Thesis subject

DNA polymerases at work: single-molecule observations of DNA synthesis in real time

PhD thesis Carel Fijen, 21 March, 2018

This thesis focuses on the characterization of DNA polymerases with single-molecule
techniques. More specifically, I aimed to study polymerase processivity and fidelityrelated
conformational changes using assays based on Förster Resonance Energy
Transfer (FRET) on a total internal reflection fluorescence (TIRF) microscope.

Chapter 2 reviews some of the recent applications of single-molecule FRET (smFRET)
to study DNA and DNA binding proteins, in particular DNA polymerases. The
chapter begins with an introduction of FRET, employed to measure distance changes
in the 1-10 nm region, and introduces the two most common fluorescence-based
implementations of single-molecule techniques: confocal microscopy and TIRF
microscopy. The chapter concludes with a short discussion on FRET-based structural
modelling, parts of which are applied in practice later in this thesis.

In chapter 3, I report the development of a short, fluorescently labelled DNA sensor
to probe DNA polymerization at the single-molecule level. The sensor is a simple
primer-template combination labelled with donor and acceptor fluorophores suitable
for FRET. The advantage of this assay is that polymerases do not need to be labelled
with any fluorophore. I show that the FRET efficiency of the sensors changes
significantly upon polymerization of the 25 nucleotide template, and I present time
traces showing polymerization of single sensors by three different polymerases (E. coli
DNA Polymerase I (KF), human Polymerase Beta (POLB) a nd t he α s ubunit o f
bacterial Polymerase III (POLIIIα)). Based on these traces, I can measure polymerase
speed and pausing: KF and POLIIIα extended the primer in ~1.0-1.5 s, but POLB was
far slower (tens of seconds). I foresee applications for these sensors in the singlemolecule
field, where they can be used to characterize the processivity of other
polymerases, but also for ensemble experiments in which native polymerases need to
be tested for activity.

I take a closer look at POLB in chapter 4. This polymerase is involved in DNA repair,
and I a ddress t he q uestion w hether r esolving t he c onformational dynamics of the
enzyme can shed new light on fidelity-related mechanisms. Previous work on both
KF and POLB showed that the polymerase “fingers” domain binds a nucleotide and
subsequently transfers it to the active site (a conformational change known as “fingers
closing”). For KF, it was shown that the fingers domain does not entirely close when
non-complementary nucleotides are present, suggesting that nucleotides are screened
for complementarity with the templating base during fingers closing. To see whether
POLB employs a similar mechanism, I designed an smFRET assay with an immobile
donor fluorophore on the DNA primer and an acceptor fluorophore on the fingers
domain. Using this approach, I can visualize fingers c losing i n t he p resence o f t he
correct nucleotide in single POLB-DNA complexes. Incorrect nucleotides (noncomplementary
dGTPs and complementary rUTPs) did not induce fingers closing.
Instead, w e o bserved a s light s hift i n t he m ean F RET efficiency of the open
conformation (from E* ≈ 0.55 to E* ≈ 0.62), while a fully closed conformation
corresponds to E* ≈ 0.75. I find evidence for a partially closed, fidelity-related
conformation of the fingers subdomain. Simultaneously, I find that high
concentrations of incorrect nucleotides (1 mM and 3 mM) stabilize the POLB-DNA
complex by lowering the POLB dissociation rate. In contrast, for KF, a destabilizing
effect was shown previously. The mechanism behind this stabilization remains
unknown, but I hypothesize that with the abundance of incorrect nucleotides in the
cell, DNA repair is much faster if high levels of these nucleotides do not promote

In chapter 5, I introduce novel nanofluidic devices for high-throughput singlemolecule
imaging. These devices are completely made of glass. I present two designs:
one design with a parallel array of nanochannels for equilibrium studies, and another
with a single, T-shaped nanochannel for mixing studies allowing access to nonequilibrium
conditions. A channel height of 200 nm confines movement of the
molecules such that they do not move out of focus. With the implementation of
parallel flow control, the devices can be driven with conventional syringe pumps. I
achieve a high temporal resolution on our emCCD camera due to stroboscopic
excitation (1.5 ms excitation in 10 ms frame time). I s how t hat w e can t rack s ingle
molecules at low concentrations for extended periods of time. The track length
depends on the flow speed, but ranges from several frames t o t ens o f f rames.
Moreover, at higher concentrations, I achieve hundreds of thousands of localizations
within 10 minutes. These localizations allowed me to construct flow profiles, which
confirms that the flow in the nanochannels is laminar. I also calculate that, at low flow
rates and with the small DNA molecules I used, motion due to flow is of the same
order of magnitude as motion due to diffusion. I illustrate this concept by mixing DNA
hairpins in a primarily open configuration with a high-salt solution in the mixing
channel: the FRET signature of the hairpins changes abruptly towards an equilibrium
of primarily closed DNA hairpins. After fine-tuning the conditions, this so-called
“diffusive” mixing is employed to trigger single-molecule reactions: I successfully
polymerize my previously described DNA sensor inside the channel. I believe that
these nanofluidic devices are a promising platform for studying non-immobilized
single molecules at high throughput and high temporal resolution.