Energy transfer at heterogeneous protein-protein interfaces to investigate the molecular behaviour in the crowding environment

Accepted Manuscript Energy transfer at heterogeneous protein-protein interfaces to investigate the molecular behaviour in the crowding environment

Chikashi Ota PII: DOI: Reference:

S1386-1425(16)30726-0 doi: 10.1016/j.saa.2016.12.010 SAA 14823

To appear in:

Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy

Received date: Revised date: Accepted date:

19 August 2016 11 December 2016 12 December 2016

Please cite this article as: Chikashi Ota , Energy transfer at heterogeneous protein-protein interfaces to investigate the molecular behaviour in the crowding environment. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Saa(2016), doi: 10.1016/j.saa.2016.12.010

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Energy

Transfer

at

heterogeneous

protein-protein

interfaces to investigate the molecular behaviour in the

PT

crowding environment

SC

Advanced R&D Center, HORIBA, Ltd., Minami-ku, Kyoto 601-8510, Japan

PT E

D

MA

NU

a

RI

Chikashi Otaa

Chikashi Ota

CE

Corresponding Author:

Advanced R&D Center,

AC

HORIBA, Ltd.,

2 Miyanohigashi, Kisshoin, Minami-ku, Kyoto 601-8510, Japan Tel: 81-75-325-5037 Fax: 81-75-315-3806 E-Mail: [email protected]

1

ACCEPTED MANUSCRIPT

Abstract Investigation of the behaviour of proteins in crowded environments is crucial for understanding the role of proteins in biological environments. In this study, the behaviour of bovine serum albumin (BSA) in crowded (highly concentrated)

PT

environments was investigated using time-resolved fluorescence spectroscopy as a

RI

model system. By using energy transfer as a molecular ruler, the crowding effect was

SC

clearly observed in the time resolved spectra. In addition, by using both time resolved anisotropy measurement and Raman spectroscopy, more detail insights from

NU

conformational and dynamic points of view were described. Consequently, it was revealed that in the highly concentrated solution, most of the BSA molecules are in the

MA

fast-reversible oligomeric state and the association at the “hard” and “soft” interfaces between protein surfaces occurred in a highly crowded environment with the aid of a

D

charge-charge and short-range attractive interface. From both the conformational and

PT E

dynamic aspects, the detail spectroscopic understanding of the behaviour of BSA in the

AC

CE

crowding environment was obtained.

Highlights

・Spectroscopic understanding of protein behaviour in crowded environments was obtained. ・Using energy transfer as a molecular ruler, the crowding effect was described. ・The crowding effect was clearly observed in the time resolved fluorescence spectra. ・BSA molecules were in the fast-reversible oligomeric state in high concentration. 2

ACCEPTED MANUSCRIPT

1. Introduction Proteins are one of the most important biomolecules for life, and each protein plays an important, unique role in a crowded environment such as a cell or other biological environments. Protein forms a unique conformation such as oligomerization or

PT

aggregation in the crowding environment. However, characterizing the state of proteins

RI

in such crowded environments is an important subject that has not been completely

SC

understood. Previous studies of crowding mainly focused on the excluded volume effect with the main conclusion that the crowding effects lead to more compact structure.1,2 In

NU

contrast, some recent reports have reported that the molecular behaviour of proteins in crowded environments is different from that in vitro (the ideal solution) and the effect of

MA

crowding in more realistic environments with protein crowders tends to destabilize rather than to stabilize native states.4-6 Deeper understanding of their behaviour may

D

provide the useful knowledge in the biomedical area such as the amyloidosis or in the

PT E

industry such as the quality problems due to the aggregation in the biopharmaceutical drug.

CE

Fluorescence spectroscopy is a useful method for investigating proteins.7-9 Especially, fluorescence energy transfer is a useful tool to assess the molecular

AC

behaviour in an ångström resolution since it is sensitive to the distance between molecules. In the crowding or highly concentrated environment, the molecules are packed together and the distance between them is almost the same size as the molecule, therefore, energy transfer could be useful for the study of the behaviour in the crowding environment. In proteins, three aromatic amino acids, namely, phenylalanine (Phe), tyrosine (Tyr) and tryptophan (Trp) are fluorescent. Since the fluorescence of Phe is relatively weak due to its low quantum yield, the Tyr-Trp or Trp-Trp energy transfer is 3

ACCEPTED MANUSCRIPT

common.10,11 Serum albumin is the most abundant protein in the bloodstream.10 Although the basic properties of BSA in dilute solutions (in vitro) have been investigated in detail, the behaviour of BSA in crowded environments has not been investigated thoroughly. In

PT

this study, the behaviour of BSA in a crowded environment was investigated using

RI

energy transfer by time-resolved fluorescence spectroscopy as a model system. In

SC

addition, time-resolved anisotropy measurement and Raman spectroscopy were also employed to investigate the behaviour in a different aspect. The spectroscopic

NU

understanding of the behaviour of BSA in a crowded environment was described in

2. Experimental Methods

D

2.1. Materials

MA

detail.

PT E

BSA (>98%) was purchased from Sigma Aldrich (Product no. A3983-109, St. Louis, MO, USA). A buffer solution of BSA was prepared by mixing

CE

tris(hydroxymethyl)aminomethane (CAS no. 1185-53-1, lot no. 35433, Nacalai Tesque, Kyoto, Japan) and 35% HCl (code 18320-15, 500 mL, lot no. V5P7712, Nacalai

AC

Tesque). The concentration of the buffer solution was 20 mM. The water for the buffer was purified with a Milli-Q laboratory water purifier (Millipore, Billerica, MA, USA). The final pH of the solutions was adjusted to 8.0 using a LAQUA F-72 pH meter (HORIBA, Kyoto, Japan).

2.2. Fluorescence spectroscopy Fluorescence spectra were measured in a quartz cuvette (Hellma Analytics, 4

ACCEPTED MANUSCRIPT

Müllheim, Germany) using a Delta Flex system (HORIBA Jobin Yvon IBH, Glasgow, Scotland). Excitation was accomplished at 280 nm using a picosecond laser diode, NanoLED N-280. The range of the emission wavelength was set to be from 305 to 500 nm. Fluorescence decays and time-resolved fluorescence spectra were measured using

PT

the time-correlated single photon counting (TCSPC) technique. A Delta Flex system

RI

equipped with a NanoLED N-280, producing <1.2 ns pulses at 280 nm with a repetition of 1 MHz, was used for measurements. The instrument response function (IRF, fwhm

SC

~140 ps) was measured using dilute colloidal silica, namely, LUDOX® (Sigma-Aldrich).

NU

The optical geometry is explained in supporting information 1. For a dilute solution of less than 1 mg/mL, right-angle geometry was employed, as shown in Figure S1(a). For a

MA

higher concentration, over 1 mg/mL, oblique incidence geometry was employed to expand the concentration region to the high concentration range, as shown in Figure

D

S1(b).11 The illuminated surface was set to be oriented about 30° from the incident

PT E

light to decrease the amount of the reflected light into the monochromator. This geometry will work well to avoid the re-absorption or multi-scattering effect. Test

CE

solutions of BSA were prepared at concentrations of 1.5, 7.4, 15, 74, 147 M and 0.74, 1.47 mM (0.1, 0.5, 1.0, 5.0, 10, 50 and 100 mg/mL). All measurements were performed

AC

at room temperature (~20°C). The fluorescence decays were analysed and fitted with multi-exponential functions, according to11 ,

(1)

where I(t) is the total fluorescence decay, and ai is the amplitude of each decay component. The average lifetime is calculated according to .

(2) 5

ACCEPTED MANUSCRIPT

The normalised concentration of each decay component (fi) is calculated as described below: .

(3)

PT

All the fluorescence decays were analysed using DAS 6.0 software (HORIBA Jobin Yvon IBH, Glasgow) by the iterative reconvolution method. The accuracy of the

RI

fitting result was evaluated using the reduced -square () value.

SC

Fluorescence anisotropy measurements were also performed using the Delta Flex system. The fluorescence anisotropy decay r(t) was determined by simultaneous

NU

analysis of the horizontal Ih(t) and vertical Iv(t) emission intensity components,

,

(4) (5)

D

.

MA

according to10

PT E

The anisotropy is given by

(6)

CE

.

2.3. Raman spectroscopy

AC

Raman spectra were measured with a LabRam HR Raman microscope equipped with the Labspec 6 software (HORIBA Scientific, Kyoto, Japan). The spectrograph of the system had the Czerny–Turner configuration and the focal length was 800 mm. Raman spectra were measured in the back-scattering geometry. A thermoelectrically cooled Synapse charge-coupled device camera (HORIBA, Ltd.) was used as the detector. The entrance slit of the spectrometer was set to 100 m. The dispersive element had a grating of 600 lines/mm, which provides a wavenumber resolution of ca. 2 cm–1.

6

ACCEPTED MANUSCRIPT

For the concentration-dependent measurements, the excitation was accomplished with the 532-nm line (20 mW at the sample surface). The scattered light was collected using a multipass cell holder (HORIBA Jobin Yvon) with a visible macro lens (focal length = 40 mm). A 115-QS micro cell (Hellma Analytics, Müllheim, Germany) was used for the

PT

micro cuvette. Test solutions of BSA were prepared at concentrations of 1.5×10-1,

RI

2.9×10-1, 5.9×10-1, 8.8×10-1, 1.2, 1.5 mM (10, 20, 40, 60, 80 and 100 mg/mL) for the

SC

comparison of the fluorescence measurement. All measurements were performed at room temperature (~20°C). The Raman shift was calibrated by means of the atomic

NU

emission lines of a neon lamp. Raman spectra were normalized at the maximum

MA

intensity in the measurement range.

2.4. Spectroscopic analysis

D

Principal components analysis (PCA) is a multivariate analytical method which is

PT E

useful for resolving the spectrum of a pure component from several types of spectra.12 This method has a unique property to detect the spectra of minute chemical species in

CE

local environments or under different interactions from those of dominant bulk species. The accuracy of the predictive ability of the model is evaluated by rank analysis using a

AC

reduced-eigenvalue (REV) plot. REV has the similar statistical meaning as the normal eigenvalue (EV), however, REV has more accuracy for the evaluation of minute eigenvalues because it takes into account the degree of freedom.11 A REV plot is calculated according to the following equation:

REV j 

EV j

N  j  1M  j  1 ,

(7)

where j is the factor level, and the parameters N and M represent the size of the

7

ACCEPTED MANUSCRIPT

spectral matrix (N × M). Of note is that the analysis of basis factors by use of eigenvalue analysis does not take into account concentration (intensity) information. Therefore, the cross-validation (CV) technique, which takes into account both spectral and intensity information, can be employed to cross-check the accuracy of the results of

PT

REV.

RI

Root-mean-square prediction error of cross-validation (RMSECV), which is a CV

SC

method with leave-one-out cross-validation,13 is also calculated according to the following equation: 

n

(c j  c j ) 2 j 1 n

NU

RMSECV 



,

(8)

MA

where ĉj is the predicted concentration, cj is the actual concentration, as determined by the reference method, and n is the number of samples used in the calibration model.

D

The accuracy of the factor analysis of PCA increases using both REV and RMSECV.

PT E

The classical least squares (CLS) method is the simplest technique to calculate the concentration (C) of each model species based on the pure-component spectra (K),

CE

according to the following equation:12 (9)

AC

A = CK + R.

Here, A is the actual spectra, and R is the residue matrix where the nonlinear responses to C are stored. The CLS method has a powerful advantage to provide the results of the real chemical meaning; however, if the model spectra are not proper or the number of the model is not correct, the calculated result often loses quantitative accuracy. In such a case, PCA can work complementarily and support the CLS method. Using the number of chemical species and the loading that PCA provides in the CLS method, the calculated result from the CLS method would be more realistic. In addition, since the 8

ACCEPTED MANUSCRIPT

residual component in the loading is rejected by PCA, the calculated concentration is more accurate due to “PCA Noise Filtering”. 14

3. Results

PT

3-1. Steady-state fluorescence spectrum

RI

BSA has 27 Phe, 21 Tyr and 2 Trp in a molecule. 15 On the basis of the knowledge

SC

as shown in supporting information 2, the fluorescence spectrum of a single Trp of HSA excited at 290 nm is not affected by the concentration effect. Instead, 280 nm

NU

excitation is selected with the aim of using energy transfer effect mainly from Tyr to Trp. Since the molar absorption coefficients of Tyr and Trp are larger than that of Phe

MA

at 280 nm and the quantum yield of Phe is relatively low,11, 16 the fluorescence from Phe can be neglected. Figure 1 show the steady-state fluorescence spectra of BSA

D

excited at 280 nm at concentrations from 0.1 to 100 mg/mL. The fluorescence

PT E

spectrum of BSA has a broad band with a maximum at around 340 nm. Apparently, there are no significant changes in the spectra and no concentration induced energy

CE

transfer effect apparently as the concentration increases, which may indicate that there are no specific changes in the structure or the microenvironment for aromatic residues

AC

such as Tyr or Trp. In addition, these data also imply that the distortion of the spectrum due to the re-absorption effect can be neglected by use of the oblique incidence geometry.

3-2. Fluorescence lifetime measurement Time-resolved fluorescence spectroscopy can facilitate a more detailed understanding of protein structures and their functions.17,18 In addition, the 9

ACCEPTED MANUSCRIPT

time-resolved fluorescence spectroscopy has the advantage of resolving the scattering component from the fluorescence one. To investigate the detailed behaviour of BSA, fluorescence lifetime measurement was performed from 0.1 to 100 mg/mL. The lifetime at each wavelength from 305 to 500 nm was measured and analysed by the global fitting

PT

method.19 To resolve the scattering component from the fluorescence components, one

RI

of the decay parameters was fixed as 0.1 ns, which is the same as the time resolution of the instrument. To fit the fluorescence decay with a multi-exponential function with a

SC

small 2, four decay components, including the fixed one (0.1 ns), were needed. Figure

NU

2 shows the fitting results of (a) the decay curve of the fluorescence lifetime at 335 nm and (b) its 2 of the BSA solution at 0.1 mg/mL. The fluorescence decay has one short

MA

lifetime component of 1.2 ns and longer components of 3.6 ns and 6.9 ns. In Figure 3, (a) the decay components of each concentration and (b) their

D

normalised concentrations are shown. In addition to the fixed lifetime, τfixed (0.1 ns), the

PT E

fluorescence decay of each concentration has components of ca 4.0 ns (τ1), ca 7.0 ns (τ2) and one shorter lifetime component of ca 1.0 ns (τ3). The average lifetime of each

CE

concentration is ca 2 ns. The error bars in the figures are the three standard deviations. As the concentration increases, there is a small increase in the lifetime of longer

AC

components, ca 4.0 ns and ca 7.0 ns (Figure 3(a)). The difference of the lifetime can be rooted in different local environments as is explained by some previous studies.20-22 As shown in Figure 3(b), there is a small change in the normalised concentrations of the two longer components. The normalised concentration of τ1 shows a relative decrease; on the other hand, the normalised concentration of τ2 shows a relative increase of over 1 mg/mL. These changes in the lifetimes and in each normalised concentration may indicate that there is a small change in the local environment of the aromatic amino 10

ACCEPTED MANUSCRIPT

acids, such as Tyr and Trp, due to the concentration change. Figure 4 shows the decay-associated spectra (DAS) of each species at each concentration (0.1, 1.0, 10 and 100 mg/mL). Each decay component of (a) 0.1 (fixed), (b) ca 1.0, (c) 4.0 and (d) 7.0 ns is shown in Figure 4. The intensities of the spectra are

PT

normalised. Figure 4(a) and (b) show the unique concentration-dependent changes in the

RI

spectra, while Figure 4(c) and (d) show the typical intrinsic fluorescence spectra of the

SC

proteins and do not show any specific concentration-dependent changes, which indicates that there is no specific change in the structure or the local environments of

NU

some parts of the protein. Figure 4(a) shows a sharp positive band at 305 nm and a small negative band around 355 nm from lower to higher concentrations. Since these

MA

components have a short lifetime, the same as the time resolution of the instrument, the sharp positive band can be assigned to the scattering component. On the other hand, the

D

small negative band may be due to intramolecular energy transfer, which shortens the

PT E

lifetime. In other words, the first components can include the species of the high energy transfer efficiency. Figure 4(b) also shows unique concentration-dependent changes

CE

from lower to higher concentrations. It is of interest that, as the concentration increases, the intensity of the negative band around 360 nm also increases. This negative band

AC

may also be due to the energy transfer mainly from Tyr to Trp induced by the concentration change and this result could imply not the structural or conformational change but the molecular contact between neighbour molecules in highly concentrated solution causes the energy transfer. This insight is supported by the concentration dependence measurement of human serum albumin (HSA) by the time resolved fluorescence spectroscopy, which indicates that the concentration effect (mainly excluded volume effect) does not change the behaviour of a Trp in a protein in the

11

ACCEPTED MANUSCRIPT

highly concentrated solution as shown in supporting information 2. Although the knowledge from the time-resolved fluorescence measurement is of high interest, other approaches from different aspects will be necessary for the deeper understanding. Therefore, to understand the behaviour of a BSA molecule in a highly

PT

crowded environment in detail from different aspects, additional measurements by

RI

fluorescence anisotropy decay analysis and Raman spectroscopy were carried out for

NU

3-3. Time-resolved anisotropy measurement

SC

each concentration.

Figure 5 shows (a) the time-resolved fluorescence anisotropy decay of the BSA

MA

solution at 100 mg/mL and (b) its  value. The fitted parameters consist of two lifetime components of 0.81 and 55 ns. A shorter rotational correlation time of around 0.1 ns can

D

include both the fast segmental motion of the molecules, such as Tyr or Trp, and the

PT E

depolarisation effect resulting from the energy transfer. A longer rotational correlation time can be assigned to the overall rotational motion of the BSA molecule.

CE

Figure 6(a) shows a longer rotational correlation time ( ) for each concentration. The error bars in the figure are the three standard deviations. As the concentration increases. This increase may be due to the increase in the viscosity of the

AC

increases,

solution or the increase in the molecular size resulting from oligomerisation. The hydrodynamic radius (R) of the BSA molecule can be calculated according to the following equation:11 ,

(10)

where η is the bulk viscosity, T is the temperature and k is the Boltzmann constant. The bulk viscosity of the solution can be corrected by use of the Ross–Minton equation:23 12

ACCEPTED MANUSCRIPT

,

(11)

where [η] is the intrinsic viscosity in mL/g, k denotes the self-crowding factor, c is the concentration in g/mL and ν is the Simha shape parameter. The k/ν was assumed to be

PT

0.45, which is a reasonable value according to the previous study.23 The calculated

RI

viscosity η is shown in supporting information 3. By using this result, the calculated R is

SC

shown in Figure 6(b).

As shown in Figure 6(b), the hydrodynamic radius R increases with concentration

NU

from 0.1 to 10 mg/mL; however, unexpectedly, R decreases at 50 mg/mL. From previous studies, it was found that oligomerisation (mainly dimerisation) starts to occur

MA

at over a few mg/mL;15,24 thus, the increase in R from 1.0 to 10 mg/mL may be due to the dimerisation as the average R starts to increase. It is of interest that, at

D

concentrations over 10 mg/mL, R decreases, while an increase in concentration could

PT E

promote oligomerisation. In considering this behaviour, the environmental effect on the highly concentrated solution should be taken into account. As the concentration

CE

increases, the distance between the molecules (d) decreases, which will restrict the dynamic motion of BSA in a molecularly crowded environment. The distances between

AC

the surfaces of the molecules are calculated and are shown in Supporting Information 4. If the distance between the molecules is larger than the molecular scale, the dynamic motion is not restricted. The distance between the molecules is ca 10 nm at <10 mg/mL and d is larger than the size of the molecule, as shown in Figure S4. Thus, the overall rotational motion of the molecule is not restricted. Therefore, R can increase as dimerisation of the molecules begins. (Figure 6) In contrast, if the distance between the molecules is smaller than the molecular scale, the dynamic motion will be restricted. 13

ACCEPTED MANUSCRIPT

When the concentration of the protein solution is over 50 mg/mL, the distance between the molecules is only ca a few nm and d is almost the same size as the molecule, as shown in Figure S4. In this case, the overall rotational motion of the molecules will be highly affected by the molecular crowding effect, mainly from the excluded volume

PT

effect; however, only small rotational motions, i.e. rotation on its axis, can be allowed,

RI

and R will be similar to the radius of a molecule. (Figure 6) This result indicates that

SC

molecules still moves around independently, even if the excess of the excluded volume

NU

effect works powerfully.

3.4. Raman spectroscopy

MA

Raman spectroscopy is a powerful method for the detailed investigation of the behaviour of molecules in highly crowded / concentrated environments because the

D

Raman spectrum is not disturbed by optical effects such as the multi-scattering effect.25

PT E

Specifically, for protein science, Raman spectroscopy has some important advantages. Not only the information about each amino acid type, but also the one about the

CE

secondary and tertiary structures of a protein can be obtained from Raman spectra. Figure 8 shows the Amide I band of Raman spectra of the BSA solution (10, 20, 40,

AC

60, 80, 100 mg/mL). The amide I band of proteins is sensitive to the strength of hydrogen-bonding interactions (C=O···HN) involving amide groups. Thus, the information about protein secondary structures can be obtained from the location of this band.26,27 In the Raman spectrum of BSA, the amide I band overlaps with the water bending band at 1640 cm–1 (Figure 8). Thus, the water Raman band was subtracted from the spectra using PCA.25 Although Raman spectroscopy is only available in higher concentration (over 10 mg/mL) compared to fluorescence, there were no specific

14

ACCEPTED MANUSCRIPT

changes to the amide I band at each concentration. In other words, this result indicates that over this concentration range, there were no significant changes in the secondary structure of BSA, supporting the result of steady state and time resolved fluorescence spectroscopy.

PT

In the finger print region (400- 2000 cm-1) of Raman spectrum, there are various

RI

bands from each functional group. Of note is that aromatic hydrophobic groups such as Tyr or Trp have a larger intensity than other amino acid residues for their large

SC

polarizability. Tyr has doublet bands at 850 cm-1 and 830 cm-1 due to the Fermi

NU

Resonance of between the ring-breathing vibration and the overtone of an out-of-plane ring-bending vibration of the benzene ring of Tyr. The band intensity ratio of the doublet

MA

I850/I830 is a known marker of the microenvironment around Tyr.28 Raman spectra in the region of 790–890 cm–1 over the BSA concentration range of 10–100 mg/mL at pH 8.0

D

were compared as shown in Figure 9(a). To calculate the ratio at each concentration, we

PT E

fitted the 830 and 850 cm–1 bands with a Gaussian–Lorentz function (the fitting results are indicated by the blue curves in Figure 9(a). As shown in Figure 9(b), the

CE

concentration dependence of the I850/I830 ratio shows that the ratio started to increase from ca. 1.0 to 1.3 at concentration ≥ 10 mg/mL and was saturated at concentrations ≥

AC

60 mg/mL. The error bars in the figures are the standard deviations. This result indicates that the microenvironment around Tyr residues changed at concentrations ≥ 10 mg/mL and the change saturated over 60 mg/mL. These concentration induced changes will be caused by the protein-protein interaction and may be due to the intermolecular interaction (side chain interaction) between Tyr and other amino acid residues. This result of Raman spectroscopy also supports the result of time resolved fluorescence spectra and the energy transfer effect, indicating the molecular contact between

15

ACCEPTED MANUSCRIPT

neighbour molecules.

4. Discussion 4-1. Excluded volume effect on the conformational and dynamic behaviour

PT

From the knowledge of a previous study and the data from the anisotropy

RI

measurements, it can be suggested that dimerisation starts to occur at over 1.0 mg/mL.

SC

In addition, it can be estimated that at over 50 mg/mL the distance between proteins is close to the molecular size, inducing short-range attractive interactions, which may

NU

cause further oligomerisation. If the dimer or oligomer of BSA molecules is formed, energy transfer can occur at the protein–protein interface. As a result, the negative band

MA

can be observed in the DAS spectra.

To investigate the energy transfer at the protein-protein interface further, the τ3

D

(second) component of the DAS is refocused. In Figure 10(a), the data set of the τ3

PT E

component of the DAS for each concentration, before normalisation, is shown. When the concentration is ca 0.1 mg/mL, the BSA molecule is a monomer; thus, the energy

CE

transfer effect does not need to be considered when analysing DAS for a concentration of 0.1 mg/mL. In contrast, the intensity of DAS for concentrations over 0.1 mg/mL

AC

starts to decrease and the spectra exhibit some negative bands, which is due to the energy transfer effect induced by formation of the dimer or oligomer. In addition, the life time of the τ3 component (ca. 1 ns) is relatively short, which may be affected by the energy transfer effect. Therefore, energy transfer effects need to be taken into account when analysing DAS for concentrations over 0.1 mg/mL. Here, DAS with no energy transfer effect is denoted as S0 and those including an energy transfer effect are denoted as S(c). The spectra induced by the energy transfer 16

ACCEPTED MANUSCRIPT

effect can be calculated as shown below: Strans = S(c) − S0.

(12)

As shown in Figure 10(b), the intensity of this negative spectrum, Strans, increases as the concentration increases, which indicates that the size of the energy transfer effect

PT

increases with the degree of oligomerisation. In other words, this negative spectrum can

RI

be assigned to the absorption spectra of the acceptor, and the acceptor absorbs more energy from the donor as the concentration increases. Therefore, the negative

SC

“absorption” spectra of the acceptor have a band at ca 330 nm, compared to the

NU

fluorescence band at ca 305 nm for the donor, as explained in detail in Figure S5. The third (ca 4 ns) and fourth (ca 7 ns) components of the DAS are mainly average

MA

fluorescence spectra of Tyr and Trp. Of note is that the fourth component of the DAS is located at a longer wavelength than the third one, which suggests that the residues of the

D

fourth components are closer to the protein surface than those of the third and interact

PT E

more effectively with the solvent because a red shift often occurs as a result of interaction with a polar solvent in the case of Trp or indole molecules.4,5,10 As shown in

CE

Figure. 4(b), the normalised concentration of τ2 (ca 7 ns) shows a relative increase at concentrations over 1 mg/mL. Using this rough assignment of the DAS, this increase

AC

can be interpreted as a result of the energy transfer at the protein surfaces. In other words, the energy of the fluorescence of the τ3 component moves to the τ2 components, which is reasonably close to the surface of the protein. On the other hand, the normalised concentration of the τ1 components exhibits a relative decrease due to the increase in the τ2 components. To estimate the efficiency of the “apparent” energy transfer effect, the intensity of the fluorescence without and with the energy transfer effect is defined as F0 (0.1

17

ACCEPTED MANUSCRIPT

mg/mL) and F(c) (>0.1 mg/mL), respectively. The energy transfer Ftrans is calculated as shown below: (13) 2

2

1

1

PT

,where F0   I 0 d , F (c)   I (c)d , I0 and I(c) is the intensity of each wavelength. In this experiment, are 305 and 500 nm.

RI

The efficiency of the “apparent” energy transfer effect, Etrans(c), can be described

SC

according to

(14)

NU

.

In Figure 11, the calculated result for Etrans is shown. As the concentration increases,

MA

Etrans increases. This indicates that intermolecular contact resulting from an increase in the concentration induces an energy transfer at the protein–protein interface. Of interest

D

is that, for concentrations over 50 mg/mL, the Etrans is almost saturated, which will

PT E

indicate that the proteins are tightly packed together. The calculated result (Figure S4) also shows the similar view that at around 50 mg/mL, the distance between the proteins

CE

is almost the same size as the proteins and the calculated result supports the analysis result of the energy transfer effect. (Figure 11)

AC

The anisotropy decay result adds the dynamic point of view that the excluded volume effect induces the molecular contact which can cause the oligomerization, however, may induce the electrostatic repulsion between molecules at the same time. Consequently, the molecules still move independently even in the highly concentrated solution over 50 mg/mL. This result can provide the possibility that in highly concentrated solution, the molecules are in the fast reversible oligomeric state or some kinds of cluster state, which is supported by some previous studies.15,24,29

18

ACCEPTED MANUSCRIPT

Raman spectroscopy can describe the interaction between molecules from the conformational point of view and support these results. The increase of the intensity ratio I850/I830 of Tyr (Figure 9) indicates the subtle intermolecular interaction of Tyr with neighbourhood other amino acid residues in highly concentrated solution with the

PT

hydrogen bonding. This result supports the view that in highly concentrated solution,

RI

the BSA molecules contact each other from the analysis result of the energy transfer

SC

effect from Tyr to Trp. In addition, the analysis of the Amide I band also shows that the secondary structure is maintained even under the excess of the excluded volume effect

NU

in highly concentrated solution.

On the basis of these knowledge, the contribution ratio of the oligomeric state,

MA

cwas calculated as shown in supporting information 6. The result shows that while the most of the BSA is in the monomeric state at 0.1 mg/mL, more than half of the BSA

D

is in the oligomeric state at 1.0 mg/mL. In addition, at 10 mg/mL, 90% of the BSA is in

PT E

the oligomeric state, and at concentrations over 10 mg/mL, the oligomerisation becomes

CE

saturated.

4-2. Energy transfer at the heterogeneous interfaces revealed by PCA

AC

Since the intermolecular contact between molecules is influenced by the energy transfer effect in the DAS, the further investigation could reveal more detail behaviour of the protein-protein interface between molecules. It is of note that the inclination of the cagainst the concentration is different for c < 10 mg/mL and c ≥ 10 mg/mL, which implies different energy transfer efficiencies. Thus, PCA was applied to the data set of the τ3 component of the DAS. To estimate the number of chemical species in the data set, REV and RMSECV were calculated, as described in the Supporting 19

ACCEPTED MANUSCRIPT

Information 7. The REV and RMSECV of the data set are shown in Figure S7(a) and (b), respectively. The results indicate that there were two main species and one minor species. The PCA loadings are shown in Figure 12(a), (c) and (e). The loading shown in Figure 12(a) corresponds to the average spectra of the data set, and the loadings shown

PT

in Figure 12(c) and (e) correspond to the spectra of the energy transfer effect. Using

RI

these loadings as model spectra, each contribution ratio, CR (relative concentration), is

SC

calculated by the CLS method, as shown in Figure 12(b), (d) and (f). As the concentration increases, the CR of the first loading decreases. In contrast, the CR values

NU

of the second and third loadings increase with the concentration. It is interesting that the negative band of the loading spectra is different and the peak of the negative band for

MA

the second loading is located at ca 345 nm, while the one for the third loading is located at ca 380 nm. This result may indicate that the energy transfer effect shown for the

D

second and third loadings is caused by the different protein–protein interfaces. In

PT E

addition, since the negative band of the third loading is located at a higher wavelength, this interface may be more stabilised by  conjugation than that for the second loading

CE

and may have larger areas.

The CR for the second loading starts to increase at concentrations over 1 mg/mL

AC

and becomes saturated at around 50 mg/mL, while the CR for the third loading starts to increase at concentrations over 50 mg/mL. This result shows that the energy transfer effect shown for the second loading is mainly dominant at lower concentrations, while that for the third loading is relevant only for higher concentrations. This difference may indicate that the energy transfer effect for lower and higher concentrations at different protein-protein interfaces are caused by the different interactions between molecules. In the lower concentration region (1 mg/mL < C < 50 mg/mL), the electrostatic interaction

20

ACCEPTED MANUSCRIPT

still works dominantly and the attractive interaction may be caused by the localized heterogeneous charge distribution at the protein surface. These interactions may induce the association at the “hard” interface, which is a point-point charged interface and may cause the fast reversible dimerization or oligomerisation as reported in the previous

PT

study.15

RI

On the other hand, in the high concentration region (C > 50 mg/mL), the distance

SC

between proteins are almost the same size as the molecule and in addition, the dynamic behaviour of the molecule is highly restricted and the disassociation of the oligomer is

NU

also limited. Hence, the short range attractive interaction such as such as the van der waals interaction or the CH-π interaction will work powerfully. These interactions may

MA

cause the association at the “soft” interface, which is shorter ranged and has larger area and may cause the oligomerisation with the aid of the crowding effect. (Figure 13). The

D

presence of different conjugation interfaces in BSA was reported in a previous study of

PT E

the analysis of BSA–ANS interactions, and the previous result supports the experimental results in this paper.30

CE

In this manner, the detailed behaviour of the BSA molecule has been comprehensively understood from both a conformational and a dynamic point of view,

AC

as shown in Figure 13.

(i) C < 1 mg/mL: Most of the BSA molecules are in the monomeric state. The behaviour of BSA is not restricted and BSA molecules can move freely. (ii) 1 mg/mL < C < 50 mg/mL: More than half of the BSA molecules are in the dimeric or oligomeric state. The dynamic behaviour of the BSA is partly restricted by the excluded volume effect. Energy transfer at protein–protein charged interfaces starts to occur. 21

ACCEPTED MANUSCRIPT

(iii) C > 50 mg/mL: Most of the BSA molecules are in the oligomeric state, which may be the fast reversible process or some kinds of clusters. The dynamic behaviour of BSA is highly restricted by the excluded volume effect. Energy transfer at the protein–protein

PT

heterogeneous interfaces occurs with the aid of short-range attractive interfaces.

RI

Conclusions

SC

The detail spectroscopic understanding of the detail behaviour of BSA in a highly crowded environment was obtained using the energy transfer effect by time-resolved

NU

fluorescence spectroscopy with the aid of the dynamic and conformational aspect supported by time resolved anisotropy measurement and Raman spectroscopy. By using

MA

these methods, it was found that in the highly concentrated solution, most of the BSA molecules moved around independently and they were in the fast-reversible oligomeric

D

state or they formed some kinds of clusters, although the dynamic behaviour of the BSA

PT E

molecules was highly restricted by the excluded volume effect. In addition, the detail analysis of the energy transfer effect in the time resolved fluorescence spectra showed a

CE

unique insight that the association at the “hard” and “soft” interfaces between protein surfaces occurred in a highly crowded environment with the aid of a charge-charge and

AC

short-range attractive interface.

Supporting Information The optical configuration for lower and higher concentrations is supplied as Supporting Information 1. The DAS of each time-resolved species (τ1-τ3) and the lifetime at each concentration (1.0, 10 and 100 mg/mL) of the HSA solution is supplied as Supporting Information 2. The concentration dependence of the calculated viscosity 22

ACCEPTED MANUSCRIPT

on the basis of the Ross-Minton equation is supplied as Supporting Information 3. The distance between the molecules of the BSA solution as a function of concentration is supplied as Supporting Information 4. Explanation of the τ3 component of the DAS on the basis of the energy transfer effect is supplied as Supporting Information 5. The

PT

calculated ratio of the monomer and oligomer of BSA and the schematic of the

RI

monomeric or oligomeric state of each concentration is supplied as Supporting Information 6. The results of PCA (REV, RMSECV) of the concentration dependence of

NU

SC

the τ3 component of the DAS spectra are supplied as Supporting Information 7.

MA

Figure captions

Figure 1. The steady state fluorescence spectrum of BSA excited at 280 nm of each

PT E

D

concentration from 0.1 to 100 mg/mL.

Figure 2. (a) The decay curve of the fluorescence lifetime of the BSA solution (0.1

CE

mg/mL) at 335 nm with the fitted result and (b) its 2.

AC

Figure 3. (a) The decay components of each concentration and (b) its normalized concentration.

Figure 4. (a)-(d)The decay associated spectra (DAS) of each time-resolved species (τ1-τ4) at each concentration (0.1, 1.0, 10 and 100 mg/mL).

Figure 5. (a) the time-resolved fluorescence anisotropy decay of the BSA solution (100

23

ACCEPTED MANUSCRIPT

mg/mL) with the fitted result and (b) its 2 value.

Figure 6. (a) the rotation correlation time ( ) and (b) the calculated hydrodynamic

PT

radius R of each concentration

RI

Figure 7. Schematic of the dynamic behaviour of BSA molecules of each

SC

concentrations.

NU

Figure 8. Overlay of the amide I band of BSA at each concentration (10, 20, 40, 80 and

MA

mg/mL)

Figure 9. (a) Raman spectra of BSA solutions in the range of 790–890 cm-1. The

D

concentrations are 20, 40, 80, 100, and 300 mg/mL. (b) The concentration-dependent

PT E

band intensity ratio I850/I830 of BSA.

CE

Figure 10. (a) the concentration dependence of the τ3 component of the DAS and (b) the

AC

subtracted spectra from the DAS at 0.1 mg/mL.

Figure 11. The concentration dependence of the calculated Etrans.

Figure 12. (a,c,e) The result of PCA loadings of data set of the τ3 component of the DAS spectra and (b,d,f) each contribution ratio CR by use of the CLS method

Figure 13. Schematic of the behaviour of BSA and the energy transfer at the

24

ACCEPTED MANUSCRIPT

protein-protein interfaces of each concentrations

Figure S1. The optical configuration for (a) lower (<1 mg/mL) and (b) higher (

1

PT

mg/mL) concentrations.

RI

Figure S2. (a) The decay associated spectra (DAS) of each time-resolved species (τ1-τ3)

SC

at each concentration (1.0, 10 and 100 mg/mL) and (b) the lifetime of the HSA solution.

NU

Figure S3. The concentration dependence of the calculated viscosity on the basis of the

MA

Ross-Minton equation.

D

Figure S4. Distance between surfaces of the molecules of the BSA solution as a function of concentration.

PT E

Figure S5. Explanation of the τ3 component of the DAS on the basis of the energy

CE

transfer effect

AC

Figure S6. The calculated ratio of the monomer and oligomer of BSA and the schematic of the monomeric or oligomeric state of each concentration.

Figure S7. Plots of (a) reduced eigenvalue (REV) and (b) root-mean-square prediction error of cross-validation (RMSECV) against the factor level of the PCA in the data set of the 2nd component of the DAS

25

ACCEPTED MANUSCRIPT

References 1.

R. J. Ellis, A. P. Minton, Cell biology: Join the crowd. Nature 2003, 425, 27-28.

2.

A. P. Minton, Excluded volume as a determinant of macromolecular structure and

PT

reactivity. Biopolymers 1981, 20, 2093-2120.

H. X. Zhou, G. N. Rivas, A. P. Minton, Annu. Rev. Biophys. 2008, 37, 375-397.

4.

R. Harada, Y. Sugita, F. Michael, Protein crowding affects hydration structure and

SC

RI

3.

R. Harada, N. Tochio, T. Kigawa Y. Sugita, F. Michael, Reduced Native State

MA

5.

NU

dynamics. J. Am. Chem. Soc. 2012, 134, 4842-4849.

Stability in Crowded Cellular Environment Due to Protein–Protein Interactions. J.

PT E

D

Am. Chem. Soc. 2013, 135, 3696–3701.

6. K. Inomata, A. Ohno, H. Tochio, S. Isogai, T. Tenno, I. Nakase, T. Takeuchi, S.

CE

Futaki, Y. Ito, H. Hiroaki, M. Shirakawa, High-resolution multi-dimensional NMR

AC

spectroscopy of proteins in human cells. Nature, 2009, 458, 106-109 7. M. R. Eftink, Fluorescence techniques for studying protein structure. Methods Biochem Anal. 1990, 35, 117–129. 8. I. Gryczynski, W. Wiczk, M. L. Johnson, J. R. Lakowicz, Lifetime distributions and anisotropy decays of indole fluorescence in cyclohexane/ethanol mixtures by frequency-domain fluorometry. Biophys Chem 1988, 32, 173–185.

26

ACCEPTED MANUSCRIPT

9. D. M. Rayner, D. T. Krajcarski, A. G. Szabo, Excited-state acid–base equilibrium of tyrosine. Can. J. Chem. 1978, 56, 1238–1245. 10. R. Boteva, T, Zlateva, V. Dorovska-Taran, AJWG, Visser, R. Tsanev, B. Salvato,

PT

Dissociation equilibrium of human recombinant interferon Biochemistry 1996, 35,

RI

14825–14830.

SC

11. J. R. Lakowicz, Principles of Fluorescence Spectroscopy, 3rd version

NU

12. E. R. Malinowski, Factor Analysis in Chemistry, 2nd ed.; Wiley-Interscience: New

MA

York, 1991.

13. A. P. Lobo, B.S. Valles, N. F. Tascón, R. R. Madrera, Calibration models for routine

PT E

D

analysis of cider by mid-infrared spectroscopy. Lebensm. Wiss. Technol. 2006, 39, 1026-1032.

CE

14. Y. M. Jung, Principal component analysis based two-dimensional correlation

15. A.

AC

spectroscopy for noise filtering effect. Vibrational Spectroscopy. 2004, 36, 267-270. Bhattacharya,

R.

Prajapati,

S.

Chatterjee,

T.

K.

Mukherjee,

Concentration-Dependent Reversible Self-Oligomerization of Serum Albumins through Intermolecular β-Sheet Formation. Langmuir 2014, 30, 14894−14904. 16. O. Warburg, W. Christian, Biochem. Z 1942, 310, 384-421. 17. F. Tanaka, N. Mataga, Fluorescence quenching dynamics of tryptophan in proteins.

27

ACCEPTED MANUSCRIPT

Biophys. J, 1987, 51, 487–495. 18. J. R. Alcala, The effect of harmonic conformational trajectories on protein

PT

fluorescence and lifetime distributions. J. Chem. Phys. 1994, 101, 4578–4584. 19. J. R. Knutson, D. G. Walbridge, L. Brand, Decay-Associated Fluorescence Spectra

RI

and the Heterogeneous Emission of Alcohol Dehydrogenase. Biochemistry 1982, 21,

SC

4671-4679.

NU

20. A. Grinvald, IZ. Steinberg, The fluorescence decay of tryptophan residues in native

MA

and denatured proteins. Biochim Biophys Acta 1976, 427,663–678. 21. D. M. Jameson, G. Weber, Resolution of the pH-dependent heterogeneous

PT E

D

fluorescence decay of tryptophan by phase and modulation measurements. J Phys Chem 1981, 85, 953–958.

CE

22. P. R. Callis, T. Liu, Quantitative prediction of fluorescence quantum yields for

AC

tryptophan in proteins. J Phys Chem B 2004, 108, 4248–4259. 23. S. J. Shire, D. S. Kalonia, Viscosity analysis of high concentration bovine serum albumin aqueous solutions. Pharm. Res. 2011, 28, 1973-1983. 24. S. Chatterjee, T. K. Mukherjee, Effect of Self-Association of Bovine Serum Albumin on the Stability of Surfactant-Induced Aggregates of Allylamine-Capped Silicon Quantum Dots. J. Phys. Chem. B 2013, 117, 16110−16116.

28

ACCEPTED MANUSCRIPT

25. C. Ota, S. Noguchi, K. Tsumoto, The molecular interaction of a protein in highly concentrated solution investigated by Raman spectroscopy. Biopolymers, 2015, 103,

PT

237-246. 26. N. C. Maiti, M. M. Apetri, M. G. Zagorski, P. R. Carey, V. E. Anderson, Raman

RI

spectroscopic characterization of secondary structure in natively unfolded proteins:

SC

α-Synuclein. J. Am. Chem. Soc. 2004, 126, 2399-2408.

NU

27. M. M. Apetri, N. C. Maiti, M. G. Zagorski, P. R. Carey, V. E. Anderson, Secondary

MA

structure of a-synuclein oligomers: characterization by Raman and atomic force microscopy. J. Mol. Biol. 2006, 355, 63-71.

PT E

D

28. M. N. Siamwiza, R. C. Lord, M. C. Chen, T. Takamatsu, I. Harada, H. Matsuura, T. Shimanouchi, Interpretation of the doublet at 850 and 830cm-1 in the Raman spectra

AC

4870-4876.

CE

of Tyrosyl residues in proteins and certain model compounds. Biochemistry 1975, 14,

29. L. Porcar, P. Falus, W. R. Chen, A. Faraone, E. Fratini, K. Hong, P. Baglioni, Y. Liu, Formation of the Dynamic Clusters in Concentrated Lysozyme Protein Solutions J. Phys. Chem. Lett. 2010, 1, 126–129. 30. D. M. Togashi, A. G. Ryder, A Fluorescence Analysis of ANS Bound to Bovine Serum Albumin: Binding Properties Revisited by Using Energy Transfer Journal of

29

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

RI

PT

Fluorescence 2008, 18, 519–526.

30

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

31

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

32

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

33

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

34

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

35

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

36

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

37

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

38

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

39

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

40

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

41

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

42

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

43

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

44