G. Vatcher, D. Smailus, M.Krzywinski, R. Guin, J. Stott,M. Tsai, S. Chan, P. Pandoh,G. Yang, J. Asano, T. Olson,A.-L. Prabhu, R. Coope1, A.Marziali1, J. Schein, S. Jones,and M. MarraGenome Sciences CentreBC Cancer Agency1University of British ColumbiaVancouver, BC, Canada

fRFLP and fAFLP:Medium-ThroughputGenotyping by Fluores-cently Post-LabelingRestriction Digestion

BioTechniques 33:539-546 (September 2002)

ABSTRACT

Genome-scale studies of populationstructure and high-resolution mapping ofgenetically complex traits both require tech-niques for accurately and efficiently geno-typing large numbers of polymorphic sitesin multiple individuals. Many high-through-put genotyping technologies require thepurchase of expensive equipment or con-sumables and are therefore out of reach ofsome individual research laboratories.Conversely, less expensive technologies areoften labor intensive so that the effort in-volved in typing large numbers of samplesor polymorphic sites is prohibitive. Here wepresent a method of fluorescently post-la-beling restriction digestion using standarddye-terminator sequencing chemistry sothat RFLP and AFLP products can be visu-alized on an automated sequencer. This la-beling method is efficient, inexpensive, easi-ly multiplexed, and requires no unusualequipment or reagents, thus striking a bal-ance between cost and throughput thatshould be appropriate for many researchgroups and core facilities.

INTRODUCTION

Restriction endonucleases have longbeen used in population studies of allelicvariation, where the presence or absenceof a cut site is indicative of a mutation inthe enzyme target sequence. In geneti-cally well-characterized organisms, di-agnostic RFLPs are often used as mark-ers for previously characterized alleles.In organisms with less well-studied ge-netics, the selective amplification ofanonymous restriction digestion prod-ucts (e.g., AFLPs) (8) has also been usedwith considerable success (5,9). Al-though they are extremely powerful inthe appropriate contexts, many existingRFLP and AFLP methods have limita-tions in throughput, sensitivity, or con-sumable cost that lessen their utility inmany individual research laboratories.Traditionally, RFLP and AFLP productshave been electrophoretically resolvedon either agarose or acrylamide slabgels, with the digested fragments visual-ized by ethidium bromide staining or ra-diolabeling. Existing fluorescence-basedRFLP and AFLP methods use labeledamplification primers (1,3,4,6) thatcould cost over $100 apiece (ResearchGenetics, Huntsville, AL, USA). Thisquickly increases the cost of an experi-ment using multiple primers and, in thecase of RFLP, only allows the visualiza-tion of a single terminal fragment, thuslosing all sequence information 3′ of thefirst restriction cut site. To increaseRFLP information content and allowhigher RFLP and AFLP throughput atlower costs, we have developed a simpleand inexpensive method for post-label-ing restriction digestion fragments using“off-the-shelf” sequencing chemistrybefore separation on an automated se-quencer. We call this labeling methodfluorescent RFLP (fRFLP).

Briefly, we amplify the target locusin a standard PCR and then digest theproduct with a restriction enzyme thatleaves a 5′ overhang. This overhangacts as a template for the single base in-corporation of a fluorescent dye-termi-nator nucleotide from a standard cyclesequencing kit. Thus labeled, the frag-ments are resolved on a capillary-basedsequencer. Multiplexing is facilitatedby the high resolution of the sequencerand its ability to read simultaneouslymultiple fluor wavelengths. This label-

Vol. 33, No. 3 (2002) BioTechniques 539

ing method can be applied in an AFLPcontext by using adapters and unla-beled primers containing a restrictionenzyme recognition sequence that canbe cut and fluorescently labeled afterthe final, selective, AFLP amplifica-tion. Fluorescent AFLP (fAFLP) prod-ucts can be multiplexed by simultane-ously using all of the fluor channelsdetected by the sequencer.

We have been working with theCEQ 2000-XL Capillary-Based Se-quencer and a CEQ Dye-TerminatorCycle Sequencing (DTCS) Kit (bothfrom Beckman Coulter, Fullerton, CA,USA), but our fRFLP and fAFLP label-ing methods should be feasible on anysequencing platform in which thereagents are packaged individually(i.e., not in a terminator dNTP-buffer-enzyme premixture). Table 1 illustratesthat the post-digestion labeling chem-istry costs approximately $0.45/reac-tion, and the electrophoretic separationcosts approximately $1.35. These costs,per site, decrease with multiplexing.

MATERIALS AND METHODS

fRFLP

Here we illustrate fRFLP by usingthe method to distinguish Attacin A alle-les generated by gene conversion in theAttacin AB cluster of Drosophila mela-nogaster (7). DNA from D. melano-gaster lines 2CPA 1, 2CPA 14, 2CPA105, and 2CPA 129 was amplified usingprimers Att98-1 and Att98-2 (7) understandard PCR conditions. PCR was car-ried out for 35 cycles with 45-s anneal-ing and extension steps at 58°C and72°C, respectively. Following amplifica-tion, 15 µL each PCR product were in-cubated with 7.5 U exonuclease I and 3U shrimp alkaline phosphatase (USB,Cleveland, OH, USA) for 1 h at 37°C toinactivate residual primer and unincor-porated nucleotides. The enzymes werethen inactivated for 15 min at 80°C.

We digested the PCR products withthe restriction enzymes MspI, MseI, andDpnII (New England Biolabs, Beverly,MA, USA) because diagnostic poly-morphisms for the Attacin AB gene con-version events disrupt cut sites recog-nized by these enzymes and because allthree enzymes cut efficiently in PCR

buffer, which obviates the need for abuffer change at this step. Five micro-liters of each PCR product were incu-bated overnight at 37°C with 5 U eachenzyme in single digestions. For sim-plicity, only one amplification productwas digested in each of these reactions,but we have had excellent success mul-tiplexing by pooling distinct PCRs at thedigestion stage (unpublished results).

The restriction digestion reactionswere labeled using components of theDCTS kit, with extra 10× sequencingbuffer ordered directly from the manu-facturer. The molar concentrations ofthe materials in this kit are proprietary,so we can only provide volume mea-sures of reagents, but we have foundthis labeling reaction to be robust overa wide range of template, terminator,and DNA polymerase concentrations.For the present demonstration, each la-beling reaction included 1 µL digestedDNA, 1.9 µL 10× Sequencing Buffer,and 33 nL sequencing DNA poly-merase. Only the fluorescent terminatornucleotide specific to each digestionwas included in the labeling reactions(ddGTP for DpnII, ddCTP for MspI,and ddUTP for MseI). Thirty-threenanoliters of fluorescent terminator nu-cleotide were used in the ddCTP andddUTP labeling reactions, but 100 nL

were required for adequate signal withddGTP. We added 17 µL distilled waterto make 20 µL total volume, and the re-actions were incubated for 1 h at 60°C.

After the incubation, we pooled thethree labeled digestions from each PCRproduct and added 15 µL stop solution(1.2 M sodium acetate, 40 mM EDTA, 4µg/µL glycogen). The fRFLP reactionswere precipitated by the addition of 200µL ice-cold 95% ethanol to each reac-tion and were then centrifuged for 20min at 1800× g. The supernatant was as-pirated off, and the pellets were washedthree times with 200 µL ice-cold 70%ethanol and then allowed to air-dry. Thedry pellets were resuspended in 40 µLdeionized formamide with 0.25 µLCEQ DNA Size Standard-600 (Beck-man Coulter) and overlayed with onedrop of mineral oil. The fragments werethen separated on the CEQ 2000-XLcapillary-based sequencer using the de-fault “Frag-4” method and analyzedwith the fragment analysis componentof the CEQ 2000-XL software, version4.3.9 (Beckman Coulter).

fAFLP

We performed AFLP on DNA ex-tracted from hydrothermal vent tube-worms, Ridgeia piscesae (Polychaeta:

Short Technical Reports

540 BioTechniques Vol. 33, No. 3 (2002)

Per-Site Cost for Multiplexed Samples

Reagent Stage Simplex 3-folda 6-folda 9-folda

Exonuclease I sample 0.060 0.060 0.030 0.020

preparation

Shrimp alkaline sample 0.111 0.111 0.056 0.037

phosphatase preparation

DTCS label labeling 0.163 0.163 0.082 0.054

10× Seq. Buffer labeling 0.116 0.116 0.058 0.039

Separation Buffer electrophoresis 0.490 0.163 0.082 0.054

Separation Gel electrophoresis 0.550 0.183 0.092 0.061

Size Standard-600 electrophoresis 0.309 0.103 0.052 0.034

TOTAL (US dollars) 1.799 0.900 0.450 0.300

aMultiplex assumes all three fluor channels are used with either one, two, or threeamplification products in each channel. It is assumed that digested reactions arepooled for a single labeling reaction when more than one product is representedin each channel and that each product queries only a single polymorphic site.Estimated per-site costs vary slightly with other multiplex structures.

Table 1. Post-Digestion Cost of fRFLP and fAFLP

Siboglinidae), collected at the Endeav-or Segment of the Juan de Fuca Ridgein the northeast Pacific Ocean. Genom-ic DNA was extracted using a standardphenol:chloroform technique (2), andAFLP was carried out using AFLP®Analysis System I and AFLP StarterPrimer kits (Invitrogen, Carlsbad, CA,USA), which follow the protocol of Voset al. (8). DNA (250 ng) from each in-dividual served as the starting template.For fAFLP, the final amplification wasperformed with a selective extension of-AGG on the EcoRI adapter primer and-CTC on the MseI adapter primer. Inthis example, we make use of the factthat the MseI adapter and primer con-tain a recognition site for the four-cut-ting restriction enzyme DdeI.

After the selective amplification, 5µL each AFLP amplification were incu-bated with 2.5 U exonuclease I and 1 Ushrimp alkaline phosphatase for 1 h at37°C, followed by 15 min at 80°C. Thereactions were digested overnight at37°C with 2.5 U DdeI (New EnglandBiolabs), and 1 µL of the digestion waslabeled with ddUTP as described earli-er. Following labeling, 5 µL stop solu-tion and 60 µL ice-cold 95% ethanolwere added to each of the fAFLP reac-tions, which were then centrifuged for20 min at 1800× g. The pellets werewashed twice with 200 µL ice-cold70% ethanol and air-dried. As in fRFLP,the dry pellets were resuspended in 40

µL deionized formamide containing0.25 µL Size Standard-600 and over-layed with a drop of mineral oil. ThefAFLP fragments were then resolved onthe CEQ 2000-XL automated sequencerusing the Frag-4 separation method.

Post-AFLP digestion with DdeI cutboth in the adapter/primer sequencesand at internal sites in the amplifiedproducts. All of these cut sites weresubsequently labeled. In addition to

supplementing genetic informationabout the samples, the presence of inter-nal cut sites in larger products can bringthe peaks down into the effective resolu-tion range of the capillary sequencer(60–600 bp on default settings). How-ever, in some cases, it may be desirableto engineer adapters and primers carry-ing recognition sites for rare-cutting en-zymes to minimize the digestion ofAFLP amplification products.

Short Technical Reports

542 BioTechniques Vol. 33, No. 3 (2002)

Figure 2. A zoomed-in view of chromatograms showing fRFLP peaks between 90 and 220 bp in lines 2CPA 1 and 2CPA 14. The high resolution of the au-tomated sequencer allows easy identification of a 9-bp insertion/deletion polymorphism [112 vs. 121 bp in the red (T-labeled) channel; 210 vs. 219 in the blue(C-labeled) channel, and 152 vs. 161 in the black (G-labeled) channel]. Note the clear distinction between 191- and 193-base fragments in the blue channel of2CPA 14. Green peaks are the CEQ DNA Size Standard-600 internal reference size standard.

Figure 1. Digestion of the Attacin A gene of D. melanogaster with only three restriction enzymes dis-tinguishes six polymorphisms among four alleles. (A) Schematic representation of the pattern of re-striction sites in the four alleles. The numbers along the top row represent the positions of the cut sitesalong the 751-bp product. (B) Expected fragment sizes in the four lines, following digestion with each ofthe three enzymes. Fragments smaller than 60 bp are not detected, but all of the other peaks were accu-rately detected and sized within 2 bp.

RESULTS AND DISCUSSION

Using only three restriction en-zymes and a single electrophoretic runof fRFLP, we typed six polymor-phisms, including a 9-bp insertion/dele-tion polymorphism, in the Attacin Agene from four lines of D. melanogas-ter (Figures 1 and 2) at a post-digestioncost of about $0.45/site (Table 1). Wealso used fAFLP to generate a genomicfingerprint for the hydrothermal venttubeworm R. piscesae (Figure 3) at a la-beling cost of $0.45, without needing tobuy a fluorescently labeled primer. Thetotal per-site cost of fRFLP varies de-pending on the price of the diagnosticrestriction enzymes and can be reducedby higher degrees of multiplexing. Wehave had success in multiplexing bypooling independent reactions beforerestriction digestion, pooling multipledigests before fluorescent labeling(data not shown), and simultaneously

Figure 3. Chromatogram showing fAFLP peaks obtained by using DdeI to digest an unlabeled R.piscesae AFLP reaction and then incorporating a fluorescent ddUTP terminator nucleotide into thecut site (red peaks). Green peaks are the CEQ DNA Size Standard-600 internal reference size standard.

electrophoresing multiple independent-ly labeled products.

Because automated sequencers candistinguish fragments that differ in sizeby a single nucleotide (Figure 2) and candetect hundreds of fragments in a singleelectrophoretic run, the only practicallimit to the degree of fRFLP multiplex-ing that can be employed seems to be inthe informatics of analyzing the output.The CEQ Fragment Analysis softwaredoes not have a pattern-recognitioncomponent that would allow for thetranslation of highly multiplexed fRFLPpeak patterns into component multi-lo-cus genotypes, but the software does ex-port a tab-delimited text file that lists,among other attributes, the estimatedsize in base pairs, and under-peak areafor every peak in each fluor channel. Itshould be straightforward to program anindependent script that analyzes theCEQ text output for expected fRFLPgenotype patterns. Without the use ofautomated genotype calling, we havefound 6-fold multiplexing, with two in-dependent PCR products digested andlabeled in each the C, G, and T channels(3–10 peaks in each channel), to provide

an adequate compromise among cost perassay, throughput, and simplicity ofanalysis. For fAFLP analysis, a macrothat generates a presence/absence tablefor all peaks in each fluor channel in agiven sample set is available from Beck-man Coulter technical support. Thismacro, along with other improvementsthat ease data analysis, is incorporatedinto the next generation CEQ 8000 soft-ware for more seamless fAFLP analysis.

The ability of automated sequencersto resolve extremely small size differ-ences accurately can also be exploited inthe examination of microsatellite re-peats. Although we have not tested thepossibility, we believe that the post-la-beling procedure used in fRFLP andfAFLP could also be used in microsatel-lite analysis if rare-cutting restriction en-zyme recognition sites are incorporatedinto the amplification primers. Non-overlapping repeat size ranges and theuse of multiple fluors could allow a highdegree of multiplexing in microsatelliteanalysis. Our labeling method couldconceivably even be applied to the inex-pensive generation of the internal sizestandard required in each well for accu-rate sample peak sizing.

Overall, both fRFLP and fAFLPprovide a simple, rapid, and inexpen-sive means of genotyping moderatenumbers of individuals at a large num-ber of loci with no need to order allele-specific or fluorescently labeled oligo-nucleotides. It is not even necessary tohave prior knowledge of the polymor-phisms that will be typed because bothmethods can be used to query anony-mous or uncharacterized sequences.The cost, ease, and flexibility of thesemethods make them appropriate formany medium-throughput applicationson a scale likely to be employed by in-dividual research laboratories.

REFERENCES

1.Arnold, C., L. Metherell, J.P. Clewley, and J.Stanley. 1999. Predictive modeling of fluores-cent AFLP: a new approach to the molecularepidemiology of E. coli. Res. Microbiol.150:33-44.

2.Ausubel, F.M., R. Brent, R.E. Kingston, D.D.Moore, J.G. Seidman, J.A. Smith, and K.Struhl. 1989. Current Protocols in MolecularBiology, Vol. 1. Greene Publishing Associatesand Wiley-Interscience, New York.

3.Bruce, K.D. and M.R. Hughes. 2000. Termi-

nal restriction fragment length polymorphismmonitoring of genes amplified directly frombacterial communities in soils and sediments.Mol. Biotechnol. 16:261-269.

4.Desai, M., A. Tanna, R. Wall, A. Efstratiou,R. George, and J. Stanley. 1998. Fluorescentamplified-fragment length polymorphismanalysis of an outbreak of group A streptococ-cal invasive disease. J. Clin. Microbiol.36:3133-3137.

5.Fishman, L., A.J. Kelly, E. Morgan, and J.H.Willis. 2001. A genetic map in the Mimulusguttatus species complex reveals transmissionratio distortion due to heterospecific interac-tions. Genetics 159:1701-1716.

6.Koeleman, J.G., J. Stoof, D.J. Biesmans, P.H.Savelkoul, and C.M. Vandenbroucke-Grauls. 1998. Comparison of amplified riboso-mal DNA restriction analysis, random ampli-fied polymorphic DNA analysis, and amplifiedfragment length polymorphism fingerprintingfor identification of Acinetobacter genomicspecies and typing of Acinetobacter bauman-nii. J. Clin. Microbiol. 36:2522-2529.

7.Lazzaro, B.P. and A.G. Clark. 2001. Evidencefor recurrent paralogous gene conversion andexceptional allelic divergence in the Attacingenes of Drosophila melanogaster. Genetics159:659-671.

8.Vos, P., R. Hogers, M. Bleeker, M. Reijans, T.van de Lee, M. Hornes, A. Frijters, J. Pot, etal. 1995. AFLP: a new technique for DNA fin-gerprinting. Nucleic Acids Res. 23:4407-4414.

9.Voss, S.R., J.J. Smith, D.M. Gardiner, andD.M. Parichy. 2001. Conserved vertebratechromosome segments in the large salamandergenome. Genetics 158:735-746.

We thank Cindy Jarvis and John Black atBeckman Coulter for their support of thiswork. We also thank Kim Juniper, the chiefscientist of the May 2001 cruise to the Juande Fuca Ridge, the crews and pilots of theCCGV John P. Tully, and the ROV ROPOSfor allowing tubeworm collection. This workis supported by the National Institutes ofHealth grant no. AI46402 to A.G.C. Tube-worm collections were supported by the Na-tional Underseas Research Program grantno. UAF01-0042 to Stephen Schaeffer,Charles Fisher, and Stéphane Hourdez. Ad-dress correspondence to . Brian P. Lazzaro,Department of Molecular Biology and Ge-netics, 107 Biotechnology Building, CornellUniversity, Ithaca, NY 14853, USA. e-mail:bplazzaro@psu.edu

Received 13 March 2002; accepted 22May 2002.

Brian P. Lazzaro, Bonnielin K.Sceurman, Susan L. Carney,and Andrew G. ClarkPennsylvania State UniversityUniversity Park, PA, USA

Short Technical Reports

Vol. 33, No. 3 (2002)Circle Reader Service No. 177