Advances in DNA Sequencing Technologies

Abstract

Recent advances in DNA sequencing technologies have led to efficient methods for determining the sequence of DNA. DNA sequencing was born in 1977 when Sanger et al proposed the chain termination method and Maxam and Gilbert proposed their own method in the same year. Sanger’s method was proven to be the most favourable out of the two. Since the birth of DNA sequencing, efficient DNA sequencing technologies was being produced, as Sanger’s method was laborious, time consuming and expensive; Hood et al proposed automated sequencers involving dye-labelled terminators. Due to the lack of available computational power prior to 1995, sequencing an entire bacterial genome was considered out of reach. This became a reality when Venter and Smith proposed shotgun sequencing in 1995. Pyrosequencing was introduced by Ronagi in 1996 and this method produce the sequence in real-time and is applied by 454 Life Sciences. An indirect method of sequencing DNA was proposed by Drmanac in 1987 called sequencing by hybridisation and this method lead to the DNA array used by Affymetrix. Nanopore sequencing is a single-molecule sequencing technique and involves single-stranded DNA passing through lipid bilayer via an ion channel, and the ion conductance is measured. Synthetic Nanopores are being produced in order to substitute the lipid bilayer. Illumina sequencing is one of the latest sequencing technologies to be developed involving DNA clustering on flow cells and four dye-labelled terminators performing reverse termination. DNA sequencing has not only been applied to sequence DNA but applied to the real world. DNA sequencing has been involved in the Human genome project and DNA fingerprinting.

Introduction

Reliable DNA sequencing became a reality in 1977 when Frederick Sanger who perfected the chain termination method to sequence the genome of bacteriophage ?X174 [1][2]. Before Sanger’s proposal of the chain termination method, there was the “plus and minus” method, also presented by Sanger along with Coulson [2]. The “plus and minus” method depended on the use of DNA polymerase in transcribing the specific sequence DNA under controlled conditions. This method was considered efficient and simple, however it was not accurate [2]. As well as the proposal of the chain termination sequencing by Sanger, another method of DNA sequencing was introduced by Maxam and Gilbert involving restriction enzymes, which was also reported in 1977, the same year as Sanger’s method. The Maxamm and Gilbert method shall be discussed in more detail later on in this essay. Since the proposal of these two methods, spurred many DNA sequencing methods and as the technology developed, so did DNA sequencing. In this literature review, the various DNA sequencing technologies shall be looked into as well their applications in the real world and the tools that have aided sequencing DNA e.g. PCR. This review shall begin with the discussion of the chain termination method by Sanger.

The Chain Termination Method

Sanger discovered that the inhibitory activity of 2’3′-didoxythymidine triphosphate (ddTTP) on the DNA polymerase I was dependent on its incorporation with the growing oligonucleotide chain in the place of thymidylic acid (dT) [2]. In the structure of ddT, there is no 3′-hydroxyl group, by there is a hydrogen group in place. With the hydrogen in place of the hydroxyl group, the chain cannot be extended any further, so a termination occurs at the position where dT is positioned. Figure 1 shows the structure of dNTP and ddNTP.

Sanger discovered that the inhibitory activity of 2’3′-didoxythymidine triphosphate (ddTTP) on the DNA polymerase I was dependent on its incorporation with the growing oligonucleotide chain in the place of thymidylic acid (dT) [2]. In the structure of ddT, there is no 3′-hydroxyl group, by there is a hydrogen group in place. With the hydrogen in place of the hydroxyl group, the chain cannot be extended any further, so a termination occurs at the position where dT is positioned. Figure 1 shows the structure of dNTP and ddNTP.

In order to remove the 3′-hydroxyl group and replace it with a proton, the triphosphate has to undergo a chemical procedure [1]. There is a different procedure employed for each of the triphosphate groups. Preparation of ddATP was produced from the starting material of 3′-O-tosyl-2′-deoxyadenosine which was treated with sodium methoxide in dimethylformamide to produce 2′,3′-dideoxy-2′,3′-didehydroadenosine, which is an unsaturated compound [4]. The double bond between carbon 2′ and 3′ of the cyclic ether was then hydrogenated with a palladium-on-carbon catalyst to give 2′,3′-dideoxyadenosine (ddA). The ddA (ddA) was then phosphorylated in order add the triphosphate group. Purification then took place on DEAE-Sephadex column using a gradient of triethylamine carbonate at pH 8.4. Figure 2 is schematic representation to produce ddA prior to phosphorylation.

In the preparation of ddTTP (Figure 3), thymidine was tritylated (+C(Ph3)) at the 5′-position and a methanesulphonyl (+CH3SO2) group was introduced at the 3′-OH group[5]. The methanesulphonyl group was substituted with iodine by refluxing the compound in 1,2-dimethoxythane in the presence of NaI. After chromatography on a silica column the 5′-trityl-3′-iodothymidine was hydrogenated in 80% acetic acid to remove the trityl group. The resultant 3′-iodothymidine was hydrogenated to produce 2’3′-dideoxythymidine which subsequently was phosphorylated. Once phosphorylated, ddTTP was then purified on a DEAE-sephadex column with triethylammonium-hydrogen carbonate gradient. Figure 3 is a schematic representation to produce ddT prior phosphorylation.

When preparing ddGTP, the starting material was N-isobutyryl-5′-O-monomethoxytrityldepxyguanosine [1]. After the tosylation of the 3′-OH group the compound was then converted to the 2’3′-didehydro derivative with sodium methoxide. Then the isobutyryl group was partly removed during this treatment of sodium methoxide and was removed completely by incubation in the presence of NH3 overnight at 45oC. During the overnight incubation period, the didehydro derivative was reduced to the dideoxy derivative and then converted to the triphosphate. The triphosphate was purified by the fractionation on a DEAE-Sephadex column using a triethylamine carbonate gradient. Figure 4 is a schematic representation to produce ddG prior phosphorylation.

Preparing the ddCTP was similar to ddGTP, but was prepared from N-anisoyl-5′-O-monomethoxytrityldeoxycytidine. However the purification process was omitted for ddCTP, as it produced a very low yield, therefore the solution was used directly in the experiment described in the paper [2]. Figure 5 is a schematic representation to produce ddC prior phosphorylation.

With the four dideoxy samples now prepared, the sequencing procedure can now commence. The dideoxy samples are in separate tubes, along with restriction enzymes obtained from ?X174 replicative form and the four dNTPs [2]. The restriction enzymes and the dNTPs begin strand synthesis and the ddNTP is incorporated to the growing polynucleotide and terminates further strand synthesis. This is due to the lack of the hydroxyl group at the 3′ position of ddNTP which prevents the next nucleotide to attach onto the strand. The four tubes are separate by gel-electrophoresis on acrylamide gels (see Gel-Electrophoresis). Figure 6 shows the sequencing procedure.

Reading the sequence is straightforward [1]. The first band that moved the furthest is located, this represents the smallest piece of DNA and is the strand terminated by incorporation of the dideoxynucleotide at the first position in the template. The track in which this band occurs is noted. For example (shown in Figure 6), the band that moved the furthest is in track A, so the first nucleotide in the sequence is A. To find out what the next nucleotide, the next most mobile band corresponding to DNA molecule which is one nucleotide longer than the first, and in this example, the band is on track T. Therefore the second nucleotide is T, and the overall sequence so far is AT. The processed is carried on along the autoradiograph until the individual bands start to close in and become inseparable, therefore becoming hard to read. In general it is possible to read upto 400 nucleotides from one autoradiograph with this method. Figure 7 is a schematic representation of an autoradiograph. Ever since Sanger perfected the method of DNA sequencing, there have been advances methods of sequencing along with the achievements. Certain achievements such as the Human genome project and shall be discussed later on in this review.

Gel-Electrophoresis Gel-Electrophoresis is defined as the movement of charged molecules in an electric field [1][8]. DNA molecules, like many other biological compounds carry an electric charge. With the case of DNA, this charge is negative. Therefore when DNA is placed in an electric field, they migrate towards the positive pole (as shown in figure 8). There are three factors which affect the rate of migration, which are shape, electrical charge and size. The polyacrylamide gel comprises a complex network of pores through which the molecules must travel to reach the anode.

Maxam and Gilbert Method The Maxam and Gilbert method was proposed before Sanger Method in the same year. While the Sanger’s method involves enzymatic radiolabelled fragments from unlabelled DNA strands [2]. The Maxam-Gilbert method involves chemical cleavage of prelabelled DNA strands in four different ways to form the four different collections of labelled fragments [6][7]. Both methods use gel-electrophoresis to separate the DNA target molecules [8]. However Sanger’s Chain Termination method has been proven to be simpler and easier to use than the Maxam and Gilbert method [9]. As a matter of fact, looking through the literature text books, Sanger’s method of DNA sequencing have been explained rather than Maxam and Gilberts [1][3][9][10]. With Maxam and Gilbert’s method there are two chemical cleavage reactions that take place [6][7]. One of the chemical reaction take places with guanine and the adenine, which are the two purines and the other cleaves the DNA at the cytosine and thymine, the pyrimidines. For the cleavage reaction, specific reagents are used for each of the reaction. The purine specific reagent is dimethyl sulphate and the pyrimidine specific reagent is hydrazine. Each of these reactions are done in a different way, as each of the four bases have different chemical properties. The cleavage reaction for the guanine/adenine involves using dimethyl sulphate to add a methyl group to the guanines at the N7 position and at the N3 position at the adenines [7]. The glycosidic bond of a methylated adenines is unstable and breaks easily on heating at neutral pH, leaving the sugar free. Treatment with 0.1M alkali at 90oC then will cleave the sugar from the neighbouring phosphate groups. When the resulting end-labelled fragments are resolved on a polyacrylamide gel, the autoradiograph contains a pattern a dark and light bands. The dark bands arise from the breakage at the guanines, which methylate at a rate which is 5-fold faster than adenines. From this reaction the guanine appear stronger than the adenosine, this can lead to a misinterpretation. Therefore an Adenine-Enhanced cleavage reaction takes place. Figure 9 shows the structural changes of guanine when undergoing the structural modifications involved in Maxam-Gilbert sequencing. With an Adenine-Enhanced cleavage, the glycosidic bond of methylated adenosine is less stable than that of methylated guanosine, thus gentle treatment with dilute acid at the methylation step releases the adenine, allowing darker bands to appear on the autoradiograph [7].

The chemical cleavage for the cytosine and thymine residues involves hydrazine instead of dimethyl sulphate. The hydrazine cleaves the base and leaving ribosylurea [7]. After partial hydrazinolysis in 15-18M aqueous hydrazine at 20oC, the DNA is cleaved with 0.5M piperidine. The piperidine (a cyclic secondary amine), as the free base, displaces all the products of the hydrazine reaction from the sugars and catalyzses the b-elimination of the phosphates. The final pattern contains bands of the similar intensity from the cleavages at the cytosines and thymines. As for cleavage for the cytosine, the presence of 2M NaCl preferentially suppresses the reaction of thymine with hydrazine. Once the cleavage reaction has taken place each original strand is broken into a labelled fragment and an unlabelled fragment [7]. All the labelled fragments start at the 5′ end of the strand and terminate at the base that precedes the site of a nucleotide along the original strand. Only the labelled fragments are recorded on the gel electrophoresis.

Dye-labelled terminators

For many years DNA sequencing has been done by hand, which is both laborious and expensive[3]. Before automated sequencing, about 4 x 106 bases of DNA had been sequenced after the introduction of the Sanger’s method and Maxam & Gilbert methods [11]. In both methods, four sets of reactions and a subsequent electrophoresis step in adjacent lanes of a high-resolution polyacrylamide gel. With the new automated sequencing procedures, four different fluorophores are used, one in each of the base-specific reactions. The reaction products are combined and co-electrophoresed, and the DNA fragments generated in each reaction are detected near the bottom of the gel and identified by their colour. As for choosing which DNA sequencing method to be used, Sanger’s Method was chosen. This is because Sanger’s method has been proven to be the most durable and efficient method of DNA sequencing and was the choice of most investigators in large scale sequencing [12]. Figure 10 shows a typical sequence is generated using an automated sequencer.

The selection of the dyes was the central development of automated DNA sequencing [11]. The fluorophores that were selected, had to meet several criteria. For instance the absorption and emission maxima had to be in the visible region of the spectrum [11] which is between 380 nm and 780 nm [10], each dye had to be easily distinguishable from one another [11]. Also the dyes should not impair the hybridisation of the oligonucleotide primer, as this would decrease the reliability of synthesis in the sequencing reactions. Figure 11 shows the structures of the dyes which are used in a typical automated sequencing procedure, where X is the moiety where the dye will be bound to.

Table 1 shows which dye is covalently attached to which nucleotide in a typical automated DNA sequencing procedure

In designing the instrumentation of the florescence detection apparatus, the primary consideration was sensitivity. As the concentration of each band on the co-electrophoresis gel is around 10 M, the instrument needs to be capable of detecting dye concentration of that order. This level of detection can readily be achieved by commercial spectrofluorimeter systems. Unfortunately detection from a gel leads to a much higher background scatter which in turn leads to a decrease in sensitivity. This is solved by using a laser excitation source in order to obtain maximum sensitivity [11]. Figure 12 is schematic diagram of the instrument with the explanation of the instrumentation employed.

When analyzing data, Hood had found some complications [11]. Firstly the emission spectra of the different dyes overlapped, in order to overcome this, multicomponent analysis was employed to determine the different amounts of the four dyes present in the gel at any given time. Secondly, the different dye molecules impart non-identical electrophoretic mobilities to the DNA fragments. This meant that the oligonucleotides were not equal base lengths. The third major complication was in analyzing the data comes from the imperfections of the enzymatic methods, for instance there are often regions of the autoradiograph that are difficult to sequence. These complications were overcome in five steps [11]

1. High frequency noise is removed by using a low-pass Fourier filter.
2. A time delay (1.5-4.5 s) between measurements at different wavelength is partially corrected for by linear interpolation between successive measurements.
3. A multicomponent analysis is performed on each set of four data points; this computation yields the amount of each of the four dyes present in the detector as a function of time.
4. The peaks present in the data are located
5. The mobility shift introduced by the dyes is corrected for using empirical determined correction factors.

Since the publication of Hood’s proposal of the fluorescence detection in automated DNA sequence analysis. Research has been made on focussed on developing which are better in terms of sensitivity [12].

Bacterial and Viral Genome Sequencing (Shotgun Sequencing)

Prior to 1995, many viral genomes have been sequenced using Sanger’s chain termination technique [13], but no bacterial genome has been sequenced. The viral genomes that been sequenced are the 229 kb genome of cytomegalovirus [14], and the 192 kb genome of vaccinia [15], the 187 kb mitochondrial and 121 kb cholorophast genomes of Marchantia polymorpha have been sequenced [16]. Viral genome sequencing has been based upon the sequencing of clones usually derived from extensively mapped restriction fragments, or ? or cosmid clones [17]. Despite advances in DNA sequencing technology, the sequencing of genomes has not progressed beyond clones on the order of the size of the ~ 250kb, which is due to the lack of computational approaches that would enable the efficient assembly of a large number of fragments into an ordered single assembly [13][17]. Upon this, Venter and Smith in 1995 proposed Shotgun Sequencing and enabled Haemophilus influenzae (H. influenzae) to become the first bacterial genome to be sequenced [13][17]. H. influenzae was chosen as it has a similar base composition as a human does with 38 % of sequence made of G + C. Table 2 shows the procedure of the Shotgun Sequencing [17]. When constructing the library ultrasonic waves were used to randomly fragment the genomic DNA into fairly small pieces of about the size of a gene [13]. The fragments were purified and then attached to plasmid vectors[13][17]. The plasmid vectors were then inserted into an E. coli host cell to produce a library of plasmid clones. The E. coli host cell strains had no restriction enzymes which prevented any deletions, rearrangements and loss of the clones [17]. The fragments are randomly sequenced using automated sequencers (Dye-Labelled terminators), with the use of T7 and SP6 primers to sequence the ends of the inserts to enable the coverage of fragments by a factor of 6 [17].

Table 2 (Reference 17)

Once the sequencing reaction has been completed, the fragments need to be assembled, and this process is done by using the software TIGR Assembler (The Institute of Genomic Research) [17]. The TIGR Assembler simultaneously clusters and assembles fragments of the genome. In order to obtain the speed necessary to assemble more than 104 fragments [17], an algorithm is used to build up the table of all 10-bp oligonucleotide subsequences to generate a list of potential sequence fragment overlaps. The algorithm begins with the initial contig (single fragment); to extend the contig, a candidate fragment is based on the overlap oligonucleotide content. The initial contig and candidate fragment are aligned by a modified version of the Smith-Waterman [18] algorithm, which allows optional gapped alignments. The contig is extended by the fragment only if strict criteria of overlap content match. The algorithm automatically lowers these criteria in regions of minimal coverage and raises them in regions with a possible repetitive element [17]. TIGR assembler is designed to take advantage of huge clone sizes [17]. It also enforces a constraint that sequence from two ends of the same template point toward one another in the contig and are located within a certain range of the base pair [17]. Therefore the TIGR assembler provides the computational power to assemble the fragments. Once the fragments have been aligned, the TIGR Editor is used to proofread the sequence and check for any ambiguities in the data [17]. With this technique it does required precautionary care, for instance the small insert in the library should be constructed and end-sequenced concurrently [17]. It is essential that the sequence fragments are of the highest quality and should be rigorously check for any contamination [17].

Pyrosequencing

Most of the DNA sequencing required gel-electrophoresis, however in 1996 at the Royal Institute of Technology, Stockholm, Ronaghi proposed Pyrosequencing [19][20]. This is an example of sequencing-by-synthesis, where DNA molecules are clonally amplified on a template, and this template then goes under sequencing [25]. This approach relies on the detection of DNA polymerase activity by enzymatic luminometric inorganic pyrophosphate (PPi) that is released during DNA synthesis and goes under detection assay and offers the advantage of real-time detection [19]. Ronaghi used Nyren [21] description of an enzymatic system consisting of DNA polymerase, ATP sulphurylase and lucifinerase to couple the release of PPi obtained when a nucleotide is incorporated by the polymerase with light emission that can be easily detected by a luminometer or photodiode [20].

When PPi is released, it is immediately converted to adenosine triphosphate (ATP) by ATP sulphurylase, and the level of generated ATP is sensed by luciferase-producing photons [19][20][21]. The unused ATP and deoxynucleotide are degraded by the enzyme apyrase. The presence or absence of PPi, and therefore the incorporation or nonincorporation of each nucleotide added, is ultimately assessed on the basis of whether or not the photons are detected. There is minimal time lapse between these events, and the conditions of the reaction are such that iterative addition of the nucleotides and PPi detection are possible. The release of PPi via the nucleotide incorporation, it is detected by ELIDA (Enzymatic Luminometric Inorganic pyrophosphate Detection Assay) [19][21]. It is within the ELIDA, the PPi is converted to ATP, with the help of ATP sulfurylase and the ATP reacts with the luciferin to generate the light at more than 6 x 109 photons at a wavelength of 560 nm which can be detected by a photodiode, photomultiplier tube, or charge-coupled device (CCD) camera [19][20]. As mentioned before, the DNA molecules need to be amplified by polymerase chain reaction (PCR which is discussed later Ronaghi observed that dATP interfered with the detection system [19]. This interference is a major problem when the method is used to detect a single-base incorporation event. This problem was rectified by replacing the dATP with dATPaS (deoxyadenosine a–thiotrisulphate). It is noticed that adding a small amount of the dATP (0.1 nmol) induces an instantaneous increase in the light emission followed by a slow decrease until it reached a steady-state level (as Figure 11 shows). This makes it impossible to start a sequencing reaction by adding dATP; the reaction must instead be started by addition of DNA polymerase. The signal-to-noise ratio also became higher for dATP compared to the other nucleotides. On the other hand, addition of 8 nmol dATPaS (80-fold higher than the amount of dATP) had only a minor effect on luciferase (as Figure 14 shows). However dATPaS is less than 0.05% as effective as dATP as a substrate for luciferase [19].

Pyrosequencing is adapted by 454 Life Sciences for sequencing by synthesis [22] and is known as the Genome Sequencer (GS) FLX [23][24]. The 454 system consist of random ssDNA (single-stranded) fragments, and each random fragment is bound to the bead under conditions that allow only one fragment to a bead [22]. Once the fragment is attached to the bead, clonal amplification occurs via emulsion. The emulsified beads are purified and placed in microfabricated picolitre wells and then goes under pyrosequencing. A lens array in the detection of the instrument focuses luminescene from each well onto the chip of a CCD camera. The CCD camera images the plate every second in order to detect progression of the pyrosequencing [20][22]. The pyrosequencing machine generates raw data in real time in form of bioluminescence generated from the reactions, and data is presented on a pyrogram [20]

Sequencing by Hybridisation

As discussed earlier with chain-termination, Maxamm and Gilbert and pyrosequencing, these are all direct methods of sequencing DNA, where each base position is determined individually [26]. There are also indirect methods of sequencing DNA in which the DNA sequence is assembled based on experimental determination of oligonucleotide content of the chain. One promising method of indirect DNA sequencing is called Sequencing by Hybridisation in which sets of oligonucleotide probes are hybridised under conditions that allow the detection of complementary sequences in the target nucleic acid [26]. Sequencing by Hybridisation (SBH) was proposed by Drmanac et al in 1987 [27] and is based on Doty’s observation that when DNA is heated in solution, the double-strand melts to form single stranded chains, which then re-nature spontaneously when the solution is cooled [28]. This results the possibility of one piece of DNA recognize another. And hence lead to Drmanac proposal of oligonucleotide’s probes being hybridised under these conditions allowing the complementary sequence in the DNA target to be detected [26][27]. In SBH, an oligonucleotide probe (n-mer probe where n is the length of the probe) is a substring of a DNA sample. This process is similar to doing a keyword search in a page full of text [29]. The set of positively expressed probes is known as the spectrum of DNA sample. For example, the single strand DNA 5’GGTCTCG 3′ will be sequenced using 4-mer probes and 5 probes will hybridise onto the sequence successfully. The remaining probes will form hybrids with a mismatch at the end base and will be denatured during selective washing. The five probes that are of good match at the end base will result in fully matched hybrids, which will be retained and detected. Each positively expressed serves as a platform to decipher the next base as is seen in Figure 16.

For the probes that have successfully hybridised onto the sequence need to be detected. This is achieved by labelling the probes with dyes such as Cyanine3 (Cy3) and Cyanine5 (Cy5) so that the degree of hybridisation can be detected by imaging devices [29]. SBH methods are ideally suited to microarray technology due to their inherent potential for parallel sample processing [29]. An important advantage of using of using a DNA array rather than a multiple probe array is that all the resulting probe-DNA hybrids in any single probe hybridisation are of identical sequence [29]. One of main type of DNA hybridisation array formats is oligonucleotide array which is currently patented by Affymetrix [30]. The commercial uses of this shall be discussed under application of the DNA Array (Affymetrix). Due to the small size of the hybridisation array and the small amount of the target present, it is a challenge to acquire the signals from a DNA Array [29]. These signals must first be amplified before they can be detected by the imaging devices. Signals can be boosted by the two means; namely target amplification and signal amplification. In target amplification such as PCR, the amount of target is increased to enhance signal strength while in signal amplification; the amount of signal per unit is increased.

Nanopore Sequencing

Nanopore sequencing was proposed in 1996 by Branton et al, and shows that individual polynucleotide molecules can be characterised using a membrane channel [31]. Nanopore sequencing is an example of single-molecule sequencing, in which the concept of sequencing-by-synthesis is followed, but without the prior amplification step [24]. This is achieved by the measurement of ionic conductance of a nucleotide passing through a single ion channels in biological membranes or planar lipid bilayer. The measurement of ionic conductance is routine neurobiology and biophysics [31], as well as pharmacology (Ca+ and K+ channel)[32] and biochemistry[9]. Most channels undergo voltage-dependant or ligand dependant gating, there are several large ion channels (i.e. Staphylococcus aureus a-hemolysin) which can remain open extended periods, thereby allowing continuous ionic current to flow across a lipid bilayer [31]. If a transmembrane voltage applied across an open channel of appropriate size should draw DNA molecules through the channel as extended linear chains whose presence would detect reduce ionic flow. It was assumed, that the reduction in the ionic flow would lead to single channel recordings to characterise the length and hence lead to other characteristics of the polynucleotide. In the proposal by Branton, a-hemolysin was used to form a single channel across a lipid bilayer separating two buffer-filled compartment [31]. a-Hemolysin is a monomeric, 33kD, 293 residue protein that is secreted by the human pathogen Staphylococcus aureus [33]. The nanopore are produced when a-hemolysin subsunits are introduced into a buffered solution that separates lipid bilayer into two compartments (known as cis and trans): the head of t