14

Advances in DNA diagnostics Joel H Graber*t, Maryanne J O'Donnell , Cassandra L Smith*§ and Charles R Cantor *#

The key advances in DNA diagnostics during the past year are techniques which will lead to advanced throughput without sacrificing sensitivity: miniaturization of samples to reduce material cost and preparation time, and parallelization through use of measurement arrays. The most promising gains have come in the areas of DNA arrays and mass spectrometry, where differential sequencing measurements are now possible.

Addresses *Center for Advanced Biotechnology, Boston University, 36 Cummington St, Boston, MA 02215, USA re-mail: jhg@darwin.bu.edu §e-mail: cls@darwin.bu.edu #e-mail: crc@enga.bu.edu :~Sequenom Inc., 11555 Sorrento Valley Rd, San Diego, CA 92121, USA; e-mail: maryanne@sequenom-san.com

Current Opinion in Biotechnology 1998, 9:14-16

http://biomednet.com/elecref/0958166900900014

© Current Biology Ltd ISSN 0958-1669

Abbreviations DE delayed extraction MALDI matrix-assisted laser desorption/ionization MR mass resolving power MS mass spectroscopy PROBE primer oligo base extension SAGE serial analysis of gene expression TOF time of flight

I n t r o d u c t i o n The availability of DNA sequence information, from both humans and other organisms, is increasing worldwide at an ever quickening pace. One of the great challenges of modern molecular biology is the integration of the new genetic information into diagnostic procedures, which can be implemented in rapid, cost-effective methods for identification and treatment of disease-related phe- notypes. There is a fundamental question of nearly all of the work presented in this review: "given a known genetic sequence, how can an unknown sample (target) be rapidly, sensitively, and robustly assayed against the known sequence (probe)?"

The largest effort in DNA diagnostic development is being directed towards differential sequencing, the quest to identify mutations or polymorphisms in a known target sequence where changes are possibly as small as a single base substitution. These measurements are further complicated by the high probability of such mutations occurring in heterozygous samples.

In this review we discuss recent advances in DNA diagnostics, with a particular emphasis on those techniques that we feel will lend themselves to the development of cost-effective, robust techniques for clinical use.

D N A c h i p s : m i n i a t u r i z a t i o n a n d p a r a l l e l i z a t i o n Possibly the most mature modern DNA diagnostic tool is the so-called ' D N A chip', a small device that holds a regular array of DNA molecules. The DNA molecules can be synthesized in place on the array, or synthesized separately and then chemically attached to the surface. The DNA molecules that are attached to the surface can be the known probe sequence, in which case the targets are hybridized to the probes, or alternatively the unknown target sequence, which is assayed by known probes. Th e fundamental principle of most DNA-chip measurements is the highly selective nature of DNA dot, ble helix hybridization. Perfectly complementary sequences can be made to hybridize with a much greater efficiency than imperfectly matched sequences. Measurements are most often made via fluorescence through tags incorporated into either the probe or target molecules.

At Affymetrix, for example, high density DNA arrays are created in place on the chip, utilizing photo-lithographic oligonucleotide synthesis. These arrays are specifically designed for use with a known genetic sample, which, in principle, allows a far smaller number of probes to be used than would be required for a general use array, in which all possible n-mers are created. In order to assay the entire 16.6 kilobase human mitochondrial genome [1"], chips were created with up to 135,000 different 16 base long hybridization probes in a single arra'~: In the 4L probe configuration, hybridization probes are created for each sequential 16-met in the mitochondrial genome. The ninth base is varied among all possible bases, creating four possible probes for hybridization at each position through the genome. These four probes are arranged on the array in a simple geometric pattern, allowing a simple and highly automatable interpretation of sequence. Use of a two-color fluorescence scheme allowed hybridization of an unknown sample to be performed simultaneously with a control sequence.

Each base in the mitochondrial sequence occurs in 16 diffcrent sets of probes; therefore, a substitution mutation will appear not only as an altered base at its exact read position, but will also result in a footprint of reduced or eliminated hybridization, in which the hybridization efficiency is affected in all probes which contain the mutated base. Compound mutations appear

Advances in DNA diagnostics Graber et aL 15

as broadened or separate footprints. Using chips of this type, 179 of 180 polymorphisms tested were successfully identified. Samples of the entire mitochondrial genome were analyzed approximately 25 times faster than with conventional electrophoretic analysis, with 99% accuracy in the direct readout of bases in a test sequence.

The chief advantages of this technology are the speed of measurement, and the ease with which analysis lends itself to automation. One disadvantage of this technique is the large amount of sample material which must be created in order to provide hybridization targets to all probes in the array. This method is also susceptible to any sequence- specific features that affect hybridization efficiency, such as melting temperature variation, and secondary or tertiary structure effects.

An ahernative approach to the use of DNA arrays is the creation of miniature gel pads [2,3",4,5]. The gel pads arc physically separated by the surface of the glass slide, and the glass surface between the pads is coated to be hydrophobic in order to eliminate, or greatly reduce, any interaction between adjacent pads. Probe or target sequences are attached at the array site by crosslinking the molecule into the gel material. The individual pads usually contain a single probe type. The three dimensional nature of the gel pads provides for hundreds times larger amounts of attached material, making the array sites more like 'miniature test tubes', than surface sites. Recent results [3"] show the ability to perform procedures such as ligation or PCR in situ in the gel pads (see [6,7] for further recent results on chip arrays).

Mass spectrometry: high sensitivity and high throughput The use of mass spectroscopy (MS) as a tool for nucleic acid analysis was only made possible through the de- velopment of 'gentle' ionization techniques, electrospray ionization and matrix assisted laser desorption/ionization (MALDI). In MALDI, nucleic acid fragments are sus- pended in a crystalline matrix which has an absorption maximum near the wavelength of an incident laser pulse. The most common form of MALDI device in use is the time of flight (TOF) configuration, in which the nucleic acids are desorbed and ionized by the incident laser pulse, and then subjected to an intense electric field which accelerates the fragments proportionally to their mass to charge ratio. The time required to reach the detector is measured; it is proportional to the square root of the charge to mass ratio. The primary advantages of MALDI-TOF MS are the speed of measurement (<1 sec), the lack of need for special labels for detection, and much greater accuracy, resolution, and sensitivity than the equivalent electrophoretic techniques.

The primary limitation, to date, in MALDI-TOF MS has been the size of the samples which can be sensitively measured. This is most severe for DNA fragments, where

the inherent fragility of the DNA molecule has limited most measurements to 50-100 bases, with a mass resolving power (MR) of 1 part in 1,000 or less. The potential requirement of uniquely identifying the smallest simple mutation, for example, A ~ T transversion (change in mass=9 daltons), limits fragments to a mass of 9,000 daltons, or about 30 nucleotides.

The existing capabilities of MALDI-TOF MS are easily adequate for basic differential sequencing schemes, and several groups are developing such schemes [8",9,10",11, 12,13"'] based on this technique. The best results to date in the use of MALDI-TOF MS for differential sequencing is the primer oligo base extension (PROBE TM) technique developed at Sequenom, Inc. [8",9,10"]. In PROBE TM, a primer is annealed to the solid-phase immobilized target template immediately upstream from the region to be investigated for mutations or polymorphisms. Polymerase extension is then carried out in the presence of a selected mixture of deoxyribonucleoside triphosphates and dideoxyribonucleoside triphosphates, which terminate extension. The exact composition of the mixture of normal and dideoxy bases is determined by the local sequence and potential mutations. Knowledge of the wildtype sequence allows exact prediction of the expected length and mass of the resulting oligomer; mutations or polymorphism are nearly always uniquely characterized by changes in the mass. With PROBE rl'M, mutations can be detected up to several bases away from the 3' end of the primer, in contrast with several of the other techniques for MALDI-TOF MS differential sequencing, which are capable of assaying only a single base change.

PROBE TM was used to investigate various mutations, including those of the cystic fibrosis transmembrane conductance regulator [8°]. In addition, PROBE TM was used as a method to determine the size of microsatellite repeat regions [10"]. The instability of microsatellite repeat lengths has been definitively associated with several diseases, for example Huntington's Disease. Microsatellite sizes are also used in linkage analysis.

Several recent results point the way to further im- provements of the MALDI technique, specifically with regards to issues of improved throughput and sensitivity. A very significant result is the combination of array technology with MALDI analysis [13"']. The surface preparation of the array includes the creation of small etched pits with approximately nano-liter volume at each array site. The surface pit concentrates the location of crystalline material into an area of equal or smaller size than then incident laser spot size. This configuration provides two enormous advantages over previous MALDI investigations: firstl3; detection without the search for the 'sweet spot' (a region of crystalline matrix material which provides strong analyte signal when sampled by the laser); and secondl3; detection of quantities two to three orders of magnitude smaller. Samples are dispensed with a

16 Analytical bioteehnology

piezo-electric controlled capillary pipette. Reproducibility of the samples is greatly improved over manual spotting with conventional pipettes. The implications of these results are enormous with regards to automation of MALDI. The ability to obtain a sensitive, reproducible measurement from each array element without searching makes automated, rapid measurement of the entire array possible. Initial measurements have been made on a 10x 10 array of synthetic oligonucleotides, as well as on PROBE TM products, for both mutation detection and microsatellite characterization. The MALDI results are accurate and reproducible, allowing accurate and rapid genotyping of samples.

Improvement of the sensitivity of MALDI measurements will lead to the ability to further increase the size of the fragments, while maintaining the ability to detect the smallest mutations. A significant gain in MALDI- TOF sensitivity has been obtained through the use of delayed extraction (DE) MALDI-TOF [14,15"], where the accelerating field is delayed for up to several hundred microseconds after the ionizing laser pulse. Delayed extraction improves MS resolution through at least two distinct effects. The first is reduction of effects of initial velocity distribution of the fragment, by allowing the fragments to separate according to their initial velocities, the accelerating field can compensate for the distribution. The second effect of interfragment interaction due to in- creased spatial separation of fragments. DE MALDI-TOF measurements of RNA fragments resulted in a MR of 7500 [15"'], nearly an order of magnitude better than continuous field MALDI-TOE DNA analysis with DE MALDI-TOF produced spectra with MR as high as 2,000-3,000, a factor of five improvement over continuous field conditions. As a demonstration of this technology [14], the improved resolution of DE MALDI-TOF allowed sequencing using exonucleases for significantly longer lengths than was possible in the continuous field mode.

Fragmentation limits both the maximum size and the mass resolution of DNA fragments. RNA has long been known to be less susceptible to fragmentation than DNA, and building upon this, Tang et al. [16"] have investigated the effect of chemical modification of the 2 carbon, and other locations on post-source fragmentation, with results that will potentially lead to analysis of longer DNA fragments through the use of base and sugar analogs.

Mismatch cleavage As noted above, differential sequencing schemes that rely solely upon hybridization are subject to the limitations of variation in hybridization efficiency due to effects other than sequence differences. An alternative approach to differential sequencing is the use of mismatch cleavage procedures, where the probe and target are hybridized to- gether, but instead of measuring the altered hybridization efficiency, an enzyme which cuts the hybrid at mismatched

bases is used, and detection is simply the process of analyzing the resulting cleavage pattern.

Cleavage patterns are currently investigated through conventional electrophoresis, and, thus, are not by them- sclves amenable to high-throughput analysis; however, in conjunction with other technologies for increased meas- urement speed (e.g. MS or capillary array electrophoresis [17,18]) mismatch cleavage is potentially capable of becoming a working tool for diagnostics. Recent results have reported on successful mutation detection schemes based on mismatch cleavage with DNA-DNA hybrids [19,20], as well as RNA-RNA hybrids [21].

Genetic expression measurements Schemes for quantitative assessment of the abundance of mRNA associated with particular genes are potentially a very powerful diagnostic tool. In the past year, Kinzler and co-workers [22",23 "°] have released two new articles utilizing the serial analysis of gene expression (SAGE) technique. SAGE uses a molecular indexing scheme, in which individual transcripts are indentified by short (10-15 nucleotide), well defined, subsequences. The indexes from a sample are counted by classical sequencing methods and are used as a measure of relative populations of RNA transcripts in the sample. In the paper by Velculescu et a/. [22"], SAGE was used to characterize the 'transcriptome' ofSaccharomyces cerevisiae. Indexes from over 60,000 transcripts were analyzed, revealing 4,665 unique identifiers. Of these unique identifiers, 1,981 were correlated with genes of known function, while 2,684 were associated with genes of unknown function. This work clearly demonstrated the applicability of the SAGE technique for large scale expression profile measurements in eukaryotic tissues.

SAGE was also used in the work of Zhang et al. [23 "°] where expression profiles were compared in normal and tumor human cells. Normal and cancerous samples from human colorectal tissue were examined in order to determine differentially expressed genes. While most of the 49,000 identified transcripts were similarly expressed in tumor and normal cells, over 500 transcripts were identified with greater than factor of 10 suppression or enhancement. Identification of the specific genes which code the differentially expressed transcripts has the potential to enhance the understanding and treatment of cancer.

The primary limitation to the SAGE technique is the need for classical sequencing of the tags, at approximately 10 bases per tag sequenced. In order for this technology to become a widely used diagnostic tool, it will need to be implemented in conjunction with one or more of the techniques listed above for rapid analysis of sequence data.

Ordered differential display [24] is a similar technique to SAGE, in that the 3' end of mRNA transcripts is isolated and amplified for detection. The technique does not

Advances in DNA diagnostics Graber et aL 17

employ standardized conditions of PCR for all amplified species however, thus, while inter-sample comparisons can be made, intra-sample quantitations are not possible.

Sample isolation with laser capture microdissection The quality of any diagnostic genetic measurement will be intrinsically linked to the quality and/or purity of the sample under investigation. This is especially important with regards to neoplastic or tumor samples, where a differential measurement in comparison with normal tissue is a key element in diagnosis.

The recent paper of Emmert-Buck et al. [25 °] describes a very promising technique for simple and rapid isolation of tissue samples by laser capture. Inspection and capture can be performed simultaneously under a microscope. An incident laser pulse captures the tissue to an overlaid film. Post-capture manipulations of the tissue samples indicate that the capturing process does not inhibit such diagnostic procedures as PeR or enzyme activity assays.

Conclusions The past year has seen significant gain in the power of DNA diagnostics, especially in the area of differential sequencing. The resolution and fidelity of hybridization to arrays of DNA on chips has been improved. MALDI-TOF MS measurements for use in diagnostic measurements have moved from the realm of the possible to clear demonstration of working experiments.

It seems likely that we have not yet seen the optimal scheme for differcntial sequencing. The most promising future directions will be those in which the ideas of speed, sensitivity, and parallelization are combined, as demonstrated by the promising result of MALDI-TOF measurements of an array of samples [13"].

References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as:

• of special interest • - of outstanding interest

1. Chee M, Yang R, Hubbell E, Berno A, Huang XC, Stern D, • . Winkler J, Lockhart D J, Morris MS, Fodor SPA: Accessing genetic

information with high-density DNA arrays. Science 1996, 274:610-614.

Simultaneous analysis of the entire human mitochondrial genome (16.6 kilo- base) was performed on DNA arrays of up to 135,000 different probe oligonucleotides. A two-color scheme allows differential measurement of hy- bridization on a single chip. This paper is of outstanding interest because it is the most complete measurement to date utilizing several of the princi- ples we outlined at the start of the review: parallelization, automation, and miniaturization. Although it may not be the ideal mesurement scheme, it is the current standard that must be improved upon for future work.

2. Drobyshev A, Nologina N, Shik V, Pobedimskaya D, Yershov G, Mirzabekov A: Sequence analysis by hybridization with oligonucleotide microchip: identification of 13-thalassemia mutations. Gene 1997, 188:45-52.

3. Dubiley S, Kirillov E, Lysov Y, Mirzabekov A: Fractionation, • phosphorylation and ligation on oligonucleotide microchips to

enhance sequencing by hybridization. Nucleic Acids Res 1997, 25:2259-2265.

Arrays of oligonucleotides are anchored in miniature polyacrilimide gel pads, rather than on a flat chip surface. The primary advantage is a higher density of target oligomers. In addition, typical reactions (e.g. hybridization or ligation) occur with solution phase characteristics as opposed to the solid phase behavior of standard DNA arrays.

4. Gruschin D, Yershov G, Zaslavsky A, Gemmell A, Shick V, Proudnikov D, Arenkov P, Mirzabekov A: Manual manufacturing of oligonucleotide, DNA and protein microchips. Anal Biochem 1997, 250:203-211.

5. Livshits, Mirzabekov AD: Theoretical analysis of the kinetics of DNA hybridization with gel-immobilized oligonucleotides. Biophys J 1996, 71:2795-2801.

6. Wailer J, Gausepohy H, Hauser N, Jansen ON, Hoheisel JD: Hybridisation based DNA screening on peptide nucleic acid (PNA) oligomer arrays. Nucleic Acids Res 1997, 25:2792-2799.

7. Pastinen T, Kurg A, Metspalu A, Peltonen L, Syv~nen A: Minisequencing: a specific tool for DNA analysis and diagnostics on oligonucleotide arrays. Genome Res 1997, 7:606-614.

8. Braun A, Little DP, K6ster H: Detecting CFTR gene mutations • by using primer oligo base extension and mass spectrometry.

C/in Chern 1997, 43:1151-1158. The primer oligo base extension (PROBE) technique involves an isothermal extension of a primer oligomer, which is first annealed to a region of inter- est in the genome. Careful selection of the mixture of dNTPs and ddNTPs present during the extension reaction allows accurate characterization of several bases downstream from the 3" end of the primer.

9. Little DP, Braun A, Darnhofer-Demar B, Frilling A, Li Y, Mclver RT, K6ster H: Detection of nET proto-oncogene codon 634 mutations using mass spectrometry. J Mol Mad 1997, 75:745-?50.

10. Braun A, Little DP, neuter D, M011er-Mysok B, K6ster H: Improved • analysis of microsatellites using mass spectrometry. Genomics

1997, 46:18-23. The authors use MALDI-TOF mass spectrometry and the PROBE technique (see Braun eta/. 1997 [8°]) to accurately characterize the size of tandem repeat microsatellites. Limiting the dNTPs in the mixture to those nucleotides which occur in tandem repeat regions allows the extension reaction to pro- ceed up to the end of the repeat region. The mass differences between varying length microsatellites correspond to entire repeat units, which vary between -600 daltons for dinucleotide repeats and -1200 daltons for tetra- nucleotide repeats. These mass differences are easily within the resolving power of current MALDI-TOF instruments.

11. Haft LA, Smirnov IP: Single-nucleotide polymorphism identification assays using a thermostable DNA polymerase and delayed extraction MALDI-TOF mass spectrometry. Genome Res 1997, 7:378-388.

12. Monforte JA, Backer CH: High-throughput DNA analysis by t ime-of-f l ight mass spectrometry. Nat Mad 1997, 3:360-362.

13. Little DP, Cornish TJ, O'Donnell MJ, Braun A, Cotter RJ, K6ster H: • * MALDI on a chip: analysis of low- to sub-femtomole quantities

of synthetic oligonucleotides and DNA diagnostic products dispensed by a piezoelectric pipette. Ana/Chem 1997, in press.

The authors have developed and tested a chip based 10× 10 array of oligonucleotides, utilizing nanoliter volume pits in the chip. The array of oligonucleotides can be completely automatically scanned, which eliminates the usual search for the MALDI 'sweet spot', which has previously limited the attainable throughput. Combination of this array technology with other diag- nostic techniques, such as PROBE (see Braun et al. 1997 [8",10"]), present a very promising avenue to high throughput, high sensitivity, diagnostic DNA measurements.

14. Smirnov IP, Roskey MT, Juhasz P, Takach ELI, Martin SA, Haft LA: Sequencing oligonucleotides by exonuclease digestion and delayed extraction matrix-assisted laser desorption ionization time-of-flight mass spectrometry. Ana/Biochem 1996, 238:19-25.

15. Juhasz P, Roskey MT, Smirnov IP, Haft LA, Vestal ML, Martin SA: • , Applications of delayed extraction matrix-assisted laser

desorption ionization time-of-flight mass spectrometry to oligonucleotide analysis. Ana/Chem 1996, 68:941-946.

Delayed extraction MALDI is performed by delaying the onset of the ac- celerating voltage for several hundred nanonseconds after the desorp- tion/ionization event. This delay has been shown to greatly increase the mass resolution attainable in continuous field MALDI, by up to an order of mag- nitude for RNA fragments. Increased mass resolution allows for diagnostic measurements of larger fragments, which in turn will allow for increased throughput in measurements.

18 Analyt ical b io techno logy

16. Tang W, Zhu L, Smith LM: Controlling DNA fragmentation • in MALDI-MS by chemical modification. Anal Chem 1997,

69:302-312. Fragmentation is one of the most severe limitations on both resolution and maximum size of DNA fragments analyzed by MALDI-TOF mass spectrome- try. This group is analyzing effects of chemical modification of the individual bases on the fragmentation rate, and has, in fact, shown that modifications of the 2' carbon of an individual base can significantly reduce fragmentation.

17. Woolley AT, Sensabaugh GF, Mathies RA: High-speed DNA genotyping using microfabricated capillary array electrophoresis chips. Anal Chem 1997, 69:2181-2186.

18. Quesada MA, Zhang S: Multiple capillary DNA sequencer that uses f iber-optic i l lumination and detection. Electrophoresis 1996, 17:1841-1851.

19. Giunta C; Youil R, Venter D, Chow CW, Somers G, Lafferty A, Kemper B, Cotton RGH: Rapid diagnosis of germline p53 mutation using the enzyme mismatch cleavage method. Diagn Mol Path 1996, 5:265-270.

20. Roberts E, Deeble VJ, Woods CG, Taylor GR: Potassium permanganate and tetraethylammonium chloride are a safe and effective substitute for osmium tetroxide in solid-phase fluorescent chemical cleavage of mismatch. Nucleic Acids Res 1997, 25:3377-3378.

21. Goldrick MM, Kimba!l GR, Liu Q, Martin LA, Sommer SS, Tseng JY-H: NIRCAO: a rapid robust method for screening for unknown point mutations. Biotechniques 1996, 21:106-112.

22. Velcutescu VE, Zhang L, Zhou W, Vogelstein J, Basrai MA, • Bassett DE, Hieter P, Vogelstein B, Kinzler KW: Characterization

of the yeast transcriptome, Ceil 1997, 88:243-251.

Serial analysis of gene expression (SAGE) is used to characterize which genes are active in yeast cells under three different conditions. As the entire yeast genome has now been sequenced, SAGE tags (14 nucleotides long) can be used to uniquely identify active genes and then locate them within the genome.

23. Zhang L, Zhou W, Velculescu VE, Kern SE, Hruban RH, • • Hamilton SR, Vogelstein B, Kinzler KW: Gene expression profiles

in normal and cancer cells. Science 1997, 276:1268-1272. Serial analysis of gene expression is used to characterize the differential activity of thousands of different mRNA transcripts in normal and cancer cells. Measurements have indicated that whereas most genes are relatively unchanged, up to several hundred are found which are either over- or under- expressed by up to lO0-fold. Information of this nature will help further our understanding of the mechanisms involved in cancer cell growth, and how they differ from normal cells.

24. Matz M, Usman N, Shagin D, Bogdanova E, Lukyanov S: Ordered differential display: a simple method for systematic comparison of gene expression profiles. Nucleic Acids Res 1997, 25:2541-2542.

25. Emmert-Buck MR, Bonner RF, Smith PD, Chuaqui RF, • Zhuang Z, Goldstein SR, Weiss RA, Liotta LA: Laser capture

microdissection. Science 1996, 274: 998-1001. Laser capture microdissection is a technique which allows simultaneous in- spection and capture of tissue samples for further analysis. Preliminary tests indicate that samples captured with this technique can be manipulated with such procedures as PCR, ligation, and so on. Sample capture sizes have been demonstrated in the 60-700 micron size range. This is very promising with regards to preferential capture of tumor cells for differential analysis with normal cells.