The impact of various chromatographic conditions on the separation of modified and unmodified oligonucleotides

(1)

The impact of various chromatographic conditions on the separation of modified and unmodified oligonucleotides

Påverkan av olika kromatografiska förhållanden på separationen av modifierade och omodifierade oligonukleotider

Lewis Frazer

Department of Engineering and Chemical sciences Chemistry, Bachelor Thesis

30HP, KEGCX6

Supervisor: Olof Stålberg – SGS DNA Supervisor: Torgny Fornstedt – KAU Examinator: Maria Rova – KAU 2021 – 06 – 18

(2)

2

Abstract

In this study, the effects of certain chromatographic conditions on various modified and unmodified oligonucleotides were investigated. At the forefront of this study was the investigation of a new Ion-pair Reversed-phase liquid chromatography (IP-RPLC) method, that had the potential to replace a previously established triethylammonium acetate (TEAA) IP-RPLC method developed for oligonucleotide separations. This method, utilising the counter ion dibutyl amine (DBA) and a Tris-buffer at pH 8, produced promising results indicating that the strong binding strength of DBA creates a hybrid IEX/RPLC separation method – the separation of oligonucleotides is dynamically based on both charge and length.

Higher concentrations of DBA appear to produce better results that include improved efficiency, increased retention and even the potential discovery of hidden impurities. In conjugation with Ultra-high-pressure liquid chromatography (UHPLC) systems, sub-2 µm particle columns and gradient optimisations, separations of complex oligonucleotides could be achieved in short analysis times. Furthermore, effective separations at the analytical level can be applied and adapted to larger scale Prep-LC, potentially also improving the

purification process of crude oligonucleotide samples. Further development and validation are, however, required for any future work with this method.

Abstrakt

I denna studie har effekten av vissa kromatografiska förhållanden på olika modifierade och icke-modifierade oligonukleotider undersökts. I framkanten av denna studie var

undersökningen av en ny IP-RPLC metod, vilken har potential att ersätta den tidigare etablerade trietylammonium acetat (TEAA) IP-RPLC metoden, vilken utvecklats för separationen av oligonukleotider. Denna metod, vilken använder dibutylamin (DBA) som motjon och en Tris-buffert vid pH 8, gav lovande resultat vilka indikerar att den starka bindningsstyrkan av DBA skapar en hybrid IEX/RPLC separationsmetod – separationen av oligonkuleotider styrs både av dess laddning och dess längd. Höga koncentrationer av DBA verkade ge bättre resultat som inkluderar hög effektivitet, ökad retention och även den potentiella upptäckten av gömda föroreningar. I samband med UHPLC systemer, kolonner med mindre än 2µm i partikelstorlek och optimiserade gradienter, separationer av komplexa oligonukleotider erhölls på korta analystider. Effektiva separationer vid den analytiska nivån kan appliceras och adapteras till storskalig preparative-LC, med potential att kunna förbättra reningsprocessen för syntetiserade oligonukleotider. Vidare utveckling och validering krävs för framtida användning av denna metod.

(3)

3

Abbreviations

ASO(s) – Antisense oligonucleotide(s) BHQ – Black hole quencher

DBA – Dibutylamine

EDQ – Eclipse dark quencher

FAM – Fluorescein Phosphoramidite HPS – High performance surfaces IEX – Ion exchange chromatography IPC – Ion-pair chromatography

IP-RPLC – Ion-pair Reversed-phase liquid chromatography MeCN – Acetonitrile

MGB – Minor groove binder ON(s) – Oligonucleotide(s)

PCR-SSO – Polymerase chain reaction Sequence-specific oligonucleotide RPLC – Reversed-phase liquid chromatography

SST/SUIT – System suitability test TEAA – Triethylammonium acetate

UHPLC – Ultra-high-pressure liquid chromatography

(5)

5

1. Introduction

1.1 Modified oligonucleotides

Oligonucleotides are formed by a short chain of nucleotide monomers (oligomer), creating either a short deoxyribonucleic acid (DNA) or ribonucleic acid (RNA) molecule. A

nucleotide consists of three parts: a nucleobase, a sugar molecule, and a phosphate group.

The five nucleobases are adenine (A), guanine (G), cytosine (C), uracil (U) and thymine (T);

A and G classified as purine bases whilst C, U and T are pyrimidine bases. When a base is paired with either the sugar molecule deoxyribose or ribose, the resulting molecule is known as a nucleoside. A nucleotide is then formed when the phosphate group binds to the sugar of the nucleoside. When multiple nucleotides bind to one another a sugar-phosphate backbone is formed, whereupon the properties of the molecule are governed by the sequence of

nucleobases. DNA consists of a deoxyribose and the bases A, G, C and T whereas RNA consists of a ribose and the bases A, G, C and U. The resulting oligonucleotide can be either single or double stranded, where the four bases can undergo hydrogen bonding via Watson- Crick base pairing (G-C, A-T/U). ^[1]

Figure 1. Short nucleotide chain of DNA with one of each base; RNA substitutes are seen in the bottom left. ^[23]

Retrieved from Polynucleotide chain of deoxyribonucleic acid (DNA), by Encyclopaedia Britannica. Copyright 2015 by Encyclopaedia Britannica.

Oligonucleotides can undergo an assortment of various modifications depending on their purpose and the situation at hand, such as antisense oligonucleotides (ASOs). ^[2] These modifications can improve the oligonucleotide’s ability to pass through the cell membrane or increase its resistance to degradation once inside the cell. ^[2] Another modification is the addition of covalently bound hydrophobic molecules, or simply labels, at the 5’ and 3’ end of the sugar-phosphate backbone. These labels are known as reporters and quenchers

respectively and the resulting modified oligonucleotide is often known as a probe. ^{[3] [4]}

Probes have a variety of uses such as in diagnostic kits for various ailments and the

hybridisation technique sequence-specific oligonucleotide polymerase chain reaction (PCR-

(6)

6

SSO). ^[4]^[5] To function, the reporter and quencher must have an overlapping spectrum. When both are bound to the oligonucleotide, the light emitted by the reporter is absorbed by the quencher; the light is consequently not detected. After the oligonucleotide has bound to the target molecule however, the reporter is released, and the emitted light is no longer absorbed.

The increase of emitted light is detected and is proportionate to the amount of the target molecule, such as a certain DNA sequence. The energy released does not necessarily have to be in the form of light. Heat can also be used as the detectable signal using certain

modifications such as a black hole quencher (BHQ), and in some cases is preferable over light. Scorpion probes are more complex and further modified; the reporter and quencher are still at the 5’ and 3’ ends however are placed closely together with the formation of a hair-pin structure. ^[6] Attached to the quencher is a blocker, which links the quencher to the 5’ end of a PCR primer. When the hair-pin structure known as the probe region hybridises with the target molecule, it unfolds, separating the quencher from the reporter. The emitted light from the reporter is consequently no longer absorbed by the quencher and the increase of fluorescence can be detected, in the same vein as with dual-labelled probes.

The separation of these probes has, however, proven to be more challenging than their unmodified counterparts. The sugar-phosphate backbone of the oligonucleotide has a negative charge, owning to the phosphodiester bond between the nucleosides. ^[1] As the nucleobases are bound inward to each other, the sugar-phosphate backbone is subsequently pushed outwards toward the surrounding environment. This occurs when the secondary structure of a double stranded oligonucleotide is formed but is also possible with single strands, which can form various three-dimensional structures with itself. ^[1] The outwards facing sugar-phosphate backbone has hydrophilic properties which allow it to interact with water in the surrounding environment. The addition of modifications at either end of this backbone then complicates the separation, as the added molecules are hydrophobic. ^[4]

1.2 Chromatographic separation

Liquid chromatography (LC) is a powerhouse standard in the separation of pharmaceutical analytes and has proven itself effective, especially with certain modifications, at separating oligonucleotides. ^[7] In reversed-phase liquid chromatography (RPLC) the stationary phase is non-polar with hydrophobic properties, thus separating non-polar analytes from the polar mobile phase. ^[8] With the addition of ion-pair reagents (IP-RPLC) the retention of the target oligonucleotide, and thus the overall separation, can be modulated. ^{[8] [9]}^[25] By utilising ion pair chromatography (IPC) in combination with ultra-high pressure liquid chromatography (UHPLC), effective separations of oligonucleotides can be achieved within minutes. UHPLC enables separations to be conducted at high pressure, high temperatures and with small- particle size UHPLC columns for the most efficient ON separations. ^[8] Ion exchange chromatography (IEX) is another form of separation that differentiates analytes based upon their charge. ^[8] However, samples consisting of long ON chains can cause the system to crash as the method has difficulty in separating the molecules. Various “hybrid” methods of

IEX/RPLC, or mixed modes, have been studied in order to generate ON separation methods that draw upon the strength of both modes whilst negating their weaknesses. ^[10]

The mechanism of IPC is complex and various theoretical models exist that attempt to accurately describe this mode of separation. ^{[8] [11]}^[13]^[22] The simplified theory of this

mechanism involves the attraction of the ion-pair reagent to the stationary phase. The reagent molecule consists of a large, hydrophobic region and an ionic region oppositely charged to the target analyte. ^[8] As oligonucleotides consist of a long, negatively charged phosphate-

(7)

7

sugar backbone the chosen ion-pair reagent is therefore positively charged. When it adsorbs to the stationary phase through hydrophobic interactions, a positive charge amasses in the surrounding area around the stationary phase. This allows the ion-pair reagent to

electrostatically interact with the charge of the other analytes in the sample. The retention of the oligonucleotide increases as it binds to the ion-pair reagent, effectively cancelling out the effect of its negative charge. ^{[8] [12]}

Sterically, oligonucleotides can bind to the stationary phase in various positions. This can alter the retention and lead to a loss in selectivity, as various smaller peaks can belong to a single target analyte. Through modulation the retention of a target analyte can be normalised, thus leading to increased selectivity and a more accurate measurement of purity. Ion-pairing reagents can increase retention through electrostatic interactions, contributing to a more singular, complete peak. However, the addition of labels in probes further complicates the retention mechanism as the oligonucleotide gains increased hydrophobicity at each end of the molecule. This should theoretically lead to various levels of retention, as the molecule binds hydrophobically to the stationary phase via its labels in various steric positions.

The result of this would be a loss of interaction between the stationary phase and the rest of the oligonucleotide, leading to a decrease in the impact of the molecular length or charge on retention.

The separation of various impurities from the target ON would then become even more challenging. Shortmers (N-1 bases) and longmers (N+1 bases) are the main impurities commonly found in crude ON samples such as the one presented in figure 9. Both of these impurities are a result of synthesis failure during the first stage of oligonucleotide production.

[7] [15] Shortmers occur when a nucleotide fails to attach to the ON product whilst longmers occur when an extra nucleotide is added, creating a longer ON product than intended. ^[7]The separation of these impurities, especially N-1 and N+1 ONs, can be challenging as they can share almost identical length and/or charge as the target ON. Thus, an effective separation method would require a high level of selectivity in order to differentiate between the target ON and the similar impurities – hybrid methods that could dynamically separate molecules based on both charge and length are therefore of investigational importance.

By utilising strong ion-pairing reagents the hydrophobic influence of the labels can be supressed, thus allowing the length of the oligonucleotide to have a greater influence the retention. ^[4] It is this combination of electrostatic interaction and label suppression that allows IP-RPLC to effectively separate modified oligonucleotides and increase selectivity by normalising the retention of the target analyte. Furthermore, the separation method benefits from faster analysis times when combined with UHPLC systems, utilising sub-2 µm particle column technology and high pressure. ^[14] As stronger ion-pair reagents require a greater amount of organic modifier to elute the retained analyte, the overall chromatographic separation should benefit from shorter retention times and a greater peak capacity due to an increased number of theoretical plates and efficiency (see 1.3).

1.3 Chromatographic parameters

There are a variety of parameters in chromatography that can be altered with to change the outcome of chromatographic separations. The chosen parameter settings are often those deemed most optimal for the separation, with selectivity, resolution, and analysis time

amongst the most important aspects of chromatographic results. Those of most interest in this

(8)

8

study were the choice of ion-pair reagent, the effect of counter ion concentration and organic modifier gradient changes, and the capabilities of UHPLC using sub-2 µm particle columns.

1.3.1 Buffer pH and choice of counter ion

The pH value of the mobile phase is of great importance for the separation of analytes.

Degradation can occur if the pH is not at an optimal, stable level for the respective molecule.

[1] To increase analyte stability, an optimal pH is chosen for both the sample solution and mobile phase to keep the pH level as stable as possible. The buffering capacity is therefore important to minimise the effect of pH changes. ^[8]Furthermore, the choice of column can limit the experimental pH range as low and high pH values often lead to degradation of the stationary phase.

As previously stated in the introduction, the net charge of an ON is negative due to its phosphate-sugar backbone. ^[1]This charge, however, can change due to the surrounding pH and will therefore no longer be determined by the base-chain length, which facilitates the negatively charged phosphor group bonds. ^[5]The bases G and T will increase the negative charge of the ON at higher pH values, whilst the bases C and A will produce a positive charge at lower pH values. ^[5]To ensure that the net charge remains solely based on the number of bases, an intermediate pH level needs to be upheld. In the previous experiments relevant to this study, a Triethylammonium-acetate based (TEAA) mobile phase with a pH of 7 was used to separate ON samples.

Figure 2. The buffer effectiveness of DBA-Tris (left) and TEAA (right) for the entire pH range. The pKa values shown are 11.3 (DBA), 8.1 (Tris-buffer), 10.8 (TEA) and 4.8 (Acetic acid). The green areas show effective buffering zones. From Schill, G., (1976). Analytical Pharmaceutical Chemistry (5^th edition). Uppsala University.

[24]

However, the above figure shows that the TEAA buffer actually performs poorly around the pH range ON separations are conducted at; 7 – 10 pH. The proposed Dibutylamine (DBA) Tris-buffer on the other hand only performs poorly at a pH level below that of the 7-pH threshold, thus producing better results for the entirety of the effective pH range.

Furthermore, the partition coefficient K (or LogP) determines the ratio of solute in both the mobile and stationary phase during a separation. ^[8] A higher LogP value indicates that there is a greater concentration of solute in the stationary phase, and a lower LogP value indicates a greater solute concentration in the mobile phase; a value of 1 indicates a phase-equilibrium.

The LogP value of DBA is twice that of TEA, 2.8 and 1.4, respectively. ^[16][17] This means that DBA has a greater affinity for the stationary phase, increasing the amount of ON molecules that can retained by the electrostatic interaction and subsequently increasing the separation retention.

(9)

9

While there are many ion-pair reagents that can be used in the separation of ONs ^[4][9][18], DBA was chosen for this study based upon its perceived improvements over TEA and evidence of more efficient ion-pair facilitated separations. ^[19] After producing promising results, DBA was kept as the ion-pair reagent for the entirety of the study for further analysis.

An addition to the DBA-mobile phase was a 10 mM Tris-buffer at pH 8, which is preferably within the 0.5 pH range of the Tris pKa value of 8.3 – this was to increase the buffering capacity and uphold a high pH to facilitate ON retention. ^{[5] [8]}

1.3.2 Organic modifier

An organic modifier is added to the mobile phase to elute analytes from the column. ^[8] The choice of organic modifier depends on various factors such as effectiveness, cost, and environmental impact. The addition of organic modifiers to the mobile phase can be of great importance as it usually leads to an increase in selectivity and shorter retention times. The choice of organic modifier for this study was acetonitrile (MeCN), as it was an already pre- established mobile phase component due to its relatively low toxicity, low cost, and high effectiveness. Previous work utilising MeCN as an organic modifier in IP-RPLC includes the study of carboxylic acid retention behaviour by A. Tilly-Melin et al. (1979). ^[25]

1.3.3 Gradient

As the separation solvent consists of two mobile phases A and B, the use of a gradient allows for greater control over the elution of analytes. ^[8]This is achieved by increasing the fraction of the mobile phase that contains the organic modifier (B) overtime, which in turn lowers the fraction of mobile phase A, and can be applied in various manners:

• Isocratic elution – The initial mobile phase composition remains unchanged during the entire course of the separation.

• Linear gradient – The fraction elution buffer increases with a linear rate over the entire course of the separation.

• Step gradient – The fraction elution buffer increases at various rates over the course of the separation. This method generally gives the greatest control over analyte elution.

Another important aspect is the gradient starting point (starting %B), which can increase or decrease the overall gradient slope in conjunction with the gradient end point (ending %B) – this is the essence of a step gradient. The gradient starting point should be chosen based on the desired gradient slope and care should be taken not to elute analytes immediately with a high fraction of elution buffer.

Initial tests in this study revolved around the development of optimised gradients for the separation of various ON samples. Certain ON molecules were inseparable from similar impurities without modification to the method gradient. For already purified ON samples, linear gradients could be effective whilst step gradients were required for samples with certain impurities – these include both crude and purified samples. Several examples of method gradient modifications, and the interaction with DBA, are presented in 3.1.

1.3.4 Column

(10)

10

There are a variety of UHPLC columns that can be used for oligonucleotide separation at the analytical level. The column properties are an important factor in the separation of analytes and can vary significantly depending on the model and manufacturer; these parameters include:

• Column proportions – The inner diameter and length of the column.

• Packing material – Acts as the stationary phase in RPLC consisting of particles made from various materials, which retain analytes as they pass through the column in the mobile phase. The particles contain pores in which the analytes are retained in.

• Particle size – The size of the particles used to create the inner packing material.

• Pore size – The size of the pores imbedded in the packing material. This determines the total surface area the analytes can be retained on; smaller pores result in a larger surface area which increases the amount retained analyte.

UHPLC allows the use of columns with sub-2 µm particles that can withstand the high pressures required to separate analytes. This improves the overall efficiency of the separation, by increasing the number of theoretical plates and resolution, whilst simultaneously

decreasing the analysis time. ^[8]Two UHPLC columns manufactured by Waters, the Acquity C18 and Acquity Premier C18 columns, were chosen for this study – the results of a

separation comparison for two ON samples are presented in 3.5.

1.3.5 Temperature

Separations of ON molecules have produced better results at higher temperatures ^[5], as this supresses the ONs ability to form three-dimensional structures with itself and other ONs during the separation process – DNA denaturates at higher temperatures ^[1][20] and ONs that cannot form complexes result in a more normalised retention. As such, a high temperature of 60^oC was chosen for all subsequent separations and remained so as to minimise the number of variables during the investigation of other factors.

1.3.6 Flow rate

The flow rate controls the volume of mobile phase that passes through the column, defined by ml/min. UHPLC systems allow for higher operating pressures and faster flow rates, due to the small particle sizes employed in the stationary phase, that can increase peak resolution and decrease analysis times. ^[8][21] After initial scouting experiments, a flow rate of 0.4 ml/min was chosen for the development of the separation method. This allowed for quick ON separations and sharper peaks, without degrading the columns by operating at or near to the pressure limit provided by the manufacturer.

1.3.7 Resolution

The separation resolution is determined by a number of other parameters, the efficiency, retention, and selectivity of the separation method. ^[8][21] Resolution is defined as the following expression:

𝑅𝑠 = ∆𝑡_𝑟/𝑤_𝑎𝑣

(11)

11

where ∆𝑡_𝑟 is the difference in retention time between the two peaks and 𝑤_𝑎𝑣 is the average width of both peaks. ^[8]

Alternatively, the expression can be written as the following:

𝑅𝑠 = 2 ∗ (𝑡_𝑟2− 𝑡_𝑟1 𝑤₂− 𝑤₁)

where tr and w2 are, respectively, the differences in retention and base-width of the two peaks.

[21] The resolution can thus be affected by a number of different factors, such as pH, particle size, temperature, flow rate and gradient slope. A resolution of >1.5, also known as baseline resolution, is desired for quantitative analysis and is often the threshold for an acceptable separation; for further method development purposes, >2.0 is often desired instead. ^[8]

Achieving a high resolution for the separation of complex ON samples is one of the most important aspects of this study, as it is an indicator of an overall efficient and successful separation that can furthermore be applied to larger scale preparative-LC processes.

1.3.8 Asymmetry

Peak asymmetry, often calculated as the asymmetry factor, determines the peak shape of analytes. ^[8] As the solute passes through the column it tends to take on the form of a Gaussian shape, and when there is a partition coefficient equilibrium a perfect Gaussian peak occurs.

This, however, rarely occurs, as the concentration of the solute in each phase changes during the separation, subsequently altering the partition coefficient. ^[8]There are variety of

isotherms that describe the peak shapes that can occur during a separation, the three most common being:

• K decreases, results in tailing of the peak

• K = 1, results in a symmetric, Gaussian peak

• K increases, results in fronting of the peak

The asymmetry factor is calculated using the following expression:

𝐴_𝑆 = 𝐴 𝐵

where A and B is the peak tailing and fronting at 4.4% of the total peak height, respectively. A preferred asymmetry factor should be as close to 1 as possible, where asymmetrical peaks can cause issues with selectivity, resolution, and quantitative purity analysis by interfering with nearby eluted peaks.

1.3.9 Selectivity

Selectivity is used to describe the separation of two analytes by the differentiation of either chemical properties or structural forms. It is an important factor in ON separation, as the differentiation between target ON peaks and impurities is required for efficient separations and the purification of crude ON samples. Selectivity is measured simply by calculating the average adjusted retention time between the two peaks of interest. ^[8]The selectivity of ONs in the proposed DBA hybrid method is determined by altering the method’s ability to separate analytes based upon charge or length. Changes to the counter-ion concentration, mobile phase

(12)

12

pH, and % organic modifier should alter the retention mechanism or ON properties, subsequently altering the selectivity.

1.3.10 Efficiency and theoretical plates

In order to generate the best possible separations of ON samples, a high column efficiency is required. This relates to the column’s ability to separate components of a sample and produce narrow, symmetrical peaks that elute from the packed stationary phase. ^[8][21]Efficiency can be determined by calculating either the plate height H or number of theoretical plates N.

These values relate to the zone broadening that can occur if the mobile phase passes unevenly through the column, affecting the solute equilibrium between the mobile and stationary phase. ^[8][21]Inefficient separations will lead to peak broadening and negatively impact the resolution of separated analytes. By utilising columns with small particle sizes, the distance required for the analytes to diffuse through the column is minimised and thus, increases the column efficiency and peak resolution. ^[21]

Column efficiency is dependant on the number of theoretical plates, determined by the following expression:

𝑁 = 𝐿 𝐻

where N is the number of theoretical plates, H is the plate height and L is the column length

[8]. From this expression, an efficient column requires a large number of plates, which can be achieved by decreasing the plate height or increasing the column length. Greater efficiency will allow the separation of more peaks and with greater resolution and symmetry, a useful advantage when separating target ON molecules from similar impurities – especially in impurity rich, crude samples.

1.3.11 Retention time

The retention time, tr is defined by the time required for an analyte to reach the detector from the moment of sample injection. An adjusted retention time, tr’, can be calculated with the following expression:

𝑡_𝑟^′ = 𝑡_𝑟− 𝑡_𝑚

where tm is the time required for the mobile phase to pass through the column - the adjusted retention time is then the additional time required for the analyte to elute from the column after the mobile phase. The retention time of ONs is a useful parameter that can be utilised in the investigation of the effect of DBA concentration on retention, impurity determination and analysis time optimisation in conjugation with UHPLC.

1.4 Aim of bachelor thesis study

The aim of this study was to investigate the effects of various chromatographic parameters on the separation of modified and unmodified oligonucleotides and investigate the effectiveness of a different IP-RPLC separation method. If a greater and more efficient separation of these

(13)

13

ON samples can be achieved, the analysis of their respective products can thereby also be improved. This would not only benefit the analytical steps of oligonucleotide production but rather the whole process in its entirety. A greater detection of impurities and a more accurate determination of product purity would lead to an end-product of higher quality. Faster analysis times with optimised conditions using an UHPLC system would result in a greater amount of analysis opportunities, improving production efficiency. Furthermore, promising separation methods at the analytical scale could be applied to larger-scale steps in production such as preparative-LC. This can be achieved by using larger columns that share the same properties as their analytical counterparts and identical chemical conditions.

(14)

14

2. Method

2.1 Chemicals and samples

Table 1. The various ON samples provided for this study are listed below in order of nucleotide length. The symbol “*” denotes the oligonucleotide as a scorpion, “**” as an unmodified primer.

Sequence (5’ – 3’) Length

ON-1 – (6-FAM) – 14 bases – (BHQ1+) 16-mer ON-2 – (6-FAM) – 18 bases – (MGB-EDQ) 20-mer

ON-3 – (FAM) – 28 bases – (BHQ1) 30-mer

ON-4 – 44 bases** 44-mer ON-5 – 46 bases** 46-mer ON-6 – (Amine) – 26 bases – (BHQ2) (Spacer 18) – 17 bases* 46-mer ON-7 – (Phosphate) – 57 bases – (ddC) 59-mer Poly-T Suit** 16-, 18- and 19-mer

All oligonucleotide samples were produced in-house and provided for analysis by SGS DNA, Västerås. These samples will henceforth be named as ON-1 – ON-7 in descending order; the Poly-T Suit mix shall remain unchanged. All of the samples presented in this report (with the exception of ON-5 presented in figure 10) have been purified during the purification process of oligonucleotide production. This results in most samples producing very few or even a single peak during the separation analysis, as most of the impurities have been removed.

Purification of difficult samples and non-optimised purification methods can produce samples that retain a noticeable level of impurities (seen, for example, in figures 3, 4 and 15), albeit at a smaller scale than those observed in fully crude products, as seen in figure 10.

An in-house buffer consisting of 10 mM Tris-HCl at pH 8 was used to create the mobile phases. HPLC gradient grade (>99.9%) acetonitrile was purchased from Fischer Chemical and dibutylamine (>= 99.5%) from Sigma Aldrich. Milli-Q water with a conductivity of 18,2 Ω was created using a Milli-Q IQ 7000 from Merck. BioPerformance certified Trizma base (>= 99.9 %) was purchased from Sigma Aldrich and HCl (37%) from Merck KgaA Germany was used to alter the pH of the mobile phases. Tri-sodium phosphate dodecahydrate also from Merck KgaA Germany was used to create a 10 mM phosphate buffer at pH 8.

2.2 Columns and instruments

Table 2. The following columns that were used for the separations in this study are listed below.

Column Dimensions Particle size Pore size Waters Acquity C18 BEH 2.1 x 50 mm 1.7 µm 130 Å Waters Acquity Premier C18 BEH 2.1 x 50 mm 1.7 µm 130 Å

The separation system used in this study was an Acquity UHPLC system from Waters; as such, columns also produced by Waters were chosen for this study due to their increased compatibility with the UHPLC-system. The Waters Premier Acquity column utilises high performance surfaces (HPS) that can lead to an increase in certain parameters such as efficiency and reproducibility. ^[26] Certain fraction samples were also analysed, and these were collected using a Waters Prep-LC system. Data was processed and retrieved using the Empower 3 chromatography data system in conjugation with a Waters LAC/E box.

(15)

15 2.3 Mobile phase preparation

The TEAA-based mobile phases used in previous separations (presented in 3.1) were prepared in-house and were available prior to the start of this study. A consisted of 0.1 M TEAA (20%) at pH 7 whilst B consisted of 0.1 M TEAA (20%) at pH 7 with 80% MeCN.

The DBA-based mobile phases were prepared with 3 different concentrations of DBA; 25 mM, 50 mM, and 75 mM. The initial 800 ml stock solution had a DBA concentration of 0.5 M at pH 8, consisting of MilliQ H20, MeCN and DBA – the pH was adjusted with 37% HCl.

Mobile phases A and B with 25-, 50- and 75 mM DBA were prepared respectively from this stock solution, following the compositions provided in table 3:

Table 3. The compositions of the second batch of mobile phases consisting of various concentrations of DBA are presented below.

Mobile phase 0.5 M DBA Stock (ml) MeCN (ml) 10 mM Tris-HCl (ml)

A. 25 mM DBA pH 8 50 50 900

B. 25 mM DBA pH 8 50 775 175

A. 50 mM DBA pH 8 100 25 875

B. 50 mM DBA pH 8 100 750 150

A. 75 mM DBA pH 8 150 0 850

B. 75 mM DBA pH 8 150 725 125

Mobile phase A, thus, consists of either 25-, 50- or 75 mM DBA with 7.5% MeCN and mobile phase B consists of either 25-, 50- or 75 mM DBA with 80% MeCN. The subsequent 75 mM DBA-phosphor mobile phases utilised in 3.6 were prepared in the same manner as presented above in table 3. All mobile phases were sonicated using a VWR ultrasonic cleaner before use with the UHPLC system.

Important to note is that the initial 50 mM DBA mobile phases used in the separations presented in 3.1 were later calculated at a slightly lower DBA concentration of around 46 mM, due to a small volume error during the mobile phases’ preparation. This does not apply to the 25- and 75 mM separations, nor the other 50 mM separations not presented in 3.1.

Based upon the results obtained in this study, however, the impact of this difference in DBA concentration only leads to the possibility of achieving even greater separations than those presented in 3.1 at 50 mM, had the concentration been at the intended, higher amount.

(16)

16

3. Results and discussion

3.1 Alternative DBA-based mobile phase results

The following results show a comparison of various oligonucleotide sample separations obtained with an established TEAA mobile phase and the investigated DBA-Tris alternative.

Important to note is that the right chromatograph (blue colour-coded) presented in figure 5 and the separation presented in figure 6 are from previous analyses conducted in-house prior to this study. These results have been presented alongside the results gathered in this study in order to visually compare the difference in results between the two separation methods.

Figure 3. Separation of ON-3 using 0.1 M TEAA mobile phases, pH 7. Column specifications: Acquity C18, 1,7 µm, 2.1x50mm. The gradient applied was 8% B 0-1.5 min, then up to 45% B between 1.5 – 19 min; 50 mM DBA.

Figure 4. Separation of ON-3 using DBA-Tris mobile phases, pH 8. Column specifications: Acquity C18, 1,7µm, 2.1x50mm. The gradient applied was 5% B at 0 min, 30% B at 3 min, 52% B at 10 min, 60% B at 13,5 min and 80% B at 19 min; 50 mM DBA.

As shown in figure 4, a change in retention can be observed between the two methods. Due to the binding strength of the DBA-based mobile phases, the gradient method was modified, and a larger amount of organic modifier was required in order to separate the target ON from its closest impurity (seen at 11.215 and 11.033 minutes respectively). This results in the target ON eluting at a higher %B with the DBA method when compared to the standard TEAA system. The consequence of this is an increase to the ON signal height and number of

(17)

17

theoretical plates, creating a more efficient separation method with the possibility of separating a greater number of analytes during the given analysis time. The overall time required is also decreased, from around 17 minutes to 12 minutes.

Figure 5. A side-by-side comparison of ON-7; on the left using DBA-Tris pH 8 mobile phases and on the right using the established 0.1 M TEAA pH 7 mobile phases. The DBA-based method used a Acquity C18, 1,7µm, 2.1x50mm coloum while the previous TEAA-based method used an Acquity C18, 1,7µm, 2.1x100mm coloum.

(The TEAA separation has been scaled to fit within the DBA separation). The applied gradient for the DBA separation was 5% B at 0 min, 30% B at 1 min, 70% B at 3 min and 75% B at 5 min; 50 mM DBA.

In figure 5, two identical samples of ON-7 have been investigated on Waters Acquity columns, varying only in column length (50 mm and 100 mm). ON-7 is a complex 59-mer, where many of the bases are randomly synthesised resulting in a myriad of possible

combinations for the end product. Due to these differences in the ON base-chain, there is a difficulty in creating an optimised method that can successfully generate an efficient,

normalised separation. This difficulty can be observed in the ordinary TEAA method, where the peak is displayed with a large spread as the separation method has an affinity for

hydrophobicity. The DBA method, however, acts as a form of a hybrid IEX/RPLC method which can dynamically separate ONs based on both length and charge. This results in a vastly improved separation with a high peak capacity and number of theoretical plates, increasing the separation efficiency and decreasing analysis times. The sample used in this experiment was previously purified using a larger scale Prep-LC, where impurities such as shortmers have been removed. However, if this method can successfully be applied to crude ON-7 products, these impurities should be easier to identify and separate from the target ON. This would increase the number of impurities that could successfully be removed, thus increasing the overall purity of the end product.

(18)

18

Figure 6. An ordinary separation of ON-1 using the 0.1 M TEAA pH 7 method on an Acquity C18, 1,7µm, 2.1x100mm column. The applied gradient was 10% B at 0 min, 35% B at 2.8 min, 35% at 3 min (flow rate lowered from 0.35 ml/min to 0.208 ml/min at 3 min) and then 53% B at 19 min with a flow rate of 0.208 ml/min.

Figure 7. The separation of ON-1 using the DBA pH 8 method on an Acquity C18, 1,7µm, 2.1x50mm column.

The applied gradient was 5% B at 0 min, 30% B at 3 min, 52% B at 10 min, 60% B at 13,5 min and 80% B at 19 min; 50 mM DBA.

The comparison between the TEAA method and DBA method for the separation of ON-1 is quite similar, with the only difference in column choice is the length (100 mm and 50 mm respectively). The DBA method, however, appears to produce a greater peak height and potentially reveals small amounts of impurities around the base of the target ON peak (the peak to the right of the main ON is a system peak, discussed further in 3.6). An attempt to generate a better separation method which required a shorter analysis time was successful, due to the capabilities of UHPLC instrumentation, utilising higher system pressures and shorter columns designed for the Waters UHPLC system. This separation is presented in figure 8 below:

(19)

19

Figure 8. The separation of ON-1 using the DBA pH 8 method on an Acquity C18, 1,7µm, 2.1x50mm column.

The applied gradient was 5% B at 0 min, 35% at 1 min, 55% B at 3 min and 75% B at 5 min; 50 mM DBA. To achieve this separation, the only modification applied to the previous separation presented in figure 7 was a gradient optimisation. The analysis time was first reduced to a total of 7 minutes, and the % values of mobile phase B required to achieve an efficient separation were subsequently changed accordingly.

This separation continues to generate a high peak height and narrow peaks, resulting in an efficient ON-1 separation within a fraction of the time required for the previous TEAA method. The target ON peak elutes here within of time frame of around 12 seconds, compared to the TEAA method of 1 minute – 1/5^th of the time originally required.

Furthermore, a greater separation from the system peaks was also achieved; however, this will not be required in future methods if these peaks can be successfully removed. It is believed that with further optimisation, a separation of ON-1 of similar or improved quality could be achieved within a shorter overall analysis time. This is an example of the

effectiveness of DBA-Tris mobile phase in conjugation with the capabilities of UHPLC and sub-2 µm columns. ^[14]

Figure 9. The separation of ON-4 and ON-5 using the DBA pH 8 method with 75 mM DBA; column used was a Waters Acquity Premier C18 BEH, 1,7µm, 2.1x50mm. The applied gradient was 5% B at 0 min, 20% B at 2 min, 40% B at 6 min and 45% B at 7 min.

(20)

20

Figure 9 presents an example of a successful separation of two unmodified ONs, ON-4 and ON-5. These primers are 44- and 46-bases long and are similar in length and composition so that their separation can prove difficult. When the DBA method is applied in conjugation with an UHPLC system, an efficient, acceptable separation within a short analysis time can be achieved; the resolution between these two peaks was calculated at around 1,6.

Figure 10. The separation of a crude sample of ON-5, using the DBA pH 8 method with 75 mM DBA; column used was a Waters Acquity Premier C18 BEH, 1,7µm, 2.1x50mm. The applied gradient was 5% B at 0 min, 20% B at 2 min, 48% B at 6 min and then 80% at 7 min. System peaks appear after the gradient is raised to 100% B to wash the column.

An example of a crude product separated using the developed DBA method can be seen above in figure 10. The multitude of peaks that appear before the target ON-5 peak at 5.686 minutes are shortmers, while the peaks that appear afterwards are longmers. Crude products are purified at a larger scale using preparative-LC (Prep-LC) to remove the sample of impurities, with the intent of leaving as much of the target ON as possible to create an end product of high purity. Through the development of improved separation methods at the analytical UHPLC scale, successful methods can then be applied and adapted to the larger scale purification method. By increasing the peak efficiency and selectivity of impurities from target analytes with the DBA-based method, shortmers and other impurities will be easier to identify and remove during the purification process. This should improve the overall efficiency of purification, generating ON end products with a higher level of purity.

3.2 Effect of varying DBA concentrations

In the following figures (figure 11, 13, 15 and 16), separations of various samples are presented in an overlay chromatograph. These separations show the results from identical separation methods that varied only in the concentration of DBA in the mobile phases. The colour indication is as follows; black = 25 mM, blue = 50 mM, green = 75 mM. The intent of these experiments was to systematically investigate the effect of DBA concentration on the separation of various ON samples.

A system suitability test (SST, or “SUIT”) is often done on new separation systems (changed column, mobile phase etc.) before samples are analysed. This is done to provide insight to the new system’s separation capabilities and consists of a simple mixture of oligonucleotides

(21)

21

with similar lengths. It is important that an acceptable separation can be achieved on the new system and thus, methods that provide such results and can even be optimised are desired.

The SUIT mixture used in this study consisted of 3 oligonucleotides that differed only in length – 16, 18 and 19 bases, respectively.

Figure 11. The separation of Poly T suit using 25-, 50- and 75 mM DBA, respectively. Column specifications;

Waters Acquity Premier C18 BEH, 1,7um, 2.1x50mm. The applied gradient was 5% B at 0 min, 30% B at 1.5 min, 35% B at 6.5 min and 65% B at 7.5 min. The system peaks were eluted by increasing the %B to 100, to flush out the column with high levels of organic modifier.

In figure 11, the retention of the Poly-T Suit mix increases with the DBA concentration. The resolution of the separation also appears to drastically increase at higher DBA concentrations.

To determine the resolution achieved in a Poly-T separation, the measurement is made by the program between the last two peaks, T-18, and T-19.

Figure 12. The resolution of Poly-T Suit plotted against the DBA concentrations 25-, 50- and 75 mM

From the data presented in figure 12, the resolution increases in an almost linear fashion with the increase to DBA concentration, with an R²-value of 0.998. As an acceptable resolution requires a value of 1.5, and a preferred value of at least 2.0 for further method development, a

1.7564611

2.604947

3.333246 y = 0.0315x + 0.9881

R² = 0.9981

0 0.5 1 1.5 2 2.5 3 3.5 4

0 10 20 30 40 50 60 70 80

Resolution

DBA concentration (mM)

(22)

22

separation using 25- and 50 mM DBA would be adequate for these intentions, respectively.

Thus, while 75 mM DBA does improve the resolution further, it is not actually required for Poly-T Suit separations. Furthermore, utilising lower DBA concentrations in the case of Poly- T Suit separations benefits from a greater robustness to gradient changes, results of which are presented in 3.3.

Interesting to note for this sample is that contrary to the other samples in this experiment, the peak height decreases with increasing DBA concentration. Furthermore, peak widening can also be observed at higher DBA concentrations. This indicates that a more efficient separation can be achieved at lower DBA concentrations for the separation of Poly-T Suit, as the number of theoretical plates is higher.

Figure 13. The separation of ON-1 using 25-, 50- and 75 mM DBA, respectively. Column specifications; Waters Acquity Premier C18 BEH, 1,7um, 2.1x50mm. The applied gradient was 5% B at 0 min, 35% B at 1 min, 55%

B at 3 min and 75% B at 5 min.

ON-1 is notoriously difficult to separate on a variety of systems, due to its modified base- chain which increases the ONs overall hydrophobicity. A hypothesis about the behaviour of ONs such as ON-1 is that the increased hydrophobicity due to the modifications results in it forming three-dimensional structures with itself. As there is a possibility that only certain amounts of ON-1 form structures with itself, this could lead to a variation in the retention and subsequent peak widening. As stated in 1.2, the hydrophobicity of such modifications can be supressed by utilising strong ion-pair reagents. Furthermore, a larger amount of organic modifier is required to eluate the analytes due to the ion-pairing strength of DBA. This results in an overall more efficient separation, with an increase to peak height and number of

theoretical plates as the varied retention of ON-1 is supressed.

As previously observed with Poly-T Suit in figure 11, ON-1 gains increased retention at higher DBA concentrations. However, ON-1 also appears to benefit from an increased peak height and number of theoretical plates. The calculated column efficiency (N) is presented below in figure 14:

(23)

23

Figure 14.The Column efficiency of the ON-1 separation plotted against the DBA concentrations 25-, 50- and 75 mM

Once again, the column efficiency increases in an almost completely linear fashion with the concentration of DBA in the mobile phases, with an R²-value of 0.999.

These results indicate that the most efficient separation of ON-1 is achieved with a high concentration of DBA in the mobile phases. Important to note, however, is the occurrence of increased tailing at higher DBA concentrations observed in both the separation of Poly-T Suit and ON-1. This could be due to the increase in ion-bonding strength at higher DBA

concentrations, requiring a larger amount of organic modifier to eluate the analytes.

Figure 15. The separation of a two-fraction sample from a ON-6 prep sample using 25-, 50- and 75 mM DBA, respectively. Column specifications; Waters Acquity Premier C18 BEH, 1,7um, 2.1x50mm. The gradient used was linear; from 5% B at 0 min to 80% B at 15 min. Thereafter the column was flushed with 100% B.

The sample presented above in figure 15 is a spiked mixture of two fractions from ON-6.

This ON, defined as a scorpion, is extremely complex with several modifications, and was prepared from two fractions generated using Prep-LC. This sample was of interest to study as the target ON, seen here as the largest peak, shares an almost identical mass number with an

3.44E+04

4.90E+04

6.41E+04 y = 594.53x + 19432

R² = 0.9999

0.00E+00 1.00E+04 2.00E+04 3.00E+04 4.00E+04 5.00E+04 6.00E+04 7.00E+04

0 10 20 30 40 50 60 70 80

Column efficiency (N)

(24)

24

impurity. This creates a difficult separation and the results generated using the DBA method requires further investigation with mass spectrometry to determine how much, if any, of the impurity was successfully separated from the target ON. The second largest peak belongs to another known impurity, with a less similar mass number to the target ON. As observed in previous separations, the retention of ON-6 increases with DBA concentration. Furthermore, the peak height and number of theoretical plates also increase; these results indicate that even heavily modified scorpions can be more efficiently separated with higher concentrations of DBA. Note the appearance of a potential impurity peak, from the second-largest peak of an already known impurity, at higher concentrations of DBA.

Figure 16. The separation of ON-7 using 25-, 50- and 75 mM DBA, respectively. Column specifications; Waters Acquity C18 BEH, 1,7um, 2.1x50mm. The applied gradient was 5% B at 0 min, 30% B at 1 min, 70% B at 3 min and 75% B at 5 min.

As previously mentioned in 3.1, ON-7 has a multitude of possible combinations and a long chain of 57 bases. Due to the length of this 59-mer ON, conventional IEX methods often have difficulty with the separation of ON-7 samples. The proposed hybrid IEX/RPLC method utilising DBA can however separate ON-7 with incredibly efficiency, shown in figure 5. As the selectivity is controlled by both the ON charge and length, this method can separate ON-7 regardless of its long length with increased efficiency due to the electrostatic interaction generated by the ion-pair reagent.

The results for this separation, however, vary slightly from many of the other ON samples.

While the retention and peak height increases with increasing DBA concentration, the highest number of theoretical plates is instead acquired at 50 mM DBA, not at 75 mM as previously seen. This is most likely due to the appearance of an impurity from the target ON peak at 75 mM DBA, leading to peak distortion and widening. These results are presented further in 3.4.

The calculated column efficiency from the separation of ON-7 is presented below in figure 17:

(25)

25

Figure 17.The Column efficiency of the ON-7 separation plotted against the DBA concentrations 25-, 50- and 75 mM

The universal increase to ON retention at higher concentrations of DBA is potentially due to the interaction with DBA and the organic modifier in the mobile phase. At higher

concentrations of DBA, the electrostatic interaction with the oligonucleotide should increase, thus requiring larger amounts of organic modifier to eluate the sample. This results in an increase to retention and can be utilised if the retention of analytes should need to be regulated. Furthermore, the advantages of using a high DBA concentration, such as more efficient separations, can be retained whilst the retention and overall analysis time can be decreased. This can be achieved by optimising the gradient method, an example of which is presented in 3.1. Important to note is that the above ON-7 separation was not performed on a Premier Acquity C18 column unlike the other three ON samples previously presented in this section.

3.3 Gradient robustness

The following chromatographs show the effect of small changes to the gradient on the separation of an oligonucleotide primer. These tests were conducted in order to investigate the robustness of the separation method using the DBA-based mobile phases. The poly-T primer was chosen as it contains the simplest oligonucleotides out of the provided samples, with three molecules separated in close proximity to each other. This would provide

information on how the oligonucleotides were affected by small gradient changes not only individually, but also in relation to each other, forming a good baseline for the future analysis of more complex samples. The following gradients were used in the study of Poly-T Suit mix gradient robustness at 25- and 75 mM DBA, based upon previous results obtained with the denoted “standard” gradient presented in figure 11:

0.00E+00 5.00E+03 1.00E+04 1.50E+04 2.00E+04 2.50E+04

0 10 20 30 40 50 60 70 80

Column efficiency (N)

(26)

26

Table 4. The standard gradient applied to the separation of Poly-T Suit mix

Time (min) %A %B

0 95 5

1.5 70 30

6.5 65 35

7.5 35 65

End-wash 0 100

Table 5. The modified gradient applied in the gradient robustness study, with +1% B at 6.5 min

Time (min) %A %B

0 95 5

1.5 70 30

6.5 64 36

7.5 35 65

End-wash 0 100

Table 6. The modified gradient applied in the gradient robustness study, with -1% B at 6.5 min

Time (min) %A %B

0 95 5

1.5 70 30

6.5 66 34

7.5 35 65

End-Wash 0 100

Figure 18. The separation of an oligonucleotide primer (Suit Poly-T) using 25 mM DBA mobile phases on a Waters Acquity Premier C18, 1,7µm, 50x2.1 mm column. The colour indication is as follows; black = standard gradient, blue = 1% increase to total % MeCN, green = 1% decrease to total % MeCN.

From the results shown in figure 18, the retention of the analytes acts as anticipated when the gradient is changed. By raising the gradient by 1% B, the analytes elute sooner as the increase to the total %B is now at a greater rate. By lowering the gradient by 1% B, the analytes

(27)

27

instead eluate later as the rate of which the %B increases during the separation time frame decreases. The longest ON (rightmost peak) appears to experience the greatest change in retention due to the gradient change whilst the shortest ON (leftmost peak) is the least affected.

As stated in 3.2, the resolution for Poly-T Suit tests is determined by the program, which calculates the resolution between the last two peaks, T-18, and T-19. The difference in resolution is presented in the following table:

Table 7. The calculated resolution for the Poly-T Suit mix for each gradient, using 25 mM DBA

Gradient (%B) Resolution

Standard 1.701

+1 1.625

-1 1.652

The standard gradient was chosen from previous experiments as it generated the best

separation results. The resolution appears to be negatively impacted through a change of +/- 1% B, though neither change causes the resolution to drop below the 1.5 threshold.

Figure 19. The separation of an oligonucleotide primer (Suit Poly-T) using 75 mM DBA mobile phases on a Waters Acquity Premier C18, 1,7µm, 50x2.1 mm column. The colour indication is as follows; black = 1%

decrease to total % MeCN, blue = 1% increase to total % MeCN.

In figure 19, the retention of the analytes using 75 mM DBA acts in the same manner as the previous test using 25 mM DBA in figure 18. When the total %B is increased the retention is decreased, with an opposite effect when the total %B is decreased. Furthermore, the longest ON continues to be the most affected by the change to the gradient, with the shortest ON affected least. Due to a limited sample amount, a standard gradient test could not be conducted; however, the calculated resolution for the separations shown in figure 19 is presented in the following table:

(28)

28

Table 8. The calculated resolution for the Poly-T Suit mix for each gradient, using 75 Mm DBA

Gradient (%B) Resolution

+1 2.865

-1 3.048

As previously discussed in 3.2, the resolution of Poly-T Suit is vastly improved when using a higher concentration of DBA in the mobile phases. However, it appears in figure 19 that the resolution is more sensitive to changes to the gradient at a higher DBA concentration than at lower concentrations, as seen in figure 18. This increased sensitivity is also observed in the retention of the ON molecules at higher DBA concentrations.

These results indicate that the overall robustness in gradient stability decreases with an increase to the concentration of DBA in the mobile phases. This must then be accounted for in subsequent experiments and in the further development of this hybrid separation method.

When utilising mobile phases with high concentrations of DBA, care should be taken in the preparation of the mobile phases and choice of gradient in order to ensure as little variation as possible.

3.4 Appearance of potential impurity peaks at higher DBA concentrations

Initially when investigating the separation capability of DBA mobile phases, certain purified probe samples such as ON-1 and ON-7 appeared to consist of a singular, pure peak alongside the two peaks presented in 3.6. However, after the concentration of DBA was raised to 75 mM and systematic DBA concentration tests were concluded, various peaks can be observed around the main peaks that share similar absorbance spectra. Similar results could be

obtained by changing the method’s gradient; however, this was not always reproducible. At a higher DBA concentration results containing these possible impurity peaks could, however, repeatably be obtained.

Figure 20. Peak base of ON-1 using 75 mM DBA; column used was a Waters Acquity Premier C18 BEH, 1,7µm, 2.1x50mm.The applied gradient was 5% B at 0 min, 35% B at 1 min, 55% B at 3 min and 75% B at 5 min.

(29)

29

Figure 21. An integrated version of the separation shown in figure 20, where the potential impurity can be more clearly observed appearing at the start of the peak. The applied gradient was 5% B at 0 min, 35% B at 1 min, 55% B at 3 min and 75% B at 5 min; 75 mM DBA.

Figure 22. The corresponding absorbance spectra for the two peaks shown in figure 21; the first spectrum (red) represents the small, potential impurity peak whilst the second spectrum (blue) represents the main ON peak.

(30)

30

In the absorbance spectra presented in figure 22, the peak appearing before the main peak appears to share a similar spectrum with the exception of two small differences. A signal at 291.5 nm is no longer significant enough to be registered and instead, a small peak can be observed towards the end of the spectrum at 649.6 nm – this may be evidence of a shortmer.

This could potentially indicate that an impurity lacking a component is hidden beneath the pure oligonucleotide peak and cannot be observed normally due to a difficulty to separate two almost identical molecules. At a higher concentration of a strong ion-pair such as DBA in the mobile phase, the binding of the sample molecules could thus be more selective, separating small amounts of similar impurities from the target molecule. If an impurity lacks a certain component that results in a small change to ON length, standard separations may be unable to differentiate the ONs from each other. With the addition of DBA however, the mechanism of separation instead potentially acts as a length-charge hybrid system. If the absence of a certain component results in small decrease to length but a significant change to the molecule’s charge, then DBA-based separations at certain concentrations could still

successfully differentiate and separate the two molecules based on charge instead of length.

It is important to note that these results require further investigation and validation before any conclusions can be made about the presence of hidden impurities and thus, the purity of purified ON products. A comparison with established TEAA methods and mass spectrometry investigation would potentially provide adequate results in order to validate these potential impurities. As the selectivity of analytes can be altered by changing ion-pair reagents ^[4], it is possible that these peaks could be obtained in other separation methods such as TEAA-based mobile phases, but with various retention times and binding mechanisms.

Figure 23. Peak base of ON-7 using 75 mM DBA; column used was a Waters Acquity C18 BEH, 1,7µm, 2.1x50mm. The applied gradient was 5% B at 0 min, 30% B at 1 min, 70% B at 3 min and 75% B at 5 min.

(31)

31

Figure 24. An integrated version of the separation shown in figure 23, where the potential impurities can be more clearly observed appearing at the start of the peak. The applied gradient was 5% B at 0 min, 30% B at 1 min, 70% B at 3 min and 75% B at 5 min; 75 mM DBA.

Figure 25. The corresponding absorbance spectra for the three peaks shown in figure 24; the first two spectra (red and blue) represent the two potential impurity peaks that appear before the main peak, respectively. The

third spectrum (green) represents the main ON peak.

(32)

32

Similar results to the separation of ON-1 were obtained for ON-7, another purified probe sample at 75 mM DBA. Two peaks appear before the target oligonucleotide and both share a similar absorbance spectrum, with a smaller but prominent signal at 258.2 nm alongside background interference. As previously discussed with the separation of ON-1, the effect of a higher concentration of DBA in the mobile phase on the discovery of potential impurities appears to also apply to this sample. Once again, these results require further investigation and validation with methods such as mass spectrometry.

Interesting to note in the case of this ON sample, is the possibility of countless variants of shortmers from the synthesis of ON-7. As previously mentioned, ON-7 has multiple bases that are synthesised randomly, thus creating a vast amount of similar end product

combinations. This means that for every end product, there are also multiple shortmer impurities that are produced through synthesis failures. If these two impurity peaks with shorter retention times belong to ON-7 shortmers, this could prove the strength of the DBA- method to separate impurities from the target ON molecule. As a hybrid IEX/RPLC system, this method could dynamically separate ONs based upon both charge and length, with the ability to differentiate and separate shortmer impurities from a target ON that can consist of a vast number of combinations.

(33)

33

3.5 Comparison between Acquity and Acquity Premier columns

The following chromatograms show the effect of applying the same separation method on a standard Waters Acquity column and a Waters Premier Acquity column.

Figure 26. The separation of ON-1 using 50 mM DBA mobile phases; the applied gradient was 5% B at 0 min, 35% B at 1 min, 55% B at 3 min and 75% B at 5 min. Figure 26. A) presents the separation conducted on a Waters Acquity C18 BEH, 1,7µm, 2.1x50mm column whilst figure 26. B) presents the separation conducted on the Premier counterpart.

Table 9. Data from the comparison of a ON-1 separation using a standard Acquity and Acquity Premier column;

theoretical plates (N), asymmetry factor (As) and retention time (Rt).

Column N (*10⁴) As Tr (min)

Acquity 4.34 2.02 3.70

Acquity Premier 5.33 2.29 3.59

A

B

(34)

34

From the data collected it appears that, when the separation method is applied to the Premier column, both the number of theoretical plates and the asymmetry factor increases, whilst the retention time decreases. Interesting to note is the appearance of a potential impurity within the tail of ON-1 when using the Premier Acquity column.

Figure 27. The separation of ON-7 using 50 mM DBA mobile phases; the applied gradient was 5% B at 0 min, 30% B at 1 min, 70% B at 3 min and 75% B at 5 min. Figure 27. A) presents the separation conducted on a Waters Acquity C18 BEH, 1,7µm, 2.1x50mm column whilst figure 27. B) presents the separation conducted on the Premier counterpart.

Table 10. Data from the comparison of a ON-7 separation using a standard Acquity and Acquity Premier column; theoretical plates (N), asymmetry factor (As) and retention time (Rt).

Column N (*10⁴) As Tr (min)

Acquity 2.33 1.43 2.26

Acquity Premier 2.70 1.76 2.32

Once again, the number of theoretical plates and asymmetry factor appear to increase when the method is applied on the Premier quality, however in the case for this oligonucleotide the retention appears to increase on the Premier column instead of decreasing.

A

B

(35)

35 3.6 System peaks and mobile phase investigation

A pair of peaks can be observed in most of the chromatographic results when using DBA as an ion-pair, appearing after the main oligonucleotide peak at higher concentrations of organic modifier. The results of an attempt to determine their origin and subsequent removal are presented here, with tests done on a modified mobile phase utilising a phosphor buffer in place of Tris-HCl.

Figure 28. The absorbance spectra of the two “system peaks” which were observed in various separations; an example can be seen in figure 27. B) after the main oligonucleotide peak between 3.6 and 4 minutes.

Figure 29. The separation of ON-1using 75 mM DBA-Phosphor mobile phases and a Waters Acquity Premier C18 BEH, 1,7µm, 2.1x50mm column. The applied gradient was 5% B at 0 min, 35% B at 1 min, 55% B at 3 min and 75% B at 5 min.

The impact of various chromatographic conditions on the separation of modified and unmodified oligonucleotides