Recently The Journal of Molecular Recognition featured a review of the optical biosensor literature published in 2008
written by Rebecca L. Rich and David G. Myszka, both at the University of Utah. The current review follows a series of
several similar undertakings. With these efforts Rich and Myszka have literally evaluated thousands of papers with a
keen eye for the strengths and shortcomings of experimental work with, in particular, biosensors based on surface
plasmon resonance (SPR). Certainly I, and probably many others, have had a sobering experience from reading these
papers, and I find (and still find) many of the suggestions made here valuable, if not always implementable in reports
on our own work. What we learned from this latest reading is, however, another matter. According to Rich's and
Myszka's series of reviews it would appear that not much progress was made in the way researchers design, execute,
and report their experiments with SPR sensors. In short, if you share the views of Rich and Myszka, it must be a bore
to read through all of these papers looking for excellence. Perhaps this explains a striking development with their
writing style. Starting out from a fairly academic approach (Rich and Myszka, [2000]) the reports over the literature from
2007 (Rich and Myszka, [2008]) and, in particular, 2008 (Rich and Myszka, [2010]) are kept in a colloquial style. The
2008 report contains some rather odd features such as the insights of David G. Myszka's mother, a stanza from the
late Michal Jackson's I am bad, and a mysterious call to grab your pole and put on some sunblock because you're
about to get schooled on the Moby Dick. I am not sure what Moby Dick and the biosensor literature share and I am
not in any way enlightened by Rich and Myszka. This would be endurable, however, was it not for the fact that Rich's
and Myszka's writings are in the want to form a report with harsh criticisms of the scientific work of named individuals.
In the abstract, Rich and Myszka are not even shy of calling a large number of the reports abysmal. In this context
their capricious writing appears to leave behind an impression of hauteur that outshines whatever serious message can
be found in the review.
A look on the formal scientific justifications for Rich's and Myszka's grading of the scientific papers on biosensor
literature also calls for concerns. Four of the ten papers selected for the honor roll appear to pertain to Technology
Validation. It is, of course, no surprise that such a literature excels in making nice experiments with instruments as
this is indeed the focus of the work. This section also includes a report by Rich et al. ([2008]) (reference 1362 in Rich
and Myszka, [2010]) on the binding of IgG to Protein A and Protein G. These are extraordinarily well-characterized
interactions, and for this reason well-suited for validation studies. I am not underestimating the challenges of making
good validation studies, but it seems hard to compare the quality obtained from a recent study on the long-known
binding of IgG to Protein A with, e.g., the endeavors of studying the interaction between apolipoprotein A-V and
members of the low-density lipoprotein receptor families (Nilsson et al., [2008]) as reported by Nilsson et al. (reference
624 in Rich and Myszka, [2010]). The validation study by Rich et al. basks in abundant and high-grade supplies of
protein and reports something useful, but nothing particularly novel or surprising. By contrast, the undertaking by
Nilsson et al. faced the challenges of providing complex reagents and handling a less-characterized experimental
system. Nevertheless they managed to provide a consistent set of data, including the SPR biosensor experiments.
While the quantification of dissociation constant for one of the interactions probed in the study (ApoA-V binding sortilin)
could have been described in greater detail, the major point about the experiments (as shown in the paper's Figure 1,
i.e., the ability of heparin, neurotensin, or pre-receptor associated protein to inhibit this binding) is well supported by the
SPR biosensor data. Even so this paper was only graded with a D. This is the most common grade given by Rich and
Myszka, which, in their words, are for papers that give biosensor technology a bad reputation. As outlined above,
however, there is definitely interesting findings from Nilsson and others' work with SPR and I cannot agree that their
efforts compromise SPR technology. For the sake of clarity, I must reveal a slight conflict of interest regarding my
impartiality since the study by Nilsson et al. involves colleagues from my own institution (Aarhus University). However, I
note that Rich and Myszka do not hesitate to grade their own paper with an A in the midst of work by many others
claimed - also by them - to be abysmal. In this view, at least I feel my style of self-appraisal less blatant.
A similar point can be made about the work by Gehman et al. ([2008]; reference 1075 in Rich and Myszka, [2010]),
however, with the important difference that this study was singled out by Rich and Myszka as an example of papers
that include some data, but are absurdly awful. In the report by Gehman et al., an antimicrobial peptide from the
Australian tree frog was demonstrated by use of a SPR biosensor and other means to release an immobilized lipid
layer. Almost needless to say this setting will never generate sensorgrams following single exponentials. Accordingly
an initial association between the injected peptide and anionic lipids is observed followed by a release of the peptides
and lipids leading to a lower response level than before the injection of peptides. In short, some of the immobilized
lipids were removed as a consequence of the exposure to the peptides. This is a quite ingenious way of making use of
label-free detection to confirm protein-lipid interaction, which also solves the challenge of making these observations
with a cell membrane-like environment by presenting the lipids in an ordered layer. The interpretations by Gehman et
al. are limited to what is doable (that is, no estimates of affinities or kinetics, but a comprehensive description of the
outcome of exposing the lipid layer to the antimicrobial peptides). In my translation of the biological implications, the
experiment provides compelling evidence that the Australian tree frog has powerful means of destroying the lipid
membrane of potential microbial pathogens. It would only be justified if the experimental set-up catches the interest of
researchers wanting to use biosensors for a study of protein-lipid interactions, and indeed this has already happened
(Biverstahl et al., [2009]; Dennison et al., [2009]). I will leave it to the interested reader to look up all of the greetings
by Rich's and Myszka's to papers they grade with an F, which include the report by Gehman et al. However, the
following statement Without seeing the data, we have to wonder if the authors of F papers did the experiments right
or, in fact, did it at all appears as a statement of consequence. I am wondering whether Rich and Myszka mean to
suggest that Gehman and others are involved in cases of scientific dishonesty. As should be clear from my writings
above, the study by Gehman et al. appears to a non-specialist such as myself as solid and trustworthy. I cannot say I
have the same feeling concerning Rich's and Myszka's grading of the scientific literature.
Finally, a point of scientific interest is indirectly brought up by Rich and Myszka with their statements: It's apparent
that most users whose papers got D's fail to understand that a sensorgram should be a single exponential (p. 17 in
Rich and Myszka, [2010]) and when referring to additional complexity not accounted for in an 1:1 binding model used
by Kalyuzhniy et al. (Kalyuzhniy et al., [2008]; reference 172 in Rich and Myszka, [2010]), that Rather than invent an
unlikely binding scenario to account for these secondary components, the authors realized the complexity most likely
arose from biologically irrelevant artifacts (p. 6 in Rich and Myszka, [2010]). I cast no doubts on Kalyuzhniy and
others' interpretations, but I feel that Rich and Myszka use this study to express the opinion that only simple 1:1
interactions are worth a study. In an earlier review Rich and Myszka stated that If the model does not fit, then Step 2
should be to go back and redesign the experiment to improve the quality of the data. Step 2 is not to start fitting with
complex models (p. 365 in Rich and Myszka, [2008]). Whether this is possible must, of course, critically depend on
the biology and chemistry of the system under scrutiny. I cannot agree that sensorgrams not following single
exponentials imply that the study is in any way flawed. It is now widely appreciated that biologically relevant proteinligand interactions are not following a simple 1:1 interaction with implications both for molecular biology and drug
development (Mandell and Kortemme, [2009]). For example, heterogeneous ligand interactions are probably likely to
explain observations on the important functions of at least certain types of proteins, e.g., integrins (Vorup-Jensen et al.,
[2005]) and other membrane-bound receptors in the immune system (Collins et al., [2002]). Significant progress has
been made by Peter Schuck and colleagues in dealing with the resulting sensorgrams in an unambiguous way (Svitel
et al., [2003]; Svitel et al., [2007]). Furthermore, I am not certain that, for instance, avoiding avidity binding in studies of
antibody-epitope interactions by immobilizing antibodies on the sensor chip surface would capture all the relevant
aspects of antibodies binding certain antigens. This is particularly true for large molecules such as IgM, where we and
other have suggested that conformational changes in connection with avidity binding could form an important part of
the biology of this molecule (Pedersen et al., [2010]). In our view this is a direct consequence of the large dimensions
of the molecule and a similar line of thinking could probably be extended in the case of other large proteins. To study
epitope-IgM binding with biosensors is consequently from the outset expected to be difficult and not easily reduced to a
simpler approach if the aim is to understand how intact IgM binds. A role for biosensors in such investigations would be
interesting, but, with Rich's and Myszka's writing in mind, critically dependent on the development of appropriate
models for the analysis. I agree with an editorial (Van Regenmortel, [2002]) in The Journal of Molecular Recognition
that the paradigmatic statement structure determines function should more likely be replaced by binding
determines function. However, it must then follow that we are permitted to study and report on even complex binding
without any request to reduce such systems to uninformative simplicity.
In conclusion, it seems to me that the efforts of Rich and Myszka were not well spent. Their criteria for grading the
literature on optical biosensors are ambiguous and a closer read of papers scorned for the way SPR was used shows
that the studies have few, if any, significant errors. In their evaluation of the use of SPR biosensors, Rich and Myszka
apparently disregard other experimental evidence not involving biosensors. However, with Gehman and others' study
as a striking example, this can cause a misleading interpretation of what use is made of the data from the biosensor
and whether such use is appropriate or not. Experimental data that to Rich and Myszka are so weirdly shaped it's
hard to tell what is going on (p. 17 in (Rich and Myszka, [2010])) could perhaps make sense to others with a better
capacity for coping with complexity or, quite simply, with a biological insight on the questions studied. I always liked to
think that Rich and Myszka actually read the studies tabulated in their reviews, but I am no longer certain that this is the
case. This is a disappointment in a world of science, which is already flooded with superficial analyses of scientific
value and impact. However, it is not necessary to read thousands of papers to arrive at the conclusion that optical
biosensors now reveal the complexity of macromolecular interactions, which beforehand was lost with more crude
methodologies (Price and Dwek, [1974]). Efforts to unravel these phenomena not only deserve to be published but also
require a more substantial treatment than offered by Rebecca L. Rich and David G. Myszka.
REFERENCES
Biverstahl H, Lind J, Bodor A, Maler L. 2009. Biophysical studies of the membrane location of the voltage-gated sensors in the
HsapBK and KvAP K(+) channels. Biochim. Biophys. Acta 1788(9): 1976-1986. Links
Collins AV, Brodie DW, Gilbert RJ, Iaboni A, Manso-Sancho R, Walse B, Stuart DI, van der Merwe PA, Davis SJ. 2002. The
interaction properties of costimulatory molecules revisited. Immunity 17(2): 201-210. Links
Dennison SR, Harris F, Phoenix DA. 2009. A study on the importance of phenylalanine for aurein functionality. Protein Pept.
Lett. 16(12): 1455-1458. Links
Gehman JD, Luc F, Hall K, Lee TH, Boland MP, Pukala TL, Bowie JH, Aguilar MI, Separovic F. 2008. Effect of antimicrobial
peptides from Australian tree frogs on anionic phospholipid membranes. Biochemistry 47(33): 8557-8565. Links
Kalyuzhniy O, Di Paolo NC, Silvestry M, Hofherr SE, Barry MA, Stewart PL, Shayakhmetov DM. 2008. Adenovirus serotype 5
hexon is critical for virus infection of hepatocytes in vivo. Proc. Natl. Acad. Sci. U.S.A. 105(14): 5483-5488. Links
Mandell DJ, Kortemme T. 2009. Computer-aided design of functional protein interactions. Nat. Chem. Biol. 5(11): 797-807.
Links
Nilsson SK, Christensen S, Raarup MK, Ryan RO, Nielsen MS, Olivecrona G. 2008. Endocytosis of apolipoprotein A-V by
members of the low density lipoprotein receptor and the VPS10p domain receptor families. J. Biol. Chem. 283(38): 25920-25927.
Links
Pedersen MB, Zhou X, Larsen EK, Sorensen US, Kjems J, Nygaard JV, Nyengaard JR, Meyer RL, Boesen T, Vorup-Jensen T.
2010. Curvature of Synthetic and Natural Surfaces Is an Important Target Feature in Classical Pathway Complement Activation.
J. Immunol. 184(4): 1931-1945. Links
Price NC, Dwek RA. 1974. Principles and Problems in Physical Chemistry for Biochemists. Clarendon Press: Oxford.
Rich RL, Myszka DG. 2000. Survey of the 1999 surface plasmon resonance biosensor literature. J. Mol. Recognit. 13(6): 388-407.
Links
Rich RL, Myszka DG. 2008. Survey of the year 2007 commercial optical biosensor literature. J. Mol. Recognit. 21(6): 355-400.
Links
Rich RL, Myszka DG. 2010. Grading the commercial optical biosensor literature-Class of 2008: The Mighty Binders . J. Mol.
Recognit. 23(1): 1-64. Links
Svitel J, Balbo A, Mariuzza RA, Gonzales NR, Schuck P. 2003. Combined affinity and rate constant distributions of ligand
populations from experimental surface binding kinetics and equilibria. Biophys. J. 84(6): 4062-4077. Links
Svitel J, Boukari H, Van Ryk D, Willson RC, Schuck P. 2007. Probing the functional heterogeneity of surface binding sites by
analysis of experimental binding traces and the effect of mass transport limitation. Biophys. J. 92(5): 1742-1758. Links
Van Regenmortel MH. 2002. A paradigm shift is needed in proteomics: structure determines function should be replaced by
binding determines function . J. Mol. Recognit. 15(6): 349-351. Links
Vorup-Jensen T, Carman CV, Shimaoka M, Schuck P, Svitel J, Springer TA. 2005. Exposure of acidic residues as a danger signal
for recognition of fibrinogen and other macromolecules by integrin alphaXbeta2. Proc. Natl. Acad. Sci. U.S.A. 102(5): 1614-1619.
Links