272 Biochimica et Biophysica Acta, 957 (1988) 272-280 Elsevier
BBA 33247
Secondary s t ruc ture of 11 S globulin in aqueous solut ion invest igated
by F T - I R derivat ive spect roscopy
S.B. D e v a,, , J.T. Kel le r a and C.K. R h a b
a National Center for Biomedical IR Spectroscopy, Battelle Memorial Institute, Columbus, OH and b Biomaterials Science and Engineering Laboratory, Department of Applied Biological Sciences, Massachusetts Institute of Technology,
Cambridge, MA (U.S.A.)
(Received 25 April 1988)
Key words: 11 S globulin; Secondary structure; Protein denaturation; FT.IR; Deconvolution; Derivative spectroscopy
Secondary structure of 11 S globulin, a major storage protein of soybean seeds, has been investigated in aqueous solution by FT-IR spectroscopy. Conformationai changes in the native protein upon thermal and chemical denaturation have been monitored by observing changes in the frequency position and peak intensity of the various bands. The frequency of the Amide 1 band of the native protein shifts by 4 cm-i from 1 643 era-i to 1 647 cm-I when denatured, while the corresponding intensity of the Amide ! band compared to the native protein, decreases by 30 and 67%, respectively, for the urea and thermally denatured proteins, indicating gross eonformational changes in the secondary structure. Trifluoroethanol, an a-helix promoter shifts the Amide i band from 1 643 era-1 to 1 651 c m - i typical of a-helix, with a corresponding increase in intensity by 14% relative to the native protein. Derivative spectroscopy, allowing resolution of overlapping bands, shows that the native protein mainly consists of r-sheet, r-turns and disordered structure with very little a-helix. On denaturation, ~-sheet disappeared almost completely with urea, while this is less so with thermal denaturation.
Introduction
A knowledge of the secondary structures of proteins in their native or denatured state is ex- tremely important because it allows one to gain insight into the ordered, unordered and/or dis- ordered structures. Vibrational spectroscopy is one of the important techniques for investigation of
Correspondence (and * Present address): S.B. Dev, Depart- merit of Applied Biological Sciences, Massachusetts Institute of Technology, Bldg. 56-138, Cambridge, MA 02139, U.S.A.
such structures. 11 S globulin, also known as glycinin, is one of the major reserve proteins of soybean seeds. Although a great deal of work has been done [1,2] on the isolation, purification and characterization of this protein in the native and denatured state, study of detailed secondary s'truc- ture has been very few, and the reported values of the a-helix, r-sheet, and random coil contents have been substantially different [3-5].
We have, therefore, studied the secondary structure of 11 S globulin by Fourier Transform Infrared Spectroscopy (FT-IR) by taking ad- vantage of the microcirde cell that has ATR optics. This enabled us to obtain excellent IR spectra in aqueous solution, and use mathematical tech- niques to enhance resolution of the overlapping
for comparison with the second-order derivative spectra in some cases. For the derivative spectra, we present the spectra in a slightly different form than is conventional. For a second-derivative spec- trum, the major positive peaks of the original spectrum should approximately coincide with the corresponding negative peaks of the derivative spectra. The resol,.:tion-enhanced peaks in the derivative spec*,ra ~may not, however, be recog- nized in the original spectra, since they appear as very weak peaks or as minor shoulders. In the program, the results of the second derivative are multiplied by - 1 , so that the match is now be- tween the positive peaks of the original with the positive peaks of the derivative spectra. Because the negative peaks of the derivative spectra are now redundant, they have been left out in the final plot and only the positive bands are plotted out. We have found that this particular method of presentation makes the comparison of peaks direct and straightforward. This procedure shows up the bands similar to the discrete areas obtained by constrained deconvolution method where one of the initial boundary conditions imposed is one C,t" non-negativity of the peaks [9]. It has to be emphasized, however, that the derivative results will refer only to the position of the peaks and no attempt has been made to relate these to areas mentioned above or to the areas obtained by band-fitting of spectral profiles.
Results and Discussion
The digitally water-subtracted undeconvoluted spectrum of 10~; soy protein is shown in Fig. 1. As expected, very strong ~brations appear at 1643 and 1546 cm -1 arising out of C ffi O vibration (Amide I), and N - H bend and C-N stretch (Amide II), respectively. Amide III band, much weaker than the Amide I and II bands in IR, appears at 1242 cm-1. There is also a very strong band at 1454 cm -! which is attributed to 8(CH2) and 8(CH3). A cursory glance at Fig. 1 im- mediately reveals that there are a number of shoulders which have not been resolved. In other words, it is essential to enhance such weak signals without which freer features of the spectrum, that contains further information on the secondary structure, might be lost.
0.155- m
0.145--
0.135 Z
~ 0.125--
.8 o .115- A
o.Io5: 0O95 -
0 085 -
I | • •
1800 1600 14~::) I
1200 Wavenurnber
Fig. 1. Spectrura of 10~ native protein. Resolution: 8 cm-~; scan: 2000; slrectra-tech microcircle cell with ZnSe cryst'd; absorbance sc~de in all figures are relative. All spectra are
water-sebtracted.
There are a number of ways in which th:,s can be carded out. These include derivative spec- troscopy which invariably enhances weak signals. Other methods, which are also well-established, such as Fourier self deconvolution (FSD), Con- strained Deconvolution or band-fitting have ad- ded a new dimension for treating vibrational spec- tral profiles of biological macromolecuY, es [10-12]. The peaks in the FT-IR spectra, be ffLey overlap- ping or not, arise out of inherent vffrations and, hence, all the above procedures should, ideally, lead to the same results. In actual practice, how- ever, the final results may be sli&htly different, each method having its own limitation [10-12]. What is important, however, is to have a spectrum with a very high S / N for any resolution enhance- ment technique to be meaningful. Since the rever- sal of the deconvolution procedure should lead back to the same original spectrum, these methods cannot be, strictly speaking, independent of each other. However, except in the most simple cases, it may be impossible to deduce exactly the analytical function that will allow easy conversion between one technique and another. This will certainly be the case for experimental spectra which always have some noise present. Current band-fitting pro- grams usually carry out full optimization on the peak positions derived from the FSD or derivative spectra.
Unless mentioned otherwise, we have used 'single pass' smoothing, meaning smoothing only once, for the derivative spectra presented here.
TABLE i
PEAK POSITIONS AND CORRECT INTENSITIES OF THE AMIDE I AND AMIDE 11 BANDS
Nature o[ the 11 S Peak frequency Correct intensity Amide I globuhn ~:ample {cm -1 ) (mau) Amide II
Native 1643.3 70.78 1.25 1546.9 56.57
In 4 M urea 1643.3 52.36 1.30 after 1 h 1546.9 40.43
In 4 M urea 1647.2 49.87 1.02 after 5 h 1546.9 48.68
Thermally denatured 1647.2 23.08 1.08 (95 o C) 1546.9 21.34
in 40~ TFE 1651.1 80.80 3.71 1546.9 2i .76
change in the intensity of Amide I relative to the native
- 2 6
- 30
- 6 7
+14
275
For the actual peak frequency and correct inten- sity of the three most strong peaks in the undecon- voluted spectra, we have ased the standard 'center of gravity" ,program. Table I lists the Amide I and II frequem;ies for the 11 S globulin under different conditions,, together with the ratio of two Amide bands and per ,cent change in intensity. Subtraci- ton fact, or certainly affects the peak intensities. However, it has been found that if the criteria cho~el~ "or the correct subtraction is satisfied (dis- cussed, in the section on 'collection of spectra'), the rat!~o of the intensity remains fairly constant.
The digitally subtracted undeconvoluted spec- tra for the 10% protein in 4 M urea (1 and 5 h), 40% TFE and also the heat-denatured protein are sho~'a in figs. 2-5. A composite figure is shown in fig. 6. As mentioned earlier, spectra for 4 M urea
0.125-
0.115-
8 o.~o5-
_ .13
b 0,095-
° -j 0 0 8 5 -
0,075 1800
I 1 6 0 0
, i ~ i 1,400 1200
W o v e n u m b e r
and 40% TFE were also run to enable us tor proper digital subtraction and also for reference peaks for the TFE. These also help to identify the
0 085-- u
0075-
0.065--
~ 0.055-- g -
o.o45- <
0.035-
0025-.. , ,,,
/
1 8 0 0 1 6 0 0 1 4 0 0 1 2 0 0 W o v e n u m b e r
Fig. 3. Spectrum of protein in 4 M urea after 5 h. Sample left undisturbed in the cell once inserted into the spectrometer at
the start of the urea-denaturation experiment.
0.110 ~
0.090--
~ O.070-- a
~ 0050 - ./3 <
0.030-
O.010- T
1800 1600 1400 Wevenumber
Fig. 2. Spectrum of protein in 4 M urea after 1 h. Fig. 4. Spectrum of protein in 40~ trifluoroethanol (TFE).
276
0.108~ O1Od~
0.100~ 0096~
o.°92-I ~ o.oee--
o.o~- < o . ~ -
0.076" 0.072----
I;~'~ ' 1600 ' 1400 ' 1200 Wa~number
Fig, 5, Spectrum of protein which has been heat-denatured at 95 oC.
bands in the spectra of protein together with any other component (in this case urea or TFE). The bands for the urea appear at 1659 cm -l , 1624 cm -1, 1601 cm -!, 1462 cm -l , 1 161 cm -1 and those for TFE at 1454 cm -l , 1 416 cm -t, 1373 cm-t and 1277 cm-~, respectively. The principal bands for the undeconvoluted spectra correspond- ing to Figs. 1, 3, 4 and 5 are given in Table II.
Figs. 7-10 show the results of the second de- rivative spectroscopy for native protein, protein perturbed by 4 M urea after 5 h, thermally dena- tured protein and, finally, the effect of adding a helix promoter. It is obvious in each case that there is a considerable increase in the information content in the derivative spectral profiles. To make the comparison easier, we have superimposed the undeconvo!uted spectrum of the protein in each case. It is clear that any shoulder in the unde-
TABLE II
PEAK POSITIONS IN UNDECONVOLVED SPECTRA (cm - I )
Native Protein in Thermally Protein in protein 4 M urea (5 h) denatured protein 40% TFE
Fig. 7. (a) Native protein. (b) Second derivative spectra of (a) x ( - 1 ) ; only positive values are shown; binomial smoothing
[7].
nconvoluted spectrum is matched exactly by a corresponding peak in the derivative spectra (Figs. 7-10). The band positions for the resolved curves are given in Table III. This method of presenta-
/ / \ \ o,oo: oo o: / \
i \
" ' Ir O ' 15 o Wavenumber
Fig. 6. Composite spectra of protein. (a) Native protein. (b) Native protein, perturbed by 4 M urea, after 5 h. (c) Native
protein, perfurbed by 40~$ TFE.
0.085~
0.075q
0.065~
o.o5%! i 0.045q
0,035q
0
1800 1600 1400 1200 V~venumber
Fig. 8. (a) Native protein perturbed by 4 M urea (5 h). (b) Second derivative of (a) x ( - 1 ) ; ozdy positive values are
shown; smoothing as before.
277
(1 h)
Native Assignment Protein perturbed protein by 4 M urea
tion of the derivative spectra brings out more clearly the nature of the overlapping bands of protein, thereby revealing the finer secondary structures.
Maddams et al. [13] have discussed derivative IR spectroscopy comparing various methods to obtain derivative spectra and also the effect of bandwidth and bandshape on the resolution en- hancement by derivative spectroscopy. Second de- rivative spectra of proteins have been used to relate various peaks to the a-helix, //-sheet and r-turn of the protein [14] and to study protein-de- tergent interaction by combining FSD and deriva-
I 1800 1600
,, = •
' 1~:~o 12~ Wavenumber
Fig. 9. (a) Thermally denatured protein (95 o C). (b) Second derivative spectra of (a) × ( - 1 ) ; only positive values are
shown; smoothing as before.
tive spectra [15]. Earnest et al. [16] have been able to detect localized vibrational modes from tae retinal chromophore of the purple membrane bacteriorhodopsin, although it is only 1% (by ,0vt.) of the total membrane, by the same techniqu,::s.
Certain interesting features emerge fror,l the frequency and intensity data given in Ta/ale I, although the frequencies correspond to the unde- convolved spectra. First, we notice from tLe com- posite spectral profile of Fig. 6 that the l,~mide I bands for a and b are skewed towards tl~e higher frequencies, indicating highly disordered ~tructure, as in the case of fl-lactoglobulin [17]. The Amide I
0 . 1 0 0 -
0.080--
0.060- E
~ 0.040- <
0020-
TABLE II1
PEAK POSITIONS IN 2ND DERIVATIVE SPECTRA gem - ! )
O
16CD 1500 ' I '
1800 1700 WQvenumber
Fig. 10. (a) Spectrum of protein in 40% TFE, an a.helix promoter; (b) Second derivative spectra of (a) × ( - l y , only positive values are shown. TFE bands discussed in the text are
clearly seen; smoothing as before.
278
vibration for the native protein at 1643 cm -1 shifts to 1647 cm-l in 4 M urea after 5 h, and the corresponding intensity decreases considerably. The band at 1643 cm -1, although composite in nature, as well be shortly found out, is principally owing to the ~-sheet which is being gradually destroyed by urea.
The addition of TFE, an a-helix promoter, has shifted the Amide I band by 8 cm -1 to 1651 cm -~ with a consequent increase of 14~ in the intensity relative to the Amide I band, showing generation of a-helix - again as would be ex- pected. The very high value in the Amide I/Amide II ratio, 3.7, as opposed to 1.25 in the native state, indicates gross change in secondary structure. Ad- dition of such promoter also confirms peak assign- ment where a non-a-helical peak must move to- wards the general range of helix frequency or, in case of the presence of some helix, the intensity should necessarily increase, showing generation of new helical segments. We also notice that, in the thermally denatured protein, the intensity of the peak characterizing the fl sheet has decreased by 67% relative to the native Amide I peak. There is one more striking feature, however, in that the Amide I peak appears exactly at the same frequency as for the 4 M urea peak after 5 h. This seems to confirm, at least on a gross scale, the observation by Pfeil and Privalov [18], and Priva- lee [19] that protein denatured by heat, pH, and chemicals is 'thermodynamically similar' and that one state can be transformed into another with very little cooperative thermodynamic effect, thus implying that the denatured states are structurally similar- disordered and unfolded.
The cell containing the protein and the dena- turing agents was left in the spectrometer, and the data processing carried out from time to time. Table I shows that the Amide I intensity decreases by 26 and 30~, respectively, after 1 and 5 hours, relative to the nature, and that the corresponding Amide I/Amide II ratio reduces from 1.30 to 1.02. These values indicate that the action of the 4M urea in hydrogen bond-breaking is not in- stantaneous, and that conformational change, activated chemically, in secondary structure oc- curs over a finite time period. Table I shows that the percentage change in intensity for the ther- marly and chemically denatured protein is very
different. This would suggest that the degree of denaturation by 4 M urea is considerably less than that obtained at 95 ° C.
Recently, Kate et al. [20] have used the Amide I / I f ratio for quantitative estimation of a-helix content of albumin, and have investigated the thermal denaturation kinetics of this protein. They show that while the p-sheet and random structure contents increase with time of denaturation, the percentage of a-helix decreases. One of us (S.B.D.) has recently discussed [21] the various problems associated with quantitative determination of sec- ondary structures of proteins from spectroscopic data where an enormous variation can occur for the same protein. This will be so even when X-ray data are available.
Thermal denaturation of globular proteins has also been recently investigated by FT-IR spec- troscopy from the Amide Ill spectral region [22]. Although Amide III band had certain advantages, it is a very weak band in IR compared to the Amide I and II bands. It is now well-established that for the study of protein secondary structure by vibrational spectroscopy in aqueous solution, some resolution enhancement of the spectral pro- files must be carried out, and that the band ratio of Amide I to Amide II, even for the undeconvo- luted spectrum, reflects, at least qualitatively, con- formational change. We, therefore, feel that the results of spectral deconvolution and the ratio of the Amide I and II intensity can give substantially more information on the conformational changes of chemically or thermally denatured proteins than would be possible from the Amide III band alone. Previous studies [2,5] based on CD and ORD have also indicated widely differing values of the a-helix and t-structure contents of the 11 S globulin protein, varying between 5-20% and 23-25% for the helix and the t-structure, respectively.
Table III shows the results of band resolution due to the application of second derivative spec- troscopy. Because of the non-availability of the X-ray data for this protein and other limitations that exist in the current state-of-the-art of quanti- tative secondary structure prediction from spec- troscopic measurements [21], we can only make, at best, semi-quantitative observation. The assign- ments given in Table III are based on commonly published IR markers for proteins [10,17].
It can be seen from Tables I and III that maximum changes in peak intensities and frequen- cies occur near the Amide I band and the 1 455 cm-! region. The native protein is mostly made up of anti-parallel fl-sheet (1638 cm-l) , fl-tums (1687 cm -1) and disordered structures (1660 cm-1) with very little a-helix. Unfortunately, the Amide I vibration for a-helix, for different pro- teins, can be anywhere between 1650-1660 cm-1, together with the fact that for one protein a frequency that corresponds with the a-helix may represent a disordered structure in another pro- tein. Bovine serum albumin and a-casein, e.g., [17] are cases to support this, While the 1656 + 1 cm- i peak is due to the a-helix in bovine serum albumin, a peak at 1655 cm-~ refers to the dis- ordered structure in a-casein. Combining the spec- tral features of the original and the deconvolved bands in the Amide I region (Fig. 8), together with the assignments of the second derivative peaks (Table III), it is clear that urea has considerably decreased the contribution due to the fl-sheet compared to that in the native state (Fig. 7). All that remains of the 1638 cm-1 band is only a very minor shoulder (Fig. 8). Overall, this is also re- flected in Table I in the intensity reduction of the Amide I band which is mostly made up of fl-sheet. When the protein is thermally denatured, it is seen that the derivative spectra (Fig. 9) is characterized by fl-sheet and disordered structures, whereas the band due to the turns hardly appears. It has been mentioned before that the final structure of the protein after denaturation, irrespective of the pro- cess involved, is the same as ,evident from the appearance of identical frequency of the undecon- voluted spectra; it is now clear that, at a finer level, this is not the case.
There has been considerable debate [1,23] on the subunit structure of 11 S globulin and its role in the denaturation of this protein by urea, SDS and heating. It has been proposed [23] that this protein is made up of four acidic and four basic subunits and also of three kinds of intermediary subunits. At least six disulfide bonds are believed to maintain the subunit structure and that eight different subunits are involved in the denaturation of 11 S by urea or SDS. Heat denaturation of 11 S [1] showed two major components which, in turn,
279
were found to contain one monomer and at least three kinds of oligomer. There is controversy about the exact number of subunits in 11 S globulin, since Badley et al. [24] have proposed from the study on small-angle X-ray scattering and electron microscopy that this protein consists of six acidic and six basic subunits arranged in two hexagonal rings.
Since the experimental conditions of heat and urea dentauration of this protein reported before [1,23] are not identical to ours, the conformation of the denatured protein is not necessarily the same in all cases. In fact, analysis of the secondary structure of 11 S globulin suggests to us that the resultant change in conformation in the case of urea and thermal dentauration is different. It is also possible that the extent of denaturation for 4 M urea vis-a-vis 95 °C for this protein is not the same. Monitoring the S-S linkage is not very effective in IR because of rather weak peaks.
Referred ~o the derivative spectra (Fig. 10), addition of TFE shows a strong Amide I peak at 1653 cm- 1, an indication of a-helical segments, as one would expect. We also notice that there are only very weak peaks accompanying this band, which are at 1675 cm -! and 1681 cm -~. The 1476 cm -1 peak, assigned to the phenylalanine residue, seems to become prominent with the ad- dition of urea which we interpret as the residue getting more and more exposed with the adition of the denaturant. In the thermally denatured protein this is not evident in the undeconvohted spectrum and it is only a minor shoulder at -- 1470 cm-1. In the derivative spectroscopy, we have occasion- ally lowered the threshold and /o r expanded the ordinate scale considerably to check for minor peaks. Since IR adsorption profile, in general, is not a single Lorentzian or Gaussian, or, as in these specific cases, not even symmetrical, no attempt has been made to relate the original intensity and bandwidth to those of derivative spectra.
In conclusion, therefore, we show that the sec- ondary structural characteristics of the native and, thermally a~d chemically denatured 11 S globulin can be monitored very effectively by FT-IR de- rivative spectroscopy. The native protein mainly consists of anti-parallel /]-sheet, fl-turn, dis- ordered structure and very little a-helix. Addit[ol~
280
of an a-helix promoter, TFE, produces the char- acteristic a-helical peak at the expense of the fl-sheet and the disordered structure.
Acknowledgements
The work was partly supported by a grant to Battelle from t ~ Division of Research Resources (No. RR-013676), NIH. S.B.D. thanks Mr. T. Hutson of the National Center for his encourage- ment, and Mr. Robert Seebes of the Spectra-Tech for the loan of a microcircle ce~l. Mr. Rong Chen is thanked for sample preparation.
References
1 Yamagishi, T., Yamauchi, F. and Shibasaki, "K. (1980) Agfic. Biol. Chem. 44, 1575-1582.
2 Koshiyama, I. (1972) Intern. J. Peptide Protein Res. 4, 167-176.
3 Fukushima, D. (1968) Cereal Chem. 45, 203-223. 4 Koshiyama, I. and Fukushima, D. (1976) Intern. J. Peptide
Protein Res. 8, 283-289. 5 Ishino, K. and Kuda, S. (1980) Agric. Biol. Chem. 44,
537-543. 6 Thanh, V.H. and Shibasaki, K. (1976) J. Agric. Food
Chem. 24, 1117-1121. 7 Marchand, P. and Marmet, L. (1983) Rev. Sci. Instrum. 54,
1034-1041. 8 Savitsky, A. and Golay, M.J.E. (1964) Anal. Chem. 36,
1627. 9 Thomas. G.J., Jr. and Agard, D.A. (1984) Biophys. J. 46,
763-768.
10 Byler, D.M. and Susi, H. (1986) Biopolymers 25, 469-487. 11 Mantsch, H.M., Casal, H.L. and Jones, R.N. (1986) in
Spectroscopy of Biological Systems, Vol. 13 (Clark, R.J.H. and Hester, R.E., eds.), pp. 1-46, John Wiley & Sons, New York.
12 Thomas, G.J., Jr. (1986) in Spectroscopy of Biological Systems, Vol. 13 (Clark, R.J.H. and Hester, R.E., eds.), pp. 233-309, John Wiley & Sons, New York.
14 Susi, H. and Byler, M. (1983) Biophys. Biochem. Res. Commun. 115, 391-397.
15 Dev, S.B., Rha, C.K. and Walder, F. (1984) J. Biomol. Struct. Dyn. 2, 431-442.
16 Earnest, T.N., Roepe, P., Rothschild, K.J., Das Gupta, S.K. and Herzfelcl, J. (1987) in Biophysical Studies of Retinal Proteins. (Ebrey, T.G., Frauenfelder, H., Honig, B. and Nakanishi, K., eds.), pp. 133-143, University of Illinois Press.
17 Koenig, J.L. and Tabb, D.L. (1980) in Analytical Applica- tions of FT-IR to Molecular and Biological Systems (Durig, J.R., ed.), pp. 241-255.
18 Pfeil, W. and Privalov, P.L. (1976) Biophys. Chem. 4, 33-40.
19 Privalov, P.L. (1979) Adv. Protein Chem. 33, 167-241. 20 Kato, K., Matsui, T. and Tanaka, S. (1987) Appl. Spec. 41,
861-865. 21 Dev, S.B. (1987) J. Biol. Phys. 15, 57-61. 22 Anderle, G. and Mendelsohn, R. (1987) Biophys. J. 52,
69-74. 23 Kitamura, K., Takagi, T. and Shibakasi, K. (1976) Agr.
Biol. Chem. 40, 1837-1844. 24 Badley, R.A., Atkinson, D., Hauser, H. et al. (1975) Bio-
