Sep 192011

Given that videogames are often demonized by research (and “research”) blaming them for everything from rudeness to the epidemic of youth violence, gamers often take a great deal of cheer from research attaching positive outcomes to videogame play. One such article that recently attracted some attention was work suggesting that playing videogames could correct amblyopia (often called “lazy eye”) in adults (1). Of course, given how negative results get oversold, it’s worth asking whether these have been, too. The paper appeared in the open-access journal PLoS Biology, so let’s open it up and take a look.

The fundamental problem that the authors are out to solve is that, while amblyopia can generally be corrected if it is treated in childhood, success tends to be rarer in adults. Knowing that video games have proven useful in improving adults’ abilities to perform a wide variety of visual tasks, these researchers decided to ask whether they could help treat amblyopia.

Figure 1 shows their experimental design. First they screened and assessed a group of adults with amblyopia. Then they divided these individuals into three groups. One group (10 individuals) played a total of 40 hours of Medal of Honor: Pacific Assault with the normal eye patched. An additional 3 individuals were assigned to a group that played SimCity Societies for an equal amount of time (it is unknown whether the author’s controlled for Societies‘ well-known liberal bias), again with the normal eye patched. The final seven individuals were given twenty hours of ordinary visual challenges (watching movies, reading, etc.) with the normal eye patched (occlusion therapy or OT), in order to ensure that patching alone wasn’t causing any observed improvements. Most individuals from the last two groups, following an intermediate assessment, then went on to play 40 hours of MOH.

As the authors note, there are several limitations to this study immediately apparent. The sample size was small, individuals were not assigned to groups randomly, and both participants and researchers knew what kind of treatment they were getting. This does not mean we should disregard the results. However, they do need to be taken with a grain of salt until the findings can be replicated in a larger sample.

And there is good cause to try to replicate these findings. Figure 2 is, unfortunately, something of a symbol party (the symbols and colors identify individual subjects by their type of amblyopia), so we’re better off focusing only on panel D, at lower right. The first item in panel D is a logMAR chart, used to measure visual acuity, and it probably looks familiar to you. Each line on the chart represents 0.1 logMAR units, and as you can see, the lower the score, the better your vision. The panel to the right of that shows the averaged data from all twenty individuals after OT and videogame therapy (VG). Here they are showing the percentage improvement in acuity in crowded conditions (the whole chart) or in isolation (a single letter). OT did not produce any improvement in acuity, while 40 hours of VG therapy produced an average 30% improvement in acuity. The other two graphs here indicate that improvement in acuity was unrelated to baseline acuity, and that the crowding index (the loss of acuity due to the presence of other letters) did not change substantially due to therapy.

This is a critical figure because, as the authors state, “reduced visual acuity is the sine qua non of amblyopia.” Substantial improvements in acuity, therefore, represent a major goal of therapy. Perceptual learning, in which participants make subtle visual judgments using their amblyopic eye, has been shown to improve acuity in adult amblyopes as well. If videogames can produce a comparable improvement, however, they may prove just as efficacious because they encourage therapy (=play) through fun.

Panels A-C of Figure 2 show the raw results and percentage improvements for each individual group. Two additional points are worth noting. Panel B shows that the 20h of occlusion therapy were ineffective, but the subsequent 40h of MOH improved acuity in all continuing individuals. However, it should be noted that while the gaming took place at the research location, the occlusion therapy was done on the individual’s own time and self-reported. This study therefore does not control for the benefits of a monitored and enforced eye-exercise regimen.

Panel C is also of interest. Although the group here is small (and the data correspondingly noisy), it appears that their acuity was improved by both SimCity and MOH. This was somewhat unexpected, because in the past positive visual effects produced by action video games have not been replicated by non-action games. Understanding why that’s not the case here may help provide some additional insights into the mechanisms by which games improve acuity in these patients. I haven’t played SimCity Societies, but having played previous SimCity iterations I know that these games often require the player to integrate a variety of visual information (traffic flow, electricity, dynamic economies) simultaneously, which may underlie the observation. Had these subjects actually played videogame chess, their improvement might have been less.

The authors went on to test the subjects’ vision in various ways. Figure 3 shows a test of positional acuity, and is rather badly made, but gets the point across that positional acuity (assessed using the funky little chart in panel A) improved in the game-playing group (panel B). This included both increases in “sampling efficiency”, related to a fitted number of correct positions extracted (out of 8) (panel C and E-SB2) and decreases in “internal noise”, or the degree to which the individual’s own eyes interfere with his assessment of position (panel D and E – SA5). The results in panel E compare improvements in efficiency and internal noise, with the three labeled graphs comparing results in the non-amblyopic eye (NAE) to the amblyopic eye (AE) before and after videogame treatment.

The authors also decided to test the effect of the games on spatial attention, as they report in Figure 4, by briefly showing the subjects a field of dots (at a size where they could be easily seen), followed by a checkerboard pattern and asking them to report the number of dots seen (panel A). Not all the individuals had an appreciable difference between the non-amblyopic eye and the amblyopic eye prior to the VG treatment (panel B). However, the degree of improvement in spatial attention tended to be greater the worse the initial condition was (panel C), including for SimCity players (symbols surrounded by dotted circles). For the worst-off subjects (dotted circle in panel B), significant improvements in accuracy and response time were observed (panel D-F).

Finally, the authors tested the stereovision of some subjects using a standard test (Figure 5). Again, substantial improvements were noted in all those tested (which excluded subjects with strabismus), to the degree that some of them were effectively cured.

These results show that playing video games produced dramatic improvements in vision for adults with amblyopia by a variety of measures. However, this study had many limitations, and nobody should go around prescribing (or self-prescribing) videogames as amblyopia therapy just yet. The sample size here was very small, and because of the way groups were assigned the various populations differed in non-trivial ways (the MOH group, for instance, was younger and more male than the others). The conditions for occlusion therapy were very different from those used in the videogame therapy, which could have contributed to the different outcomes. Even if a more comprehensive trial shows similar results, more work will be necessary to identify the best course of treatment, which I note is unlikely to take the form of a 24-hour Modern Warfare 3 binge fueled by Bawls and pizza.

That said, these results appear to justify a larger, more complete study, which we can certainly hope to see in a few years from these authors.

1) Li, R., Ngo, C., Nguyen, J., & Levi, D. (2011). Video-Game Play Induces Plasticity in the Visual System of Adults with Amblyopia PLoS Biology, 9 (8) DOI: 10.1371/journal.pbio.1001135

Apr 122011

ResearchBlogging.orgThe classic neuropathological hallmarks of Alzheimer’s disease are the appearance of amyloid plaques composed primarily of amyloid beta (Aβ) peptides, and neurofibrillary tangles composed mainly of hyperphosphorylated tau protein. For many years, research into treatments for Alzheimer’s disease proceeded on the hypothesis that the plaques were toxic to the surrounding neurons. More recently, however, evidence has shown that soluble Aβ oligomers may be the primary toxic species. A recent paper in Proceedings of the National Academy of Sciences supports this hypothesis by showing that Aβ oligomers isolated from the brains of Alzheimer’s sufferers cause neuronal degradation and improper phosphorylation of tau (1). This paper is open access, so open it upand read along.

Jin et al. isolated dimers of Aβ from homogenates of human brains from Alzheimer’s patients. Dimers were separated from monomers and higher-order oligomers by size-exclusion chromatography in the presence of a strong detergent that typically breaks up folded proteins and repeating aggregates. This separates the dimers and higher oligomers from each other, and also dissociated weakly-interacting peptides (due to the effects of the detergent). As you can see from Fig. 1A, this produced fractions that contained either detergent-stable Aβ dimers (AD-TBS) or normal cortical proteins (cont-TBS) in identical solution conditions. They also created synthetic dimers by mutating Aβ to contain a cysteine that could form a covalent linkage between peptides (Aβ40S26C). They then used these various materials to treat primary cultures of neurons (that is, neurons that were obtained by directly harvesting them from an animal), with the dimers reaching a final concentration 0.5 nM in the growth medium.

Fig 1B establishes that, among the materials studied here, Aβ dimers are uniquely responsible for the appearance of tau “beads” along the neurites of the cultured cells after 3 days (the widespread dots in the final column of images). This effect is quantified in 1C, which shows that the dimer-containing fractions produced a dramatic increase in this clumping. According to the authors, these easily-visible clumps are only one symptom of widespread problems with the cells’ cytoskeletons. This sort of cytoskeletal trouble is expected because tau’s function is to stabilize and assist in the formation of microtubules from tubulin. The upshot of this figure is that continuous exposure to Aβ dimers (Fig 1D establishes that the dimers persist through the treatment period) appears to cause some sort of trouble with tau, which may reflect the incipient formation of the famous tangles.

The natural follow-up question is whether tau is necessary for this cytoskeletal derangement. The fact that the cultured neurons must mature, with a correlated increase in tau expression, for Aβ dimers to have an effect suggests that it must be. To check this, the authors used RNA inhibition to knock down tau levels. Fig 2A demonstrates that tau, but not tubulin, expression was altered using the tau-specific RNAi (but not the scrambled cont-RNAi). The cytoskeletal damage caused by both the natural dimer and the Aβ40S26C synthetic dimer were suppressed by tau-RNAi (Fig. 2B). At least at this timescale, it therefore appears that normal tau expression levels are necessary for this toxic effect of Aβ dimers. However, as tau in neurofibrillary tangles never breaks down, it seems like a longer exposure to Aβ under these conditions should produce similar toxic effects eventually.

The complementary experiment, is shown in Fig. 3, using a hybrid construct where human tau was fused to a fluorescent protein. As you can see from these images, under control conditions (columns labeled EGFP), cells treated with Aβ monomers and dimers have only subtle differences after two days, and beading is only evident after three days of treatment. When tau is overexpressed (columns labeled tau-EYFP), the cytoskeletal issues are obvious a day earlier. The tau-EYFP appears to be distributed in the same places as normal tau (fourth row), so the EYFP tag probably isn’t responsible for this effect, and the normal behavior of monomer-treated cells is reassuring. However, the EYFP tag may make tau more susceptible to some kind of dysregulation. Because this experiment both increases the total amount of tau and introduces the human protein, the reason for the enhanced susceptibility is difficult to determine. A control experiment in which rat tau-EYFP was expressed in the same construct would have been very helpful in clarifying this point.

As I mentioned above the formation of the tangles is associated with tau becoming highly phosphorylated. Jin et al. therefore made an effort to confirm that this was happening in their cultured cells, using antibodies that would recognize some specific sites in the tau protein that receive phosphate tags. Fig. 4 summarizes the results, indicating that human tau expressed in rat neurons becomes highly phosphorylated at serines 202, 205, and 262. For some reason, the endogenous rat tau did not become significantly phosphorylated at S262; this may have something to do with the apparently enhanced toxicity of Aβ dimers in the presence of human tau.

The paper’s final figure tests whether antibodies directed against specific sites in Aβ can prevent the observed cytoskeletal degradation. They found that two antibodies that bound to the N-terminus of Aβ significantly suppressed the effect of the dimers over the three-day timespan (fourth and fifth columns of A). However, an antibody directed towards the C-terminus of Aβ42 did not have much effect. Fig. 5C suggests that this is because this antibody simply didn’t bind to much of the Aβ in solution, either because most of the isoforms are shorter or because the C-terminus is protected in some way.

These results clearly link cytoskeletal disruptions caused by tau to the presence of soluble Aβ dimers, linking the two well-known pathological hallmarks of Alzheimer’s disease. That soluble oligomers, rather than fibrils, were responsible for the effect does not necessarily prove that the plaques aren’t important, for two reasons. The first is that, while these results clearly demonstrate dysregulation and aggregation of tau, true neurofibrillary tangles did not appear, and until we can assemble a full chain of events leading from beads to tangles the case, though strong, is still unproven. Secondly, as I’ve discussed previously, research has shown that the plaques can release soluble oligomers into the surrounding neural tissue and will therefore serve as reservoirs of toxic protein even if the fibrils themselves are completely inert.

That Aβ dimers can derange tau regulation in cultured neurons is not a new finding; similar results were reported last year using synthetic dimers (2). Zempel et al.‘s experiments used Aβ concentrations up to 5 μM, but Jin et al. show that naturally-obtained dimers have toxic effects at much, much lower concentrations. As the Zempel et al. paper suggests (consistent with much previous work), dysregulation of calcium levels caused by Aβ oligomers may be how they cause these effects. It is not presently clear why natural oligomers should be four orders of magnitude more potent than the various kinds of synthetic dimers at causing the effect; an understanding of this difference may be crucial in developing a suite of effective treatments for the disease.

1) Jin, M., Shepardson, N., Yang, T., Chen, G., Walsh, D., & Selkoe, D. (2011). Soluble amyloid β-protein dimers isolated from Alzheimer cortex directly induce Tau hyperphosphorylation and neuritic degeneration. Proceedings of the National Academy of Sciences DOI: 10.1073/pnas.1017033108 OPEN ACCESS

2) Zempel, H., Thies, E., Mandelkow, E., & Mandelkow, E. (2010). Aβ Oligomers Cause Localized Ca2+ Elevation, Missorting of Endogenous Tau into Dendrites, Tau Phosphorylation, and Destruction of Microtubules and Spines. Journal of Neuroscience, 30 (36), 11938-11950 DOI: 10.1523/JNEUROSCI.2357-10.2010

Oct 112010

ResearchBlogging.orgClostridium difficile is an intestinal pathogen that causes diarrhea in hospitals and other healthcare settings (including nursing homes). Present as a commensal bacterium in a significant fraction of the population, C. difficile is usually rather harmless, its numbers suppressed by competition with the intestinal flora. When its competitors are decimated by antibiotics, however, C. difficile flourishes, releasing toxins that cause inflammation and diarrhea, which can be dangerous because the individuals suffering these effects are often already ill. There has been conflicting information, however, as to which of C. difficile‘s toxins are necessary to cause disease. A paper in the recent Nature (1) aims to resolve the question.

The two best-characterized C. difficile toxins (TcdA and TcdB) have the same general arrangement and function (and ~45% identical AA sequence). An N-terminal glucosylating domain attacks the cytoskeleton of host cells by inactivating Rho GTPases, a C-terminal domain mediates binding and uptake by the host cells, and a protease domain in the middle releases the glucosylating domain to do its work. Since these proteins appear to serve redundant functions, one might expect that both would support virulence. However, preceding work in the field has variously identified TcdA or TcdB as a key virulence factor (2,3). Differences in methodology and materials have contributed to the confusion, in part because different kinds of cells seem to be more or less susceptible to particular toxins, and different strains of C. difficile might have different behaviors.

Kuehne et al. aim to relieve some of the confusion by removing a subset of these confounding factors. In a single strain of C. difficile they inactivated the genes for either TcdA, TcdB, or both by inserting introns into them. An intron would be no problem for a eukaryote, but bacteria can’t handle them, so this has the effect of eliminating the expression of the gene. They then tested the toxin mixtures shed by the bacteria against cultured human and monkey cells. As expected, A-B- bacteria (with both toxins knocked out) showed no toxicity towards the cells, but A-B+ and B+A- variants were toxic towards both kinds of cells to roughly the same degree. This suggests that both toxins are sufficient for virulence.

This implication was largely backed up by a subsequent experiment in hamsters. The animals were dosed with an antibiotic and then infected with C. difficile spores of a single strain. Colonization occurred (in every case but one) within three days. The hamsters infected with A-B- C. difficile remained asymptomatic until the end of the experiment, but the recipients of the other strains all perished within a week. The A+B- group survived somewhat longer, but not dramatically so; again, this supports the interpretation that both proteins are sufficient for virulence.

This contrasts with an earlier study published in Nature (2) where it was shown that deletion of the B toxin protected hamsters from C. difficile-associated disease, using very similar protocols. Kuehne et al. attribute the differences in their results to the hamsters or genetic variation in the C. difficile strains used. While the virulence of the B- strain in this experiment was slightly attenuated, all colonized hamsters still died in relatively short order, and in human beings the situation might well be reversed, since cultured human cells are more vulnerable to toxin A.

The results of Kuehne et al. largely agree with earlier experiments (3) and with what one would naturally expect of two very similar toxins being released by the same organism. While susceptibility to a particular toxin may vary with characteristics of the host species or cell type, it seems likely that both toxins are capable of supporting virulence. While it is to be hoped that additional research will clarify the reasons for the discrepancy between these two experiments, efforts to treat C. difficile-associated disease by attacking the toxins should proceed with the assumption that both must inactivated. Thanks to their functional and sequence similarity this will hopefully not be too much of a complication.

1. Kuehne, S., Cartman, S., Heap, J., Kelly, M., Cockayne, A., & Minton, N. (2010). The role of toxin A and toxin B in Clostridium difficile infection Nature, 467 (7316), 711-713 DOI: 10.1038/nature09397

2. Lyras, D., O’Connor, J., Howarth, P., Sambol, S., Carter, G., Phumoonna, T., Poon, R., Adams, V., Vedantam, G., Johnson, S., Gerding, D., & Rood, J. (2009). Toxin B is essential for virulence of Clostridium difficile Nature, 458 (7242), 1176-1179 PMCID: PMC2679968 OPEN ACCESS

3. Voth, D., & Ballard, J. (2005). Clostridium difficile Toxins: Mechanism of Action and Role in Disease Clinical Microbiology Reviews, 18 (2), 247-263 PMCID: PMC1082799 OPEN ACCESS

Aug 262010
ResearchBlogging.orgIf you’re going to study the role an enzyme plays in a biological pathway, it’s often useful to “kill” it with a mutation. For example, the proline cis-trans isomerase cyclophilin A (CypA) needs a particular arginine residue for its chemistry, so mutations that remove or alter that functional group, like R55K and R55A, should destroy the protein’s function and have effects on the related pathways that help illustrate its role. The hydrophobic pocket it uses to bind substrates is made by residues like H126, F113, and W121. Growing or shrinking those residues should alter the shape of the pocket and change binding or activity, leaving the enzyme “dead”.

Using model reactions and various binding assays, researchers have previously examined a number of these mutants (4,7) and found that they diminish isomerase activity and alter inhibition. However, a detailed study of the effects of the mutations on CypA’s catalytic cycle has not been performed. Former Kern lab members Daryl Bosco (now a professor at UMass Medical) and Elan Eisenmesser (now at UCHSC) examined these mutants in greater detail to see how they really behaved. I also contributed some data at the last minute, when the third reviewer requested we study an additional mutant, prompting a scene that I promise was not too much like that Downfall parody. In every case we found that these enzymes, although significantly impaired, weren’t as dead as they had seemed.

You had me at “dunno”

CAN bound to CypA, from PDB structure 1AK4 (5).
CypA residues are labeled in black, CAN residues in red.

One key aspect of this work is that it involves a physiological substrate of CypA, namely the N-terminal domain of the HIV-1 capsid protein (CAN). Mature HIV-1 virions contain CypA that is bound to proline 90 of CAN. The absence of CypA dramatically reduces their ability to infect their target cells, which we know from experiments with mutant CA proteins as well as ones involving the CypA inhibitor cyclosporin (3). What we don’t know about the system is exactly what CypA does for HIV-1. The crystal structure (right) of CAN in complex with CypA appears to only capture the trans isomer configuration (5), but for reasons I have discussed previously on this blog, that’s not particularly informative. We know, largely from Daryl and Elan’s previous research on the system (2), that when CAN is floating free in solution CypA will catalyze isomerization, but in the context of a fully assembled capsid that situation could conceivably change.

This leaves us with three possibilities for CypA’s function in the capsid. Catalysis of cis-trans isomerization of the proline bond could be important. Or, maybe all capsid needs is for CypA to bind at P90, and catalysis is irrelevant. And perhaps neither of these functions matters and CypA just needs to be hanging around for some other reason. To address these possibilities, Saphire et al. performed an elegant series of experiments where they sneaked an engineered CypA protein into another part of the capsid by fusing it to a protein called Vpr. When they replaced the normal CypA sequence with a mutant (H126A) that was supposed to abrogate both binding and catalysis, HIV-1 could still infect CD4+ cells (6). But, how sure can we be that H126A, or any other mutant, is actually “dead”?

You can’t measure what you can’t see

The problem with proline isomerization, from a biochemist’s standpoint, is that it’s a difficult reaction to detect. While switching between isomerization states may have structurally significant effects, there’s no direct spectroscopic signal to tell you whether a proline bond is in the cis or trans conformation. Even if there was, most proteins have many prolines and so the signal of the bond you care about might be difficult to separate from the bonds you don’t.

You can get around this difficulty using a coupled reaction with a model substrate. CypA catalyzes the isomerization of tetrapeptides of the form AXPF pretty efficiently. As it turns out, sequences like this are also good substrates for the protease chymotrypsin, but there’s a catch. Chymotrypsin only cleaves substrates where the proline bond is in the trans configuration. So, what you can do is take a substrate like succinyl-Ala-Ala-Pro-Phe-p-nitroanilide, add a tiny amount of cyclophilin, and then dump in a huge amount of chymotrypsin. With enough chymotrypsin, the peptide that’s already trans will be cleaved before the solution stabilizes, causing a color change (due to the pNA) that can be measured with a conventional spectrophotometer. Then you can monitor the conversion of the remaining substrate from cis to trans, because there’s so much chymotrypsin that cleavage after isomerization is essentially instantaneous.

This works reasonably well, but it has some limitations. You’re stuck with a model peptide that may not behave very much like your particular protein substrate. You’re only following the cis-to-trans reaction, and even that comes with limited detail. Also, performing the experiment takes some careful work, because if you add too much of your CypA the reaction will end before the solution turbulence settles, and if you add too little, the intrinsic cis-trans isomerization will interfere with your catalytic measurement.

Although proline isomerization is a difficult reaction to follow by spectrophotometry, it’s actually quite convenient to assay by NMR. Because CypA catalyzes the reaction in both directions, it’s impossible to exhaust the substrate. The kinetics can therefore be measured at equilibrium using NOESY and ZZ-exchange experiments (2). Of course the experiment is limited by our ability to express isotopically-labeled substrate proteins, but provided we can do that and visualize the active site in our spectra, then we can observe catalysis of the native substrate. When you perform this experiment on these various “dead” forms of CypA using CAN as a substrate, it becomes evident they’re still active after all.

Night of the living “dead” enzymes

Panels B-G of Figure 3 in this paper directly show that every single one of the CypA mutants catalyzes CAN isomerization in solution (1). These spectra show peaks representing the chemical shifts of the nitrogen and hydrogen atoms of CAN‘s backbone amide groups in the presence of a small amount of CypA, so we are not looking at P90 directly. Fortunately, the chemical shift of the G89 amide is dependent on the isomerization state of P90. If the G89-P90 bond is in trans, G89 shows up as the large peak at lower right in these panels, but if the bond is in cis you get the small peak at upper left.

If you don’t wait very long between determining the 15N chemical shift (y-axis) and the 1H chemical shift (x-axis), you get something that looks like panel A. If, however, you pause between determining the 15N shift and the 1H shift, you get cross-peaks representing the portion of CAN proteins that started the experiment in trans and ended it in cis, or vice-versa. The presence of these cross-peaks in the CAN/CypA samples, and their absence in the CAN-only sample (panel A), proves that catalysis is occuring. I’ve blown up the figure for H126A on the right to make things a little clearer. In this case the cross-peaks were pretty weak, but still in evidence.

There was more variability when it came to affinity, the strength with which CypA binds the CAN substrate. I’ve shown a complete titration for H126A on the left. As you can see, progressive addition of H126A causes the free CAN peaks to disappear while the new bound CAN peak grows in. This behavior is characteristic of slow chemical exchange on the NMR timescale, and indicates a high-affinity binding interaction. WT CypA binds CAN with a KD of 13 µM, and H126A probably has similar affinity. Note also that the bound state has a single peak for each residue, while the free state of CAN has separate cis and trans peaks. This indicates that the cis and trans isomers are interconverting rapidly on the enzyme, and constitutes additional evidence that H126A CypA is catalytically active.

This pattern was not repeated for all the mutants, however. H126A and W121Y had affinity similar to WT, while R55A, R55K, and F113W had significantly higher KD (lower affinity). You can see this clearly from the titrations in Figure 5. For each of these mutants, adding CypA to CAN caused the CAN peaks to move around in the spectrum, rather than disappearing and reappearing (R55K had a mixture of behaviors because the NMR timescale also depends on chemical shift). This peak shifting is characteristic of fast chemical exchange on the NMR timescale and indicates relatively low affinity. This wasn’t the only change for those mutants.

Shifts in the rate-limiting step

An enzyme that doesn’t have very high affinity for its substrate isn’t necessarily in trouble. The NMR titrations of the R55A and R55K mutants indicate that their KDs are near 1 mM (Table 1). This is comparable to the affinity the WT protein has for the AAPF peptide, which gets catalyzed pretty efficiently. What does seem strange about this result is that the ZZ-exchange spectra are very similar.

The presence of single peaks for residues of CAN bound to WT suggests that the isomerization step is fast. Using relaxation-dispersion techniques, Daryl established that the net process rate (kct+ktc) for CAN on WT CypA was about 2200 /s. From the ZZ-exchange spectra we know that the total catalytic cycle goes a great deal slower (closer to 75 /s), from which we can deduce that isomerization is not the rate-limiting step. An analysis of the lineshapes suggested that the unbinding rate (koff) was about 45 /s, which is close enough to the catalytic rate to indicate that this step is rate-limiting.

But if koff is rate-limiting for this reaction, and the koff for R55A and R55K is dramatically increased (as it must be, with lower affinity), we ought to be seeing a higher rate in the ZZ-exchange experiments, or maybe even not seeing independent cross-peaks at all. How can this reaction be going slowly enough to be seen by this technique? As it turns out, the titrations hold the key. When CAN is saturated with R55A CypA, you can clearly see independent cis and trans peaks in the bound state (Figure 6, partially reproduced at right). This means that cis-trans interconversion on the enzyme has gotten much slower.

In fact, the presence of those two peaks means that we can use the ZZ-exchange experiment again, this time to determine the on-enzyme interconversion rate directly. The answer we get is about 20 /s, which is, within error, equivalent to the rate of the full cycle for this mutant. That means the rate-limiting step is no longer the unbinding of substrate, but rather the isomerization step itself. There’s only a minor change in overall catalytic efficiency, but this is the result of large changes in the rates of the individual steps that happen to cancel each other out.

Implications for the study of CypA-associated pathways

The best evidence available at the time supported the decisions Saphire et al. made in setting up their experiment. Previous work had clearly shown that an H126Q mutation of CypA significantly reduces the protein’s incorporation into virions (4,7). Saphire et al. made an H126A mutation on this basis and seemingly assumed that the activity would be similar (6). Unfortunately, the evidence from the NMR spectra is that H126A binds to the capsid protein perfectly well and also catalyzes its isomerization. This does not prove that the catalytic function of CypA is important for HIV-1 infectivity. However, on the basis of the existing experiments that possibility cannot yet be ruled out.

More broadly, these results demonstrate that claims about CypA’s role in biology cannot be based on mutant studies alone. The mutants discussed here alter, rather than abolish, CypA’s catalytic activity towards a biological substrate. Even those that appear not to bind the substrate in certain assays still display catalysis, because the strength of binding that is required for successful catalysis is considerably less than what is required for, say, a successful co-precipitation. The standard for assessing how a mutation has changed an enzyme’s behavior needs to be one that pays attention to the various steps of the reaction and how changes in particular rates can compensate one another. Experiments that rely on overexpression of a CypA mutant are particularly vulnerable to erroneous interpretation, because adding more enzyme is always an efficient way to compensate for a loss of activity and binding.

CypA and its related domains are very highly conserved across all vertebrates, yet its function was preserved even when apparently critical residues were dramatically altered by mutation. Our existing knowledge of protein sequences is limited to just a few examples from a relatively tiny number of species, and our structural and biochemical data encompass just a fraction of that. Assessments based on these databases are likely to underestimate the functionally viable sequence space. Descriptions of function based on model systems are also suspect. That goes for this system too — the findings about CypA activity made with respect to CAN are not necessarily any more generalizable than the chymotrypsin assay results. CypA can bind an enormous number of potential targets, and what is true for one may not be true for another. Whenever possible, catalytic activity and binding affinity ought to be verified directly on the substrate of interest. Otherwise, you might find that an enzyme you thought was dead is still stumbling along.


1) Bosco, D.A., Eisenmesser, E.Z., Clarkson, M.W., Wolf-Watz, M., Labeikovsky, W., Millet, O., & Kern, D. (2010). “Dissecting the Microscopic Steps of the Cyclophilin A Enzymatic Cycle on the biological substrate HIV-capsid by NMR” Journal of Molecular Biology DOI: 10.1016/j.jmb.2010.08.001

2) Bosco, D.A., Eisenmesser, E.Z., Pochapsky, S., Sunquist, W.I., and Kern, D. (2002) “Catalysis of cis/trans isomerization in native HIV-1 capsid by human cyclophilin A.” Proceedings of the National Academy of Sciences, 99(8), 5247-5252. DOI: 10.1073/pnas.082100499

3) Braaten, D., & Luban, J. (2001). “Cyclophilin A regulates HIV-1 infectivity, as demonstrated by gene targeting in human T cells” The EMBO Journal, 20 (6), 1300-1309 DOI: 10.1093/emboj/20.6.1300 OPEN ACCESS

4) Dorfman, T., Weimann, A., Borsetti, A., Walsh, C.T., & Göttlinger, H.G. (1997). “Active-site residues of cyclophilin A are crucial for its incorporation into human immunodeficiency virus type 1 virions.” Journal of Virology, 71 (9), 7110-3 PMCID: PMC192007 OPEN ACCESS

5) Gamble, T.R., Vajdos, F.F., Yoo, S., Worthylake, D.K., Houseweart, M., Sundquist, W.I., & Hill, C.P. (1996). “Crystal Structure of Human Cyclophilin A Bound to the Amino-Terminal Domain of HIV-1 Capsid” Cell, 87 (7), 1285-1294 DOI: 10.1016/S0092-8674(00)81823-1 OPEN ACCESS

6) Saphire, A.C.S., Bobardt, M.D., & Gallay, P.A. (2002). “trans-Complementation Rescue of Cyclophilin A-Deficient Viruses Reveals that the Requirement for Cyclophilin A in Human Immunodeficiency Virus Type 1 Replication Is Independent of Its Isomerase Activity” Journal of Virology, 76 (5), 2255-2262 DOI: 10.1128/jvi.76.5.2255-2262.2002 OPEN ACCESS

7) Zydowsky, L., Etzkorn, F., Chang, H., Ferguson, S., Stolz, L., Ho, S., & Walsh, C. (1992). “Active site mutants of human cyclophilin A separate peptidyl-prolyl isomerase activity from cyclosporin A binding and calcineurin inhibition” Protein Science, 1 (9), 1092-1099 PMCID: PMC2142182 OPEN ACCESS