Isolation of one or multiple cell types from a heterogeneous population is an integral part of modern biological research and routine clinical diagnosis and treatment. Purification of specific cells is essential for basic cell biology research, cellular enumeration in specific pathologies and cell-based regenerative therapies. The central principle of separating any cell type from a population is to utilize one or more properties that are unique to that cell type. The most widely used cell isolation and separation techniques can be broadly classified as based on adherence, morphology (density/size) and antibody binding. The high-precision single-cell isolation methods are usually based on one or more of these properties while newer techniques incorporating microfluidics make use of some additional cellular characteristics. The recent improvements in cell isolation procedures vis-à-vis purity, yield and viability of cells have resulted in significant advances in the areas of stem cell biology, oncology and regenerative medicine among others.

A cell isolation procedure can either be a positive selection or a negative selection – the former aims at isolating the target cell type from the entire population, usually with specific antibodies while the latter strategy involves the depletion of all cell types of the population resulting in only the target cells remaining. Both types of isolation methods have their advantages and disadvantages. Due to the use of specific antibodies targeting a particular cell type, positive selection yields a higher purity of the desired population. On the other hand, it is more complicated to design an antibody cocktail to deplete all the non-target cells making negative selection less efficient vis-à-vis purity. Furthermore, a cell population isolated through positive selection can be sequentially purified through several cycles of the procedure, a benefit that negative selective cannot provide. However, positively selected cells carry antibodies and other labeling agents that may interfere with downstream culture and assays – if that is a concern, it is preferable to use a negative selection method [9].

To isolate a particular cell type from a heterogeneous population depends on the unique properties of that cell type. In this review, we discuss four broad categories of cell isolation techniques based on the following cellular characteristics [10] :

• Surface charge and adhesion – This feature determines the extent of attachment of cells to plastic and other polymer surfaces and can be used to separate adherent cells from suspension/free-floating cells.
• Cell size and density – The physical properties of size and density are commonly used for the bulk recovery of cells; either by sedimentation, filtration or density gradient centrifugation.
• Cell morphology and physiology – Different cell types can be distinguished by shape, histological staining, media-selective growth, redox potential and other visual and behavioral properties which can then be harnessed to isolate those cells.
• Surface markers – Specific binding of surface antigens to either antibodies or aptamers can selectively capture cells of the specific surface phenotype. The captured cells are subsequently detected with the help of measurable probes – usually fluorochromes and magnetic particles – with which the antibodies/aptamers are labeled.

In addition, two or more of the above principles can be combined to increase the specificity of isolated cells further – usually, such compound techniques consist of a label-free (the first three in the list) method along with a label-incorporating method.

Choosing a cell isolation method for an experiment depends on the following criteria [10] :

• The exploitable characteristics of the cells
• The amount of stress – mechanical/chemical/physiological – that the cell type can endure and still remain viable
• The levels of cell purity and yield needed
• The acceptable risk of contamination (zero in case the cells are needed for culture)
• The type of isolation method – whether negative or positive
• Any specific requirements of the downstream applications e.g. cell culture, nucleic acid/protein extraction etc.
• The time and costs of labor, reagents, automation etc.

Table 1 summarizes the different cell isolation techniques [10].

Most cell separation procedures are preceded by the removal of as much extraneous material as possible, especially when freshly isolated tissue is the starting material. Some of the common procedures are enumerated below.

• Digestion/disintegration of solid tissues – Pieces of whole tissues need to be first broken down to a single-cell suspension. Table 2 summarizes the common methods. Commercial kits are available, for example, neural and tumour tissue dissociation kits from Miltenyi Biotec [11-13]. Collagenase and TrypLE from Thermo Fisher / Gibco are popular choices [14, 15]. Experimental artifacts may be introduced during the digestion/disintegration, for example, liberase introduces transcriptional artifacts [16] and enzymatic digestion activates microglia [17].
• Separation of plasma from blood – Plasma is the liquid portion of the blood containing clotting factors, albumin, salts, minerals etc. which can potentially interfere with cellular assays. Therefore, a buffy coat is usually preferred over whole blood: the latter collected in anti-coagulant coated tubes and then centrifuged at 200 x g with brakes off. The WBC rich band or the buffy coat sandwiched between the upper plasma and lower RBC layers is carefully aspirated out [18].
• Removal of dead cells and debris – Primary cells and cultured cells often need to ‘clean up’ of dead cells and debris to ensure that the downstream steps flow smoothly. Low-speed centrifugation at 200-300 x g, gravity sedimentation, and nylon mesh sieving are some easy methods to achieve this [18]. Specific kits such as Dead Cell Removal kit [16] can also be used.

Based on whether cells need to attach to a solid surface for survival and growth, they can be classified as follows:

• Adherent – These cells require a suitable surface to attach to thrive, e.g., macrophages, fibroblasts, mesenchymal stem cells, etc. This attachment occurs through the interaction between the cell adhesion molecules (CAMs) on the cell surface and the corresponding ligands on the culture surface. The CAM family important in the context of in vitro cell adhesion is the integrin family which can attach to collagen, fibronectin and other extracellular matrix proteins.
• Suspension cells – These cells naturally do not require an attachment surface and occur in suspension in the body, usually in fluids like blood and lymph. Examples include lymphocytes, granulocytes, and other immune cells. The in vitro growth and selection of these cells also occur in suspension. Embryonic and adult stem cells also do not require anchorage and can grow in suspension in serum-free media.
• Cancer cells that have lost contact inhibition and anchorage dependency. In serum-free conditions without any anchorage facilities, cancer cells can form spheroids and amorphous aggregates.

Ever since mammalian cell culture became a staple of biomedical research and diagnostics, the fabrication of culture vessels – in particular, the polymer surfaces – have also undergone significant improvements. Till the 1950s, only Pyrex® glass flasks and dishes were available which severely limited the scope of culturing primary cells (as the latter do not attach to glass surfaces) and also incurred additional costs of maintenance and sterilization. The first disposable plastic culture vessels came out in the 1960s and were made from polystyrene, which offered optical clarity, malleability, and ease of sterilization. However, simple polystyrene could not solve the problem of non-adhesion of cells due to its hydrophobic nature. The solution is used to treat the polystyrene surfaces to make them more hydrophilic [23]. There are three main ways of achieving this:

• Oxidization – Soon after molding, the polystyrene surfaces are treated with chemical oxidizers or ionizing radiations which generate oxidized ions and graft into the polystyrene chains. This treatment renders the polystyrene surface positively charged to which the negatively charged serum proteins bind and the resulting spread of the serum provides a better surface for the cells to attach. Various oxidizing agents have been tested, e.g., ozone, sulfuric acid, potassium hydroxide, hydrochloric acid, UV radiation among others. Nowadays, the method employed is exposure to corona discharge under atmospheric pressure or gas-plasma under vacuum [24].
• Coating with proteins – The limitation of oxidized polystyrene is that it can only support the growth of cells in a serum containing media. To support adherent culture systems that require serum-free conditions requires a protein coating that can attach to the specific adhesion receptors on cells. The common extracellular matrix proteins used for coating are collagen, fibronectin, chondroitin sulfate, etc. [24, 25].
• Coating with poly-lysine – Poly-lysine is a cationic polymer which provides a positively charged culture surface for the negatively charged ions of the cell membrane to bind. The advantage of using poly-lysine is that it interacts non-specifically with all adherent cell types and thus poly-lysine coated plates can be used for different cell types and cultures [25].

In the parlance of cell culture technology, all the above types of plates are termed as ‘tissue/cell-culture treated.’ Conversely, some cells need to be grown in suspension for specialized cultures, e.g., the formation of embryoid bodies or tumor spheres. Ultra-low attachment plates are available for such applications – these surfaces are coated with a hydrophilic gel that neutralizes any surface charge and prevents cells from adhering and keeps them in suspension [26].

Cell culture systems are broadly classified as adherent and suspension depending on the anchorage dependency of the cells. Isolation of both adherent and free-floating cells can be tailored to the specific cell type by adding the requisite growth factors and suitable cell culture plates. We describe three examples below:

• Adherent cultures – example: macrophages, fibroblasts, mesenchymal cells

  Macrophages are routinely isolated from bone marrow or peripheral blood. Right after the isolation of mononuclear cells, they can be seeded on coated polystyrene plates along-with serum and monocyte/macrophage differentiation cytokine cocktail. After 5-7 days, the cells differentiate and form an adherent monolayer while the unwanted cells remain in suspension and are discarded [27]. Microglia, brain macrophages, can be similarly isolated [28].

• Suspension cultures – examples: stem cells, embryoid bodies, tumorspheres.

Cells naturally growing in suspension and cells that have lost anchorage dependency can be separated from the adherent cells by culturing in ultra-low attachment plates in the absence of serum. The desired cells either grow in a single-cell suspension or aggregate to form floating spheroids. The adherent cells die out without the support of an attachment surface [29, 30]. Hoffmann M et al, for example, obtained human airway epithelial cells by removing the adherent fibroblasts [21].

Adhesion-based methods are technically simple, reproducible and mostly cost-effective which yield a good crop of the cells. The limitation is that the purity of the recovered cells is low and there is always a risk of cross-contamination with other cells as well as with bacteria.

Specific applications include [27, 29, 30] :

• The routine isolation of blood and bone marrow cells especially macrophages.
• The isolation and characterization of cancer stem cells, e.g., colonospheres, mammospheres, neurospheres have helped in identifying and expanding the rare cancer stem cells.
• The separation of adult stem and progenitor cells from the differentiated lineage positive cells.

Separation techniques relying on the size and density of cells are frequently used to separate specific cell types from bulk peripheral blood (PB) and bone marrow (BM). The most commonly used methods in this category include density gradient centrifugation and filtration. They are also called ‘bulk sorting’ methods since they help isolate relatively large cell populations in a short duration. Although bulk sorting methods are an essential part of several clinical and biotechnology procedures because of high yields, the purity and homogeneity of the harvested cells are quite low compared to other cell separation procedures.

The most frequently used density-based separation method is gradient centrifugation - the biological sample is centrifuged in a suitable gradient medium at the appropriate speed till the different cell types are fractionated into different layers or phases depending on their respective densities. For example, Aizarani N et al first removed nonviable cells, then separated primary human hepatocytes and non-parenchymal cells, from dissociated liver tissue before FACS sorting, to build a human liver cell atlas through single-cell RNA-seq using Percoll gradient centrifugation [31]. Density gradient centrifugation is frequently used as an initial step to enrich for certain cell populations before more sophisticated isolation procedures such as FACS, MACS or single-cell sorting as well as organelle and nucleic acid extraction.

The sedimentation rate of cells in a suspension is directly proportional to the centrifugal force (g) applied and inversely proportional to the viscosity of the medium: at constant g-force and viscosity, the sedimentation rate would depend on the size of the cells and the difference between the densities of the cells and the medium. The relationship between these variables is best explained by the equation below [32] : v=((d² * (ρ_p-ρ₀)) / (18 * η)) * g; where v = sedimentation rate, d = diameter of the cell, g = centrifugal force, ρ_p = cell density, ρ₀ = medium density, η = medium viscosity.

[enlarge]

Figure 1. The fractionation of blood cells on the ficoll gradient [1].

The medium used to separate the cells through density gradient centrifugation is made up of gradients that encompass the densities of all the possible cell types in the mixture. During the centrifugal process, each cell type will sediment to that part of the gradient whose density equals that of the cell, that is, the isopycnic point. As per the equation above, when the density of a medium gradient phase and a cell type is equal, the sedimentation rate is zero, and the cells will settle in that phase instead of the bottom of the tube. Therefore, the cells are separated only by density, irrespective of size. The heavier cells usually pellet down while the lightest ones, along with the dead cells float on top of the gradient [32]. Figure 1 shows the schematics of isopycnic gradient separation [1].

The gradients used in density-based fractionation are mainly of two types – continuous and discontinuous [33].

• A discontinuous gradient is made of distinct bands of different densities increasing from the top to the bottom. The separation medium is diluted to a series of different densities and then carefully layered in decreasing order of density starting from the bottommost layer to the topmost. This kind of gradient has sharp interfaces between the bands and therefore is used when a sharp band of cells is required at the relevant interface(s). Discontinuous gradients are widely used to fractionate blood cell populations.
• A continuous gradient is characterized by a smoothly increasing density from top to bottom – in fact, it can be considered to be made of an infinite number of interfaces. Creating a continuous gradient is more complex and requires creating solutions of the lowest and highest densities and then mixed to form a linear gradient the spans the range between the limits of the two starting solutions. However, a continuous gradient provides a greater resolution since the wide range of available densities increases the chance of an isopycnic banding of cells at their precise density. Furthermore, to separate cells that do not differ much in their densities, a narrow range of gradient can be created. Discontinuous gradients are used to isolate Leydig cells and intestinal epithelial cells in addition to BM cells.

The ideal gradient medium should have the following properties [34] :

• Sufficient density range for the isopycnic banding of several cell types
• Physiological ionic strength, pH an osmolarity
• Low viscosity
• Non-toxicity
• Inability to penetrate biological membranes
• Ability to form continuous and discontinuous gradients
• Easily removable from isolated cells
• No effect on downstream assay or culture procedures

Table 3 summarizes some common density gradient media. Other commonly used media include histopaque 1077 or 1119 [35, 36].

The most common application of density gradient centrifugation is blood fractionation. Therefore the generic protocol for the fractionation of blood using Ficoll is outlined here:

1. Collect the blood/buffy coat in anti-coagulant coated tubes and dilute with equal volume of PBS.
2. Dispense suitable volumes of Ficoll into centrifuge tubes and carefully layer the diluted blood over the gradient media. Do not let the blood and Ficoll mix!
3. Centrifuge at 400-500 x g for 30-40 minutes at room temperature with the centrifuge brakes off.
4. Aspirate the desired cell fraction (refer to figure 1): the upper layer of plasma → platelets; plasma and Ficoll interface → mononuclear cells; layer below Ficoll → granulocytes; pellet → RBCs.
5. Wash the cells with PBS several times to get rid of plasma and Ficoll before using the cells for any subsequent analysis.

Important points to consider to ensure optimum gradient fractionation:

• The choice of anticoagulant affects the final quality and yield of cells
  • A purer monocyte and lymphocyte preparation is obtained with EDTA as opposed to heparin which can affect the proliferation rate
  • Citrates are the better choice when the cells are needed for DNA and RNA extraction
  • Citrates also increase monocyte yield while prior de-fibrination of blood results in low yield
• It is always preferable to fractionate blood soon after it is collected. If that is not possible, then the blood should be stored at room temperature for no more than 24 hours. Longer storage times can reduce lymphocyte yield and alter surface immunophenotype.
• Old blood samples should be avoided as they have a significant content of lysed RBCs which skew the gradient and make accurate isolation of specific cells difficult.
• The gradient media should be stored at the ambient temperature ad used before the expiry date as stamped by the manufacturer.

Advantages:

• The procedure is technically simple and cost-effective; lab personnel can be easily trained.
• It can be scaled up or down as per requirement with minimal adjustments.
• The yield of cells obtained, especially for blood samples, is high.

Limitations:

• The purity of the different cell fractions obtained is low, especially when fractionating blood.
• The procedure is time-consuming and low throughput.

• Fractionation of blood mononuclear cells for clinical and research applications
• Exclusion of dead cells from a cell culture harvest– the dead cells will pellet out, and the viable cells will be present in the culture medium and gradient medium interphase.
• Separation of plasma from blood cells for clinical and diagnostic uses.
• Specialized gradient media have been developed to isolate cells from tissues such as the liver, pancreas, lungs, testes, and intestines.

Filtration-based cell separation is a size-based method wherein cells smaller than the pore size of the specific filtration device pass through and larger cells are trapped. Filtration was first used successfully for cell separation by S.H. Seal in 1964 who could separate large tumor cells from the smaller blood cells [43].

There are a few commercially available filtration devices that have been specifically developed for cytological purposes. The starting sample can be blood or cultured cells prepared in the filtration buffer recommended by the company. The process of filtration in these devices can either passive or active – the former usually involves gravity or capillary action and the latter employs a motorized vacuum pump. Table 4 summarizes the cytological filtration devices [44].

Advantages:

• Simple, easy to perform and reproducible
• High throughput and yield
• Filters containing the captured cells can be directly used in downstream assays

Limitations:

• Poor specificity and purity – rare cells are frequently lost, and false positives are common
• The surface phenotype of some cancer cells is lost during the process, which affects downstream assays

Cell filtration is almost exclusively applied to the field of cancer biology – specifically in the isolation of circulating tumor cells (CTCs) which are known to be larger than healthy cells. To study different aspects of CTCs, the filtration process can be simultaneously combined with the following assays [44] :

• Immuno-staining can be done on the filter trapped cells directly for the identification of suspected CTCs by specific biomarkers. In addition, the origin of CTCs – epithelial or mesenchymal – can also be determined using lineage-specific probes.
• The Screencell® systems can also be connected to culture vessels and a microscope to directly culture the trapped tumor cells in the filters and keep them under observation.
• FISH analysis can be directly performed on the filter to assess chromosomal rearrangements.
• The Screencell BM® system has inbuilt collection tubes to perform DNA and RNA analysis of the trapped CTCs immediately.

A specific cell type can be selected over unwanted cell populations by culturing them in a medium that provides some selective advantage to the desired cell type. In routine cell culture, a particular cell type can be enriched by adding specific growth factors and cytokines – typical examples include enriching populations of a specific lineage and differentiation stage. In this review, we will discuss the use of selective media that allow specific cells to grow and inhibit others through antibiotics or specific growth inhibitors. Selective media is often used for isolating transfected cells and hybridomas [45].

Gene function analysis very often requires the creation of mammalian cell lines that either overexpress a transgene or a mutated/knockout gene. To select out the stable transfectants and expand a homogenous clone, a selectable marker is incorporated into the gene vector and a reagent that requires the activity of the marker is included in the medium. Therefore, only the cells that express the selectable marker have a growth advantage and can easily become the dominant clone within a few weeks of culture [45]. There are basically two approaches to use a selective growth medium, based on the type of selectable marker

• Antibiotic resistance – If the transfected cells express an antibiotic resistance gene, the specific antibiotic is added to the medium at a concentration that is lethal to the cells which do not express the resistance gene. Common antibiotics used for mammalian cell selection include bleomycin, puromycin and hygromycin [45].
• Metabolic/biosynthetic enzymes – This strategy can be best explained by the HAT medium selection often used in hybridoma technology. HAT stands for hypoxanthine-aminopterin-thymidine: aminopterin blocks de-novo DNA synthesis and hypoxanthine and thymidine provide the raw materials for the alternative ‘salvage pathway’. The key enzyme needed for the salvage pathway is HGPRT – hypoxanthine guanine phosphoribosyltransferase and only cells expressing the HGPRT gene can survive in the HAT medium [46].

Table 5 summarizes common selective markers used in mammalian culture and the basis for selection [45].

Growth selection of specific clones is routine in most cell and molecular biology labs and offers the following advantages:

• relative ease of performance
• reproducibility
• adequate cell yield

The limitations of selective media-based isolation are as follows:

• the risk of contamination
• the emergence of spontaneous resistant clones that do not carry the gene of interest
• the high costs of some reagents and specialized media
• the time and labor required for the expansion and maintenance of the relevant clones

Laser dissection technology, first described in the early 20th century [47], basically involves cleaving areas or cells of interest from tissue sections using a narrow beam laser. A more refined, microscope-based form of laser dissection technique – Laser capture microdissection (LCM) – was later developed in the NIH by Liotta and Emmert-Buck [48] to isolate pure cell populations from heterogeneous tissues on the basis of their natural morphology or specific histological/immunological staining (immuno-LCM is discussed below). LCM has enabled researchers to separate tumor from healthy tissues, stem cells from stroma and epithelial cells from parenchyma with high precision. Coupled with high precision surgery techniques, LCM can even be used to isolate single cells (discussed below) [49]. Sophisticated LCM devices have been developed for a variety of applications such as genomics, proteomics, and molecular characterization of cancer cells, tissue regeneration, diagnostic pathology, and basic cell biology.

[enlarge]

Figure 2. The basic mechanics of laser capture microdissection (LCM) using an IR-laser beam [2].

The basic LCM instrument consists of an inverted microscope, a laser diode, a laser control unit, a joystick-controlled stage, a CCD camera, and a computer for additional controls and image visualization [50]. The LCM systems can be both manually and robotically operated. Based on the type of laser, LCM systems are of two types – infrared (IR) and ultraviolet (UV). The minimum diameter of the laser beam of the LCM microscope is 7.5 μm, and the maximum diameter is 30 μm. The tissues can be maximally heated to a temperature of 90°C and that also for a few milliseconds which therefore leaves the cellular structure and macromolecules intact. There are two main types of LCM platforms based on the type of laser used – UV and IR. A combined automated IR/UV system is also available as the Arcturus Veritas™ instrument [51]. Figure 2 outlines the methodology of LCM [2].

• Infra-red LCM – The IR-LCM was first developed by Emmert-Buck et al [48] at the National Institutes of Health and was introduced as the PixCell system by Arcturus Engineering. The IR-LCM platform is based on the placement of a thin transparent thermoplastic ethylene vinyl acetate (EVA) film over a tissue section. After the tissue area of interest is identified visually, the cells are fixed to the film with the help of a short duration IR laser pulses. The cell-film adhesion is stronger than the cell-slide adhesion which allows the specific cells to be plucked out. The removed cells are subsequently transferred to a microcentrifuge tube containing the requisite buffer solutions for the downstream assays [48].
• Ultra-violet LCM – UV-LCM was invented by Schütze and Lahr in 1998 and commercialized by PALM Zeiss Microlaser Technologies [52]. The tissue is first mounted on a 6 μm thick membrane which is then placed on a microscope glass slide. A narrow UV laser beam is directed onto the specific area after visual identification, and the unwanted cells surrounding the desired cells are ablated. The desired cells are then pulled up and captured into an overhanging cap. Since the UV-LCM platform does not involve any non-specific adherence of the desired cells, it is preferred over the IR-LCM for some applications. Leica LMD7 system uses gravity to collect the captured cells [53].

A variety of source materials can be utilized for LCM technology: examples include live cells from cell cultures, formalin-fixed paraffin embedded tissues [53], frozen tissues, fresh tissues, metaphase spreads, and blood smears. Tissue sections (frozen or paraffin embedded) are the most commonly used samples and for optimal capture and analysis.

• The correct temperature should be used while cutting the sections using the cryostat/microtome
• The tissue sections should be between 5-15µm thick. Thinner than 5µm will be too fragile and thicker than 15 µm will not dissect properly.
• The sections should undergo the suitable staining protocol and adequate dehydration for proper visualization and storage.

Frozen tissue is largely preferred over formalin-fixed tissues as it preserves RNA, DNA, and proteins without any cross-linkages/changes in conformation - the latter being the main drawback of formalin-fixed tissues. Nevertheless, formalin-fixed tissue sections are the standard of pathology labs worldwide and often used for LCM based cancer cell biomarker studies [50].

LCM has the obvious advantages of being high throughput, precise and adaptable to different tissue sources and applications.

• Depending on the laser voltage and tissue structure, as many as several thousands of cells can be collected in a short period.
• Due to the high precision laser beams, the damage to adjacent tissues is minimal, and one tissue section can be probed several times
• LCM microscope and platforms are easy to operate and can be easily synergized with other assays.
• A wide range of tissues can be subjected to LCM including regular hematoxylin & eosin stained slides, archival sections and immune-stained frozen/fresh tissues among others.
• The lasers used are essentially low powered and therefore do not destroy the structure and integrity of macromolecules. LCM collected cells can be successfully used for downstream assays requiring functionally active nucleic acids and proteins.

However, LCM is not without limitations:

• For removal of selected cells post laser targeting; the tissue sections cannot be coverslipped. Without mounting medium and coverslip, the refractive index of the tissue is altered, and that makes it difficult to visualize the desired cells. This problem is of particular concern in tissues with amorphous structures – such as tumors and lymphoid tissues – where without any ‘architectural markers,’ capturing specific cell types becomes very difficult. However, this problem can be circumvented to a certain extent by using specialized staining techniques and immune-staining
• Most LCM platforms employ lasers whose minimal spot size is 7.5µm, and this is not precise enough to isolate single cells. For single-cell LCM from tissues, newer platforms have been developed and will be discussed below. If cytological preparations are used, however, then even the regular lasers can dissect out single cells.
• Another technical challenge that results due to tissue drying (especially frozen tissues) is the lack of adherence of the dissected tissue onto the film or inverted cap. In the case of IR-LCM, if the power of the laser is low, it may not be sufficient to ‘melt’ the membrane to the tissue [48, 50, 52].

LCM has gained popularity in recent years in the study of the genomics, transcriptomics, and proteomics of distinct cells from tissue samples. It has been especially valuable in cancer research since:

• malignant tissues are highly heterogeneous in terms of cell composition
• cancer-initiating cells and cancer stem cells are poorly characterized and need novel biomarkers for identification
• neo-plastic transformation involves sequential mutations and clonal evolutions.

Some specific biomedical application of LCM have been enumerated as follows [54] :

• Cancer diagnostics – Since there is a paucity of cancer biomarkers, new approaches to detecting tumor cells and clonal evolution is to detect mutations and deletions. LCM has been used to isolate cells from tumor tissues and use them for molecular analysis and tumor genotyping. LCM has been successfully used to detect gene alterations, deletions, and loss of heterozygosity in various malignancies like prostate cancer, melanomas, glioblastomas, lymphomas, etc.
• Cancer chemotherapy evaluation – Biomarkers are used to monitor the success of chemotherapy, evaluate new chemotherapeutic drugs, track relapse and identify patients at high risk. LCM can be used to randomly select cells from tumor tissues and analyzed for these biomarkers to identify persistent clones that escape chemotherapy.
• Biomarker discovery – When coupled with genomic and transcriptomic profiling, LCM can be used to detect differentially expressed genes in malignant tissues relative to healthy tissues. Multi-assay platforms incorporate LCM, whole genome microarrays and RT-PCR have been able to identify novel biomarkers in invasive glioblastoma, breast cancer, and ovarian cancer cells.

In addition, LCM has also been employed in developmental biology, embryology, and xeno-grafting technology.

Cellular surfaces are populated with a multitude of ligands – proteins, carbohydrates, and glycoproteins - which perform various biological functions and give every cell a unique surface phenotype. Many ligands are antigenic and elicit an immune response in an allogeneic body. Therefore specific antibodies raised against cell surface antigens can be used to target the cells expressing those antigens, for example, neuronal cells, T or B lymphocytes, stem cells, and a myriad of other cell types. Antibody-mediated cell detection and isolation techniques became commonplace with the discovery of the cluster of differentiation (CD) markers – surface receptors usually involved in cell signaling and adhesion. The CD nomenclature system was first devised to classify the monoclonal antibodies (mAbs) generated by different laboratories against white blood cells (WBCs) surface epitopes [55]. The system was later extended to different cell types, and at present, more than 370 CD antigens are known [55]. The unique CD profiles of different cell types are used to define and separate different populations within a tissue, isolate rare cells like adult stem cells, cancer stem cells, etc. and differentiate between control and treated cells depending on the experimental design.

The detection of cells is possible through a detectable probe with which the antibodies are labeled; the most widely used probes are fluorophores and magnetic nanoparticles and will be the ones discussed in this review. Antibody-based detection consists of first incubating the cells with the labeled antibodies and running the cells through a suitable detection system. In addition to antibodies, surface molecules can also be detected by aptamers – single-stranded oligonucleotides that can bind to the several different targets simultaneously and detected by fluorescent or magnetic probes like antibodies. Immunopanning is another technique that uses specific antibodies coated on plastic petri dishes to enable the adhesion of cells with cognate antigens. Trevino AE et al harvested neurons with an anti-Thy1/CD90 antibody (from BD Biosciences ( 550402), astrocytes with an anti-HepaCAM antibody from Bio-Techne ( MAB4108) from dissociated neural spheroids and human fetal brain tissue through immunopanning [56].

The Fluorescence Activated Cell Sorter (FACS) was invented by Bonner and Herzenberg [57] for sorting viable cells with the help of fluorescent probes. The first commercial machines were introduced by Becton Dickinson Immunocytometry Systems in the 1970s [58], and over the years these machines have been upgraded to highly sophisticated and high throughput systems that can simultaneously measure as many as 12 fluorescent colors. The populations stained with different fluorophore-tagged antibodies can be separated by the different fluorescent signals that they generate – in terms of wavelength and intensities. FACS is used exclusively for the positive selection and isolation of cells. K Yoshida et al, for example, sorted each DAPI−CD45−CD31−EpCAM+ bronchial epithelial cell for somatic mutational analysis [59]. L Cantuti-Castelvetri et al sorted transfected cells with EGFP or GFP marker [60].

• A single-cell suspension is prepared in phosphate buffer saline (PBS) containing serum usually at the concentration of 10⁶/100µl.
• The cells are incubated with the labeled antibody cocktail on the ice and away from direct light for the suitable duration. An indirect staining procedure is followed, wherein an unlabeled primary antibody and a labeled secondary antibody are used, to amplify the fluorescence signal strength. Several molecules of the secondary antibodies (and therefore the fluorophores) can bind to a single molecule of the primary antibody through the latter’s Fc region. Therefore for each antigen-antibody binding, the signal is amplified through several fluorophore moieties.
• The cells are washed to remove any unbound antibodies and re-suspended in the FACS buffer (PBS with 1-10% serum and 0.1% sodium azide preservative) and stored on ice till sorting [61].

When a fluorophore absorbs light at a particular wavelength, its electrons move from a resting state to a maximum energy state known as excitation. It then undergoes a conformational change wherein the electrons fall back to the resting state, and the excess energy is released as the fluorescence or emission. The difference between the wavelengths of the excitation and emission maxima is called a Stokes shift. This cycle of excitation and emission occurs several times and is read out as the fluorescent signal. Since the emitted light has less energy than the excitation, the emission wavelength of any fluorophore is longer than its excitation wavelength and therefore has a different color. It is important to excite fluorophores at their excitation maxima to increase the intensity of emission. Table 6 lists the spectral properties of commonly used fluorophores.

There are two main aspects of FACS instrumentation – the fluidics which allows the cells to flow in a stream and the optics which detect the cells. Figure 3 describes the schematics.

[enlarge]

Figure 3. Schematic representation of fluorescent activated cell sorting (FACS) [3].

• FACS fluidics

  The objective of the fluidic system is to order the cell suspension into a uniform stream of single cells so that they can be efficiently detected and then sorted. The fluidics system consists of a central core through which the sheath fluid flows under high pressure and encloses the sample fluid. Since the sheath fluid is under higher pressure, it moves faster than the sample and thus creates a drag on the sample fluid. It is this hydrodynamic force on the sample fluid that creates the single stream and allows the cells to be analyzed individually as they pass through the illumination source.

• FACS optics – the lasers and filters

  After a single stream is formed, each cell passes through the different laser beams and is sorted according to its light scattering and fluorescence emission properties. Each laser produces a single wavelength of light at a specific frequency and depending on the wavelength; they are classified as ultraviolet, violet, blue, yellow/green, red and infra-red. Once a laser beam excites the fluorophores, they emit signals at different wavelengths which then pass through specific optical filters. Each filter, in turn, allows only specific wavelengths to pass through and these specific wavelengths are captured by photo-diode detectors and quantified. Most of the modern flow cytometers are equipped with 3-5 lasers and several optical filters per laser and can therefore simultaneously detect multiple signals.

  Apart from the fluorescence, the light scattered by the size and granularity of the cells are also captured and are helpful parameters for analysis. The wavelengths generate photon signals that are captured by the photo-multipliers (PMTs) and converted into the proportional number of electronic pulses which are recorded as ‘events.’ These pulses are digitized by the electronic processing system and stored as FCS files. An entire pulse is quantified by measuring its height and area which correspond to the signal intensity and its width which represents the duration of the cell in a laser beam.

• The electrostatic sorting of fluorescently labeled cells

  Apart from the fluidics and signal detectors, FACS instruments are additionally equipped with a) the machinery to generate cell droplets (usually by high-frequency vibrations) and b) charged plates to deflect these droplets into specific collection tubes. There are five main steps of cell sorting:

  • The cells pass through the laser beams, and the specific fluorescent signals are detected as described in the previous section.
  • The stream containing the cells is broken down into droplets and each cell is captured in a droplet.
  • The droplets pass between the deflection plates.
  • The charged droplets (those enclosing cells) are deflected in the electrostatic field and collected.
  • The uncharged droplets pass through and are collected in the waste chamber [63, 64].

Flow cytometry-based sorting of cells is now a standard procedure in many research and clinical laboratories. With the ever-increasing sophistication of FACS instruments and analysis software, more and more applications are turning to FACS for analyzing as well as isolating various cell types.

• FACS is a highly sensitive and high throughput procedure for isolating cells from heterogeneous populations.
• It is the ideal method of simultaneously sorting multiple populations based on just their immuno-phenotype.
• FACS is a highly versatile technology that can separate cells based not only on surface markers but also cell size and granularity, cell cycle status, intracellular cytokine expression, metabolic status, etc.

The technology of FACS has the following limitations and challenges

• Although FACS can help sort several populations simultaneously, the sorting process itself is slow. This slow sorting is because high flow rates are detrimental to the cells due to shear forces; thus to ensure cell viability and maintain a reasonable sorting speed, the stream flow rate has to be fine-tuned.
• The recovery of most FACS sorters is around 50%-70% making it necessary to begin the sort with a high number of cells. This requirement poses a challenge when rare cells like stem cells and CTCs need to be sorted.
• As the sophistication and precision of FACS instruments increase, so does the probability of technical errors. The operation and maintenance of FACS sorters are expensive and also requires highly skilled personnel for handling, troubleshooting, and repairs.
• Since FACS requires single-cell suspension, information regarding the tissue architecture and inter-cellular interactions are not available.
• Even though advanced FACS sorters provide the option of detecting as many as 10-12 colors simultaneously, it raises the problem of ‘spillover’ of the different fluorophores into non-specific channels. The spillover occurs due to the overlapping spectra of the fluorophores and lowers the resolution between populations that have closely related immune-phenotypes [63-65].

FACS has contributed significantly to the rapid advancements in the fields of hematopoiesis, stem cell biology and oncology in the last couple of decades. In addition, high-speed sorters have eased the recovery of specific blood cell populations in clinics. Some specific applications are enumerated below:

• FACS is the backbone of hematopoiesis research. Much of our knowledge of the biology of different hematopoietic cells, adult stem cells and progenitors and the hierarchical structure of the hematopoietic system come from highly specific cell sorting and analysis.
• The isolation of different kind of cancer cells like CSCs, metastatic cells, CTCs, and blasts Ion the basis of unique immuno-phenotypes is largely carried out by FACS.
• Enrichment of transfected cells, a standard procedure in most biomedical laboratories, has also come to depend on FACS.
• High-throughput speed sorters in clinics are used routinely to separate blood cell fractions such as lineage negative or CD34+ populations for ex vivo manipulation and/or transplantation [63, 64].

[enlarge]

Figure 4. Immuno-magnetic separation of cells using a magnetic separator (a) or MACS column technology (b) [4].

Immuno-magnetic separation of cells is based on the deflection of cells in a magnetic gradient field. The magnetic properties of the cells can be intrinsic, e.g. iron containing RBCs or on account of superparamagnetic particles coated with antibodies directed against specific antigens. Based on the type of magnetic particles used, immunomagnetic separation are mainly of two kinds – Dynabeads®-based [66] and MACS™ technologies [67]. Immuno-magnetic techniques can be used for both positive and negative selection of cells. The labeling procedure of cells is similar in principle to the protocol outlined earlier. The labeling strategy can be either direct (using primary antibodies coated beads) or indirect (using unconjugated primary antibodies and conjugated secondary antibodies). For example, Venkatesh HS et al obtained mouse neurons with Neuron Isolation Kit from Miltenyi, which indirectly labels non-neuronal cells like astrocytes, oligodendrocytes, microglia, endothelial cells, and fibroblasts (except erythrocytes), with biotin-conjugated antibodies specific for non-neuronal cells in combination with anti-biotin microbeads and depletes magnetically labeled cells to obtain highly pure unlabeled neuronal cells [11]. Herb M et al isolated macrophages from mouse peritoneum by magnetic CD11b MicroBeads from Miltenyi Biotec [68]. CD11b-coated microbeads (130-093-636, Miltenyi) was also used with the QuadroMACs separator to isoloate microglia [42]. Figure 4 shows a simplified schematic for immunomagnetic separation [4].

Dynabeads® are hollow, spherical superparamagnetic polymer beads available in sizes ranging from 1-3µm. These beads exhibit magnetic properties only in a magnetic field and have no residual magnetism once the latter is removed. Dynabeads® can be coupled with antibodies (for example, ThermoFisher 11445D, with CD4 antibody, for the isolation of mouse CD4⁺ T cells [69] ) or other ligands to targeting of specific cells or immunoprecipitation reactions. The procedure of using these beads is straightforward – the cells are incubated with the suitable antibody-coated beads cocktail and placed in a magnetic field to separate the bound and the unbound cells. For example, Genet G et al purified endothelial cells from mouse lungs through anti-rat immunoglobulin G-coated magnetic beads precoupled with rat anti-mouse platelet/endothelial cell adhesion molecule-1 [70]. In case the target cells are in the unbound fraction (negative selection), they are merely aspirated out. In the case the isolation is positive, the bead-bound target cells are washed and eluted out. Dynabead® separation can be performed in conical tubes, micro-tubes and even multi-well plates and the latter can be fitted into customized magnets.

The MACS technology was first developed by Miltenyi Biotec™ in the 1990s [67] and since then has become a staple in cell culture laboratories and clinics for the rapid and bulk isolation of desired cells. The target – or non-target cells depending on the selection strategy – are labeled with 50nm magnetic microbeads conjugated antibodies and then subjected to a magnetic field. Boettcher S et al isolated hematopoietic stem and progenitor cells from mouse whole bone marrow with anti-CD117 (c-Kit) magnetic beads from Miltenyi (#130-091-224) [71]. Chopra S et al positively selected CD14⁺ cells with Miltenyi 130-050-201 from human blood/buffy coats and isolated neutrophils through negative selection with Miltenyi 130-097-658 from mouse bone marrow [72]. Garrett-Bakelman FE et al used Miltenyi 130-097-048, 130-097-057, 130-097-055 to collect CD4, CD8, CD19 cells, repectively, from the NASA twin astraunauts [73]. There are three different platforms available for providing the magnetic gradient:

• Magnetic separators – These are powerful magnets into which tubes of suitable sizes can be fitted. This approach is the simplest MACS procedure wherein following incubation, the tubes are simply placed inside the magnet for the stipulated time. Thereafter, the unbound fraction is drained out, and the bound cells remain in the tube – the target cells are present either in the bound or unbound fraction.
• MACS columns – These columns consist of a matrix of ferromagnetic spheres which vastly amplify the magnetic field when placed inside a magnetic separator. The cells are injected into the columns and can move freely in the inter-spherical spaces. The labeled cells are retained in the column in suspension (not bound to the magnetic spheres) while the un-labeled cells flow through and can be easily collected. The bound cells can also be eluted out once the column is removed from the separator. Columns provide two advantages compared to simple separators for positive selection – a) the increase in magnetic gradient increases the sensitivity of detection, especially of the minimally labeled cells and b) since the labeled cells do not bind to the magnetic spheres, there is less stress on the cells.
• autoMACS™ – This is an automated version of the MACS column separation and can be customized for multi-sample sorting.

Advantages

• MACS is a high throughput, selective and rapid method for isolating target populations or removing undesired cells.
• The MACS columns help isolate rare cells with high specificity.
• MACS can be scaled up or down depending on the desired yield and the downstream applications.
• It is a highly versatile technique and can be extended to a wide range of cells and over various platforms including microfluidic devices.

Disadvantages

• In case sorting with Dynabeads®, the latter need to be eluted out owing to their large sizes as it may affect downstream assays.
• Separating cells using magnet separators is not very efficient as draining out the unbound cells still leaves many unbound cells in the tube along with the bound cells. To improve the purity of the sample, the sorting step may need to be repeated several times.
• For delicate cells, binding of the magnetic particles may cause mechanical shear. MACS columns should be used instead of separators to minimize the shear [4, 67].

• Automated MACS is used routinely in clinics to obtain specific blood populations in bulk – it gives a higher yield than FACS.
• Magnetic separation is the preferred technique to de-bulk the blood (or any other tissue) of undesired cells before turning to more sophisticated sorting techniques (like FACS) for the isolation of specific cells.
• MACS is also used to enrich for certain populations like T-cells, monocytes, etc. before in vitro culture [4]. For example, Greenwood DJ et al isolated human monocyte-derived primary macrophages by first obtaining white blood cells through Ficoll-Paque centrifugation, then positive selection with anti-CD14 magnetic beads (130-050-201) and LS columns (130-042-401) from Miltenyi [74].

Aptamers are single-stranded oligonucleotides – DNA or RNA – that can bind to highly specific targets based on structural conformation. Aptamers were first generated using a procedure called SELEX – Systemic Evolution of Ligands by Exponential enrichment – which involves the sequential binding of the oligonucleotides to target molecules [75]. The targets of aptamers have evolved from simple ligands like surface proteins – the first DNA aptamer was designed for the human thrombin [76] - to more complex ones like RBC membranes and even whole cells [77]. Although both DNA and RNA aptamers have similar functionality, DNA aptamers have the advantage of being more resistant to nucleases and are easier to synthesize as there is no need for an additional transcription step [78].

[enlarge]

Figure 5. A concise graphical representation of the Cell-SELEX technique [5].

SELEX was first described by Gold and Tuerk in 1990 [75] and has gone through several modifications in the following years. Figure 5 outlines the basic steps of SELEX [5].

• The target purified protein/cells is incubated with an oligonucleotide library. The library is made of 10¹⁴-10¹⁵ random oligonucleotide sequences between 15-70 nucleotides long.
• The DNA/RNA bound target complexes are separated from the unbound sequences.
• The bound sequences are then amplified by PCR or RT-PCR as required to enrich for the specific aptamers.
• The enriched oligonucleotide pool is incubated again with the target protein.
• Steps 2 and 3 are repeated till the target-specific oligonucleotides are highly enriched - around 15-20 enrichment cycles are normally used.
• The enriched aptamers are cloned in suitable vectors and sequenced.
• The specific aptamer sequences are then chemically synthesized and finally tested against the target to select the aptamer(s) with the highest specificity and affinity.

Once aptamers are selected for whole cells or specific biomarkers, they can be conjugated with different biosensors like fluorophore, luminophores, nanoparticles, etc. for rapid detection of target cells. Depending on the signal output, the aptamer-based assays are classified into four types [78] :

• Direct binding – It is the simplest assay where aptamers labeled with a single moiety of a fluorophore or a luminophore directly bind to the target and the signal can be easily read out with a suitable detector.
• Target induced structural switch – In the absence of the target, the fluorophore-labeled DNA aptamer forms a partial duplex with another (non-specific) aptamer labeled with a quencher. When the target is introduced, the aptamer binds to the target, thereby releasing the quencher from the fluorophore and triggering the latter’s signal.
• Sandwich binding – The specific unlabeled aptamers immobilized on a solid phase first capture the target cells, which is followed by the addition of the biotinylated version of the same aptamers. The signal is generated through further binding of streptavidin-conjugated HRP to the biotin molecules as in ELISA.
• Target induced dissociation – DNA aptamers coiled with gold nanoparticles (AuNPs) are used for this assay. As soon as the aptamers see the target, they uncoil and bind to the target cells and release the AuNPs. The latter are precipitated in a salt solution bringing about a color change.

Aptamers offer several advantages, especially over the antibodies-based isolation methods

• The preparation of aptamers is completely in vitro and thus obviates the use of live animals and highly sensitive hybridoma cultures that are required for the generation of monoclonal antibodies.
• The repertoire of specific aptamers that can be synthesized for targeting any cell is virtually limitless.
• The chemical synthesis of aptamers and the scale-out PCRs are easy to perform and cost-effective.
• The aptamer assays are highly reproducible, and variability is generally very low between different batches [77, 78].

There are some potential limitations as well:

• Due to the vast number of aptamers generated, there is always the risk of non-specific binding. Therefore several clones need to be tested for specificity which can be time-consuming.
• The cell yield is very low which does not make it an ideal method for the isolation of rare cells.
• The safety and bio-compatibility concerns of aptamers have not been addressed so far, therefore it not yet approved for clinical applications [78].

• DNA aptamers have been largely used for the diagnosis and imaging of cancer cells including pancreatic, colon, breast, prostate and glioblastoma cancers.
• There is a paucity of suitable surface biomarkers for detecting rare cells like stems cells, cancer stem cells, CTCs, etc., which makes characterization difficult. Aptamers, therefore, have a lot of potential in the isolation of these specialized cell populations. Aptamers have been created for separating of embryonic stem cells (ESCs) from differentiated cells, mesenchymal stromal cells and various immune cells [77, 78].

Certain cell isolation techniques make use of more than one cell characteristic (as classified above) and are therefore the combination of at least two different techniques. The objective is to either combine the advantages of both techniques or overcome the limitations of one of them. The combination is usually that of an immuno-label-based technique with a label-free technique utilizing a cell property like density, size, morphology, etc.

Erythrocyte rosetting or E-rosetting is the phenomenon where the RBCs are arranged around a central cell to form a flower-like cluster. This formation occurs due to the specific binding between a ligand on the central cell and the corresponding receptor on the RBCs. The best example of this is the binding of the T-cell surface protein CD2 to the sugar-based LFA-3 homolog on the surface of sheep RBCs leading to agglutination [79]. The target cells increase in density following the rosette formation and can then be separated either by sedimentation or density gradient centrifugation. The specificity and sensitivity of the rosette technique have been improved by adding an antigen-antibody binding step, which also provides the advantage of increasing the repertoire of cell types which can be targeted and purified.

This immuno-density principle has been adapted by StemCell Technologies™ in their RosetteSep™ technique [80, 81]. It is a rapid and easy cell procedure for the isolation of purified cells directly from whole blood using their patented tetrameric antibody complex (TAC) technology. The protocol first entails the incubation of blood with the specific antibody-based enrichment cocktail, during which time the antibody complexes cross-link the target cells to the RBCs forming ‘immuno-rosettes.’ Immuno-rosetting increases the density of the target cells, and they can be pelleted out along-with the RBCs following a density gradient centrifugation.

Immuno-rosetting can be adapted to both positive and negative isolation tactics. In case the enrichment cocktail aims for negative selection, the unwanted and pelleted cells are discarded while the desired cells can be collected from the gradient medium/plasma interface. If the selection is positive, the desired cells are present in the pellet – therefore after the pellet is retrieved, an RBC lysis step has to be performed to isolate the target cells. Immuno-rosetting offers the advantages of speed, ease of performance, high yield and direct isolation from blood. The only drawback is that the individual isolation of different populations from blood becomes complicated using this method; the preferred techniques for multiple isolations would be FACS and MACS [80].

The microdissection of specific cells from stained or unstained tissue sections with the help of a precise laser beam (LCM) has been discussed in detail above. The precision of LCM can be further improved by using immuno-stained sections which allow the capture of cells based not just on morphology and tissue location, but also the immunophenotype. To this end, a rapid immuno-staining procedure has been developed for frozen tissue sections for subsequent microdissection and downstream assays [82, 83]. This novel method, aimed at adapting the immunohistochemistry technique for immuno-LCM, is significantly different from the typical immuno-staining protocols in the following ways:

• Immediately after sectioning, the tissue sections are fixed in a cold solution of methanol or acetone.
• The staining steps with the primary and secondary antibodies are performed for a very short duration (90-120 seconds).

These alterations aim to prevent any drying of the sections through the staining process to prevent the detachment of sections during the laser capture.

Immuno-LCM provides a rapid and high throughput method of isolating multiple cell types from a tissue based on their distinct immuno-phenotypical characteristics. It is, therefore, a very suitable approach to study [82, 83] :

• different types of tumor cells in complex malignant tissues
• the stem cell niches in terminally differentiated tissues.

Microfluidics is a technology for manipulating fluids at the micrometer scale. It has been adapted in clinical and biomedical applications, where it is frequently known as the lab-on-a-chip (LOC) technique [84, 85]. The cell preparations are loaded onto the microchips and subjected to an external force which can sort different cell populations on the basis of specific physical and/or biochemical properties. Micro-fabricated channels, chambers, and valves form the core ‘machinery’ of these microfluidic cell sorting systems. Compared to conventional cell sorting tools like automated FACS and MACS, microfluidic cell sorting offer:

• Faster sorting rates and higher output
• Simpler operating procedures, portability, and reduced costs
• Reduced biohazard risks
• Improved purity of sorted samples

The ultra-small dimensions of microfluidic technologies make it easier to manipulate cells and enable faster detection which make them perfect for in-situ testing. Microfluidic or LOC devices are capable of integrating as well miniaturizing multiple laboratory procedures which is highly desirable in the ever-advancing biomedical and biotechnological translational research. Microfluidic sorting has great potential in isolation highly pure cells for use in biomedical and clinical applications. At present, microfluidic devices integrated with FACS are used in isolating stem cells, lymphocytes and circulating tumor cells (CTCs) [84, 85]. Microfluidic sorting can be classified in two ways.

• Active or passive – Active sorting uses some kind of an external force field like electric or magnetic in order to separate the cells. Passive sorting, on the other hand, relies on cell mass or density and requires either gravity or some mechanical force to sort out the different cells.
  [enlarge]

  

  Figure 6. Schematic representation of the main microfluidic principles – a) hydrodynamic, b) acoustic, c) electrophoretic, d) optic, e) magnetophoretic, f) filtration [6].
• labeled or un-labeled – The cells can be sorted on the basis of their inherent properties e.g. the hemoglobin content of RBCs make them inherently paramagnetic, which can be used to separate RBCs from other cell types in a suitable magnetic field. The other approach is to use antibodies tagged with suitable labels targeted against specific cell antigens – such methods are usually the micro-version of FACS and MACS technologies.

This review will briefly discuss the different physical principles employed to sort cells in microfluidic devices. A recent review [6] has tabulated the main features of these different microfluidic principles and an abridged version of the same is presented as table 7. Figure 6 outlines the six microfluidic techniques [6].

Hydrodynamic force – the force generated by moving fluid – is used in several microfluidic devices to sort cells based on their sizes passively. These devices consist of straight, spiral or curved micro-channels, into which the cell suspension is injected and then fractionated due to inertial and drag forces exerted by the fluid movement. When the fluid flows past the cell surface (as well as the channel walls), it exerts an inertial lift force on the cells; this is counterbalanced by the drag or fluid friction which acts opposite to the relative motion of an object moving with respect to a surrounding fluid. The lift and drag forces are generated and controlled by sequentially contracting and expanding the channel diameters. The greater is the size of the cells, the greater is the lift force required to move the cells; therefore cells of different sizes have different flow trajectories when subjected to controlled hydrodynamics and migrate to different channels [86].

Acoustophoresis refers to the movement of an object in an acoustic pressure gradient generated by ultrasonic waves. Acoustic microfluidic devices have been developed for spatiotemporal manipulation of cells based on their different properties like size, density, fluorescence, etc. through one of the following types of acoustic waves:

• Bulk acoustic standing waves – Bulk acoustic standing waves occur when a microfluidic channel is excited by ultrasound to the resonance at which the applied wavelength matches the spatial dimensions of the microfluidic channel. The magnitude of this acoustic force is directly proportional to cell volume while the direction of this force depends on the cell and fluid densities.
• Standing surface acoustic wave – Devices operating on the SSAW principle form a standing wave along the floor of the channel using inter-digital transducers which are installed on both walls of the channels.
• Traveling acoustic waves – These are non-standing surface acoustic waves that emanate at the fluid surface and travel to the microfluidic cavity with the help of transducers [87].

These techniques involve the application of an electric field that results in the migration of cells based on their surface charges – either inherent or bestowed by the fluorophore-tagged antibodies. This approach is similar in principle to FACS as it charges aerosol droplets for electrostatic sorting. Electrokinetic mechanisms are of 3 types:

• Electrophoresis – The movement of particle or cells towards an oppositely charged electrode under the influence of a direct current (DC) in a uniform electric field constitutes the basis of electrophoresis. Most cells possess a slight negative charge and therefore migrate to the positive electrode. In addition, the fluorescent or magnetic probes used to label the cells also respond to the electric field and are thus used to separate cells by surface charge.
• Di-electrophoresis (DEP) – DEP, such as the DEPArray technology [88], refers to the movement of cells in a non-uniform electric field due to their polarizability to an alternating current (AC), . Rather than the surface charges, it is the electrical permeability of the cells relative to the fluid that determines the response to the AC. Higher is the relative permeability of the cells – which in turn depends on the size and other properties of the cells –greater is the migration towards the stronger fields. On a microfluidic scale, electrophoretic forces are applied by positioning electrodes at strategic locations along the sorting channels.
• Electro-osmosis – Electro-osmotic flow refers to the movement of a fluid due to the electrically induced migration of solvated ions in the fluid. This approach mitigates the potentially adverse effects of applying a continuous electric current such as generation of aerosols or cytotoxic by-products like H₂O₂ [89].

Cells can be sorted in a magnetic field by their electromagnetic properties – either inherent as in case of iron-rich RBCs or through magnetic nano-bead coated specific antibodies. Separating one cell type from a heterogeneous population by the former’s magnetic signature requires a simple chip design wherein a magnetic gradient is set up using a ferromagnetic nickel wire. High-throughput sorting of several populations calls for more complex microfluidic devices; all of them have the typical design of a central flow channel that splits into several vertical comb-like side channels, fitted with magnets of varying (field) strengths. The cells are first incubated with magnetic beads coated antibodies – the different target populations are each labeled with different sized beads. Greater the size of the magnetic particles, stronger the magnetic field required to deflect them. Therefore, these multiple populations can be sorted into the different side channels depending on the size of the particles they are carrying and the strength of the channels’ magnetic fields [90].

Optical manipulation and sorting of cells require a focused laser beam which can trap cells owing to the difference between the refractive indices of the cell and its surrounding fluid. The difference in refractive index results in optical scattering which pushes the cells away from the source of light; at the same time, the gradient of the radiation pressure attracts the cells to the point of highest intensity or focusing maxima. When the gradient overcomes the light scattering, the cells travel towards the maxima and are trapped into ‘optical tweezers.’ Optical sorting principles have been used to adapt FACS on a miniaturized scale. Microfluidic optical tweezer devices have also been designed for high precision single-cell isolation (see later). Optical cell sorting microsystems have also been devised that utilize more than one wavelength of laser beam to sort several populations. The different optical tweezers can separate the cells based on the cell size, fluorescent intensity (inherent or from labeled antibodies) and the laser power [91].

Microfiltration is an entirely passive sorting procedure that makes use of cell size and their ability to pass or not pass through micropores. Cells smaller than the pores can filter through and be collected in a micro-well, while the larger cells are trapped in the ‘sieve’, for example, 1 70-um filter for mouse embryonic fibroblasts [92], Fisher 40μm cell strainer 22363547 [93]. Microfiltration devices have been designed to sort several populations at once by using arrays of micropores of different target sizes or chambers in a microfluidics device [94]. Furthermore, the precision of this procedure can be improved by labeling different cell populations with specific antibodies tagged with density beads [95].

The conventional cell-based assays study specific cell populations and therefore read out the average response from that population. However, this average readout is not representative of every cell, and thus the apparent limitation of population-based assays is that relevant but rare cells/sub-populations are lost in the crowd. In any given population of cells – especially in an in vivo microenvironment – the individual cells have unique characteristics as a result of different cell cycle status, exposure to stimuli, epistatic gene regulation, etc.

There are two major objectives of studying cells at the single-cell level:

• Identification of rare cells in a heterogeneous population
• Studying genomics, transcriptomics and proteomics at the level of the basic biological unit – the cell [96].

[enlarge]

Figure 7. Schematic representation of common single-cell isolation techniques [7].

The principle of isolating single cells is similar to that of general cell isolation procedures – identifying unique biological characteristics of the cell type that can help separate it from all the other cell types in a complex population. The properties can be physical such as size, density, electric charge or expression of specific proteins which can be detected using labeled probes. As with general cell isolation techniques, single-cell isolation techniques also need to have a high throughput, purity, and yield of the target cells. Figure 7 summarizes commonly practiced single-cell isolation techniques [7].

The isolation of single or very few cells (1-4) through the serial dilution of the cell suspension is termed as limiting dilution. It is used extensively in microbiological and mammalian cell studies to create monoclonal cultures. Limiting dilution is performed since the conventional micro-pipetting approach cannot aliquot cells at such low densities. On the other hand, serial limiting dilution increases the probability of obtaining a single cell in an aliquot as per the Poisson's distribution [97].

The principles and methodologies of FACS have been discussed in detail above. Isolating single cells as opposed to specific populations involves the same principles – with a difference only at the end stage output. Instead of collecting cells in tubes, single cells are dispensed in multi-well plates or mini-tubes at one cell per well/tube. To enable single-cell sort, a standard sorting instrument needs to have two main modifications

• Adaptable collection platforms with specifications that allow fitting of FACS tubes, micro-tubes, 384-, 96-, 24-, and 6-well plates, as well as calibrated slides.
• An inbuilt Computerized Cell Deposition Unit (CCDU) which is an XY coordinate system that allows accurate placement of cells during the single-cell sort [98].

Laser capture microdissection (LCM) has been described earlier. The basic technology of LCM has been scaled down to isolating single cells from solid tissues by the Zeiss PALM MicroBeam system which makes use of contact-free laser pressure catapulting (LPC) to capture cells. A short defocused infra-red laser pulse is used to ignite a local plasma below the previously cut cell following which the plasma impulse catapults the cell against gravity into a suitable collector. To further increase the precision, Zeiss has integrated the PALM MicroBeam system with optical tweezers (PALM MicroTweezers) – high precision instruments that use a focused laser beams to create a force field called the optical trap which can move microparticles including cells, depending on the latter's dielectric properties [99].

Micromanipulators consist of an inverted microscope fitted with ultra-thin glass capillaries on a motorized stage. These micro-pipettes, in turn, are connected to an aspiration and release unit. The cell sample – usually in the form of a single-cell suspension – is taken into a well plate of specified dimensions and placed under the eyepiece. By visual examination, the specific cells are selected, and each cell is pulled up through one capillary using suction. The aspirated volume can be transferred to the suitable collection tubes for downstream assay or culture [100]. For example, Gkountela S et al selected individual CTC cells or clusters from a Parsortix cassette with a 30 um glass capillary on AVISO CellCelector Micromanipulator [101].

Microfluidic devices have scaled down standard cell sorting techniques to microscopic dimensions. There are four main microfluidic devices – often used in conjunction with FACS or MACS – that have successfully been adapted to isolate cells at the single-cell level [7] :

• Droplet-based devices – These devices use oil filled channels in which an aqueous droplet can be trapped and then isolated. The probability of a droplet containing a cell follows Poisson's distribution as in the limiting dilution technique.
• Pneumatic membrane valves – These devices use pressurized air to move elastomer membrane or valves that cover the microfluidic channels. Whenever air pressure deflects the membrane, it opens or closes the channels. A cell detection system (e.g., fluorescence, electric charge, etc.) is needed to control the pneumatic pump, and whenever a cell is detected, the resulting air pressure opens the valve and lets the cell enter the fluidic channel.
• Hydrodynamic traps – These devices follow a passive sorting approach which uses cell size for separation. Hydrodynamic force propels the cells one after the other into the traps of specific sizes which allow only one cell to enter at a time. The throughput of this system can be increased by increasing the number of traps.
• Optical tweezers – The principle of optical tweezers has been discussed above. The microfluidic device is fitted with a CCD (charge coupled device) camera that first captures the image of the target cells (recognized either by morphology or immuno-staining). The optical tweezers are then used to move the target cells with high precision into a suitable channel or trap.

The isolation of single cells has to be interfaced with the requisite analysis tools that have been specifically designed for a single cell. The four downstream assays performed with single cells are briefly discussed below:

The major challenge in the single-cell genome sequencing is the meager amount of genomic material available. Amplification of such amounts of DNA (~6 pg per cell) with traditional PCR methods would result in one of the two following errors [102, 103] :

• Allelic dropout (ADO) – ADO is defined as the random non-amplification of one of the alleles present in a heterozygous sample. This is a common problem in single-cell genotyping and results in a high rate of misdiagnosis.
• Preferential amplification – It refers to random over-amplification of one of the alleles in comparison to the other.

  To override these concerns, two novel techniques have been developed for single-cell genome amplification:

• Multiple displacement amplification (MDA) – This is a non-PCR based amplification that requires a high fidelity DNA polymerase like the bacteriophage 29 pol and hexamer primers. Amplification is carried out at a consistent temperature of 30°C and results in large sized products with a low error rate
• Multiple annealing and looping-based amplification cycles (MALBAC) – This a linear amplification method that uses special primers which enables the looping of the amplicons. This prevents exponential DNA synthesis and amplification bias [102, 103].

The analysis of the single-cell transcriptome is performed with RNA-sequencing rather than cDNA microarray due to the higher sensitivity and lower sample requirements of the former. As direct RNA sequencing is not possible yet, a 3 step procedure is performed - a) RNA reverse transcription into first-strand cDNA, b) second-strand synthesis and cDNA amplification, and c) cDNA sequencing. There are two major amplification strategies used:

• SMART-seq – Switching mechanism at 5′ end of the RNA transcript or SMART is a PCR based amplification technique that uses the M-MuLV reverse transcriptase (RT). The RT adds three to four cytosines to the 3′ end of the first cDNA strand which can attach to a universal PCR primer thus ensuring strand specificity and the amplification of only the full-length transcripts.
• in vitro transcription (IVT) – IVT refers to the template-directed synthesis of RNA molecules from short oligonucleotides to several kilobases long. It usually uses the T7 coliphage RNA polymerase and requires a template that includes the promoter for T7 RNA pol upstream of the sequence of interest. IVT offers the advantages of higher specificity and fidelity compared to the PCR based methods. [102, 103].

The major technical challenge of single-cell proteomics is the same as encountered with single genomics and transcriptomics – the ultra-low amount of biological material available for analysis. Additionally, unlike nucleic acids, there is no possibility of amplifying protein molecules. Therefore the traditional protein analysis techniques like gel electrophoresis, immunoassays, chromatography, and mass spectrometry cannot be considered for single-cell proteomics. In recent years, microfluidic devices for isolating single cells have been integrated with various protein assays, also on miniaturized platforms. Examples include:

• single-cell mass cytometry
• single-cell ELISA on antibody-coated microchips
• single-cell western blots [102, 103]

There are two main approaches to culturing single cells

• Dispensing cells in individual micro-wells of multi-well plates – either manually after limiting dilution or as part of automated cell sorting.
• Loading on microfluidic chips – Specialized microfluidic devices have been designed which consist of small micro-wells to trap single cells using gravity and then transferring the cells through channels into larger micro-wells. These microfluidic culturing chips are usually made of polydimethylsiloxane that offers gas permeability and optical transparency. The CellRaftTM system developed by Cell Microsystems is an example of a microfluidic cell culture device [104].

Since the main objectives of the single-cell isolation are to identify and study rare cells and perform –omics analysis at higher precision, the main fields that single-cell technologies have benefitted are oncology and stem cell biology. Some of the specific applications and advancements have been enumerated below [105, 106] :

• Cancer biology – the characteristic trait of cancer cells is their genome instability, sequential mutations, and evolution of neoplastic clones. These distinct clones differ in their proliferation and metastatic potential as well as in their response to chemotherapy. Analysis of tumors at the single-cell level is therefore especially useful in diagnosis and therapeutics
  • single-cell genome analysis of cancer cells has helped elucidate the intra-tumor heterogeneity and identified primary tumor cells, circulating tumor cells (CTCs), metastatic tumor cells and cancer stem cells (CSCs).
  • Studies on the copy number variations and mutations in individual cells has helped determine tumor evolution patterns.
  • The detection and sequencing of the individual circulating tumor cells (CTCs) has improved our understanding of metastasis and has offered new ways of early tumor detection and relapse.
  • Lastly, a combination of single-cell isolation, proteomics, clonal expansion and in vivo transplantation has helped narrow down on the elusive CSCs that persist and are responsible for relapse and metastasis – examples include colorectal cancer, breast cancer and leukemia.
• Stem cell biology – Stem cells are undifferentiated cells that self-renew and differentiate into cells of specific lineages. single-cell approaches are necessary to isolate these rare cells from the heterogeneous populations and harness their potential in autologous regenerative therapies.
  • single-cell isolation and subsequent culture and/or in vivo transplantation has helped identify the real stem cells from the less potent progenitors – the best example being the hematopoietic system.
  • single-cell proteomics can help identify new biomarkers of the rare adult stem cells and further improve their characterization and expansion.
  • single-cell analysis has also shown the heterogeneous nature of the stem cells themselves, e.g. in the case of neural stem cells.

Since the past decade or so, the direction of biomedical research has been going steadily from cell populations to single cells. This, in turn, has driven research into the development of high precision and high throughput technologies for the isolation, culture and –omics analyses of single cells. Concurrently, the need to miniaturize and integrate cell isolation platforms with various assays has led to an upsurge in microfluidic systems. Although the lab-on-chip devices offer high precision, speed, and ease of operation, one single device cannot perform a broad range of applications.

[enlarge]

Figure 8. The working principle of the single-cell printer (SCP) technology – a) the SCP machine, b) microchip dispenser, c) microchip dispenser nozzle, d) imaging cells with the camera and e) target substrate of cell printing [8].

The solution to this is the single-cell printer or SCP which combines an inkjet printer like principle with an optical system for the detection and ‘printing’ of single living cells. Cells are captured in a picolitre-sized droplet which is then deposited by the printer at defined locations for downstream analysis. What makes this technique more advanced than other single-cell manipulations is that it is spatially flexible and can deposit or ‘print’ the cells on several surfaces like a micro-well, microscope slides, micro-tubes, etc. in a single experiment [8]. Figure 8 outlines the four major parts of an SCP [8].

• The micro-dispensing chip - The dispenser chip is a silicon and glass micro-fabricated chamber into which the cell suspension is loaded from a reservoir using capillary action. The silicon membrane of the chip is connected to a piezoelectric actuator which causes a displacement of liquid inside the chamber, leading to the formation and ejection of droplets from the chip nozzle. Depending on the nozzle diameter (10-100µm), these droplets range from 100 pL to 250 pL in volume.
• The vision system – This consists of a highly sensitive camera with a high spatial resolution of up to 0.8µm per pixel. The camera is targeted onto the dispenser chip nozzle and takes a picture before each droplet is generated. In this way, the system algorithm extracts the position of each cell and predicts the number of cells that will be present in every droplet.
• The micro-pneumatic shutter – For single-cell isolation, it is essential that each droplet that is ‘printed’ onto the suitable substrate contains a single cell. Unfortunately, it is not possible to control the number of cells per droplet. However, since the distribution of cells in the droplets is random, the system can separate single-cell-containing droplets from those which contain no cells or several cells. The micro-pneumatic shutter system is installed directly below the chip’s nozzle and working in conjunction with the imaging system; it allows only single-cell droplets to pass through and be printed while shunting away the other droplets into the waste chamber.
• The robotic stage - The droplet generator, vision system, and pneumatic shutter are linked in a compact ‘print-head’ which is mounted on the robotic stage. The stage can be controlled in 3 axes and can be positioned relative to the print-head for printing the cells onto several types of substrates like slides, micro-wells, PCR tubes, etc.

The SCP technology offers speed, precision, flexibility, and cost-effectiveness, in addition to being a label-free isolation procedure. The applications of SCP technology are potentially immense and can revolutionize our understanding of cell and molecular biology at the level of the single cell. SCP has already been used for molecular and genetic characterization of individual tumor cells and has provided valuable insights into clonal heterogeneity [107]. This technique can be further extended to stem cell analysis, developmental biology, lineage mapping and other applications using the single-cell analysis.