A 502-Base Free-Solution Electrophoretic DNA Sequencing Method

Subscriber access provided by SIMON FRASER UNIV BURNABY

Article

A 502-Base Free-solution Electrophoretic DNA Sequencing Method using End-attached Wormlike Micelles Stephen B. Istivan, Daniel K. Bishop, Angela L. Jones, Shane T. Grosser, and James W Schneider Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.5b02931 • Publication Date (Web): 10 Oct 2015 Downloaded from http://pubs.acs.org on October 19, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

A 502-Base Free-solution Electrophoretic DNA Sequencing Method using Endattached Wormlike Micelles Stephen B. Istivan1, Daniel K. Bishop2, Angela L. Jones1, Shane T. Grosser1, and James W. Schneider1* 1

Department of Chemical Engineering, Carnegie Mellon University, Pittsburgh PA 15213

2

Department of Biomedical Engineering, Carnegie Mellon University, Pittsburgh PA

15213

Keywords:

DNA electrophoresis, DNA sequencing, ELFSE, Micellar electrokinetic

chromatography (MEKC)

Abstract

We demonstrate that the use of wormlike nonionic micelles as drag-tags in end-labeled free-solution electrophoresis (“micelle-ELFSE”) provides single-base resolution of Sanger sequencing products up to 502 bases in length, a nearly two-fold improvement over reported ELFSE separations. “CiEj” running buffers containing 48 mM C12E5, 6 mM C10E5, and 3 M urea (32.5°C) form wormlike micelles that provide a drag equivalent to an uncharged DNA fragment with a length (α) of 509 bases (effective Rh = 27 nm). Runtime in a 40-cm capillary (30 kV) was 35 mins for elution of all products down to the 26-base primer. We also show that smaller Triton X-100 micelles give a read length of 103 bases in a 4-min run, so that a combined analysis of the Sanger products using the two buffers in separate capillaries could be completed in 14 mins for the full range of lengths. A van Deemter analysis shows that resolution is limited by diffusionbased peak broadening and wall adsorption. Effects of drag-tag polydispersity are not observed despite the inherent polydispersity of the wormlike micelles. We ascribe this to a stochastic size-sampling process that occurs as micelle size fluctuates rapidly during the runtime. A theoretical model of the process suggests that fluctuations occur with a time *

author to whom correspondence should be addressed

1

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

scale less than 10 msec, consistent with the monomer exchange process in nonionic micelles. The CiEj buffer has a low viscosity (2.7 cP) and appears to be semi-dilute in micelle concentration. The large drag-tag size of the CiEj buffers leads to steric segregation of the DNA and tag for short fragments and attendant mobility shifts.

Introduction

Electrophoretic separations of DNA are required for countless types of medical, biological, and forensic analysis. Historically, electrophoretic separations of DNA have been performed using a polymeric sieving matrix that provides for a length-dependent electrophoretic mobility. In the absence of the sieving matrix (“free solution”), DNA fragments will migrate much more quickly, but with little dependence on their length and therefore no useful resolution. This is due to the invariance of the ratio of charge to friction, which establishes electrophoretic mobility, for ssDNA fragments greater than about 20 bases in length.1 In 1992, Noolandi proposed to break the scaling symmetry of charge to friction by attaching a protein, virus, or nanoparticle to the end of the DNA that possesses a significantly different charge-to-friction ratio so that resolution could be achieved in a free-solution separation.2 In principle, the method promised very rapid DNA separations, without complications brought about by injecting viscous polymer solutions into capillaries. Later work by Slater and coworkers examined the theoretical limits of this “end-labeled free-solution electrophoresis (ELFSE)” method3 and used the protein streptavidin as an end-label to separate Sanger sequencing products in capillary electrophoresis.4 They were able to distinguish fragments up to about 110 bases in length with single-base resolution. While this was not sufficient to be competitive with gelbased separations, it did demonstrate the feasibility of ELFSE for high-resolution DNA analysis. Later work focused on using larger end labels (“drag-tags”) composed of linear and branched polymers,5-10 but these did not possess size and monodispersity required to extend the read length in ELFSE. The Barron group has recently developed a biosynthesized protein polymer composed of a repeating 7-residue block for use as an ELFSE drag-tag, which provided a read length of 280 bases, the longest reported to date.11 Here, a series of drag-tags with 27, 36, 54, and 72 blocks were used as large drag-

2


Page 2 of 27

Page 3 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


tags. Drag-tags above 36 blocks in size were not viable as they interfered with the action of polymerase during the cycle sequencing reactions. While the post-extension attachment of drag-tags is an alternative, there remains an upper limit to the size of polymer or protein drag-tags that can be synthesized with precise control of their size polydispersity. This has been the principal barrier to accessing higher read lengths using the rapid, gel-free ELFSE methods.12

Our group has developed a process (“micelle-ELFSE”) where DNA oligonucleotides are covalently linked to long-chain alkanes prior to separation in micelle electrokinetic chromatography (MEKC),13 using running buffers that contain 1-3% nonionic surfactant.14-18 During electrophoresis, micelles bind transiently to the end-grafted alkane, forming a temporary DNA-micelle complex with the micelle acting as a drag-tag. Use of a large excess of micelles over alkylated DNA ensures that the migrating DNA is attached to a micelle drag-tag for nearly the entire runtime.14 As we show here, the main advantage of the micelle-ELFSE approach is that when properly implemented it allows for use of large drag-tags without suffering polydispersity-based band broadening. Even though populations of micelles are generally polydisperse in size, the size of an individual micelle fluctuates rapidly owing to the rapid exchange of surfactant monomer from the solution phase to micelle phase.19 Likewise, the end-alkylated DNA fragments will exchange between the two phases, and therefore among distinct micelle drag-tags, during the runtime. This statistical size-sampling process, if sufficiently rapid compared to the runtime, would ensure that each alkylated DNA fragment experience a highly uniform drag to minimize polydispersity-based band broadening. Additionally, the micelle drag-tags are present only in the running buffer, and are not attached to the primers to be extended in PCR or Sanger reactions, so there is negligible interference with enzyme activity. Running buffers can be formulated to provide a drag-tag size appropriate for a desired separation and different drag-tag sizes can be applied to a single extension product without the need for multiple attachment steps. The surfactant running buffers used here have low viscosity (less than 3 cP) and are easily introduced into thin capillary tubes. We choose to end-alkylate DNA primers prior to extension by an activated linkage of commercially available, amine-terminated oligomers to long-chain

3



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

fatty acids. We note that an on-resin approach, using alkylated phosphoramidites, would simplify the workflow for larger-scale applications of micelle-ELFSE.

Here we apply the micelle-ELFSE method to the analysis of Sanger sequencing products and observe read lengths of over 500 bases using a commercial bench-top CE instrument. We discuss some the parameters of interest when formulating effective running buffers for micelle-ELFSE and identify the main sources of band broadening observed.

Materials and Methods

Alkylation of DNA Oligonucleotides. The method of DNA alkylation is based on a procedure from Boutorine et al.20 Prior to electrophoresis, 5’-amine-terminated DNA oligomers are end-alkylated by a simple bioconjugation method linking them to octadecanoic acid (all reagents Sigma-Aldrich unless stated). Briefly, an activating solution containing the following reagents is prepared: 20 uL of 5 mM octadecanoic acid in dimethyl sulfonic acid (DMSO), 1 µL of triethylammonium acetate (TEA) in DMSO, 5 µL of 1 M 4-dimethylaminopyridine (DMAP) in DMSO, 5 µL of 0.5 M thiamine pyrophosphate (TPP) in DMSO, and 5.0 µL of 0.5M dipropyl disulfide (DPDS) in DMSO. The activating solution is heated to 80°C and allowed to stand at room temperature for 20 min. In a separate vial, 10 nmole of commercially available, 5’-amine terminated DNA oligomer (Integrated DNA Technologies) is precipitated in a 5.0 µL of a 0.15 M solution of cetytrimethylammonium bromide (CTAB), pelletized, and dried. 36 µL of the activating solution is added to the dry pellet and shaken for several hours. The product is precipitated in 1.0 mL of 2.0 wt% LiClO4 in acetone, pelletized and washed three times with acetone. The pellet is resuspended in 100 uL of 0.1 M triethylammonium acetate (TEAA) and purified by reversed-phase HPLC. HPLC is performed using a Waters 4.6 x 250 mm Symmetry300 C18 HPLC column (Milford, MA) with a 1 mL/min flowrate and linear gradient from 0.1M TEAA to pure acetonitrile over 30 mins. Unmodified DNA elute near 10 min while alkylated DNA elute between 20 to 24 min. Yields vary from 40-70% and the product molecular weight is confirmed by MALDI mass spectroscopy.

4


Page 4 of 27

Page 5 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Sanger Sequencing Reactions. Sanger sequencing reactions were carried out using the Thermo SequenaseTM Cy5 Dye Terminator Sequencing Kit (GE Life Sciences. Piscataway, NJ). Alkylated primers were extended in the presence of M13mp18 single stranded DNA template purchased from New England Biolabs (Ipswich, MA) using a thermocycler (SmartCycler, Cepheid, Sunnyvale, CA). All material concentrations, temperature profiles, and protocols were carried out as outlined in the Thermo SequenaseTM kit manual. Total reaction volumes were 25 µL, and ethanol precipitation was chosen for purification. Purified products were dried by vacuum centrifugation, suspended in 30 µL of Hi-Di Formamide (Promega, Madison WI), heated to 90°C for 5 mins, and stored at -20°C prior to use.

Buffer preparation. The nonionic surfactants Triton X-100, pentaethylene glycol monododecyl ether (C12E5), and pentaethylene glycol monodecyl ether (C10E5) were purchased from Sigma-Aldrich and used as received. 1-2 ml of running buffer containing 48 mM C12E5, 6 mM C10E5, and 3 M urea (Electrophoresis grade, Sigma-Aldrich) was prepared by combining in 1x TBE (pH 8.0, 10x concentrate from Sigma-Aldrich) and shaken for 1 hour. After shaking, 0.1% (v/v) PoP-6 (Thermo Fisher, formerly Applied Biosystems) was added to the solution, and the solution was vortexed for 15 sec. The solution was then centrifuged for 15 mins and transferred to a buffer vial for electrophoresis runs. A 10% (v/v) solution of PoP-6 on 1x TBE was prepared by adding an appropriate volume of PoP-6 to 1x TBE (pH 8.0) and vortexing for 15 sec.

Capillary electrophoresis. All micelle-ELFSE experiments were carried out in a Beckman P/ACE MDQ (Beckman Coulter, Fullerton, CA) equipped with laser induced fluorescence (LIF) detection (488/635 ext., 512/680 ems.). Fused-silica capillaries with 30 µm I.D. (Polymicro Technologies) were cut to 40 cm in length and a window burned following manufacturer specifications. Newly prepared capillaries were pressure rinsed at 20 psi using a 10% (v/v) mixture of PoP-6 in 1x TBE for 20 mins followed by a 5 min rinse with micelle containing electrophoresis buffer. Before individual experiments capillaries were rinsed for 5 mins with buffer at 20 psi. Capillary tips and electrodes were

5



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 27

then soaked in pure water for several seconds, and alkylated DNA fragments injected at 4 kV for 30 sec. Electrophoresis was carried out at 32.5°C and voltages between 10-30 kV with LIF detection (635 nm/680 nm).

Results and Discussion

In ELFSE separations, the electrophoretic mobility of tagged DNA is modeled accurately by the working equation:4  L    L +α 

µ = µ0 

(1)

where µ is the mobility of the DNA-tag complex, µ0 is the mobility of untagged DNA, and L is the number of DNA bases in the tagged DNA fragment. The size of drag-tag is expressed by the value of α, which can be envisioned as the length of a hypothetical uncharged DNA fragment that would possess the same drag as the drag-tag. Meagher and coworkers have shown that, in the limit of band broadening due to diffusion alone, the read length increases substantially with α, from a value of 129 bases for α = 30, to a value of 312 bases for α = 200 (E = 300 V/cm).12 The main reason for this improvement is an increase in the spacing between the centers of peaks belonging to DNA fragments that differ by a single base (∂t/∂L), which increases linearly with α. The Barron group has synthesized a series of polypeptide drag-tags with α varying from 70-152.11,21,22 The α=70 drag-tag gave a read length of 265 bases which is the longest yet reported.11 Larger drag-tags blocked enzymatic extension and therefore read lengths could not be measured. In our earlier work, we showed that Triton X-100 (TX-100) buffers gave α =75 in micelle-ELFSE but they were not interrogated using Sanger products.15

To achieve significantly longer read lengths, we chose to use surfactant formulations that produce longer, wormlike micelles to increase the α value observed in micelle-ELFSE. While TX-100 forms small, oblate ellipsoids,23 nonionic surfactants with longer hydrophobes (C12E5) tend to yield extended wormlike micelles.24 At high concentrations, these micelles can overlap and entangle to form a gel that we expect to sieve DNA rather than act as isolated drag-tags.25 To prevent this, we also added nonionic

6


Page 7 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


surfactant with a shorter hydrophobe (C10E5) to serve as an end-cap of the cylinders to limit their length. After some screening, we found that the formulation of 48 mM C12E5, 6 mM C10E5 in 3 M urea (T=32.5°C) gave the best combination of high α without excessive peak broadening (“CiEj buffer” hereafter). It was also necessary to add 0.1% PoP-6 to the running buffer to suppress EOF and ensure proper injection.

Micelle-ELFSE experiments were carried out using a commercial CE instrument that has access to high electric fields (Beckman Coulter P/ACE MDQ, up to 1000 V/cm). This instrument lacks the optics required for detection of the FRET-coupled dye sets commonly used for Sanger analysis, so we instead used a kit with a single dye terminator and independently created Sanger products for each base termination. These were injected as four distinct samples to yield four electropherograms. There was no detectable EOF drift with consecutive runs so their traces were overlaid to assess the read length (Figure 1). The sequence of the template is listed above the composite electropherogram, demonstrating very good agreement with the order of the peak elution up to 502 bases. We note that the run of “A” near 4.75 mins exhibits a minimum precisely where the G trace shows a peak, and the template has a single G interrupting the A run at the corresponding fragment length (499 bases). The total runtime was 32 minutes, including elution of all fragments down to the 26-base primer.

For the analysis of shorter fragments, much faster runs are realized using smaller micelle drag-tags. Figure 2 shows an electropherogram for the T-terminated channel of the same products using a TX-100 running buffer. Here, resolution of fragments is achieved from about 26-100 bases in less than 4 minutes. This suggests a means to reduce total analysis time by running the same sample in two different buffers, one with small micelle dragtags and one with larger ones. Such an approach would reduce the total runtime to about 14 minutes, albeit at the expense of using two capillaries. An analysis of the design tradeoffs presented by this approach, using non-linear optimization methods, has been published by our group.17,18

7



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 27

Even though the micelle drag-tags interact transiently with the end-alkylated DNA fragments, their elution time agrees well with eq (1) as expected for elution of DNA with covalently attached drag-tags. For regression purposes, eq (1) can be rewritten as: t=

lc ld  α  1+  µ0V  L 

(2)

where t is the elution time of a fragment containing L bases, lc is the capillary length, ld is the capillary length between injection and detection, and V is the applied voltage. Figure 3 is a plot of t versus 1/L for the data of Figure 2, showing good linearity for longer fragments but deviation for shorter ones in the case of the CiEj buffer. We believe the small deviation is due to steric segregation of the DNA from its drag-tag, which occurs when the DNA is small compared to the drag-tag so that the two entities are hydrodynamically distinct.26 Fits to the large L regions of the data give α = 509 for the CiEj buffer and α = 73 for the TX-100 buffer. The wormlike micelles of the CiEj buffer mixture give a values that are 6-7 fold higher than those previously used in the ELFSE analysis of Sanger products.11 The larger values of α are also consistent with the more pronounced deviation from linearity for the CiEj buffers in Figure 3, as the hydrodynamic segregation is expected to persist to longer lengths of DNA attached to larger drag-tags.

Performance of these separations is benchmarked by the resolution factor Rm, which compares the width at half maximum (FWHM) of electropherogram peaks to the temporal spacing between components differing by a single base in length (∂t/∂L): 12 Rm =

σ t 8ln ( 2)

(3)

∂t ∂L

Here, σt is the temporal peak variance and the absolute value of the peak spacing is taken since ∂t/∂N is negative in ELFSE runs. Rm < 1.0 has generally been set as a criterion for base calling purposes,11,12,27 but we find here that accurate base calling is achieved up to Rm = 1.5 ~ 1.75 as shown in Figure 4. We predict the theoretical plate height (H), and therefore the peak variance, by a van Deemter analysis:

H=

σ x2 ld

=

A 2D + + Wu ld u

D=

D1 L +α

8


(4)

Page 9 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


where u is analyte velocity, A is a constant related to variance due to finite injection width and detection window, and W is a constant related to the extent of wall adsorption. Following Ren et al., 4 the diffusion coefficient (D) for the DNA-tag complex is approximated by eq (4) with D1 = 4.43 x 10-6 cm2/s. Given that σx2 = σt2u2, the temporal variance is given by: σt =

A 2Dld Wld + 3 + u u2 u

(5)

and |∂t/∂L| calculated from eq (2). A was found to be negligible so fits of the Rm vs. N plots can be obtained using only W as a fitted parameter. Agreement is excellent, with the TX-100 data fit by W = 3.3 x 10-4 sec and the CiEj buffer data fit by W = 2.1 x 10-4 sec. To check for consistency, we also used the regressed value of W to predict the experimentally obtained H at various velocities u as dictated by DNA length (Figure 5). The agreement here is also very good, and the assumption that A is negligible is validated since the fits would not be improved by increasing H by a constant value.

Ren et al. have shown that, for covalently attached drag-tags, the van Deemter expression of eq (4) should modified to include a polydispersity term:4

H=

σ x2 ld

=

A 2D + + Wu + Bld ld u

(6)

where B is a constant parameter accounting for the effect of drag-tag polydispersity. Figure 5 shows that plate height data for the CiEj buffer are fit very well considering only diffusion and wall adsorption terms, indicating that drag-tag polydispersity does not significantly impact peak broadness. However, the theory of eq (6) should be modified for micelle-ELFSE as the drag induced by the end-attached micelle fluctuates during the migration of DNA fragments. Micelles composed of small-molecule surfactants undergo rapid fluctuations in size as surfactant from the solution phase exchanges with that in the micelle phase. 19 The alkylated DNA fragments will also exchange among distinct micelles, which may have different sizes. Both phenomena amount to a size-sampling process where migrating DNA interact with a range of micelle sizes during elution through the capillary, and the elution time established by the average of the sizes sampled. Since the pool of micelles available is the same for all DNA, DNA fragments of

9



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 27

the same size should have identical mobility provided a thorough sampling has been achieved. In other words, any differences in drag-tag size are temporary since the tags are not covalently bound to the DNA.

Mathematically, we describe the process using a standard-error approach where the observed polydispersity of the drag-tags (σα,obs) is set by the inherent polydispersity (σα) and the number of samples, n:

σ α ,obs =

σα

(7)

n

We posit that the size-sampling events have a constant frequency ν, so that n increases linearly with time. The contribution to plate height is given by:

H=

σ t2,obsu 2 ld

2 2 2 2 u 2  σ α  ∂t  σ α  1  =    =   u ld  ν t  ∂α  ν  L +α 

 1  ∂t = t  ∂α  L + α 

(8)

We define a fraction polydispersity for the drag-tags (fp = σα/α), to yield a modified van Deemter expression for micelle-ELFSE: 2  f p2  α   2D1 A H= = + + W +   u ld ld u L + α  ν  L + α  

σ x2

(9)

It is interesting to note the both the wall-adsorption term and the size-sampling term depend linearly on u. Band broadening due to wall adsorption occurs when some analytes bind for a longer time to the wall than others do; since binding is a stochastic process these differences average out with time.28 Binding to the micelle phase is a similar process except that the degree of broadening is a function of DNA length. This is due to the larger difference in mobility between a tagged DNA and free DNA for shorter fragments; indeed, the size-sampling term is length-invariant and indistinguishable from W for vanishingly small L.

To check for polydispersity-based band broadening, we replot the data of Figure 5 with a modified x-axis that should give a linear plot if micelle polydispersity were the dominant source of band broadening (Figure 6). Since the plot is completely random, especially 10


Page 11 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


compared with the plot of Figure 5 and its fit to eq 4 with a constant W, we conclude that the resolution of these micelle-ELFSE runs are limited by a combination of diffusion and wall adsorption effects, and that drag-tag polydispersity is not a significant contributor. We note that the values obtained for W (Figure 4) are very similar for both the TX-100 buffer and the CiEj buffer, despite the fact that wormlike micelles are generally much more polydisperse in size than spherical or ellipsoidal micelles.29 If polydispersity effects were a main contributor to peak broadness, we may expect a strong buffer dependence for the non-diffusive peak broadening terms.

The size-sampling term highlights the key attribute of the micelle-ELFSE method that gives access to high α without suffering polydispersity-based band broadening. Any inherent polydispersity (fp) can be mitigated by a sufficiently high frequency of sizesampling. Dynamic light scattering measurements on this CiEj buffer give broad size distributions with fp ≈ 0.25-0.35. The size sampling term must be less than 10% of W to be insignificant, which for L = α would have ν greater than about 10 msec-1. Micelle relaxation dynamics are characterized by a two time scales: a short one (τ1) related to the rate of monomer transfer to and from the micelle and a longer one (τ2) related to the complete dissolution of the micelle. (τ1) is believed to be on the order of microseconds30 and (τ2), which is more easily measured, has a value on the order of seconds for nonionic surfactants.19 It appears that the frequency of size sampling observed in micelle-ELFSE is related to the faster monomer exchange process, which if true, suggests that much faster runtimes could be achieved without suffering polydispersity-induced band broadening. Of course, this would require improved capillary coatings or other modifications to reduce the wall adsorption term.

Some heuristics have emerged during formulation of the CiEj buffer. It is certainly important to have a fairly high concentration of micelles to ensure that the DNA fragments are tagged for most of the run.14 This is crucial since any migration of untagged DNA fragments will yield no separation but will generate diffusional band broadening. Broadening will be even more pronounced in this case since the tag is not attached to the DNA to slow its diffusion. We also found that a moderate amount of urea

11



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 27

was needed to give sharp peaks. The beneficial role of urea is unclear, but it may cause micelle relaxation kinetics to increase in frequency or to reduce the incidence of DNADNA binding during the run. Addition of urea tends to decrease micelle size,31 so higher amounts of cylinder-forming surfactant were needed to compensate for the higher levels of urea. Added urea also increases viscosity, which reduces mobility and increases runtime. Use of 3 M urea appeared to best balance these effects, providing a viscosity of 2.7 cP at 32.5°C. The resolution was also very sensitive to temperature, with even 2°C changes giving rise to reduced read lengths. Generally, optimal performance was realized with operating temperature about 7-8°C below the cloud point of the buffer (40°C for the CiEj buffer used here). Cloud point and micelle size are strong functions of urea concentration so screening of several buffer compositions and temperatures was needed to yield these results. We also created buffers containing oil-in-water nanoemulsions,32 phospholipid vesicles, and phospholipid bicelles,33 but in each case the peaks were extremely broad and we believe this is due to polydispersity-based broadening due very slow size fluctuations for these larger microstructures.

We have ascribed the deviations from the ELFSE theory (eq (2)) in Figure 3 to hydrodynamic segregation of the DNA and drag tag owing to the small size of the DNA fragment relative to the drag-tag. Desruisseaux et al.26 predict the mobility of the DNAtag composite in the so-called “steric segregation” regime: µ0 2R ln ( L) −1= h µ b L

(10)

where Rh is the hydrodynamic radius of the drag-tag and b is the length of a DNA monomer (0.43 nm). Figure 7 is a plot of the data in Figure 3 that did not fit the ELFSE theory (L= 44~74 bases). The fit is very good, with a slope of 126 corresponding to Rh=27 nm. Assuming a cylindrical shape of the wormlike micelles, we estimate their length to be 54 nm, well within the range of lengths reported for C12E5 surfactant micelles in urea-containing buffers at various temperatures.31 As shown in Figure 8, we have repeated these runs using electric fields between 250~750 V/cm and found no change in the mobility for any length of DNA studied so we do not believe any fieldinduced segregation or stretching 34-36 occurs under these conditions.

12


Page 13 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


It is well known that wormlike micelles can entangle and form gels at high concentration, and these systems have even been used as sieving matrices to electrophoretically separate DNA.25 Since the order of elution of DNA is reversed for sieving compared to ELFSE, there is no sieving observed using this CiEj buffer for ssDNA in the range of 26-500 bases in length. Other buffers, with less of the end-capping surfactant C10E5, gave very broad peaks whose presence could be due to background sieving. The role of the endcapping surfactant is likely to maintain a high concentration of wormlike micelles without creating an entangled network that may give rise to sieving. For Rh = 27 nm (Figure 7), we estimate an aggregation number of 700 and an overlap concentration of 14 mM, assuming all surfactant is C12E5. Concentration-dependent dynamic light scattering data confirms that the overlap concentration is near 14 mM for this buffer (see Figure S1, Supporting Information). Since the CiEj buffer has 48 mM C12E5, there is likely some overlap but not enough to severely increase the viscosity (2.7 cP as judged by capillary viscometry) and the wormlike micelles are expected to be in the semidilute regime. We cannot rule out some degree of entanglement between the rods, but such entanglements may be short lived due to micelle relaxation processes. The CiEj buffer contains about 2 vol% surfactant, which is comparable to the concentrations used by Wei and Yeung25 in their study of wormlike, C16E6 micelles as a dsDNA sieving matrix. In their case, no end-capping surfactant was used so the micelles had aggregation numbers in excess of 10,000, greatly predisposing them to entangle. Chubynsky and Slater37 proposed a refinement to ELFSE theory that accounts for the reduced hydrodynamic interaction experienced between monomer segments near the ends of the migrating DNA and those in the interior of the chain. This leads to a non-linear dependence of migration time with inverse fragment length, as predicted by: t = t0

∫

1 0

(

x −1/4 1− x

)

−1/4

dx

(11)

−1

−1/4 1+α ∫ 0( L) x −1/4 (1− x ) dx

While eq (11) gave reasonable fits to the short-fragment data of Figure 3, it did not fit data for the longer fragments. Experimental data for the longer fragments were very

13



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

linear when plotted as time versus inverse length, while eq (11) had significant curvature in this region (see Figure S-2, Supporting Information). We note that we did not directly obtain values for t0 in this work since all DNA fragments were alkylated and therefore tagged by micelles. Instead, we allowed α and t0 to freely vary and still the fits were not better than the linear fit on Figure 3. We do believe that so-called “end effects” are at play in micelle-ELFSE. We have shown that attachment of micelles to both ends of DNA gives rise to a greater hydrodynamic drag than twice what is observed when using a single micelle drag-tag,15 as predicted by consideration of end effects.10 Use of a combination of alkylated and unalkylated fragments in micelle-ELFSE may clarify the role of end effects on elution times.

Conclusions

We have demonstrated that the use of transiently attached micelle drag-tags provide read lengths in excess of 500 bases in end-labeled free-solution electrophoresis. The micelleELFSE method gives access to large drag-tags (α = 509) without suffering polydispersity-based band broadening due to the rapid sampling of micelle sizes during the elution of DNA fragments. Resolution is limited by both diffusion and wall adsorption processes. An important, additional advantage of micelle-ELFSE is that the drag-tags are added to the running buffer following PCR. As such, a single PCR product can be analyzed using different drag-tags in different capillaries to greatly reduce runtime for long read lengths. We expect that it will be straightforward to formulate buffers that span a range of α between 70 and 500 so that buffers are designed to achieve a desired read length in an optimal runtime. While we have used Sanger products as stringent test case for the method, the long read lengths achieved also make possible the application of micelle-ELFSE as a high-speed, low-cost alternative to gels for STR analysis38, miRNA detection16, and routine fragment length analysis.

14


Page 14 of 27

Page 15 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Acknowledgements

This work was supported by the National Science Foundation (CBET-0932536) and the DSF Charitable Foundation.

Conflict of Interest Disclosure

The authors declare no competing financial interest.

Supporting Information Available

Additional information as noted in the text. This material is available free of charge at http://pubs.acs.org.

15



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 1. Overlay of four micelle-ELFSE electropherograms collected sequentially using a Beckman P/ACE-MDQ capillary electrophoresis instrument (40-cm capillary) with electrokinetic injection at 100 V/cm for 30 sec, separation at 750 V/cm, 32.5°C, in a buffer containing 48 mM C12E5, 6 mM C10E5, and 3 M urea in 1x TBE (“CiEj buffer”). Samples are C18-end-alkylated Sanger sequencing products using an M13mp18 template; T-terminal products are red, C-terminal are blue, A-terminal are green, and Gterminal are black. The expected fragment sequence is listed above the traces.

16


Page 16 of 27

Page 17 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 2. CE electropherograms for the T-terminated Sanger products of Figure 1 in the CiEj buffer (above) and 48 mM TX-100 (below). The upper trace was collected using a 40-cm capillary and the lower using a 30-cm capillary. Separation voltage was 30kV for each with the same injection conditions as Figure 1. Selected fragment lengths are labeled for comparison.

17



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 3. Plot of elution time vs. reciprocal DNA length for the data of Figure 2. Fits are to eq (2) using data for longer fragments to avoid the steric segregation regime. CiEj: α = 509, µ0 = 2.8 x 10-4 cm2/V-s; TX-100: α = 73, µ0 = 2.6 x 10-4 cm2/V-s.

18


Page 18 of 27

Page 19 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 4. Resolution factor Rm vs. fragment length (L) for the data of Figure 2. Fits are to eq (3) and (5) with A set to zero, D1 set to 4.43 x 10-6 cm2/sec according to ref4. α and µ0 values are taken from Figure 3; the only adjustable parameter is W. CiEj buffer: W = 2.1 x 10-4 sec; TX-100: W = 3.3 x 10-4 sec.

19



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 5. Plate height (H) versus velocity for the CiEj trace of Figure 2. The line is a plot of eq (4) using the parameters of Figure 4.

20


Page 20 of 27

Page 21 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6. Plot of plate height versus reduced velocity as predicted by the micelle polydispersity model of eq (9). Eq (9) predicts a linear dependence if micelle polydispersity is the dominant source of peak broadening.

21



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 7. Reduced mobility vs. ln(L)/L for the short-fragment data (44-74 bases, CiEj buffer) of Figure 3. Eq (10) predicts a linear dependence for steric segregation of drag-tag and DNA. Fit is linear with slope = 126 and y-intercept = -0.68, corresponding to a tag hydrodynamic radius (Rh) of 27 nm (b =0.43 nm).

22


Page 22 of 27

Page 23 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 8. Reduced mobility vs. reciprocal length for T-terminated sequencing products separated by micelle-ELFSE in a 40-cm capillary at V = 30 kV, 27.5 kV, 25 kV, 20 kV, 15 kV, and 10 kV. The excellent agreement among the data at these electric fields shows that field-induced segregation does not occur. Electrophoresis conditions given in Figure 2.

23



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

References Cited (1) Hoagland, D. A.; Arvanitidou, E.; Welch, C. Macromolecules 1999, 32, 6180-6190. (2) Noolandi, J. Electrophoresis 1992, 13, 394-395. (3) Mayer, P.; Slater, G. W.; Drouin, G. Anal. Chem. 1994, 66, 1777-1780. (4) Ren, H.; Karger, A. E.; Oaks, F.; Menchen, S.; Slater, G. W.; Drouin, G. Electrophoresis 1999, 20, 2501-2509. (5) Vreeland, W. N.; Desruisseaux, C.; Karger, A. E.; Drouin, G.; Slater, G. W.; Barron, A. E. Anal. Chem. 2001, 73, 1795-1803. (6) Vreeland, W. N.; Meagher, R. J.; Barron, A. E. Anal. Chem. 2002, 74, 4328-4333. (7) Vreeland, W. N.; Slater, G. W.; Barron, A. E. Bioconjugate Chem. 2002, 13, 663-670. (8) Haynes, R. D.; Meagher, R. J.; Won, J. I.; Bogdan, F. M.; Barron, A. E. Bioconjugate Chem. 2005, 16, 929-938. (9) Won, J. I.; Meagher, R. J.; Barron, A. E. Electrophoresis 2005, 26, 2138-2148. (10) Meagher, R. J.; McCormick, L. C.; Haynes, R. D.; Won, J. I.; Lin, J. S.; Slater, G. W.; Barron, A. E. Electrophoresis 2006, 27, 1702-1712. (11) Albrecht, J. C.; Lin, J. S.; Barron, A. E. Anal. Chem. 2011, 83, 509-515. (12) Meagher, R. J.; Won, J. I.; McCormick, L. C.; Nedelcu, S.; Bertrand, M. M.; Bertram, J. L.; Drouin, G.; Barron, A. E.; Slater, G. W. Electrophoresis 2005, 26, 331350. (13) Terabe, S. Anal. Chem. 2004, 76, 240a-246a. (14) Grosser, S. T.; Savard, J. M.; Schneider, J. W. Anal. Chem. 2008, 79, 9513-9519. (15) Savard, J. M.; Grosser, S. T.; Schneider, J. W. Electrophoresis 2008, 29, 2779-2789. (16) Goldman, J. M.; Zhang, L. A.; Manna, A.; Armitage, B. A.; Ly, D. H.; Schneider, J. W. Biomacromolecules 2013, 14, 2253-2261. (17) Fahrenkopf, M. A.; Mukherjee, T.; Ydstie, B. E.; Schneider, J. W. Electrophoresis 2014, 35, 3408-3414. (18) Fahrenkopf, M. A.; Ydstie, B. E.; Mukherjee, T.; Schneider, J. W. Comput. Chem. Eng. 2014, 64, 63-70. (19) Patist, A.; Kanicky, J. R.; Shukla, P. K.; Shah, D. O. J. Colloid Interface Sci. 2002, 245, 1-15. (20) Boutorine, A. S.; Grimm, G. N.; Helene, C. Mol. Biol. (Mosk.) 2000, 34, 804-813. (21) Albrecht, J. C.; Kotani, A.; Lin, J. S.; Soper, S. A.; Barron, A. E. Electrophoresis 2013, 34, 590-597. (22) Lin, J. S.; Albrecht, J. C.; Meagher, R. J.; Wang, X.; Barron, A. E. Biomacromolecules 2011, 12, 2275-2284. (23) Tanford, C.; Nozaki, Y.; Rohde, M. F. J. Phys. Chem. 1977, 81, 1555-1560. (24) Lin, Z.; Scriven, L. E.; Davis, H. T. Langmuir 1992, 8, 2200-2205. (25) Wei, W.; Yeung, E. S. Anal. Chem. 2001, 73, 1776-1783. (26) Desruisseaux, C.; Long, D.; Drouin, G.; Slater, G. W. Macromolecules 2001, 34, 4452. (27) Meagher, R. J.; Won, J. I.; Coyne, J. A.; Lin, J.; Barron, A. E. Anal. Chem. 2008, 80, 2842-2848. (28) Giddings, J. C. Dynamics of Chromatography; Marcel Dekker: New York, 1965. (29) Kato, T. In Surfactant Science Series, Vol. 112; Marcel Dekker:New York, 2003, pp 485-524.

24


Page 24 of 27

Page 25 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(30) Griffiths, I. M.; Breward, C. J. W.; Colegate, D. M.; Dellar, P. J.; Howell, P. D.; Bain, C. D. Soft Matter 2013, 9, 853-863. (31) Bianco, C. L.; Schneider, C. S.; Santonicola, M.; Lenhoff, A. M.; Kaler, E. W. Ind. Eng. Chem. Res. 2011, 50, 85-96. (32) Wang, L. J.; Mutch, K. J.; Eastoe, J.; Heenan, R. K.; Dong, J. F. Langmuir 2008, 24, 6092-6099. (33) Mills, J. O.; Holland, L. A. Electrophoresis 2004, 25, 1237-1242. (34) Lau, H. W.; Archer, L. A. Phys. Rev. E 2010, 81, 031918. (35) McCormick, L. C.; Slater, G. W. Electrophoresis 2007, 28, 3837-3844. (36) McCormick, L. C.; Slater, G. W. Electrophoresis 2007, 28, 674-682. (37) Chubynsky, M. V.; Slater, G. W. Electrophoresis 2014, 35, 596-604. (38) Butler, J. M.; Buel, E.; Crivellente, F.; McCord, B. R. Electrophoresis 2004, 25, 1397-1412.

25



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

FOR TABLE OF CONTENTS ONLY

26


Page 26 of 27

Page 27 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


332x237mm (72 x 72 DPI)


A 502-Base Free-Solution Electrophoretic DNA Sequencing Method

Recommend Documents