation for the high efficiency with which the electric field was able to dissociate tissues. Importantly, the transcriptional changes observed immediately following exposure to the electric field returned to baseline after just a 60-min recovery period in control conditions, suggesting any changes triggered by such exposures are quickly reversible. Looking collectively at the extensive dataset, Welch and colleagues generated when evaluating this novel approach to tissue dissociation, this electric field–based methodology seems particularly promising, given its ability to provide near complete dissociation to single cells in a very short amount of time with minimal alterations to cell viability or phenotype.

Figure 2.

Nonenzymatic processes for tissue dissociation. (A) Schematic representation of electrical dissociation from a previous study⁵⁸ under a Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/). Tissue is loaded into device between two parallel electrodes, which are used to apply an electric field which dissociates them. A square wave frequency can be applied to periodically reverse the polarity of the electric field to facilitate better results. (B) Schematic representation of sonication tissue dissociation protocol from a study⁵⁹ under a Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/). Biopsy samples are placed with PBS into 1.5 mL tubes, up to 24 of which can be loaded onto a floating tube holder. The floating tube holder is then placed into an ultrasonication bath for 30 min at 37 °C (can also be performed at room temperature) and the sonication will break down tissue through the application of shear stress. (C) Workflow for Bulk Lateral Ultrasound nonenzymatic dissociation taken from MacMullan et al.⁶⁰ (copyright reproduced with permission). Tissues are harvested from a mouse and then dissociated either using standard enzymatic methods with the GentleMACS^TM Octo Dissociator or with the author’s Bulk Lateral Ultrasound System, which dissociates the tissue at low temperatures through the application of ultrasound. The resulting single-cell suspensions were then analyzed and compared. (D) Schematic of TissueGrinder tube-based nonenzymatic dissociation taken from a previous study⁶¹ under a Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/). Tissue is loaded onto the grinding gear of a TissueGrinder tube along with PBS and then ground mechanically with the Tissue Grinder device according to a programmed protocol for the tissue. The resulting solution is then centrifuged to obtain a single-cell suspension.

In addition to their highly innovative electrical dissociation device, Welch et al. also developed another nonenzymatic method to dissociate intact tissue using ultrasound sonication (Figure 2B). Although ultrasound had not previously been used for tissue dissociation, its utility in a wide range of applications in related fields for manipulating cells and other biological agents is well documented, and exposure to ultrasound has been shown to be gentler and induce far less pronounced transcriptional modulation compared to other mechanical and electrical methods of cell manipulation.^88,89 The theory behind the device is that the cavitation bubbles caused by ultrasonication apply a sheer force to the tissue that will gently dissociate the tissue into single cells. This assumed mechanism of action is supported by tests the authors performed with the device showing that degassing, which lowers the formation of cavitation bubbles, markedly decreased the dissociation efficiency of sonication when applied to bovine liver tissue. The investigators also found that temperature did not exert a significant effect on dissociation efficiency, with similar results being obtained when sonication was performed at room temperature and at 37 °C. Increasing sonication time did, however, increase dissociation efficiency, with the longest duration of sonication tested (30 min) yielding the highest degree of dissociation. Comparing the newly developed 24-plex sonication method to more standard high-throughput processing on a 96-well plate orbital shaker with and without enzymatic digestion with collagenase demonstrated that sonication yielded a marked improvement in dissociation efficiency of 53% ± 8% vs. 37% ± 20% with the enzymatic digestion. The dissociation efficiency could be raised further to 72% ± 10% by conducting the sonication at 37 °C and simultaneously performing enzymatic digestion within the tubes; however, this enhanced dissociation efficiency obviously comes at the expense of adding the prementioned drawbacks associated with enzymatic digestion. Further tests with the MDA-MB-231 cancer cell line established that the sonication method was more effective at dissociating the 3D aggregates into single cells than conventional methods, with a large majority being either single cells or doublets, vs. a significant number of aggregates that ranged in size from 20 to 100 µm when conventional dissociation methods were employed. While neither orbital shaker forces nor sonication forces produced any noticeable effect on the morphology of the MDA-MB-231 cells, viability was between 91%-98% for the sonication compared to only 64%-75% for the orbital shaker. It is important to note, however, that the authors did not assess cellular viability for either condition (ie, orbital shaker or ultrasound) when enzymatic digestion was occurring simultaneously. As such, it is difficult to conclude from the data presented what the optimal conditions are to obtain the highest yield of viable cells. Examining RNA content of MDA-MB-231 cells revealed that exposure of the cells to sonication did not reduce the yield of RNA, while exposure to the forces of the orbital plate shaker did. In further studies using RT-qPCR, the authors showed that only minimal changes to the transcriptome were observed for any amount of time in sonication, while significant alterations consistent with a stress response were seen in cells exposed to the orbital plate shaker. Taken together, the data demonstrates that this ultrasound-based method ultimately had significantly worse yields over longer time periods compared to the method using an electric field. However, it should be noted that ultrasound has fewer effects on the transcriptome, and its use would be far easier to scale up for high throughput, which may give this ultrasound-based method an edge in certain applications.

Another effort to use ultrasound for dissociating tissue was performed by MacMullan et al.⁶⁰ and is currently in preprint. These investigators developed a method to utilize bulk lateral ultrasound (BLU) at 8 °C to dissociate tissue in just 2 min and then compared it to commercial tissue dissociation kits that used enzymatic digestion (Figure 2C). MacMullan and colleagues tested murine brain, heart, lung, and B16 tumor tissue. The protocol employed for dissociating the tissues using the commercial kit and dissociator varied based on the specific tissue type, and results obtained for each tissue were then compared to 2 min of dissociation using the newly developed BLU device. Overall, viability and live cell yield were not found to be significantly different for most tissue types when dissociation was performed with the BLU device vs. with the commercial enzymatic method. However, when the authors analyzed the cells liberated from the tissues for expression levels of population-specific immune cell markers, they found that the BLU-dissociated cells had significantly more CD8+ T cells compared to the enzymatic method, suggesting that the BLU method is better at mitigating losses of CD8+ T cell expression that previous work had shown to be associated with enzymatic digestion.⁹⁰ Further comparisons using standard antibody panels demonstrated a significant loss of expression of myeloid cell markers in cells liberated from tissues using enzymatic methods when compared to those obtained with the BLU method. While this work has, as of the time of this writing, not yet been fully peer-reviewed, its findings that tissue dissociation can be achieved using ultrasound while incurring minimal transcriptional changes matches up with the work of Welch et al.⁵⁹ and further highlights the potential of ultrasound-based dissociation. While MacMullan and colleagues were not able to improve the efficiency or yield of dissociation over that of standard enzymatic methods, it is essential to appreciate that they also ran their device for just 2 min, and based on the other work with ultrasound dissociation, it is quite possible the yield could be further improved by simply increasing processing time.

In other studies, Scheuermann et al.⁶³ built upon their previous work and developed an enzyme-free dissociation method for archived tissue stored at −80 °C in RNAlater^® using purely mechanical dissociation provided by TissueGrinder tubes⁶¹ (Figure 2D). Tissue was collected from two human donors—a spleen and liver tissue sample from the first and a brain cortex sample from the second. Thawed tissue was processed in PBS using the grinding gear of TissueGrinder tubes. Two different methods were tested to dissociate liver and spleen tissues. In the first 70 µm filters were employed with standard tissue-specific processing programs. In the second method, modified programs were used; for the liver, the rpms were reduced by 50% in all steps, while for the spleen, the rpms were reduced by 50% and the time was reduced by 5 s for all steps. The modified spleen program was also used for the brain cortex tissue with a 100 µm filter. After this processing, the tubes were centrifuged, and the results were analyzed. The modified protocols yielded the best results, with cell yields of 4.8 × 10⁵, 2.16 × 10⁶, and 6.54 × 10⁴ cells/mL from the liver, spleen, and cortex, respectively. Cell viability was 45%, 90%, and 64% for the liver, spleen, and cortex, respectively. Although these viability numbers may not seem that remarkable, when one considers these were archival tissues stored at −80 °C in RNAlater^®, the ability to still get viable cells is of note. Unfortunately, the authors did not provide information on the size or mass of each tissue sample that was used for dissociation, data that may have put their work in a better context. The authors also performed single-cell miRNA sequencing and confirmed that despite the low RNA integrity number (RNI), the integrity of the miRNA was not compromised to the extent that would preclude its use for single-cell miRNA analysis. As this work was only intended to be the proof of concept for the workflow, the authors did not compare their results directly to other methods or obtain more than one biological source for each tissue. As such, to be able to confidently assert and assess the utility and reproducibility of this method would require further study.

Conclusions

There have been several advancements in the field of tissue dissociation since the start of the decade. Traditional enzymatic methods have continued to be optimized, while more work has been dedicated to creating effort-saving automation to simplify the processes and shorten the time of performing the dissociation. Microfluidics and nonenzymatic methods have both shown a large potential to speed up the process of isolating single-cell suspensions as well as to deliver high yield and viability samples while alleviating some issues associated with traditional methods. Despite the obvious promise of these new techniques, however, there is still much work to be done. The heterogeneity of tissues continues to be a challenge in all fields, which often requires a highly specific protocol for each given tissue, thereby making standardization difficult. While microfluidic methods and nonenzymatic methods show promise, they have not yet been tested with a large variety of tissue. As such, their wider applicability is currently still unproven. More work will need to be done to further validate these methods and broaden their applications. Furthermore, the wide variety of methods and assessments described herein also demonstrate the challenges associated with comparing methodologies between different works. For example, in the studies discussed in this Review, dissociation has been reported on the basis of viable cells recovered, cells/mg tissue, and dissociation efficacy, and the authors of the various studies have reported numbers for both total cells recovered as well as for specific cell types within a tissue. Due to the wide number of downstream applications for these techniques, as well as the differences between different tissue types, the most important qualities for assessing the usefulness of any given dissociation technique can vary dramatically depending on the intended goals of a particular work, and studies focusing on assessing the suitability of a technique for a given application may be challenging to translate to other, different usages where they might be useful.

There is no doubt that there are many promising directions for future research and more work will likely be done to further optimize more established methods, as well as exploring the utility of the more novel approaches for a wider variety of uses. Beyond this, it will be helpful to develop more tunable methods that make it easier to predictably modify a given protocol to adapt to new tissue types while still yielding high-quality results. Taking this to its limits, we envision it may ultimately be possible to measure the properties of an individual tissue and tune the protocol to the specific sample being dissociated to obtain optimal results on an individualized basis. There are also potential gains to be had by working to combine multiple different methods to improve results, as well as combining them with downstream processes on a single device to simplify and improve the overall workflow on projects that involve these downstream techniques. Microfluidic methods show a particular promise in this regard given their modular nature and their wide usage in many avenues of cellular manipulation.¹

Author contributions

Aaron Jankelow (Conceptualization [equal], Formal analysis [lead], Writing—original draft [lead], Writing—review & editing [supporting]), Graca Almeida-Porada (Conceptualization [equal], Funding acquisition [equal], Supervision [equal], Writing—review & editing [equal]), Tony Atala (Funding acquisition [lead], Writing—review & editing [equal]), and Stephen W. Sawyer (Conceptualization [equal], Supervision [equal], Writing—review & editing [equal]), Christopher D. Porada (Conceptualization [equal], Funding acquisition [equal], Supervision [equal], Writing—review & editing [lead])

Funding

This work was supported by National Science Foundation, (NSF) Engines Program #2315654 (Cell Processing ITEC).

Conflicts of interest

The authors declare that they have no competition interests.

Data availability

All data needed to evaluate the conclusions in the paper are present in the paper and/or in the materials cited herein.