Fifty Years of Molecular (DNA/RNA) Diagnostics

As the journal Clinical Chemistry celebrates its 50th birthday and enters its 51st year, we reflect on recent advances in nucleic acid diagnostics. Although the structure of DNA was also described ∼50 years ago, only 20 years have past since a practical method of amplifying DNA for the clinical laboratory appeared, well known today as the PCR. Since the advent of PCR, growth in nucleic acid diagnostics has been extraordinary and is the fastest growing segment of many clinical laboratories. Clinical Chemistry is now subtitled The International Journal of Molecular Diagnostics and Laboratory Medicine in recognition of the importance of DNA/RNA diagnostics. In this review, we offer personal perspectives on five key advances in molecular diagnostics by some of the pioneers in the field. Russ Higuchi reviews real-time PCR, which is an elegant solution for nucleic acid quantification in research and the clinical laboratory. Carl Wittwer describes DNA melting analysis as a simple and surprisingly powerful tool that can be seamlessly appended to PCR. Tom Gingeras reviews the increasing power of array technology, with feature density rivaling microelectronics miniaturization. Progress on integrated devices that combine cell separation, sample preparation, amplification, and analysis is described by Larry Kricka. Finally, the field of circulating nucleic acids is reviewed by Dennis Lo, with a myriad of clinical applications enabled by a new understanding of the biology of extracellular DNA and RNA. Many early advances in molecular diagnostics have been highlighted in Clinical Chemistry. With DNA/RNA diagnostics becoming ever more important in the 21st century, the Journal serves as an advanced window into the future of molecular clinical chemistry. —Russ Higuchi When late in 1986 I joined Cetus, the birthplace of PCR, “closed-tube” PCR using a thermostable DNA polymerase had just been achieved (1)(2). Before this, the thermolabile polymerase had to be added at each cycle. Tom White, our vice-president in charge of Research and Development, told me that what was needed now was some way to visualize the amplification product, also without opening the tube. Such closed-tube detection was termed “homogenous”. A few years later I was working with Gavin Dollinger on his idea of “tagging” things that needed to be traced (e.g., explosives, money, and pharmaceuticals) with specific, amplifiable DNA sequences (3). We noticed that very high molecular weight DNA was occasionally generated in PCRs from such tagged substances. If we could reproducibly promote such large DNA, Gavin reasoned, it might be detected directly in the tube by light scatter. We set out to use biotinylated primers and streptavidin in the PCR to try to catalyze the formation of long branched chains of amplicons. We succeeded only in forming precipitates. To determine whether DNA was in the precipitates, we added ethidium bromide to the completed PCRs and held the tubes up to ultraviolet (UV)1 light. The precipitates were fluorescent, but the fluorescence was not specific to the presence of double-stranded DNA (dsDNA). Bob Griffith, who was helping us with these experiments, was directed to try some different conditions. He showed me the gel with the expected DNA band from the control without streptavidin and mentioned that, without thinking, he had added the ethidium bromide at the beginning of the PCR. I had two reactions. One was surprise that the PCR had worked at all in the presence of ethidium bromide, a known DNA polymerase inhibitor. Indeed, PCR would have been inhibited if the concentration of ethidium bromide had been somewhat higher. The second was to realize that because PCR creates much more dsDNA than is put into it, there should be more ethidium bromide fluorescence created as well. Such fluorescence would be detectable without opening the tube. The PCRs were repeated without streptavidin. When held up to a UV light, the tube containing the amplification target glowed brightly compared with negative controls (Fig. 1A1 ). Although this “endpoint” reading of PCR was of itself very useful, we thought that taking fluorescence readings on a cycle-by-cycle basis might provide a quantitative assay with a wide dynamic range (4). The fewer PCR cycles it took to detect an increase in fluorescence, the more copies of the target DNA sequence there were to begin with. To demonstrate this “real-time” reading of the PCR (which we later called “kinetic”), we attached one end of a bidirectional fiber optic cable to the open top of a PCR tube in a thermocycler and the other to a fluorescence spectrophotometer. With excitation at 500 nm, a fluorescence trace at 600 nm was recorded (Fig. 1B1 ). The ups and downs in fluorescence that corresponded to the downs and ups in temperature from thermocycling were seen, as well as the net gain in fluorescence at low temperature attributable to the accumulation of dsDNA PCR product. To assess the quantitative performance of PCR in this mode, it soon became clear that we needed a way of reading multiple amplifications in parallel. One way, ultimately adopted in the first commercial real-time PCR instrument, was to use multiple fiber optic cables (one per tube) and a rapid multiplexing system for sequentially exciting and reading the fluorescence from each tube. I thought that a simpler way would be to use digital imaging; Bob Watson and I therefore implemented such a system using a charge-coupled device (CCD) camera from a home-built, digital gel-documentation system (5)(6). The camera was set up in a darkroom looking down over the exposed tops of the closed PCR tubes set into the thermocycler block. UV lights were used to illuminate the tubes in the block, and at every annealing/extension phase of the PCR an image was “grabbed”. At every well position in the image, the pixel values were summed. The cycle-by-cycle plot of these sums generated “growth curves” for each PCR. The raw growth curves were clearly not useful (Fig. 2A2 ) without some processing. It made sense to normalize PCRs relative to each other based on their early cycle fluorescence readings because the beginning inherent fluorescence should be the same for each tube. This made the growth curves of replicate PCRs nearly identical. A global, threshold fluorescence value could then be set (initially we referred to this as the “AFL”, or arbitrary fluorescence level). The number of cycles it took to cross this threshold, which can be interpolated to fractions of cycles, is inversely and linearly related to the logarithm of the initial number of target molecules (Fig. 2B2 and C). By comparing these cycle numbers, commonly referred to as cycles to threshold (Cts), with those from a reference set of PCRs with known initial numbers of target molecules (as a calibration curve), the starting target number in each “unknown” can be inferred. Although real-time PCR with dyes, in particular SYBR® Green I, are now well accepted, at first it was difficult to overcome concerns about nonspecific amplification. Thus, the commercialization of real-time PCR depended on the subsequent development of sequence-specific, probe-based homogeneous detection systems, in particular the dual-fluorescently labeled TaqMan® probes (8)(9). With improvements in PCR specificity, such as that from “hot-start” polymerases (10), as well as the ability to validate results by use of Carl Wittwer’s melting curve analysis (see next section), dye-alone methods, which are inherently simple and inexpensive, have gained in popularity. For the future, would it not be nice to be able to combine the high “bandwidth” (in terms of numbers of targets) of microarray-based nucleic acid quantification with the sensitivity of real-time PCR? Only the most highly expressed mRNAs are generally detectable on microarrays: many interesting genes with low expression (transcription factors, for example) remain unaccounted for in many experiments (11). There have been some attempts at a “microarray of PCRs”, but as far as I know, none has led to a practical system allowing both the individual “addressing” of primers and delivery of precise amounts of sample to the micro-PCRs. The problems do not seem intractable, however, and perhaps a solution will be available soon. —Carl Wittwer Melting of the double helix to separate random coils is a fundamental property of DNA. The melting profile of a DNA duplex depends on the surrounding buffer (particularly the salt concentration), the GC content, the length, and the sequence of the duplex. Conventional DNA melting is measured by UV absorbance, often takes hours to perform, and requires large amounts of DNA. In contrast, monitoring of DNA melting by fluorescence requires only a few minutes, and adequate amounts of DNA are conveniently provided by PCR. When performed in the same tube as PCR amplification, melting analysis provides a homogeneous, closed-tube system for product identification. With high-resolution DNA melting, sequence alterations can be easily genotyped and entire amplicons scanned for variations. We first observed DNA melting during PCR by accident. Funded by an NIH small business grant, we were adapting flow cytometry optics to monitor rapid-cycle PCR in glass capillary tubes (12). As a measure of total double-stranded product, we chose SYBR Green I, a new dye being marketed as an ethidium bromide replacement for DNA detection in gels. SYBR Green I was brighter than ethidium bromide, and its spectrum is similar to that of fluorescein, so that one instrument channel could be used for either product or probe analysis. For probe-based detection, we synthesized oligonucleotides with fluorescein and Cy5 amidites that were commercially available at the time, hoping to use fluorescence resonance energy transfer for signal generation. Our initial attempts to monitor PCR by hydrolysis of dual-labeled probes failed; therefore, instead of separating two fluorophores by hydrolysis, we labeled two adjacent probes, one with fluorescein and one with Cy5, so that the fluorophores were brought together by hybridization. In retrospect, this was serendipitous because hybridization fluorescence directly monitors probe melting, whereas probe hydrolysis does not. Real-time PCR for quantification is usually performed by acquiring fluorescence once each cycle. However, instead of looking only once each cycle, we acquired fluorescence continuously throughout PCR, including the temperature transitions. By plotting fluorescence vs temperature, we generated curious spiral curves (Fig. 33 ). Each loop of the spiral was generated from one PCR temperature cycle. With SYBR Green I, the sharp decrease in fluorescence just below the PCR denaturation temperature corresponded to the PCR product melting temperature (Tm). With adjacent hybridization probes, the major fluorescence decrease corresponded to the melting temperature of the less stable probe. We reasoned that sequence variants under hybridization probes should make the duplexes less stable and enable genotyping by melting profiles. Although originally observed during PCR, melting curve generation after PCR has become most popular. Unlike a static “dot blot” at a single temperature, an entire melting profile is obtained across the region of temperature transition. The first example of the use of this technique was published in Clinical Chemistry in 1997 for factor V Leiden (13). Factor V Leiden testing by melting analysis was cleared by the United States Food and Drug Administration (FDA) in and is one of the few in clinical The method has become a for clinical because many can be with a single probe than probes and for each are by instead of The number of that can be depends on the of the instrument and analysis melting analysis was we noticed by that genotyping was with only one probe instead of two The fluorescence of labeled oligonucleotides either or with duplex on the and the sequences of the probe and In retrospect, this is not because of the in of the with duplex the of melting curve genotyping with probes was either one or two oligonucleotides with labeled the new method requires only an probe and a DNA The and for assay development are As probe melting have become simpler over time, melting analysis of PCR has become more product melting with SYBR Green I was a method to different by melting analysis by The of melting analysis was that were not exposed to the as future However, many different had the same the of the method with real-time high-resolution melting for analysis of PCR have with the of the of that can be from this simple With high-resolution melting, a large of can be genotyped without probes only two primers in the presence of a DNA are by and that the of the melting genotyping in in is the of the melting high-resolution melting analysis is a method for The sensitivity and for are and in up to In to all other that after PCR, high-resolution melting analysis is requires after PCR, and can be performed in The is compared with real-time which are because of With only a DNA dye and high-resolution melting both product and probe melting can be obtained during the same temperature after PCR. melting is observed at low and product melting at high (Fig. ). The product can be used either to the probe or to the product for This ability to both for sequence and sequence alterations in the same the for of of its melting analysis is an method in research and clinical diagnostics. Gingeras The by in was on was used as a method to out the of of and the of as for the of important (e.g., and was clearly Although the in the showed a of a array it was the of the that In an of by by could be used to detect sequences in DNA and RNA. Such would be in diagnostics and nucleic acid I this as a way to difficult research and it in for more years as a tool in practical For years we had been working on including directly in It was important to determine whether the to to the only available at the time, was a to could be by detection of in the polymerase at a of acid However, with the of it was that genotyping more of the would be to the of To the and of the of the needed to be This sequence would to the and of However, of as of monitoring was in A with in late me to his I that an array might be to the sequence of large of the as a tool for that this would be but that it was Thus, we set out to and the first research array tool as a to a As a separate that the and genes of were the first commercial research made by Since the and applications of have The array Each of the probes was synthesized on the array as an array containing probes are with This increase in probe density has a of array each increase in probe density has been and by a or clinical For of a can be at a by use of probes in a In both genotyping and can be about the This more than the probes used in the By the feature to a increase in probe density was and such were used to detect in and to from other density for of the entire were made by the for each probe. Such were used to individual clinical by with have been used to of and by such on cell that there are many more and than are for by of these and therefore a of that be important in some It that important and clinical research will to the development of array and the the number of probes needed to different at using two probes per and the of the Thus, it will probes synthesized using to every of the on a single Although this of is not with methods, the of the can be on with and a of Such are being used today in new applications such as of of factors, of and DNA In the of these research to the of clinical diagnostics will more of array Clinical have new that could the of their and at the same are the early of to the growing of laboratory and and the of to with the It is surprise that clinical have with particular the rapid in and and the of the total system to analysis. A or is a device that all of the in an in a of and on one small device devices are made from and glass by use of adopted from the microelectronics such as and or from by The of this of to devices can be traced to the of at who a on a in the early on but was in the early This in research and development of devices from a new of the of of analysis small can many and of on a single The importance of testing and its it as a of that would from the A including sample (e.g., cell and DNA amplification of target sequence (e.g., and detection and quantification of (e.g., hybridization and individual in the were to a most PCR, a to most if not all and I our on in the We were by in the at the of who was on in We the range of applications for devices in clinical analysis and to the of in and in A of the initial phase of our was The NIH thought the and the thought it and each referred us to the other for We had with the we therefore on a to our This led to a that ultimately with We were in the early to have to the in the of of and it was this that made our very first as the we to our to commercial In the most were used to the ever on and were by our for the of our to make it we in using at the at Our early PCR of simple or in and with glass (Fig. ). were for PCR, some related to the of the used to the other individual were For in a were for from as a to analysis of (Fig. ). The in many is PCR This can be easily in a by use of capillary The next the development of was of two of the in the such as cell and PCR or amplification and capillary on a single Finally, in have all to with of sample preparation, PCR, and real-time detection of the of has been achieved for DNA on a of the of amplification, and analysis The of a new is the from idea to working then to a phase of in a laboratory and to a commercial product by the For the two are not A of for DNA and analysis by capillary as well as of DNA hybridization have been for research but a commercial system for clinical use is by the clinical With the of the clinical used in the analysis of nucleic acids is usually However, much recent has on the biology and applications of circulating nucleic acids in the of in this field in after years of to to detect in the The of these made the analysis of their In the of we noticed two in Medicine the detection of alterations in DNA from the and of with The the which and a made us whether the would also its DNA into the and of the had referred to the as being and it is to a as as a At that we not the method to DNA from and we therefore used and directly in PCR. this us to for the first time, the presence of DNA in the and of With the advent of real-time quantitative PCR, we were able to measure the and relative of such circulating DNA in The concentration of circulating DNA was that using DNA in should be a that has out to be DNA has been used to assess the of and the of the of for nucleic acid analysis has the of single the and the analysis quantitative in circulating DNA have also been detected in a of such as The development of a and DNA should the of such The of circulating DNA in has us to for circulating RNA. This was in with subsequent experiments the as the of such (Fig. ). is surprisingly its with perhaps in with the number of that can be used for monitoring In to the of circulating nucleic acids to detection and monitoring has also over the few To and DNA have been detected in the and of In the detection of DNA in the and of with and other has rapid DNA has been to be a tool for the detection, and of such It is that many other applications of circulating nucleic acids will be over the next few applications of and monitoring be about the of such circulating nucleic acids and the of such molecules Real-time of PCR. completed PCRs in tubes by UV light. tubes ethidium The first tube on the had target sequence to the the two not. fluorescence from PCRs with a fiber optic cable and a spectrophotometer. The trace is from a PCR with 20 of DNA and primers specific to a The trace is from a control PCR with the primers but without the target DNA was with from Real-time PCR for RNA. fluorescence from a camera growth curves from the starting numbers of the as described in the The threshold fluorescence from which (see are is calibration curve based on of the starting number and the growth curve This was with from monitoring of rapid-cycle PCR. SYBR Green I monitoring of PCR (12). cycles were completed in adjacent hybridization probe monitoring of PCR The of PCR product and hybridization probes are with from by probe and PCR product melting profiles. major and by a single under the probe. The probe showed melting of or on the In with the probe the product of the was than that of the A a easily with high-resolution of of of the feature and the number of probes for of a at on a single The two probes per such as and with feature For it would probes synthesized with to every of the on a single the can be by use of probes synthesized with to a of with from from the top of an integrated cell on the by the wide new by the with from the of into the Such molecules are thought to be from by with of of The of and for Clinical Chemistry double-stranded charge-coupled threshold melting total and of analysis are by the of to has a as a on these Dennis or applications on applications of circulating nucleic acids in of these have been to and Dennis is a of of melting analysis and rapid PCR are by the of Research to and from to has a as a on these and in is an of a that array is an of Molecular a that real-time PCR