Path to Becoming a Calorimeter Master Vol.10 Final Episode: Everything about Calorimeters – Professional Dialogue

by Malvern Panalytical, Thursday, 17th November 2016

In this episode, we invited Professor Fukada from Osaka Prefecture University and Professor Lee from the Institute for Protein Research, Osaka University to discuss various topics related to calorimeters. We talked about how they started their measurements with calorimeters, their ongoing research, and common questions from users.
You can download the list of references mentioned in the text.
▶Download References

1. What prompted you to start thermal measurements?

I have about 13 years of experience with calorimetry, initially self-taught from ITC. It began with research using solution NMR (nuclear magnetic resonance) spectroscopy. While investigating the binding of an enzyme to its partner using solution NMR, I discovered at the amino acid residue level that the enzyme’s fluctuation increased. Normally, we consider that protein structure becomes more rigid upon binding, right? Of course, the binding sites become rigid, but fluctuations appeared elsewhere. So from the equation ΔG = ΔH – TΔS, I thought that the system’s entropy increases as the protein becomes softer (increased fluctuation) upon binding. S, or entropy, reflects the system’s randomness, so by increasing ΔS, in this case, the conformational entropy, we could potentially lower ΔG (Gibbs free energy change), promoting binding. ΔH is enthalpy change, and T is temperature.
At that time, papers were beginning to emerge in America and Europe discussing increased protein fluctuation due to binding. This gave me confidence to propose the same idea, but at the time, other professors wouldn’t recognize it, saying that there couldn’t be fluctuation upon binding.
So, I delved into investigating this further and focused on the fact that ITC could produce ΔS. Fortunately, a MicroCal VP-ITC was available in my research lab at that time, so I started measurements using the manual. I’ve conducted ITC measurements at different temperatures, obtaining ΔH at each, and from the temperature dependence of ΔH, ΔCP (constant pressure heat capacity change) is obtained. Then, from ΔCP and the empirical equations of water entropy, the ΔS of dehydration could be determined. Using simplified equations, I could determine changes in conformational entropy from the simple calculations between ΔS from dehydration and the system’s total ΔS from ITC. As a result, I was able to prove, just like the NMR results, that the conformational entropy of the enzyme increased from ITC measurements and analysis, stabilizing the complex. Physical chemistry was a subject I favored since undergraduate, particularly entropy, as it illustrates the flow of nature.
I have been conducting thermal measurements since my thesis. My advisor was Professor Kadsutada Takahashi, who was the first to start calorimetry in the biochemical field in Japan. From there, I naturally progressed to calorimetry, and have been involved with it ever since.
Earlier, Professor Lee mentioned about not being recognized when they announced that binding results in fluctuations but actually, about 40 years ago, there were chemistry professors who said it’s odd for the entropy to be positive upon binding. Of course, if there are no influences of conformation entropy or hydration, entropy change would be negative.
Come to think of it, discussions on dehydration and protein folding started around that time, didn’t they? Entropy changes due to dehydration, for example. Professor Fukada’s work on the enthalpy change of buffer deprotonation is remarkable. (Reference 1)
Being a person from the agricultural department, I was criticized for conducting such research, with others saying, “This isn’t agricultural!”
However, because Professor Takahashi appreciated such research, that’s why I could pursue it. It was a lab that emphasized the foundation of biophysical chemistry, yet without his approval, I couldn’t have conducted the measurements.
I believe more of such fundamental research should be inquired into.
On the subject of buffers, it’s not true basics in the strict sense, but chemical researchers would not use such systems. I think the data I gather are fundamental data with practical significance.

2. Tell us about your recent research.

　

I am now focusing on the aggregation of proteins that cause various diseases. Diseases related to protein misfolding and aggregation are referred to as protein misfolding diseases. Aggregation due to incorrect protein folding can make one prone to diseases. Alzheimer’s, Parkinson’s, type II diabetes, and prion diseases are typical examples. When I began my research, reading related papers and reviews revealed more than 20 diseases related to protein misfolding and aggregation. Recently, that number has grown to over 40. Although there aren’t many patients for some of these diseases, more and more new aggregation-related diseases are coming to light, and it seems we can no longer ignore the phenomenon of aggregation.
Lately, I’ve been focusing on insulin research. Insulin remains stable when the pH is lowered to an acidic state and takes on a nearly native structure. However, as we increase the pH, dimers form hexamers, and it becomes complicated. In the absence of metal ions at pH 2, insulin exists as a monomer. When measured with DSC, gradually increasing the temperature, the native structure of course undergoes thermal denaturation, resulting in an extended structure. At this point, an endothermic reaction (upward) DSC peak is observed, and as the temperature is further increased, an enormous exothermic reaction (downward) DSC peak occurs. After that, the response returns to the baseline level.
After measurements, when the sample was taken out, it had turned into beautiful amyloid fibers. This reproducibility is well obtained, and similar reports exist from other groups. But I am examining other solvent conditions. Surprisingly, even when insulin aggregates, it can be measured with DSC. Attempts at measuring other aggregating reactions have struggled with reproducibility. Even introducing pre-formed amyloid fibers or aggregates into DSC measurements, the DSC baseline (Cp line) tends to drift down, producing non-reproducible data. Since the reason for this baseline drop is still unknown, it remains a challenge. As temperature increases, hydrophobic interactions grow stronger, leading to aggregates of thermally denatured proteins. Because aggregates are large, they tend to precipitate easily. Moreover, in DSC cells, they can stick tenaciously, possibly complicating DSC measurements. However, there is no unified viewpoint so far. Using VP-Capillary DSC suppresses aggregation to some extent, making it good for studying the thermal stability of globular proteins.
As a result, I started using ITC to study protein aggregation. Stirring the sample within the cell prevents aggregates from settling, and they don’t grow to a size where they’d settle. This enabled the measurement of dispersed aggregates in solution with reproducible ITC. A notable success example is the amyloid fibril formation of β2-microglobulin, which causes dialysis amyloidosis (Reference 2 Ikenoue and Lee et al. PNAS 2014). Similarly, for insulin, we’re conducting aggregation research using ITC. By skillfully combining ITC and DSC, I’m planning to thermodynamically study the interconversions among all possible structural states that insulin can adopt. Energy for folding, heat when transitioning from the native structure to 3D crystals, heat during amyloid fiber formation, heat when transforming from monomers to dimers or hexamers, and heat upon metal ion binding — I believe all can be measured with ITC and DSC.
Compiling such data could allow us to map out the thermodynamic energy landscape of insulin’s structural transformation and intermolecular interactions.
That’s fascinating. Insulin is a nostalgic sample for me. It was the first protein sample I used during graduate school. Back then, I would remove zinc from purchased samples, making the conditions monomeric with alkali. I measured the heat generated during the reduction of disulfide bonds in that insulin.
In those days, when modern calorimeters like we have now weren’t available, we sometimes used Dewar flasks and measured reaction heat under constant temperature with thermistor thermometers. We needed about 20 mL of sample back then.
It required a large amount indeed. Even today, I’m occasionally asked to perform highly valuable sample measurements. When using purchased proteins or peptides without expression systems, it costs a great deal per measurement, so it would be ideal if less sample were needed.

3. Development of ITC from a user perspective

I primarily used VP-ITC for my measurements. One time, when I didn’t have much sample, I went to Professor Shirakawa at Kyoto University and borrowed the iTC200 to try it out. The reduced sample volume, fast temperature equilibration, and short measurement time were indeed impressive. I am currently experimenting with the PEAQ-ITC, released after the iTC200. While each MicroCal ITC system has its own advantages and disadvantages, the reduced sample volume remains appealing. When the heat obtained is small, fitting wasn’t always great, and I had to increase concentration and measure again. Nevertheless, compared to VP-ITC, which saves sample volume, I found it astonishingly quick in both temperature equilibration and measurement completion. I could conduct more than ten measurements a day, an amazing feat. During aggregate measurements, titrating 1M NaCl results in some very high dilution heats and large titration volumes, causing the VP-ITC response to saturate. However, PEAQ-ITC titration volume is minimal, enabling all these large dilution heats to be quantified, which personally was surprising to me. Plus, having reduced sample volume means reactions conclude quicker, making it tougher for aggregates to form during the shorter measurement time. Also, aggregation occurs probabilistically, so less volume implies reduced aggregation chance. Samples that would aggregate halfway using VP-ITC were surprisingly measured cleanly and swiftly under identical conditions with PEAQ-ITC.
I also conducted amyloid fiber formation measurements using PEAQ-ITC. The reaction heat of amyloid fiber formation was observed very cleanly. Yet, it yielded different results from VP-ITC, with varying lag times for amyloid fiber formation. Even if different ITC devices produce similar amyloid fibers, I realized once again this reaction is sensitive to stirring, cell volume, and the microtopography within the cell.
Moreover, PEAQ-ITC enables automatic cleaning of both cells and syringes, resulting in very clean surfaces. When measuring aggregates with VP-ITC, insufficient cleaning prevents reproducibility, as leftover aggregates impact subsequent data drastically. Cleaning thoroughly is essential, so for VP-ITC, cleaning is done manually, using surfactants, heating, and stirring rapidly for about an hour before performing additional cleaning with accessories. On the other hand, with PEAQ-ITC, simply clicking the auto-clean function achieves clean results, a surprise. However, after aggregate formation measurements, it is advisable to use the Soak feature, where detergent is added to the cell for heating, in addition to mere cleaning. PEAQ-ITC is also very sensitive to errors. The system displays errors if the cleaning water flow rate is low or the pump pressure is low. Thus, it prevents inadequate cleaning by addressing recirculation issues of cleaning solutions.
 

4. Development of DSC from a user perspective

　

Compared to ITC, it seems DSC development hasn’t progressed as much. However, I believe significant advancements have been made with respect to sample concentration for DSC measurements. While I’m using VP-DSC now, older papers show sample concentrations taken down to 0.1 mg/mL, when papers from the 90s — not that old — cite 10 mg/mL. The sensitivity of devices has greatly improved with VP-DSC.
For bio-applications, it was 1 mg/mL from that era. 10 mg/mL is standard DSC. Those are for measuring powders and solids, placing samples in pans within a unit, different from microcalorimeters. The distinction between standard DSC and microcalorimeters is whether dilute aqueous systems can be measured. That specification was developed by Professor Privalov (Johns Hopkins Univ.) with the DASM calorimeter. It was first constructed in the late 1960s, with publications using DASM by Professor Privalov beginning in the 70s. The first MicroCal DSC system, the MC-1, was developed in 1977.
 
 VP-DSC, developed in 1996, used coin-shaped cells prior to the capillary-type VP-Capillary DSC in the 21st century.
The MC-2 which I borrowed and used needed large sample volumes, with cell capacity about 1 mL, requiring a total of about 1.5 mL. Back then, protein concentrations above 1 mg/mL were required. This standard for 1 mg/mL concentration in DSC measurements persisted for a while until the VP-DSC reduced required concentrations. Previously, only a limited number of people performed thermal measurements. Judging from that, it’s significantly spread now, I feel.
Long ago, we used the handcrafted calorimeters. The spread increased as good commercial products emerged. In the US, the recognition was high long ago, but it only began spreading in Japan around the 1990s. It genuinely spread once VP-DSC became available.

5. Is calorimetry still a high barrier? What are ΔCP values like?

　　 　　 　　 　　

It’s true that with the advent of good commercial calorimeters, the number of people conducting measurements has increased, although those who are genuinely interested but don’t actually try it may still be few. In this sense, having a collection of data focusing on biological macromolecules would be useful, right?
There was a handbook on protein structural stability written in English about 15 years ago. (Reference 3)
However, no data has been compiled over these 15 years. In the last 15 years, protein thermal stability papers using DSC are declining from a recent trend perspective. Several researchers studying globular protein structural stability and protein folding (folding reactions) have shifted to protein misfolding and aggregation studies or applied research fields.
When it comes to DSC, many people primarily want to determine Tm (thermal denaturation temperature: the temperature where native and denatured protein are halfway). Especially structural analysis researchers now evaluate thermal stability by DSC alongside structural investigations. Creating thermally stable proteins is vital for industrial use, so seeing how thermal stability rises, and ideally evaluating thermodynamic quantities to discuss them, is of interest. But it’s quite challenging. Naturally, interest in ITC exists, yet ITC requires more sample compared to DSC. If differences are discernible, Tm and KD suffice, but some might be unsure how to prioritize the thermodynamic quantities obtainable from calorimetry.
ΔCP, for instance, characterizes proteins. It’s akin to each protein having its own inherent value, like individual personalities. Using ΔCP, we try to understand there are different protein “characters.” Chains of amino acids fold into rounded tertiary structures with function, and ΔCP is assumed to relate to the number of buried hydrophobic residues inside proteins. A person rich in fat might have a high ΔCP, metaphorically.
However, even native state proteins that don’t take three-dimensional structures, known as intrinsically disordered proteins, exist. These proteins generally have few hydrophobic residues and many hydrophilic residues. As such, intrinsically disordered proteins don’t neatly fold like globular proteins; they roam freely in aqueous solutions. Consequently, the ΔCP for intrinsically disordered proteins tends to be smaller than those of globular proteins. Potentially, they lack intrinsic structures. Research thoroughly examining ΔCP of intrinsically disordered proteins via DSC hasn’t been conducted yet.
I’ve measured the ΔCP for α-synuclein, the protein responsible for Parkinson’s disease, using DSC, though. As it doesn’t take 3D structures, α-synuclein exhibits no denaturation peak, displaying an identical thermogram to the buffer, effectively yielding a ΔCP near zero. Since intrinsically disordered proteins constitute over 30% of all eukaryotic proteins, identifying them by ΔCP holds importance. ΔCP aids in delineating and distinguishing between globular and intrinsically disordered proteins.
Obtaining thermodynamic values is very difficult. Particularly for proteins, proper samples are required. Precise concentrations are essential before any measurements.
When DSC measurements yield pristinely clear baselines, obtaining ΔCP at Tm without fitting analyses may not be bad. However, the reliability when obtaining ΔCP from a single DSC thermogram is limited, so for more precision, employing different temperatures in DSC, using each temperature’s Tm and denaturation ΔH to derive ΔCP may be better, I think.
 
Yet, it’s uncertain whether that’s the true ΔCP.
You’re right; such concerns exist. I suspect there are “assumptions.” I consider myself somewhat knowledgeable now, but protein denaturation is generally discussed in terms of a two-state model. During thermal upscaling with DSC, we define pre-peak as the native state and post-peak as the denatured state, though structurally I suspect these differ by temperature. Yet, ignoring them constitutes a two-state assumption. So even if derived, as you mentioned, whether such a value constitutes the actual ΔCP is unknown. In such a two-state denaturation assumption, values emerge in this vicinity. Professor Privalov investigated the temperature dependence of ΔCP in the pre-heat denatured native state also, perhaps focusing on natural structural differences by temperature.

6. Easily explained calorimetry-obtained parameters!

　 　　

Cp signifies the heat required to elevate a substance’s temperature by 1°C under constant pressure (ΔCp’s p).larger ΔCp implies more heat needed per degree.
For alternative expression, I equate it to the ability to withstand heat.
I currently teach physical chemistry of proteins, where I personify thermodynamic parameters like Gibbs free energy (G), entropy (S), and enthalpy (H). Students seem to delight in listening. To achieve a target, G must be reduced to render ΔG negative; entropy signifies personal freedom desires, while energy for perseverance to achieve it relates to enthalpy. Comparing affinity (molecular interactions) in relationships is similarly beneficial for ΔG. If ΔCp signifies endurance, feebly ill-tempered individuals (—heating easily) have low ΔCp—low endurance (heat-insensitive). Conversely, lingering provocatively (—heating persistently) pertains to high ΔCp—high tolerance.
In that regard, Professor Fukada exemplifies high ΔCp.
Is this complimenting or criticizing? (laughs)
 
Common terminologies of physical chemistry and thermodynamics tend to discourage early on. Thus, I carefully ensure comprehension and provide easy-to-understand explanations, frequently checking student reactions.
Regarding ITC, affinity is evident in results, but some struggle with ΔH and ΔS, hence the parameters are hard to discuss. ΔH signifies hydrogen bonding interactions; S, on the other hand, depicts degrees of freedom. I explain this to students. Hydrogen bond-driven bonds yield negative ΔH (exothermic). Like with a loved one, hold hands at first (bonding = exothermic); hand-holding means enslavement, resulting in negative S. When a special someone approaches your guardian accomplice, the watcher’s duty is relinquished, leading to freedom, much like hydration interactions. If hydrophobic interactions are significant, ΔH is positive (removing guards = energy expenditure = temperature drops = endothermic) and ΔS becomes positive (free to roam = increased freedom).
Indeed. But ΔH and ΔS are not always positively correlated due to more influencing factors, making for a complicated explanation. The apparent ΔH of intermolecular reactions measured includes protonation, deprotonation, molecular structural changes beyond pure reaction contributions, affecting ΔS and ΔG. Thus, avoiding a high barrier to beginners requires simplified step-by-step explanation. Learn KD first, followed with binding ratios. Understanding reaction as well as mutation-induced affinity and binding ratio variations underpins the importance of ΔH and ΔS. Soon, differences in binding modes emerge, revealing distinct modes even when affinity and binding ratio resemble wild-type, solely via ΔH and ΔS. Analysis further unveils results like explicating dual binding sites when a fit might reveal just one.

7. Crucial aspects for obtaining reliable data

　 　　

Assuming sample integrity—appropriately knowing the sample’s authenticity (purity, concentration)—data obtained advances reliability. Given data publication, accuracy is paramount. Consequently, I often repeat measurements whenever possible, intrigued by pure figures besides interpretations.
Agreed; samples are primordial. Ensuring reproducibility is equally vital. Juxtaposing different methodologies complements finding assessments. Particularly, ITC adeptly senses both strong and weak bindings. Contrarily, crystallography typically discerns strong bindings, underlining methodology comprehension in reproducibility judgments. ITC-detected weak interactions might escape crystallography detection; such cases match expectations. Adding crowding agents such as PEG in crystallization facilitates profound crowding effects and excluded volume phenomena —interactions promoting pursuits—not infrequently considered. Conducting ITC with crowders aligned as valid must sometimes be considered.
While performing measurements, attentiveness to cleanliness and washing the cell prevails. Always after measurements, cleansing with detergents occurs. If not actively pumping, a brief placement suffices. As baseline noise stems from cell contamination, care is taken against measurement in that situation.
Correct! Instructing students stringently on cell and syringe sanitation holds. Moreover, meticulous note-taking facilitates measurements. Memorizing measurement-solvent conditions in tandem with baseline DP values-patterns aids oversight. Such insights expose equipment conditions before obtaining unreliable data when equipment malfunctions arise.

8. References for data interpretation recommendations

Names like Professors Sturtevant (Yale Univ.), Privalov, Cooper (Univ. of Glasgow), Freire (Johns Hopkins Univ.), and Ladbury (Univ. of Leeds) sound when referring to thermodynamics of DSC and ITC. Sturtevant’s 1977 composition (Reference 4), even today, serves as a canonical publication. Specifically, entropy sections alongside heat capacity narrations continue captivating many researchers. While beginners may find it advanced, slightly experienced calorimetry-minded readers grasp its profoundness.
Indeed. Before delving into how to interpret obtained data, consulting such resources proves beneficial.
At my June 2016 Malvern webinar, concluding reference lists provide utility, recommended for new users.
 
 Moreover, Nature Protocols houses roughly two ITC-related papers (References 5, 6). Both explain measurement methods, analyses, and sample preparation with exceptional clarity, Freire notably featuring initially. Ladbury’s Biocalorimetry 2, published in 2004, could use a supplied third edition by now.
 
 Nonetheless, fostering beginner-friendly textbooks remains crucial. An ITC with structural thermodynamics connecting review offers valuable reference (Reference 6), writing on conformational entropy changes as well. Reading Malvern’s “Path to Calorimeter Master” series proves beneficial.

9. What measurements would you like to conduct in the future?

With strategic ideas, many endeavors are possible using ITC. Amongst these aspirations is examining protein folding. Through pH jump or solvent condition variations, I intend to directly observe the heat during protein folding. By titrating alkali into an acid-denatured protein solution within the cell to mutual pH, measuring the refolding heat would result. Essentially the reverse to DSC, observing the heat during folding.
Particularly, a triggered folding system with enzyme induction exists. When placing denatured proteins in the cell and administering minor enzyme titrations via syringe, folding might initiate, with the ensuing heat measurable. Furthermore, as previously mentioned, I’m engaged in method development and thermodynamic establishment of aggregation reactions in ITC. Additionally, I’m devising a novel assay method for inhibiting disease-inducing protein aggregation via ITC. I plan to showcase it in an upcoming Malvern webinar.
That sounds intriguing.
Recently, I attempted measuring E. coli with DSC.
A dozen years ago, I measured yeast. Upon heat shock, a massive peak emerged.
Impressive, Professor Fukada! In my case, concerning a peptide toxic to E. coli, I came across a paper showcasing DSC E. coli membrane damage amidst lysis, showing a peak emergence at lower temperatures with peptides. So, I gave it a shot, though the data wasn’t as clean as seen in papers. I’ll retry next time.
Reproducing bacterial measurements is difficult. Growth phases generate distinct signals; results vary despite similar preparations. Living organisms are indeed complex.
Understood. We can conduct enzyme reaction Michaelis-Menten analysis via ITC. Recent presentations from Spain suggest obtaining inter-molecular exchange rate constants (kon, koff). ITC progress is advantageous.
The following application notes can be downloaded.
Enzyme Measurement

Reaction Rate Constants

In Spain, favorable impressions towards kinetic analyses endure. Deemed “past,” kinetic efforts faded due to conducting irreversibility equilibrium analyses.
ITC-measured reactions aside, DSC accommodates kinetic analysis. Upon thermal denaturation followed by aggregate formation, irreversibility emerges. Overlaying DSC thermograms with thermal denaturation models like the Lumry-Erying model enables kinetic analyses and interpretations. We are performing such analyses extensively ourselves. In developing and fitting alternative reaction models, kinetic analyses feasibly extend to other reactions, as well.
Earlier, you mentioned determining reaction rate constants. How fast a reaction is measurable?
Plausibly, in terms of ITC seconds or longer, measurable constraints apply. Anything faster appears unreasonable. Yet, utilizing reaction constants at various temperatures allows activation energies to derive detailed reaction energy landscapes aligned with equilibrium results.
Ligand bindings likely exceed this speed, and perhaps not at second-scale, yet conformation changes and slower reactions already encompass such analyses among potential involvements.
 

【Post-Dialogue Notes】

Over the last 1.5 years, we concluded our Path to Calorimeter Master series with a discussion featuring Professor Fukada from Osaka Prefecture University and Professor Lee from the Institute for Protein Research, Osaka University.
We’ve continuously aimed to bring fundamentals of calorimeters to familiarity through this series; how did it resonate?
This dialogue ventured deeper, illustrating parameter interpretation guidance gained in measurements —offering examples portraying a more profound understanding than mere textbook familiarity potentially offers.
Moving forward, Malvern continues striving to provide research-advancing technologies and insights. Lastly, heartfelt thanks to Professor Fukada for series supervision and Professor Lee for webinar collaboration.
The list of references mentioned in the text can be downloaded from the “Download” button below.

Download

Path to Calorimeter Master Back Numbers

This article may have been translated automatically