Bacterial Strain Analysis of RFLP using DNA Fragment Sizing Flow Cytometry Cheryl L' Lemanski, Xiaom

1
Bacterial Strain Analysis of RFLP using DNA
Fragment Sizing Flow Cytometry Cheryl L.
Lemanski, Xiaomei Yan, Robert C. Habbersett,
Thomas M. Yoshida, James H. Jett, Babetta L.
Marrone Biosciences Division, MS M888, Los
Alamos National Laboratory, Los Alamos, NM 87545
I-086
Bioscience
Introduction Restriction fragment length
polymorphism (RFLP) followed by Pulsed-field gel
electrophoresis (PFGE) has been the gold standard
for bacterial strain identification and
epidemiological analysis in public health
settings to date. Although the specificity and
reliability of RFLP/PFGE has made it a very
powerful tool, PFGE has several deficiencies it
requires 20 hours, micrograms of DNA, plus it is
highly non-linear, and lacks direct quantitation.
We have developed an alternative technique,
which is 100-fold faster and 10,000 fold more
sensitive than the PFGE method. In our approach,
the DNA fragments, produced by digestion of the
bacterial genome with a rare-cutting restriction
endonuclease, are extracted from the agarose plug
by GELase treatment, stained with a DNA
intercalating dye (e.g. SYTOX-Orange or
PicoGreen), and analyzed on a laboratory-built
high sensitivity flow cytometer.
Fragment size comparison among Virtual Digest,
PFGE, and HSFCM of Staphylococcus aureus Mu50
(sequenced strain)
Cell Culture
2 min 10 min 120 min 60 min 20 min 40 min 120
min 60 min 2x10 min lt8 hrs
Wash
Form Plug
PMSF Inactivation
Lysis
Proteinase K
Wash
Restriction Enzyme Digestion
RFLP GELase Staining Instrument
Reproducibility Study
GELase
Flow Cytometry
Rapid DNA preparation for RFLP-based bacterial
fingerprinting
Large DNA fragments recovered intact upon GELase
procedure
Stained bacterial/RFLP digest running through the
high sensitivity flow cytometer individually,
generating a fluorescence burst that is
proportional to the size of the fragment. A
histogram of the fragment sizes is generated and
this "fingerprint" is comparable to the PFGE
fingerprint used for bacterial strain
identification.
The automated peak finding routine performs
several operations on calibrated histograms
sharp spikes in the histogram are eliminated by
smoothing using the first derivative of the
histogram, the top of each peak is found and
reported as the fragment size. The smoothing and
derivative steps are performed by the
Savitsky-Golay kernal convolution mechanism.
Overall the routine does very well at quickly and
objectively finding the crests of well defined
peaks, but it tends to assign some false peaks
where the histogram is noisy.
Laboratory-built compact high sensitivity flow
cytometer (HSFCM)

• Conclusions
• Our unique HSFCM offers more advantages over
  traditional PEGE for bacterial RFLP
  fingerprinting at the strain level. Various
  bacterial genus, species, and strains have been
  successfully analyzed by DNA fragment sizing flow
  cytometry.
• Linear relationship between DNA fragment length
  and burst fluorescence,
• Accurate measurement of DNA fragment length from
  125 bp to 400 kbp,
• Fast analysis speed for fingerprint generation
  (10 min versus 20 hours),
• High sensitivity (10 pg DNA versus ug),
• Better resolution,
• Excellent reproducibility.

Burst size distribution histogram of 1 kbp DNA
stepladder
Accurate DNA fragment sizing by internal
standardization
Our calibration procedure uses internal staining
standards added directly to a sample of unknown
fragments. The standards, which must be
recognizable against the peaks from unknown
fragments in the sample, serve as signposts for
linear regression analysis of the observed peak
positions (of the standards) against their known
sizes. Peak positions are objectively determined
by fitting a sum of Gaussians and a background
curve to the peaks from the standards. The
derived slope and intercept are applied to the
sample with staining standards, which calibrates
the X-axis scale of the histogram. In the second
step, the histogram from the corresponding sample
- without standards - is then superimposed (on
the calibrated histogram), and the slope and
intercept are adjusted to give the best alignment
between the two histograms. The best matching
slope and intercept values are then applied to
the second file, which completes the calibration
process.
Acknowledgements This work was supported by the
Chemical and Biological Nonproliferation Program
(NN-20) of the Department of Energy, by the
NIH-funded National Flow Cytometry Resource
(PR-01315), and by the FBI Hazardous Materials
Response Unit, for which we are most grateful.
The statements and conclusions herein are those
of the authors and do not necessarily represent
the views of the FBI. Los Alamos Unclassified
Report 02-7731.