Copyright © 2012-2017 TIRF Labs, Inc. All photographic images, schemes and text located at this web site are copyrighted by TIRF Labs, Inc.
Single Molecule Detection, sm-FRET, Single Cell Analysis, Cell Membrane Studies, TIRF microscopy, TIRF accessories for fluorometers,
TIRF biosensor instruments, TIRF electrochemistry, TIRF dielectrophoresis, Real-time microarrays, iDiagnostics - cellphone based TIRF, and more>>>
Misconception of Label-Free Biodetection
Summary. There is an unrealistic dream in the area of molecular diagnostics - dream about so called “label-free” detection. The dream stems from the natural desire of minimizing the sample preparation and/or simplifying the bioassay preparation, by using a detection method, which does not require extrinsic labels. However, this nice intention, unfortunately, does not work in the area of molecular diagnostics. Only label-based methods work for applications that require femto-molar (10^-15 M) level of detection. These methods do require extrinsic labels either on the target molecule (e.g. labeled secondary antibody attaches to the target) or at the affinity reagent, as in the case of molecular beacon assay.
Perpetual motion machine has been prohibited by hard science. Nevertheless, the U.S. Patent and Trademark Office receives annually dozens of applications for invention of perpetual motion. Sensitive label-free is not prohibited by hard science. It appears that the phenomena of statistical physics make it impossible to detect small signals caused by binding of few biomolecules, if they do not contain extrinsic labels. Thermal fluctuation of the background, which is caused by numerous intrinsic labels of irrelevant molecules mask the signal of interest. Nevertheless, the attempts to improve the sensitivity of label-free continues. By 2017, U.S. Federal Government has invested into label-free estimated $20 billion; private and institutional investors added likely equal amount. Unfortunately, the results of failed attempts are too difficult, if not impossible, to publish. There is a law, which prescribes the recipients of successful projects funded by the US government to publish their results. Surprising lack of a similar law for unsuccessful projects frequently results in stepping on the same rake again and again.
NSF, NIH, other reputable U.S. and European agencies continue to solicit applications for label-free molecular diagnostics. Quartz Crystal Microbalance (QCM), Surface Plasmon Resonance (SPR), Field Effect Transistors (FET), microcavity laser resonators, nanowires, and other methods that do not require extrinsic labels are indeed capable of detecting relatively large concentrations of different bioanalytes. Typical limit of detection for label-free is at the level of micro-molar (10^-6 M) concentrations, which is million-fold away from the level necessary for molecular diagnostics. Biomarkers of diseases should be detected at the level of femtomolar concentrations (10^-15 M). Despite of numerous attempts, at the present time, only fluorescence techniques and similar methods that use extrinsic labels have demonstrated the limit of detection necessary for molecular diagnostics of diseases.
As mentioned above, the idea of label-free detection is not prohibited by hard science; it does not contradict to fundamental laws of physics. Moreover, label-free methods have been successfully applied for the detection of biomolecules with concentrations down to submicro-molar level (10-7 mol/liter). For molecular diagnostics, however, one needs a biodetection methods that are capable of measuring concentration of biomolecules down to the level (10-15 10-17mol/liter), because many of molecular markers operate in this range of concentrations.
It appears that fundamental phenomena of statistical physics are responsible for poor sensitivity of label-free methods. So far, no single molecule study has been published, which used a label-free method. On the other hand, thousands of papers have documented single molecule detection using label-based methods. Hereby, we invite a statistical physicist to investigate the mechanisms of label-free versus label-based biodetection. The figure below outlines the principles of label-free and label-based detection.
We in TIRF Labs have collected preliminary data that suggest that the thermal noise of large background signal imposes limitations on the sensitivity of label-free methods. Search of the literature gives no results for successful label-free molecular diagnostics. On the other hand, there are thousands of published scientific papers that have carefully documented the ultimate limit of detection - single molecules - in the case of label-based methods. It appears that a theoretical approach similar to the Smoluchowski equation, Fokker-Plank theory, and Kolmogorov forward and backward equations can be employed to theoretically predict the best limit of detection possible for label-free methods.
Surface Plasmon Resonance (SPR) provides an example of tremendous efforts and dramatic failures in the area of label-free. One of the leaders in SPR instrumentation is Biacore, a company founded in 1984 under the name of Pharmacia Biosensor AB, by Pharmacia and the Swedish Defence Research Agency. In 1996, the name changed to Biacore AB Corporation. The first Biacore SPR instrument was issued in 1990. In June 2006, GE Healthcare acquired Biacore for $390 million. By 2006, Biacore has saturated the market with expensive SPR instruments. It appears that the acquisition resulted in losses to GE Healthcare over the period of 10 years.
By 2000-2002, Biacore approached very close to practical limit of SPR detection - submicromolar (~10-7 mol/liter). Biacore built a benchtop machine with price tag ~$500,000 (approximately half of million US dollars), with analytical limit of detection at (~10-7 mol/liter). This limit appears to be close to that of theoretically possible. However, the diagnosing of pathogens and diseases requires, as mentioned above, more sensitive tests - capable of detecting femtomolar (10-15 mol/liter) and attomolar (10-18 mol/liter) concentrations. The concentrations of protein, nucleic acid, and metabolite molecular markers should be measured in complex matrix of biological fluids, such as blood, full of other proteins and nucleic acids. The task of biodetection for diagnostic purposes is frequently compared to the challenge of finding a needle in a haystack. In fact, the task of biodetection is even more challenging - it is like finding a unique straw with a specific shape among myriad of similar straws in the haystack. If the straw of interest is not labeled (or its counterpart - specific affinity reagent - detection antibody or probe DNA are not labeled), your chance to find the target among myriad of similar molecules is close to zero.
Unfortunately, label-free methods do not work in the range of concentrations that are clinically and diagnostically significant 10-15 -10-17 mol/liter. Why? Let us take a look into the principles of label-free. In fact, Surface Plasmon Resonance (SPR), Quartz Crystal Microbalance (QCM), field effect transistors (FET), microcavity laser resonators, nanowires, and all other allegedly “label-free” methods Do use certain labels. The problem is that these labels are intrinsic. They are naturally contained in biological analytes. In huge amounts. The problem is that they are contained in myriad of irrelevant molecules of the matrix and similar biomolecules. Why this is a problem? These intrinsic labels are not specific - they do not allow to distinguish the desired target molecule of interest, protein, or nucleic acid, at the large background of irrelevant homologs. There are thermal fluctuations of the background that totally mask the signal of interest. In the case of SPR, the label is refractive index, in the case of QCM - mass, FET - charge of molecules. The problem is that, similar to straws in the haystack, similar labels are present everywhere around the bioanalyte molecule of interest. Irrelevant labels create overwhelmingly large background. Advocates of label-free argue: “we will subtract the background”. Indeed, in certain cases you can subtract the background. But the worst problem is that the background fluctuates unpredictably. Thermal fluctuations cannot be removed. The amplitude of thermal fluctuations is larger than the anticipated signal from low concentration bioanalytes 10-15 -10-17 mol/liter. Our preliminary data suggest that label-free methods, in principle, are not capable of detecting femtomolar (10-15 mol/liter) and attomolar (10-18 mol/liter) concentrations of biomarkers.
In practical systems, in addition to thermal fluctuations there are many other processes that contribute to fluctuations of the background signal. The cumulative effect of these undesirable fluctuation pulls the limit of detection to worse. For example, in the case of SPR, large changes in the background are caused by variations of the ionic strength and other parameters of the solvent that are related to incremental change of the refractive index. Without specific, distinguishable labels, there is no way to differentiate between the SPR signal caused by binding of the target molecule and that caused by thermal fluctuations and other irrelevant processes. Our preliminary analysis shows that similar estimates are valid to all other label-free methods. A nanomolar (10-9 mol/liter) level appears to be the lowest theoretical limit of detection for all label-free methods. In practice, many additional processes, such as mentioned above changes of matrix composition, non-specific interactions with the affinity reagents, ionic strength, natural and method-specific fluctuations in solvent and solute, cause additional unpredictable fluctuations of the signal. In practical instrument, label-free methods are capable of providing the limit of detection at micromolar or sub-micromolar levels 10-6 10-7 mol/liter.
There have been numerous failed attempt of using label-free for molecular diagnostics. The failure of these attempts indirectly support our assumption that label-free detection is not suited for molecular diagnostics. After 9-11, followed by the anthrax letter attacks, the U.S. government spent over $70 billion on biodefense, including estimated $40 billion on molecular diagnostics (MDx). Naturally occurring diseases, such as avian flu, SARS, Ebola, can be even more damaging than a man-made bioweapon. Damages caused by pathogens and toxins can be enormous. 9-11 anthrax letters caused economical and societal damages comparable with the effect of a nuke explosion in the Washington D.C. area. Countermeasures, prevention, decontamination, and retaliation require rapid and accurate MDx. Inaccurate or slow results of analyses are incredibly damaging and expensive. In 2002, false positive and false negative results of traditional MDx methods almost shut down the Winter Olympics in Utah. In the period 2001-2016 false positive and false negative results of tests seriously threatened many high-profile political events and caused damages measured by billions of dollars.
After 2001, billions of dollars were invested in the development of advanced MDx. Not surprisingly, there is no analytical technique, which has not been tried for MDx. In 2003-2004, the U.S. Department of Defense carefully analyzed all existing and emerging biodetection techniques. Among these techniques were so-called label-free methods, such as SPR, QCM, field-effect transistors, resonance laser cavities, carbon nanotubes, and similar methods. All these methods that do not require extrinsic labels neither at the target molecule, nor at the affinity reagent, failed. As mentioned above, the intention of “label-free” detection stems from our natural desire to have the results rapidly, inexpensively, and with minimal labor. However, in certain cases, as a proverb says: “Nice intentions pave the road to hell.” Nice intentions of humans sometimes disagree with laws of the nature. In particular, the intention to employ so-called label-free methods have failed. Numerous times. Nevertheless, new attempt are being made. We are stepping on the same rake, again and again. Billions of dollars have been wasted. This waste of resources and propagation of error of label-free detection should be brought to the end. We hope that posting of this web page will help to rectify the situation.
In principle, label-free methods can NOT be sensitive enough for molecular diagnostics. Sub-micromolar (10-7 10-8 mol/liter) level appears to be the lowest level of detection for all label-free methods, which is not sufficient for molecular diagnostics. We believe that the misconception of label-free should be brought to the daylight and actively discussed by the entire biodetection community. We believe that such discussion will facilitate the progression of highly efficient biodetection technologies, such as iDiagnostics.
If you have thoughts and opinions about the label-free detection and the situation in MDx area in general, please share them with us. Together, we can improve the situation and accelerate the progress in MDx. Advanced MDx technologies have been developed; they do use labels. However, some of the technologies detect unlabeled analytes.
15 years and $70 billion after 9-11, accurate and rapid MDx products are not at the market yet. Accurate and inexpensive molecular diagnostics devices such as iDiagnostics, might be in medical cabinets by now, if the label-free misconception and other misfortunate aspects would not hamper the progress. Precision diagnostic devices and methods have been developed. However, they are still in laboratories, including TIRF-EC Microarray and iDiagnostics in TIRF Labs. Meantime, their way to the general public is blocked by the label-free misconception. Billions of dollars of the U.S. federal funds continue to be wasted for useless attempts. Why the results of failed attempts have not been published? Why the U.S. government agencies continue to solicit label-free MDx? We believe that we need a “Journal of Failed Research” to prevent stepping again on the same rake.