High Throughput Methods to Transfer DNA in Cells and Perspectives

Abstract

Genome sequencing led to thousands of genes to study and their molecular cloning to provide ORF collection plasmids. The main approach to study their function involves analysis of the biological consequences of their expression or knockdown, in a cellular context. Given that, the starting point of such experiments is the delivery of the exogenous material, including plasmid DNA in cells. During the last decades, efforts were made to develop efficient methods and protocols to achieve this goal. The present chapter will first give a rapid overview of the main DNA transfer methods described so far: physical, chemical, and biological. Secondly, it will focus on the different methods having reached high-throughput nowadays. Finally, it will discuss the perspectives of this field in terms of future enhancements.

1. Introduction

The most used approach to decipher proteins’ function or their interactome is to study the effects induced by the delivery of exogenous materials in living cells (deoxyribonucleic acid: DNA, ribonucleic acid: RNA, oligonucleotides, proteins, and ribonucleoproteins). Coding sequence overexpression, then gene silencing, and genome editing approaches offer a panel of induced biological modifications within cells that allowed us to increase our knowledge of most cellular processes. However, in a post-genome era, thousands of genes must be studied and exogenous material transfer into cells, including DNA, became a limiting factor. Indeed, available technologies predominantly allowed analysis at a gene-by-gene scale, and new approaches were developed to reach higher throughput. Libraries of material such as small interfering RNA (siRNA) [1, 2] and Open Reading Frame (ORF) expressing plasmids collection were developed [3, 4] to cover all proteome. To take advantage of these, concomitant High-Throughput (HT) technologies are pointed out for their transfer in cells. Plasmid DNA (pDNA) transfer in cells (by transfection or transduction) plays a central role when studying the precise biological role of proteins. For pDNAs, several efficient transfection methods were pushed to higher throughput. All these induced changes performed in cells allow not only our understanding on the biological processes of cells’ life but also have therapeutic applications [5, 6]. The huge interest in gene and cellular therapy approaches is indeed a motor in the development of highly efficient gene delivery strategies.

In this chapter, we will first give a brief overview of DNA transfer methods in cells, then a more detailed part will focus on those that reached higher throughputs and we will conclude with future expected enhancements.

2. The different methods to transfer DNA in live cells

To promote a biological effect in the cell, exogenous DNA must face several levels of pitfalls starting from the outside of the cell. First, it must cross the plasma membrane composed of a hydrophobic lipid bilayer which naturally prevents hydrophilic material such as DNA from entering the cells. In addition, DNA and the plasma membrane carry a general negative charge that impedes DNA transfer into cells by electromagnetic repulsion. Furthermore, once entered, the DNA has to face degradation mechanisms that occur in the cells. Finally, if part of the exogenous DNA succeeds in passing all these steps, the expected biological effect would be measurable. To circumvent all these, a range of approaches to transfer DNA has been developed. DNA delivery to cells can be divided into three main categories: physical, chemical, and biological methods. Among all approaches, viral ones are the most efficient but present some limitations such as the transgene size and biosafety issues. Physical and chemical methods were developed to circumvent these limitations and are not limited in the size and number of genes to be transferred. Some of these are still largely used whereas some were more a proof of concept. In this section, we briefly describe the physical, chemical, and biological methods.

2.1 Physical methods

As mentioned above, due to the plasma membrane and DNA respective properties, the transfer of DNA into cells is impaired. All the physical methods aim to directly circumvent the hydrophobic and electrochemical repulsion parameters by disrupting the integrity of the membrane and promoting a transient permeability. Physical methods then do not have a limit in cargo size and do not depend on biological mechanisms as a direct material delivery is performed [7].

An evident method is the direct microinjection of DNA in cells, which was performed in early-stage embryo [8], and then human cell lines [9]. It implies micromanipulation of a single cell under a microscope to bypass the membrane barrier using a thin glass needle to inject DNA directly into its cytosol or compartments [8, 10]. This approach is reproducible but tedious due to the need to inject each cell individually.

Electroporation method has emerged when it was shown that an electric field could promote a loss of membrane permeability by transient pore formation [11] thus allowing DNA delivery in cells [12]. DNA, target cells, and electroporation buffer laying between two electrodes are submitted to an electric pulse [13]. This pulse is divided into a high-voltage stage to create temporary pores, and a low-voltage one to allow electrophoresis of DNA through these pores [14]. Extensive optimizations (pulse voltage and duration, buffer composition) were done to balance transfection efficiency with cell viability as the requested high voltage promotes cell death [15, 16]. “Electroporators” devices are nowadays available with many predefined settings to achieve efficient transfection in almost all cell types, even hard-to-transfect ones [17].

Biolistic or micro-projectiles bombarded to the cells represent another delivery mode. The projectiles made of gold-covered by nucleic acids, penetrate into cells by high-speed bombardment [18]. First developed for plants, this approach is also efficient in mammalian cells/tissues [19]. However, the method suffers the cost of particles. Nanoparticles that can bind nucleic acids, and whose small size allows them to pass cell membranes with high efficiency, represent a cheaper alternative [20].

Femtosecond laser optoporation consists in focusing ultrashort laser pulses on a cell membrane to induce a transient perforation. This membrane perturbation allows the pDNA transfer [21]. Many cell types can be transfected using a variety of laser sources [22, 23]. Despite efficient, due to the needed laser focusing on a single cell level, its throughput is limited.

Acoustoporation or sonoporation uses ultrasounds to induce a transient plasma membrane disruption promoted by bubbles cavitation phenomenon and thus allowing gene transfer [24, 25]. The method was enhanced by the use of high-frequency waves creating reversible nanopores and furthermore promoting “molecular bombardment” on the bilayer membranes that enhances DNA delivery while limiting cell mortality [26].

Passing constriction or nano-constriction is an approach based on the mechanical deformation of cells as they pass through micro constrictions channels [27]. This controlled compression induces transient pores formation into the cell membrane and allows DNA entry from the surrounding buffer [27]. This method is expected to be universal and showed efficiency in easy and hard-to-transfect cell lines like primary and stem cells [28].

The last method, magnetofection, has been classified as a physical method. It is based on magnetic nanoparticles (MNPs) coated with transfection reagents that bind nucleic acids and promote cell entry [29]. Indeed, MNP only induces the concentration of the MNP on the cells mat when a magnetic force is applied but is per se not able to transfer DNA into cells. However, it enhances DNA delivery up to several hundred and allows to lower DNA consummation, and is furthermore efficient in hard-to-transfect cells [30, 31].

2.2 Chemical methods

Interest to develop non-viral and reproducible gene delivery methods has led to the use of chemical reagents. Chemical transfection methods represent an alternative way to bypass the membrane barrier and furthermore try to protect DNA from degradation within cells [32]. These reagents promote DNA compaction, negative charge neutralization, and cell interaction for later entry into cells. These reagents are briefly summarized here after.

Calcium phosphate co-precipitation is the cheapest method and was first described in 1973 [33]. It relies on the formation of a precipitate when the negatively charged DNA binds to calcium ions (Ca²⁺) [34]. This precipitate interacts with the plasma membrane and enters the cell by endocytosis [35]. This widely used method reaches up to 90% efficiency for easy-to-transfect cells but is impaired by the need for fresh preparation, avoiding any storage of ready-to-transfect plates [36]. The formation of an efficient precipitate depends on several parameters and this method can be toxic for certain cells such as primary ones [37]. Calcium was also shown to enhance gene delivery by other methods [38] and was then tested alone as a transfection reagent (calfection) [39]. The mechanism does not rely on the formation of a precipitate and do not need fresh preparation. Furthermore, the Ca/DNA mixture can be stored for a long period without any loss in efficacy. Intended for batch transfection of the high number of cells, it worked in a 12-wells plate format for adherent or non-adherent cell lines. The easy use, storage ability, and low cost make this method interesting whereas it was not tested so far in higher throughput.

The diethylaminoethyl-dextran (DEAE-dextran) is another reagent that showed efficiency [40]. This polycationic derivate of dextran compacts DNA to form a positively charged complex that later interacts with the plasma membrane to enter cells by endocytosis [41]. The method is simple, low cost, and efficient for many cell types however, new enhanced approaches surpassed it.

Lipofection method is based on the use of lipids and cationic lipids [42, 43]. When mixed with DNA solution, these lipids form liposomes, a kind of vesicular structure with the same composition as cellular membranes and entraps DNA in solution [44]. The formed complexes (lipoplexes) allow DNA delivery through binding to the cell membrane (due to electrostatic forces), cell entry, mainly by endocytosis [45], and release of the DNA for expression. Lipids-based transfection reagents are efficient and mostly insensitive to serum so that medium has not to be removed before transfection. Furthermore, lipofection can be used efficiently in forward or reverse mode transfection in numerous cell lines [46]. Cationic lipids are more and more efficient in DNA delivery, and furthermore efficient on suspension or adherent cells, and for increasing number of cell types, and even hard-to-transfect ones [47].

Cationic polymers are non-lipidic as deprived of a hydrophobic moiety and are then soluble in water. They use a similar mechanism: being positively charged, they interact and compact DNA under the form of polyplexes [48]. They enter the cell by endocytosis, and traffic through endosomes and cytoplasm to finally deliver DNA to the nucleus [49]. This class of reagent has the advantage to limits DNA degradation in lysosomal compartments, increasing delivery efficiency [50].

2.3 Biological methods

Biological approaches to transfer DNA are inspired by natural mechanisms. The most potent of these approaches is gene transfer by viruses. Other methods represent fields in expansion: cell-penetrating peptides or the use of exosomes or vesicular transfer. These approaches do not rely on natural products but on diverted forms to allow the transfer of a gene of interest.

Viral approaches are the highest efficient among all, even in hard-to-transfect cells [51]. To be permissive, the cells must express the receptor interacting with the virus envelope proteins. To enter in almost all cell types, a ubiquitous and widely expressed receptor is preferred. The Vesicular Stomatitis Virus G (VSV-G) protein promotes entry in almost all cell types as the Low-Density Lipoprotein Receptor family is its ubiquitously expressed receptor [52]. Its interaction with the VSV-G protein promotes membrane fusion and allows virus content to be delivered to the cells [53]. The use of a viral vector is however limited in throughput as viral particles have to be produced for each different DNA to transfer. This production involves the cloning of the gene of interest in a viral vector backbone that is later transfected into a packaging cell line to be integrated into pseudo-viral particles. Pseudo-virus are then recovered from the cell’s supernatant, concentrated, and titrated before their use for transduction of the target cells. Despite lower throughputs, viral delivery remains the most powerful way to transfer DNA in cells, even in primary cells (90% efficiency).

The fusiogenic envelope G glycoprotein of the VSV-G was also used as a reagent for gene transfer when mixed with plasmid DNA [54]. The resultant product termed “Gesicles” showed 55% transfection efficiency in HeLa cells, and 22% for hard-to-transfect human myoblast cells [55]. Whereas promising, this method did not reach HT yet.

Another interesting biological derivative used for DNA transfer is represented by proteins having natural properties to enter the cells by surface receptors dependent [56] or independent mechanisms [57]. Some natural peptides derived from these proteins, the cell-penetrating peptide (CPP) are able to enter the cell through the membrane [58, 59]. These peptides have short lengths and a global positive charge. Involved mechanisms are still unclear and depend on the CPP (direct penetration, endocytosis, or translocation via intermediate structure in the membrane lipid bilayer). Peptide from the Trans-Activator of Transcription (TAT) protein of the Human Immunodeficiency Virus (HIV) was efficiently used as a DNA carrier in HeLa cells [60]. Some others have been modified and their properties mixed with each other to promote efficient delivery of exogenous nucleic acid into cells [61]. CPP can be engineered by multiplexing peptides with distinct properties or by modifying their composition. One of the engineered CPP is the pepFect14 [62] which showed efficiency for DNA delivery in several cell types such as CHO, HEK293, U2OS, or U87 cells [63].

One last example of naturally occurring biological derivatives is the use of exosomes. First described in 1977, these nano-sized vesicles derived from plasma membrane elements, are involved in mediating messages to proximal and distant cells [64, 65]. This natural process is found in normal or pathological cells [66] and can be turned around to deliver DNA of interest [67].

3. High-throughput batch DNA transfection

To enhance the throughput of the experiments performed on cells transfected by exogenous DNA, it is interesting to do it in a HT way. However, a distinction must be done between experiments performed at HT using transfected cells, and HT transfection of cells. Indeed, depending on assay requirements, transfection of a single condition may be performed using a large volume of suspended cells that are then distributed among several individual wells for subsequent treatments and assays [68]. Alternatively, it is interesting to transfect many different plasmids, each well of transfected cells expressing different transgenes [69, 70]. This difference is generally concomitant with the way the transfection is performed: batch protocol or not.

Batch protocol allows to transfect a large number of cells that are then dispatched in separate wells for further experiments. In this case, all transfected cells in the batch share the same conditions of transfection. This protocol is generally used to limit variability in HT assays for monitoring the effect on a biological parameter under a unique transfection condition. Typically, it can be performed on adherent cells in a forward-protocol modus: cells are plated and transfected 24 h later, according to the transfection reagent’s manufacturer instructions. The day or several hours after transfection, adherent cells are suspended and dispatched in multi-well plates (96, 384, or even 1536) for further HT treatments and analysis [71]. Depending on the cells used, the batch transfection is also compatible with the suspended cells that are then directly dispatched on separate wells after transfection.

The batch protocol is not per se a HT transfer of different biological materials in cells, but rather a way to perform HT assays and treatments in separate wells. On the opposite, HT protocols can achieve a true HT transfection in which each well receive a different DNA or transfection conditions.

4. High-throughput transfection protocols

To be able to determine the behavior of cells or biological effects induced by the transfer of many different pDNAs in the cells, a real HT transfection becomes interesting. Several methods allow the management of numerous pDNA or different conditions when transfecting the cells, but to reach HT, good efficiencies are almost necessary. HT transfection can be achieved using several methods that are presented below.

4.1 High-throughput transfection using physical approaches

4.1.1 High-throughput electroporation-based transfection

As described before, electroporation is performed with buffer diluted cells and DNA, subjected to an electrical pulse that promotes membrane destabilization. Many devices, protocols, and dedicated buffers have been implemented to reach universal use. However, it seemed incompatible for HT as each separate transfection must be performed one by one in micro cuvettes. This problem has been solved by the development of new devices able to deliver an electrical pulse simultaneously in several wells on dedicated plates. Harvard Apparatus/BTX developed an up to 96 wells approach using plates with embedded aluminum electrodes. Used with the plate handler Model HT-200, it allows transfection in 8 wells simultaneously and was shown efficient in neurons [72].

Another approach based on an array of 96 suspended electrode pairs fitting on top of standard 96-well plates represents a less expensive approach [73]. Each pair of electrodes can be loaded and held 10–20 μL of transfection mixture by the surface tension. After pulse delivery, the direct addition of the cell culture medium into the array allows the electroporated cells to drop and seed into the underlying microplate. The array is reusable, and uses standard microplates and inexpensive standard buffers, reducing the cost of this approach. In addition, common liquid handling robots can achieve a 96-well transfection time of approximately 1 min. This technology could be adapted to the 384-well plate format using a more sophisticated electrode array design and concomitant robotics.

Whereas successful in almost all cell types, the electroporation method has some limitations. First, it is the most DNA-consuming one of all the HT transfection approaches described so far. Secondly, as a cell suspension is required during the electrical pulse delivery, it avoids its use on adherent differentiated cells mat. Nevertheless, its advantage in terms of success in almost all cell types, and its versatility concerning the material to transfer (not only efficient for DNA) promises electroporation to further future enhancements and use.

4.1.2 High-throughput nucleofection-based transfection

In 2001, Amaxa™, (now owned by Lonza™) launched an electroporation-derived method termed nucleofection as DNA is transferred directly into the nuclei of the cells. It shortens the time of experimentations, by suppressing the necessary nuclear import step of DNA and ensure proper expression of transgene [74]. It relies on an electroporation-based device (Nucleofector) and the use of dedicated buffer solutions to ensure nuclear transfer. The exact mechanism allowing nucleus targeting and buffer composition is kept proprietary. However, since the first published results on natural killer cells transfection [75], it has been widely used in many hard-to-transfect cells with efficiency ranging from 25 to 70% [76]. First, nucleofector devices used nucleocuvettes and were then limited in throughput. New apparatus and dedicated consumables were developed to reach higher throughput: the 96-well Shuttle® device (amaxa AG), in which cells are plated on 96 wells “nucleocuvette plates” and pulsed using Nucleofector™ programs. These plates, made of conductive polymers, allow the current delivery in each well individually. It takes less than 10 min in an automated way to process the entire plate [77]. Many optimized conditions have already been defined using nucleocuvettes depending on cell types (programs for the electrical pulses, cells number, and optimal buffer conditions) and as an advantage, these settings are transposable to nucleocuvette plates [17]. Numerous successful examples have been published ranging from 35 to 70% efficiency: primary chondrocytes [78]; dendritic cells [79], and even H9 hESC [80].

To push further the throughput a 384-well Nucleofector™ requiring 384-well Nucleocuvette™ plates was launched. The complete electrical pulse delivery process takes just one minute, and several wells are processed at the same time [81]. However, the overall process is the same as for the previous model, mixing of cells with buffer and DNA in the wells, delivery of the electric pulse, the addition of fresh medium in the 384-well Nucleocuvette™ for cell recovery, and then their dispense in a cell culture plate for later experimentations.

Whereas versatile and being efficient as electroporation in many cell types, nucleofection is still restricted to suspended cells, impairing its use on morphologically differentiated and adherent cells. Furthermore, the need to transfer transfected cells to a standard plate for further experimentation is a limiting step of the method. The cost of such approaches broadens their wide use in the scientific community due to the price of transfections kits, containing ready-to-use buffers, and nucleocuvette plates.

4.1.3 High-throughput adherent cells electroporation-based transfection

As explained above, electroporation or nucleofection are restricted to cell suspensions. To circumvent this limitation, new kind of electrodes able to deliver the electric pulse on a cell mat was developed.

One of the simplest developed approaches is an electrode device that takes place on top of standard culture cell dishes. The PetriPulser™ (from BTX) consists of 13 gold plated electrodes embedded in an isolating holder placed above the Petri dish containing the cell mat to electroporate [82]. This model fits 35 mm Petri dishes but a scaled-up model, the “Petri dish electrode” made of stainless steel electrodes, fit 100 mm diameter dishes [83]. The 2 mm distance between electrodes is the same as in most cuvettes. A model for transwell cultured cells electroporation: the BTX™ Adherent Cell Electrodes [84] presents a 5 mm distance inter electrodes and may engender adverse effects on cell viability. All these devices are reusable, lowering the cost of this approach that has however not been used so far in published works.

A sophisticated version was launched by Cellectricon™: the Cellaxess®HT. It uses dedicated 384-wells microplates and a capillary embedded microelectrodes array. Using a platform device, adherent cells seeded in 384-wells plates are washed, electroporated using transfection mixture (loaded from side donor plates), and allowed to recover with fresh medium addition. 96 wells are simultaneously electroporated by the device and throughput of 50,000 wells per day is announced by the manufacturer [85]. However, it was not really used in the academic laboratories as no work has been published except the proof of concept of the manufacturer. They simplified the method by launching the Cellaxes Elektra-Adherent Cell Electroporation System. It is also an electrode-based electroporation system optimized for the in-situ transfection of all adherent cell types, which offers superior efficiency and cell viability due to minimal cell processing and the low voltages enabled by the use of capillary electrodes laying above the cell mat. It uses 384-wells plates and delivers the electrical pulses in 96 wells simultaneously thus allowing the rapid management of the entire plate. However, whereas fully automated, the protocol is not homogeneous: some medium must be discarded from the wells to add DNA diluted in the electroporation buffer (Cellaxess Elektra Accelerator Solution). Once the pulse delivered, some fresh medium is added to the wells before returning the plate to the incubator. Such inhomogeneous protocol would render reproducibility harder to achieve. Cellaxess Elektra transfection system allows rapid optimization of the protocol as different pulse protocols can be applied in a single 384-well plate. This approach has not been widely used yet, probably due to the cost of consumables and devices, but was able to transfect primary neurons with an efficiency of up to 50% [86].

4.1.4 High-throughput electroporation-based microarray in situ transfection

Array approaches are based on spotting an array of transfection reagents and material to transfer on a planar slide where cells are later plated. Using such approaches with electroporation method was unimaginable. However, several teams pushed down this restriction by developing custom-made devices to electroporate adherent cells in a microarray manner. Two technologies are suitable for adherents cells: the delivery of the electrical pulse between the bottom and top of chip micro-wells; or between interlaced microelectrodes laying on the bottom of the dishes under the seeded cells [87, 88].

In the HT in situ cell electroporation (HiCEP) method, a microarray electroporation chip composed of 13 × 13 microelectrodes have been developed [89]. The electrodes lay under the cell cultured in a superhydrophobic microwell array chip (SMAR-chip) developed for this purpose. The electrical pulse is delivered simultaneously in the 169 wells, using for each, ten interdigital electrodes covering a 500-μm-diameter area [90]. The approach requires a dedicated platform to assemble the chip before covering it with the cells solution in a Petri dish. Before delivering the electrical pulse, the medium is removed, allowing 24 nL medium nanodrops to stay in each well of the hydrophobic matrix chip. Electroporation buffer is added and rests as nanodrop in each well after aspiration. The material to transfer is deposited by a standard microarrayer, on the top cover slide. Once reversed and placed on top of the wells using a micromanipulator under the microscope, the drops mix with the underlying buffer before electrical pulse delivery. Whereas this method is successful and promising in terms of throughput, it is not affordable for non-specialists, as many skills and specialized materials are required.

Another method was able to electroporate adherent cells, based on a glass gold electrode coated with PEI for pDNA loading [91]. Cells are plated on this electrode and the electrical pulse can be delivered using an additional top cover electrode up to 3 days post-seeding. Transfection efficiency reached 90% in HEK but was also efficient in primary fibroblasts. Although electroporation was performed in 13 mm square areas, this method allowed HT transfection using up to 169 plasmids micro-arrayed on the electrode. This method seems affordable, as it only requires a gold vaporized electrode.

4.1.5 High-throughput electroporation-based microfluidics transfection

Whereas it remains a field of specialists, microfluidic applications increased in the last decade due to their low-cost advantage, as it can be in-house designed using affordable technologies, and it deals with low quantities of reagents. Microfluidic can manipulate different solutions and mix them, and lead to cell culture and transfection chips design [92]. However, in-house designs might be difficult to reproduce, even more, if highly specialized skills are required. Furthermore, most biological experiments require a subsequent amount of transfected cells, harder to achieve using microfluidic. Despite these limitations, success in microfluidic transfection applications has been published for a wide variety of cells, and even at the single-cell level [93, 94]. First devices lacked the necessary throughput to test numerous transfections conditions in parallel, but recent advances pushed it further. In the field of transfection, two main approaches have been used with microfluidics: electroporation and nano-constriction.

Electroporation in standard 2 mm cuvettes requires high voltage that promotes cell death by a joule heating effect, a local pH change due to water electrolysis, that in turn induces the formation of bubbles promoting cells aggregation and impairing the DNA delivery efficiency [95]. Due to its efficiency, electroporation was used in microfluidic derivatives trying to circumvent some of its limitations. Embedding electrodes in a microfluidic channel can limit adverse effects on cell viability [92]. The diameter of the channel allows the electrodes to be closer to each other’s and the use of voltages as low as 1 volt [96], reduces the heating joule effect, electrolysis, and bubbles. pH modifications are still present but enhanced buffer composition improved it [97]. These microfluidics devices mostly use flowing cells transfected in a semi-continuous way [98], avoiding testing many different conditions in parallel and lowering throughput. Some devices allow transfection of adherent cells in micro-chambers using a porous substrate on which cells are seeded. The electric field is then applied through the cells (under/upper compartment). This has been successfully performed on stem cells differentiated in neurons [99]. Despite the latest improvements, microfluidic-based approaches still lack HT. However, due to the booming application of microfluidic, reaching higher throughput would be achievable and a promising way to perform transfection.

4.2 High-throughput transfection using chemical reagents

Most of the chemical transfection reagent allows two kinds of protocols: the forward and the reverse protocol. In forward protocols, DNA and transfection reagent are mixed to form transfection complexes and then distributed on previously seeded cells. Such an approach is harder to manage in a HT way as each different mixture condition implies a different container (tube or wells of multiplate wells) and necessary tedious pipetting steps. However, this kind of protocol can be manually achievable with standard molecular biology material such as multichannel micro-pipettors. An experimented user can transfect one to four 96 well plate manually in 2 h with up to 3 different pDNAs per condition [100, 101]. However, to our knowledge, the forward approach has not been automated so far to reach HT.

The forward mode has been surpassed by the reverse protocol mode. The DNA (eventually with the transfection reagent) is directly dispatched on the final wells (i.e., of a multi-well plate) or a glass slide, and cells are added directly on these deposits. This mode of transfection has several advantages: first it shortens the overall experimental time, second, it can easily be automated allowing to reach HT and good reproducibility. Suitable for such an application, liquid handling devices enable the dispense of low liquid volumes for the multiplexing of different solutions whose concentration and ratio are tightly controlled in each well. Such protocols have been developed for most of the biological material to transfer which includes DNA and follow the technological developments available to do it. An overview of these methods used for DNA transfection in a HT manner is detailed below.

4.2.1 Chemical-based high-throughput transfection

As mentioned before, lipidic transfection reagents are eligible to reverse protocol, making them suitable for potential HT approaches. This reverse mode was shown efficient on CHO cells grown in suspension in a 96 wells-plate format using PerFect Lipids (pFx-6 form lnvitrogen) as reagent [100] and even adherent cells using Lipofectamine (Invitrogen). Higher Throughput was reached using Turbofectin8 as reagent (Origene) and plasmids coding 704 different transcription factors dispensed in 384-wells plates [102].

The SMAR-chip described before in the HiCEP method [89], was also applied to HT reverse transfection but using Lipofectamine 2000 as a transfection way instead of electroporation. It allowed the efficient transfection of HEK293 (up to 65% transfected cells) in the 169 wells of the matrix. The authors aimed at producing viral particles using co-transfection of the necessary plasmids with 169 genes of interest. Proper viral packaging and sufficient viral production were shown by successful transduction of side cultured 3T3L1 cells using the supernant of the HEK producing cells.

Tavernier’s group reached a much higher throughput in 2002 using reverse transfection for HT transfection of HEK293 cells in its MAmmalian Protein–Protein Interaction Trap (MAPPIT) Arrays approaches to study protein–protein interactions [103]. Effecten reagent was used in a reverse mode protocol to transfect prey expressing plasmids in up to 384-wells plate format using classical liquid handling facilities and a mammalian ORF collection plasmids.

With the emergence of such collection, examples of microplate-based arrays of the huge collection of plasmids have grown. One of the highest throughput was reached using 6049 different human cDNA expression plasmids to study their effect on the promoter activation of the zinc-finger protein RP58, using a luciferase reporter gene [104]. 50 ng plasmid/wells were loaded on sets of 384-well plates and a HT reverse transfection of HEK293 was successfully performed using Lipofectamine 2000.

A HT transfection protocol was reached in 384-wells plates format using non-liposomal polymers (Mirus TransitX2) as transfection reagent [101]. A reverse protocol led to about 90% transfection efficiency (even in cotransfection assay). The originality of this work is the use of a tips-free acoustic delivery of reagent and DNA (Echo nanodispencer from Labcyte™). This device sends multiple droplets of 2.5 nL from a 384-wells source plate to a destination one up to 1536-wells plates. Starting from unique diluted plasmids solutions, the overall process takes less than 20 min for one plate, and transfection ready plates can be stored dry or frozen without loss of efficiency. Cells are seeding on dry or freshly dispensed plates in a reverse mode transfection. The optimized protocol would allow 20,000 human genes transfection in about 18 h on a dedicated automated platform. Nano-quantities of DNA and reagents should render this approach low cost if the nanoaccoustic dispenser was not such expensive. Nevertheless, this protocol renders transfection affordable for newbies as the tedious work of DNAs and reagents combining in each well is controlled by spreadsheet driven software [105].

4.2.2 Chemical-based microarray transfection

In 2001, DNA transfection throughput was pushed further by the use of a microarayer for the generation of transfection ready arrays of DNA [106]. In this study, 140 different plasmids DNA/gelatin mixture were deposited on glass microscope slides as 1 nL spots (of about 150 μm diameter). Effecten, a lipid transfection reagent was used to transfect cells seeded on the overall slide. Each spot led to the transfection of 30–80 HEK cells, in a DNA dose-dependent manner from 10 to 50 pg. Storage of the dried glass slides for more than 3 months did not affect transfection efficiency, allowing the matrix to be prepared in advance of use. Since this princeps study, other groups have successfully used this approach. Using the same reagent, one study transfected 16 different plasmids expressing proteins to study their cellular localization [107]. Another group used this approach for the HT screening of potential therapeutic membrane-displayed single-chain antibodies [108]. A true HT attempt was reached by the use of 1959 un-tagged ORF taken from the Mammalian Gene Collection (MGC) and expressed in HEK cells to identify genes implicated in apoptosis [109]. One similar array approach showed efficiency using Lipofectamine 2000 directly in the DNA mixture before arraying [110]. However, whereas simplified by combining the transfection reagent with DNA before dispensing, it requires about 10-fold more DNA to reach the same efficiency as the above protocols. A throughput of 2880 conditions on a complete 96-wells plate to study v-Src Mutant Protein Function was reached in HEK cells, using 30 spots of pDNAs mixtures per well of 96-wells plate [111]. Lipofectamine 2000 also showed efficiency in another microarray approach testing 600 cDNA spots on a single glass slide using reverse transfection [112]. Authors showed high efficiency in many cell types such as mouse preadipocytes (3T3L1), muscle myoblasts (C2C12), liver hepatoma (Hepa1c1c7), or macrophage (RAW-164.), human cervix epithelia adenocarcinoma (HeLa), or at bone osteosarcoma (UMR-108).

Tavernier’s group also pushed further its MAPPIT and MAmmalian Small molecule-Protein Interaction Trap (MASPIT) microplate-based array to microarrays using attractene (Qiagen), a non-liposomal lipid, as transfection reagent and a fluorescent reporter gene in place of the initial luciferase reporter [113]. Here, the ORFeome derived prey plasmid collection (15,000 cDNA) and a fluorescent reporter plasmid was mixed in 384-wells plates used as a matrix for further depositing by a microarrayer on polystyrene plates, to reach an industrial scale.

All these arrays’ methods are impressive in terms of throughput as many conditions, or different expressed genes, can be tested simultaneously in parallel cells. However, they require a consequent preparation time. DNA dilution, most of the time with gelatin, and optimally with the transfection reagent are generally performed in 96 or 384-wells plates. Once done, an arrayer robot is then plunging its tips for deposition of the DNA on several slides. The tips must be washed with detergent and then sonicated or heated to avoid cross contaminations before arraying the next DNA mixture. Finally, when the full array is printed, the slides have to be dried for 12 h to 2 days before later use and cell seeding. At the end of the experiment, a slide scanner became necessary to analyze transfected cells. The real throughput of such methods is then truly high once the arrays are ready to be incubated with the cells. Once the reagent used is efficient with the cell type requested, the throughput becomes only dependent on the liquid handling facility available in the lab. However, the method needs a certain financial investment for robotics, microarrayer platform and a scanner as the spots size and inter-distance need high resolution scanning to be analyzed.

4.2.3 Chemical-based high-throughput microfluidic transfection

As previously mentioned, microfluidic is now widely used due to the miniaturized scale it allows. Whereas it was applied to transfection using nanoconstriction or electroporation, it can also be used as a liquid and cell manipulation tool to perform transfection using chemical reagents. Schudel et al. first developed an inexpensive microfluidic-based miniaturized RNAi screening platform [114]. It relies on the use of a lipid-based transfection mixture and is low throughput as a maximum of 8 parallel transfections can be performed on this chip.

In another study, a two microchannel irrigating 8-chambers was designed on a glass slide [115]: 10 nL of a reverse transfection mixture containing gelatin, fibronectin, Lipofectamine 2000, and plasmid DNA were arrayed on a coated glass slide. This slide is mounted under the microscope, to face the microfluidic embedded chambers and showed successful transfection of Cancer LBT-N2b cells with almost no induced mortality, but the throughput was still clearly limited.

Based on the same kind of chambers design, the highest throughput was reached with a microfluidic chip of 1.6 × 5.8 cm containing 280 separate chambers. In about 10 min, the complete chip is loaded with about 600 cells per chamber of 500 μm diameter [116]. A set of valves allows the loading of different cell densities or even cell types. Once cells are loaded, the functional chip is obtained by alignment of the chambers to 280 DNA arrays (Lipid-DNA transfection mixtures) spotted on polylysine matrixes in an automated manner (about 2 h to complete). The assays showed a high transfection rate (99% efficiency) using an optimized condition but a cell line-dependent optimization is necessary. Whereas feasible, microfluidic managed reverse transfection still seems to have a long road to meet the scientific community mostly due to its required skills in the field, to be able to reproduce or use such devices.

4.3 High-throughput transfection using biological derivatives

As discussed before, some natural biologicals materials, viruses, proteins, peptides, or macromolecules have shown cell-penetrating properties and their ability to deliver different molecules to target cells either in their natural form, modified, and sometimes multiplexed by engineering. Here are some examples of such approaches that reached HT in the delivery of DNA into cells.

4.3.1 High-throughput DNA transfer using viral approaches

The main limiting factors to reach HT with viral delivery is the ability to produce these particles (i.e., biosafety cabinets class 2 or 3), in a HT manner (one independent viral production for each cDNA to transduce), and at a sufficient titer to promote efficient transduction of target cells. This production step has been shown feasible at HT in a pilot study with 1990 ORFs from the mammalian ORFeome collection [117]. In an automated platform, HEK cells were reverse co-transfected with these “gene of interest” plasmids and viral packaging ones to allow the production of the corresponding lentivirus in a 96-wells plate format (viruses transferring one ORF per wells). Supernatants (cDNA containing viral particles) were used to transduce target cells seeded in 96-wells plate format. In a similar manner, up to 16,000 cDNA were pushed to HT lentiviral production in 96-wells plates for later HT expression in target cells [118].

The previously developed SMAR-chip [89] was also used for viral particles production on the 169 matrixes embedded microwells, using reverse lipofection [119]. The method showed sufficient production to transduce 3T3L1 cell cultured in parallel to the producing cells array.

Despite these advances, such methodologies remain difficult to settle routinely due to the required material, specific skills, knowledge, and adequate biosafety facilities. To render it more accessible, some companies now propose ready-to-use kits in 96-wells microplate format to produce viruses in high titers from lab collection of cDNA [120]. Despite these limitations, this approach is highly promising as being universal for almost all cell types with high efficiency and furthermore efficient on suspension or adherent differentiated cells.

4.3.2 Protein-derivatives based HT transfection

One example of protein derivatives used is collagen derivatives, which are produced by collagen treatment or digestion. Atelocollagen, is a polymer obtained by pepsin treatment of type I collagen that shows various effects in cell and animals. Atelocollagen condenses and delivers DNA, antisense oligodeoxynucleotides, or siRNAs into cells on its own [121]. Protocol based on this polymer reached a HT microplate array level in 2001, with a collection of pDNA showing a long-term gene expression in HEK cells [122]. The array can handle long storage without loss of efficiency. Another study reached HT transfection in PC-12 cells using Atelocollagen and 288 different plasmids dispensed in 96-wells microplate arrays [123]. The advantages of these last approaches remain in the fact that atelocollagen intrinsically regroups two properties in a single bio-product: DNA condensation and cell entry of the formed complexes into cells. Furthermore, it is derived from a biocompatible natural material and per se is rarely cytotoxic for cells.

4.3.3 High-throughput cell-penetrating peptides-based transfection

Due to their potential, the use of CPP was pushed to HT transfection. The surface transfection and expression protocol (STEP method) is the only biological derivatives-based DNA transfer approach that reached such HT. It relies on the use of transferrin receptor, polylysine, adenoviral penton protein, and the HIV Tat protein to engineer some chimeric proteins. These combine functional motifs: binding of the DNA, binding to cell-surface receptors, the facilitated passage across membranes, the DNA targeting to the nucleus, and also adhesion and survival of the target cells on the arrayed spots [124]. The DNA/recombinants proteins mixtures are loaded in 384-wells source plates for standard arraying. Optimized conditions showed efficient GFP plasmid transfection efficiency (50–80%) and transgene expression in several cell types from easy to transfect HEK cells to more difficult ones such as SH-SY5Y neurons, N2A neurobalstoma cells, or PC-12 pheocromocytose cells. This method is promised for future enhancements accompanying the study of new CPPs. Indeed, many CPP have already been identified and validated leading to the creation of a dedicated database in 2012 referencing 843 CPP identified so far [125]. However, an exhaustive list is impossible to give as some are still identified nowadays and the developed database now contains 1700 unique CPPs 10 years later [126]. Some of them may represent better candidates for DNA transfer. This DNA transfection approach is also of great interest for gene therapy as it enables a kind of transduction, efficient like viral particles but without all the safety concerns for their production and use [127].

5. Conclusions and perspectives

Last human genome sequencing assembly led to more than 24,000 genes to study [128]. Many approaches to transfer DNA in cells were then pushed to HT to interrogate each gene function. While still in progress with developments of new reagents and methods, HT DNA transfer approaches are already available. The main remaining challenge is to render them cheaper and affordable for non-specialists.

Among physical approaches, electroporation methods surpass the others being efficient in all cell types. The suspended electrodes array design represents several advantages: it is low cost, usable with standard electroporators and liquid handling devices, but is currently limited to 96-wells plate format [73]. Due to the technical design, it should be amenable to a 384-wells plate format. However, it is still restricted to suspended cells. Electroporation approaches for adherents’ cells have also been developed in 384-wells plate format, but suffer from their cost and their need for expensive consumables [86].

Microfluidics devices suffer from the required skills and technologies to be assembled and used. Microfluidics combined with electroporation appears as a solution to some limitations but chamber-based devices seem too far from the standard assays format to be widely used. Applicable to all cells, microfluidic devices based on semi-continuous electroporation of flowing cells currently lack the necessary throughput [98]. The same concern is pointed out for nanoconstriction-based transfection designs [28]. However, higher throughputs would be amenable as microfluidic manipulation of cells and solutions in an automated way is possible at a high rate. Such a device would advantageously require an automated loading of pDNA from a source plate to the chip, transfection of the expected amount of cells and their dispensing on microplate wells, and then a rinsing step of the chip before starting a new cycle with the next pDNA. Indeed, these technologies are readily available and just need to be combined [129].

Methods combining microfluidic electroporation and DNA arraying seem at that time more difficult to be widely used. Indeed, many skills are necessary to prepare the functional chip: design of microfluidic device, micro arraying of the DNA, and even a micromanipulation platform to mount the complete functional chip [116]. This and the cost of the required material will limit its use in the scientific community.

Chemical-based transfection is readily available and represents the methods that reached the highest throughputs. The reverse protocol is the preferred mode with the use of lipids or cationic polymers and achieved a throughput of several thousand independent points [104]. A major limitation is that transfection occurs after a suspension step when cells are seeded. The use of the same approaches but in a microarray manner, also showed HT being possible to perform transfection on adherent differentiated cells. However, in this case, the use of microfluidics and their inconvenients impair its wide use.

Biological approaches also reached HT. Viral transduction is the most powerful tool to transfer DNA. However, biosafety concerns, and furthermore difficulties to produce viruses in arrays format avoid its wide use. CPP-based delivery is of great potential and a more important use should be expected in the next decade with the advance of our knowledge in this field.

In order to deliver an easy way to perform transfection even by novices, a fully automated transfection protocol was developed using a tipless nano-acoustic dispenser device [101]. Users just have to indicate amounts of DNA and transfection reagent to be delivered in each well using a custom spreadsheet and prepare the requested source plate. The device-controlled software performs the tedious dispensing from the source plate to destination one, based on the spreadsheet [105]. The method could be applicable to any chemical reagents and even to CPP-based approaches. This approach could also be performed in forward mode then allowing adherent differentiated cells transfection. Newer versions of the device allow 1536-wells plates as the source and can now dispense in 3456-wells plates. It then becomes possible to regroup the human ORFeome collection plasmids on less than 15 sources plates, and their transfer to about seven 3456-wells plates only. The method allows preloading of the plates and long-term storage before cell dispensing. However, the cost of the dispenser is extremely huge and still impairs its use. The future end of the patented technologies protection, expected in 2025–2030, should induce a price drop due to competitors’ and wider the use of such an approach.

Acknowledgments

The authors thank the National Institute of Health and Medical Research (INSERM) for its financial support for the publication of this chapter.