Laser Raman spectroscopy and the conformation of insulin and proinsulin

Laser Raman spectroscopy and the conformation of insulin and proinsulin

J. Mol. Biol. (1972) 70, 117-132 Laser Raman Spectroscopy and the Conformation and Proinsulin NAI-TENG Yu, C. S. LIU AND of Insulin D. C. O’SHEA...

1MB Sizes 24 Downloads 125 Views

J. Mol. Biol.

(1972) 70, 117-132

Laser Raman Spectroscopy and the Conformation and Proinsulin NAI-TENG

Yu, C. S. LIU

AND

of Insulin

D. C. O’SHEA

Schools of Chemistry and Physics Georgia Institute of Technology Atlanta, Georgia 30332, U.S.A. (Receive& 28 March 1972) Laser-excited Raman ape&a of native and denatured insulin in the solid state and aqueous solution are reported. Striking spectral changes on denaturation have been observed in the following spectral regions: (a) the amide I and III, (b) the S-S and C-S stretching, (c) the skeletal bending and (d) skeletal stretching region. Detailed analysis of these spectral changes indicates that extensive conformational changes have taken place in the oonversion of native to denatured fibrous insulin and that fibrous insulin exists in a j-conformation as proposed by Ambrose & Elliott (1951). Comparisons have also been made between the spectra of native insulin in crystals and in solution at various pH values. Small spectral differences were observed and interpreted in terms of conformational changes

and ionization effect. We also present the first Raman spectrum of proinsulin in the solid state and compare it to that of insulin. A total of eleven lines, known to be conformation-dependent, was found to be in good agreement between these two spectra. This strongly suggests that the insulin moiety of proinsulin exists in a conformation very nearly the same as insulin itself, whioh is consistent with the conclusions of Frank & Veros (1968). The differences between these two spectra are attributed to the connecting peptide, characteristic of its conformation. A graphical subtraction between these two spectra in the amide I region shows a peak at 1663 cm-l with a shoulder near 1685 cm-l. These two frequencies are indicative of a-helical structure and random-coil form, respectively, in the C-peptide.

1. Introduction In recent years there has been a number of structural investigations of proteins by laser Raman spectroscopy (Lord & Yu, 1970a,b; Lord, 1971; Koenig, 1972; Bellocq, Lord & Mendelsohn, 1972; Yu, Liu, Culver BEO’Shea, 1972). Although some structural information of considerable biological importance has been obtained, potential usefulness of this technique for the study of proteins has not as yet been fully realized. As a further effort along this line, we have selected insulin as a model protein and carried out a systematic Raman spectroscopic study of it under various conditions. It is of particular interest to compare the spectra of insulin in crystals and in aqueous solution and determine the effect of solvent molecules on molecular conformation. Insulin is ideal for Raman study because it undergoes an interesting reversible globular-fibrous transformation and does not contain tryptophan residues. Lord & Yu (1970a,b) found that tryptophan residues give rise to many intense sharp lines and that these lines are independent of protein conformation. Because of the absence of these lines, Raman spectra of insulin are expected to be easier to interpret. 117

118 First,

N-T. we wish

conformation

YU,

C. S. LIU

to demonstrate

by comparing

that

the spectra

AND

D.

C. O’SHEA

Raman spectroscopy is sensitive to protein of native

and denatured

insulin

in the solid

state. From such comparison one is able to identify those lines which are conformationdependent. With such information at hand, we can then proceed to study the spectra of insulin in crystals and in solution at various pH values and determine the nature and extent of conformational changes associated with changes in state. We present the results on the first Raman study of proinsulin. Our purpose is to determine if the insulin moiety of proinsulin exists in a conformation different from insulin itself. Recently, Frank & Veros (1968), based on their circular dichroism studies, concluded that the insulin moiety in proinsulin did exist in a conformation very nearly the same as insulin itself and that the connecting peptide took on a random-coil conformation. A preliminary report of some of the results to be discussed below has appeared (Yu & Liu, 1972).

2. Materials and Methods (a) Spectroscopic methods The sample in powder form was packed into a conical depression at the end of an l/8 in stainless steel rod. The laser beam was generated by a Coherent Radiation model 52B argon-ion laser. After passing through an interference filter which removed the plasma emission lines, the monochromatic beam at 514.5 nm wavelength was focused and directed upward onto the sample at the grazing angle so that the scattering column was a strip on the powder surface, l/8 in long and approximately 40 microns wide. The scattered light is collected by an f/l.1 lens and imaged with a 3 : 1 magnification on the entrance slit of a Spex 1401 double monochromator. The light is detected by an ITT “Startracker” (FW-I 30) which is thermoelectrically cooled to reduce the dark count rate to a few counts per second. The photomultiplier signal is amplified and processed by standard nuclear counting electronics with suppression of a large amount of background signal. An analog signal of the count rate is recorded on a strip chart recorder at a rate of 10 cm-l/mm (or 5 cm-i/min) with a time constant which gives a 1% deviation for full scale signal. Wave number accuracy is rt 1 cm-l. The laser power at the sample was kept as low as possible so long as the signal-to-noise ratio of the spectra was acceptable. This was about 160 to 300 mW. For liquid samples, a Cary Instrument’s kinematic base and backing mirror assembly (cat no. 8240150) was used to hold a capillary cell, which is 1 mm (internal diameter) x 25 mm long and made of Pyrex. The cell with a fire-polished flat end contained about 10 ~1. of solution and was held vertically. The laser beam entered the cell from the flat end and traveled through the liquid. The light scattered at 90’ to the incident beam was then collected and focused onto the entrace slit of the double monochromator. (b) Materials

and sample handling

Bovine and porcine insulin samples (twice crystallized) were obtained from Schwa+ Mann (cat. no. 4426 and 7518) and were recrystallized before experiments according to the method of Schlichtkrull (1956). Porcine proinsulin of high purity (98% single component material) was a gift from Dr R. E. Chance of the Lilly Research Laboratories, Indianapolis, Indiana. This sample was a lyophilized powder (from acetic acid) and used without further purification. Crystalline glucagon was also obtained from Schwarz/Mann (cat. no. 9542) and used without further purification. Doubly distilled, degassed water and 0.1 M-HCl (or 0.1 m-Nash) were used in preparing the solutions of insulin at various pH values. The pH values were determined by a Radiometer pH meter model 26 with a microprobe combination electrode (Fisher cat. no. 13-63992). The accuracy is &to.01 pH unit. Solution samples were centrifuged in an International Equipment refrigerated centrifuge at a speed of 7000 rev./min for 0.6 hr before injection into the capillary cell.

CONFORMATION

OF

INSULIN

AND

119

PROINSULIN

samples of denatured insulin were prepared according to the following procedures. (a) A solution of 10 mg/ml. insulin at pH 2.42 was heated at 100°C for 45 min. The solid sample was obtained by air-drying the resulting fibrous gel; (b) same as for (a) except a tenfold increase in the concentration of solution to 100 mg/ml.; (c) a solution of 10 mg/ml. insulin at pH 2.42 was seeded with 0.5 mg/ml. of denatured insulin (obtained from (a)) for 3 days at 25°C. The resulting gel was then subjected to air-drying. The denatured insulin prepared in this way is referred to as insulin fibrils in the literature (Waugh, 1944). Three

3. Results and Discussion (a) Chain conformation and Raman amide frequencies in model compounds and proteins In an earlier communication (Yu, Liu, Culver & O’Shea, 1972) the amide I at 1662 cm-i and the shoulder near 1685 cm-l in the spectrum of native insulin were assigned to the a-helical structure and random-coil form, respectively. In order to provide additional support to this assignment, we have studied the Raman spectra of crystalline glucagon in the amide I and III regions. Glucagon is a polypeptide hormone of 29 amino-acid residues (Bromer, Sinn & Behrens, 1957). It has been reported that it has 75% or-helical structure and that the C-terminal hexapeptide segment is non-helical (Haugen t Lipscomb, 1969; Schiffer & Edmundson, 1970). Our Raman spectrum of glucagon shows the amide I at 1660 cm-l with a shoulder near 1685 cm-‘. Therefore, it appears that the assignments of 1662 cm-l and 1685 cm-l to or-helical structure and random-coil form, respectively, are correct. Additional support for the assignment of a-helix comes from the work of Peticolas and his asso.. ciates (Fanconi, Tomlinson, Nafie, Small & Peticolas, 1969) on a-helical poly-L.. alanine fibers. They have observed the amide I and III frequencies of the fibers at 1660 and 1264 cm-l, respectively. In Table 1 we have listed the strongest Raman amide I and III frequencies with their conformations for various model compounds. TABLE 1 Amide I and III

frequencies (in cm-*) of m&e1 cornpounds in various conjormdio~s

Conformation a-helix

Raman

Amide I Infrared

Poly-~&mine fibers” Glucagon (crystalline)b

1660s

Polyglyoine I”

1674s

Poly-L-1ysine*

16725

1660s

16578

Amide Raman

12648

III Infrared 12625

12668

tk?-structure: Anti-parallel Parallel Random-ooil: Solvated Unsolvated

12348 1220w

1236&I 1214W

1660 1630

Calculated” Poly-n-glutamio at pH 10 Glucagon”

1685M 16368

acid’

1665B

1248B

1685

1235

Vibrational frequencies are expressed as the displacement in cm-’ of the Raman lines. In all Tables the symbol B denotes broad, M medium, S strong (Table l), sharp (Tables 2 and 3), W weak, D doublet, T triplet, V very, sh shoulder, .(X-Y) a frequency assigned to an X-Y bondstretching vibration. Figures in parentheses are relative peak intensities based on a value of 10.0 for the strongest line in each spectrum. “Fanooni, Tomlinson, Nafie, Small & Petioolas (1969) ; bpresent work; *Wallaoh et al. (1970); eMiyazawa (1967); ‘Lord $ Yu (19706).

?!lmall

et al. (1970);

120

N-T.

YU,

C. S. LIU

AND

D.

C. O’SHEA

The random-coil frequency referred to above presumably is due to the unsolvated form. We also list the frequencies of the solvated random-coil form based on evidence found in poly-L-glutamic acid at pH 10 (Lord & Yu, 19708). The difference in frequencies between solvated and unsolvated forms is attributed to the effects of hydrogen-bonding-a stronger hydrogen bond lowers the stretching frequency (amide I) and raises the bending frequency (amide III) (Richards & Thompson, 1947). The amide I and III frequencies at 1674 and 1234 cm-l, respectively, for the antiparallel &structure are derived from the Raman spectrum of the model compound polyglycine I (Small, Fanconi t Peticolas, 1970). There may be some reluctance to accept this as characterizing frequencies for the p-structure in proteins. This is because polyglycine may be considered an inappropriate model for a protein since it has symmetry elements such as planes or centres of symmetry in ,%structure. Selection rules then determine the spectroscopic activity of various vibrations. In proteins, because of the variety of side groups, symmetry is removed and there should be no symmetry restrictions on the spectroscopic activity of the various amide I and III vibrations (Lord, 1971). Moreover, the substitution of a-hydrogen in polyglycine by long-chain side groups might have an effect on the force fields of the N-H and C=O groups. It should be noted, however, a compound with just such substitution has been studied. Wallach, Graham & Oseroff (1970) have observed a strong line at 1672 cm-l in the antiparallel ,9 form of poly-r.+-lysine. Thus, it appears that the substitution of other side groups for a-hydrogen does not affect the normal vibration frequencies of the amide I vibration as much as might be expected. We have included the available infrared data in Table 1 to enable us to compare the results from the two different methods. It is evident that there is little agreement. We believe this is not due solely to experimental error. Rather, the discrepancies are due to excitation of coupled vibrations between adjacent peptide units. In the case of model compounds, symmetry can restrict modes to be Ram&n-active, infraredactive or inactive, and the differences in frequencies reflect different normal vibrations of a coupled system. In the case of proteins, however, no such symm&y exists and all vibration should be Raman- and infrared-active. However, these vibrations retain a large oscillating dipole moment or large differential polarizability, making them “infrared-intense” or “Raman-intense” vibrations. Evidence for this will be given in our spectra of the insulin molecule which we will now consider. (b) Raman spectra of native and denatured solid insulin : a demonstration of the sensitivity of Ramn spectroscopy to protein conformation The Raman spectra (as actually recorded) of native and denatured insulin (bovine) in the solid state are presented in Figure 1. These two spectra, after being redrawn to remove the noise, are superimposed for comparison in Figure 2. The vibrational frequencies in cm-l with assignments are given in Table 2. Comparison of these two spectra reveals some interesting spectral changes. From the reversibility of the biological activity of insulin (Waugh, 1954), we assume that the spectral changes observed here are due only to changes of the molecular conformation. From the detailed analysis of these spectral changes given below, one can gain some new insights into the nature of this structural transformation. The first spectral change to be discussed is in the amide I region. The amide I line has shifted from 1662 cm-’ (with a shoulder near 1685 cm-l) to 1673 cm-l and has sharpened considerably upon denaturation. A similar spectral change has also been

CONFORMATION

725 I

1

OF

054 *321\

513

I

I

I

800

It21

PROINSULIN

12'4.;;';,,

II I1

I

AND

1239 '2!2 1770/1288

963

853

600

INSULIN

I

1000

1

!

I

1200

1400

1

1600

Raman shift A cm-l FIG. 1. Raman spectra of native and denatured insulin (bovine) in the solid state. (a) Native sensitivity, 8, 1000 ots/sec zinc-insulin crystalline powder. Spectral slit width, A Q, 4 cm-‘; full scale; rate of scan, I, 10 cm-l/min; standard deviation, d, 1%. (b) Denatured insulin (heatprecipitated). The solution from which the sample was prepared hed a concentration of 10 mg insulin/ml. at pH 2.42. The method of denaturetion has been described in the text. A (r, 6 cm-l; s, 5000 cts/sec; T, 10 cm-l/min; d, 1%. The leser power, p, at both samples was 200 mW at 514.5 nm.

observed by Lord (1971) and Mendelsohn (1972) in work on the chemical denaturation of lysozyme. In that study it was found that the amide I had shifted from 1660 cm.-l (in native lysozyme) to 1675 cm-l (in S-cyanoethylated lysozyme). A sharpening of the amide I line was also observed. They interpreted the spectral changes as due to the weakening of hydrogen-bonding, and the sharpening of the amide I line as a reflection of the greater uniformity in the hydrogen-bonding. Surely, changes in the strength of hydrogen bonds in the amide groups will affect the frequency of the C=O stretching (amide I) vibration. As we remarked earlier, a stronger hydrogen-bond gives rise to a lower stretching frequency (amide I) (Richards & Thompson, 1947). This interpretation, however, is incomplete. Other factors, such as the coupling between the adjacent peptide units (both intra- and inter-chain couplings), should also play a part in determining the frequency of the amide I vibration. The evidence of strong coupling between adjacent peptide units comes from the present Raman work and the infrared results of Ambrose BEElliott (1951) They studied the denaturation of insulin by infrared techniques and found that the denaturation causes the infrared amide I to shift from 1657 cm-l to 1637 cm-‘. To

N-T.

122

YU,

C. S. LIU

AND

TABLE

D.

C. O’SHEA

2

Rantan. spectra of inmdin (bovine) (ZOOto 1800 cm-‘)

Netive

333 410 467 496

Frequencies in cm-i Denetured (CrySt8ls)

(0.9) (043) (0.8) (1.2)

515 (3.2)

265 325 420 460 480

Tentative

assignments

(solid)

(0.9) (0.8) (0.5) (0.4) (0.3)

513 (4.4)

I

Skeletal bending

v(S-S)

532 (1.5sh)

Skeletal bending

663 (1.0) 624 (2.05) 644 (3.65)

624 (2.05) 644 (3.6s)

Phe TF

668 (2.0) 678 (l.Osh)

657 (1.1) 680 (1.3)

Y(C-S)

725 (06T) 747 (O-ST) 770 (06T)

737 (0.9B)

Skeletal bending

832 (4.4D) 864 (5GD)

830 (3.9D) 853 (4.5D)

TY~

900 934 946 963

882 (1.6) 922 (0-S)

of the C-S-S-C

814 (1.4sh)

(2.0) (2.0sh) (3*2D) (2*9D)

1604 (10.0s) 1032 (3.35) 1112 (1*&h) 1128 (1.8) 1162 (0.9) 1177 (2.4) 1212 (4.65)

966 (1.6B) 1004 1020 1032 1067

(lO*OS) (2.6) (3.35) (1.3sh)

1127 (1.7) 1161 (OGh) 1175 1214 1227 1262

(2.4) (4.9) (4.3) (4eOsh)

1239 (5.Osh) 1270 (6.3) 1288 (4*7sh) 1322 (29sh) 1344 (4.0) 1367 (l&h) 1426 (2.5sh) 1450 (6.0) 1462 (46sh) 1687 1607 1616 1662

(1.3) (3.6D) (3.6D) (4.6)

I

Y(C -C) Phe v(C-N) Phe

I

v(C -N)

TF Tyr

& Phe

Amide III

(/Mructure)

Amide III Amide III Amide III

(random-coil) (e-helical) (a-helical)

1327 (2.OD) 1343 (3*1D)

CH deformation

1407 (0*4sh) 1422 (l.lsh)

Symmetrical

1460 1462 1587 1607 1616

(3.8) (3*lsh) (l.Osh) (36D) (3.5D)

1673 (8.6s) 1685 (4.Osh) 1735 (0.4B)

COs- stretching

CHP deformation Phe Phe & w Amide Amide Amide -COOH

Tyr I (N-helical structure) I (fi-structure) I (random-coil)

group

CONFORMATION

OF

INSULIN

AND

PROINSULIN

123

lnsulln

I

-----

Notwe

-

Denatured

I 600

I

Amide I

I 800

I

t 1000

I

Roman shift

I 1200

I

I -l-. 1400

1600

A cm-t

FIQ. 2. Superimposed comparison between the spectrs of native and denatured insulin (redmwn from Fig. 1). The line at 624 cm-l due to the in-plane ring vibration of phenylalenine residues is known to be conformation-independent and is used as an internal reference. This line has nearly equal intensity in both spectra.

---- Native -Denatured

Amide I Jo) ix.

(b)

FIQ. 3. Comparison

Roman

of infrared

and R&man amide I frequencies

of insulin.

124

N-T.

YU,

C. S. LIU

AND

D.

C. O’SHEA

confirm this observation, the infrared spectra of native and denatured insulin (in Nujo mull) in the amide I region were obtained in our laboratory. A comparison between infrared and Raman amide I frequencies of native and denatured insulin is shown in Figure 3. These two techniques obviously are not detecting the same amide I vibrations in insulin. In the denatured state, one coupled mode gives rise to the Raman-intense line at 1673 cm-l, while the other coupled mode appears as an infrared-intense band at 1635 cm-l. A comparison between the amide I and III frequencies of denatured insulin and those listed in Table 1 for model compounds leads us to conclude that the denatured insulin molecule exists in a /?-conformation. Whether this is parallel or antiparallel form we cannot tell from our data. On the basis of infrared dichroism studies on synthetic polypeptides anddenatured insulin, Ambrose & Elliott (1951) concluded that denatured insulin consists of extended chains lying perpendicular to the fibril axis, with a layer structure involving interchain hydrogen-bonded grids (i.e. cross-/3 structure). Thus, a mechanism to explain denaturation of insulin was suggested in which precipitation was caused by a change from intrachain to interchain hydrogen bonds. Recently, Burke & Rougvie (1972) have conducted electron microscopy, lowangle X-ray diffraction, U.V. circular dichroism and infrared dichroism studies and afhrmed the essential correctness of the Ambrose & Elliott proposal. It should be noted that because of the reversibility, the conformation of denatured fibrous insulin had been considered as the linear aggregate of only slightly distorted native insulin (Waugh, 1944; Koltun, Waugh & Bear, 1954; Reithel, 1963; Beaven, Gratzer & Davies, 1969). The present results indicate that the change in insulin upon denaturation is more drastic than had been previously thought. The next important spectral feature is the N-H in-plane bending mode (amide III). In this region the “center of gravity” of the complex line-shape has shifted from 1260 cm-l to about 1230 cm-l. Again, this same effect has been observed by Lord (1971) and Mendelsohn (1972) in their work on denaturation of lysozyme. In that case, the amide III line had shifted from 1260 cm-’ to I247 cm-l upon denaturation. On the basis of the model compounds and the agreement with the conclusions from the amide I results, we assign the line in the native insulin spectra at 1270 cm-l (with a shoulder near 1288 cm-l) to the a-helical form and the one at 1239 cm-l to both the random-coil and fl-structure. The contribution of p-structure to this line is expected to be minimal because of the small amount of /3-structure present in native insulin crystals (Blundell et al., 1971). The spectral feature at 1230 cm-l for the denatured sample agrees with the amide III frequency for the model compounds with known @tructure and with our conclusion drawn from the amide I line. In the region 450 to 700 cm-l, the S-S and C-S stretching vibrations of the C-S-S-C group appear. Lord & Yu (1970a) have shown that both the frequencies and the intensity ratio of the C-S and S-S lines depend on the dihedral angle and C-S-S bond angle. The change in the intensity ratio is probably due to a change in the polarizability of the bonds as a result of the geometrical change of the disulfide cross-links. In native insulin the S-S frequencies of the three d&sulfide linkages appear as an unresolved broad line centered at 515 cm-l (half-widths 26 cm- l), while the C-S frequencies show up at 668 and 678 cm-l. According to the earlier report (Yu, Liu, Culver & O’Shea, 1972), the intrachain disulfide bond, A6-11, may have a local geometry different from that of the inter chain disulfide bonds, B7-A7 and B19-A20. Although there is a possibility that the line at 495 cm-’

CONFORMATION

OF

INSULIN

AND

PROINSULIN

125

may also be assigned to the S-S stretching vibration, we do not do so because no S -S frequency lower than 500 cm-l has been observed in model compounds (Lord & Yu, 1970a; Yu, 1969). Even in a highly strained, closed ring system such as thioctic acid, the S-S stretching frequency appears at 512 cm-l. Recently, Mendelsohn (1972) has made a further study on those S-S-containing compounds such as L-homocystine (solid), L-homocystine (5 N.HCl), L-homocystine (0.1 N-NaOH), diethanol disulfide, formamidine disulfide (solid), formamidine disulfide (pH 2), bovine serum albumin and trypsin. The only exception observed to date is the S-S frequency of formamidine disulfide at 479 cm- l. The reason for this unusually low S-S frequency is not known. Still, on the basis of comparable model compounds we are not convinced that the 495 cm-l is an S-S stretching frequency. When insulin is denatured, the spectral changes occurring in the S-S and C-S stretching region are quite informative. The peak intensity of the S-S frequency has increased, and the line at 668 cm-l has shifted to 657 cm-l with a considerable decrease in intensity. The line at 678 cm-l has slightly shifted to 680 cm-l with no appreciable change in intensity. This observation suggests that the geometry of the two interchain disulflde links in the denatured state is different from that in native form and that the intrachain one remains in nearly the same conformation upon denaturation. The skeletal bending frequencies occur in the region below 800 cm-l and are expected to be very sensitive to protein conformation. In Table 2, seven Raman lines in the spectrum of native insulin are tentatively assigned to this category. Numerous spectral changes between 470 and 800 cm-l are clearly seen in Figures 1 and 2. For example, the lines at 495 and 563 cm-l disappear and a new line at 532 cm-l appears upon denaturation. In the region near 750 cm- I, the broad line has been resolved into three lines at 725, 747 and 777 cm-l. When insulin is denatured, these three lines coalesce into one peak at 737 cm-l. This behavior is quite similar to the sharpening of the amide I line upon denaturation. Miyazawa, Shimanouchi & Mizushima, (1958), based on normal mode vibration calculations of model compound N-methylacetamide, have assigned a line near 725 cm-r to the N-H out-of-plane bending (amide V). One is tempted to assign the triplet near 750 cm-l to specific conformational components of the native form and the single line at 737 cm-l in the spectrum of the denatured sample to the /3-conformation, as we have in the case of the amide I line. We regard this assignment as tentative, since deuteration of denatured insulin does not result in a complete disappearance of the 737 cm-l feature from the spectra. Finally, we have observed interesting changes in the 800 to 1220 cm-l region, where the C-C and C-N coupled skeletal stretching vibrations occur. Spectral changes in this region upon denaturation of a protein have been predicted earlier by Lord & Yu (197Oo). The lines at 900, 934, 946, 963 and 1112 cm-l have changed quite drastically, indicating an extensive unfolding of the protein backbone. Since the native insulin crystals were grown at neutral pH and the denatured insulin was prepared at pH 2.42, one might wonder if the spectral changes are caused partially by the presence of HCl in the denatured sample. A Raman spectrum of native insulin HCl solid is shown in Figure 6(a). The protonation of the carboxyl groups and imidazole rings does produce slight changes in this region. However, the spectra of native insulin HCl (solid) and denatured insulin are still drastically different in this region. The line at 1128 cm-l cannot be attributed to one of the characteristic ring frequencies of the aromatic side groups and is found to be independent of conformation. This line

126

N-T.

YU,

C. S. LIU

AND

D.

C. O’SHEA

was also observed in the spectra of a-chymotrypsin (Lord & Yu, 197Ob). In general, spectral lines in this region are not well understood. (c) Rarnan spectra of denatured insulin obtained under different conditions In Figure 4(a) we present another spectrum of denatured insulin, prepared under the same conditions as the sample discussed in the previous section (heating at 100°C for 45 min) except for a tenfold increase in concentration. In Figure 4(b) we show the

1673

I 200

I

I 400

I

I 600

I

I BOO

, , I 1000 1200 Roman shtfi A cm-l ,

,

, 1400

]

, 1600

/

/ 1800

/

FIG. 4. REtman spectra of various denatured insulins in the solid state. (a) Conditions for sample preparation are the same a8 for Fig. l(b) except a lo-fold increase in concentration of the solution. d u, 4 cm-l; a,2500 cts/sec; r, 10 cmDX/min; d, 1%; a, 180 mW. (b) Denatured insulin obtained by seeding. A u, 3.8 cm-‘; 8, 2600 &s/n&; r, 10 cmwl/min; d, 1%; p, 1% mw.

Ramtln spectrum of the insulin fibrils prepared by seeding at 25°C for 3 days. It has been demonstrated (Waugh, 1954 and references cited therein) that if the insulin fibril is introduced by seeding into a solution of insulin at room temperature, the fibril will grow and progressively remove native insulin from solution. By means of X-ray diffraction, Koltun et al. (1954) have examined the insulin fibrils obtained by seeding and by heating and found that the diffraction patterns were identical. Comparing these spectra to Figure l(b), a few noteworthy differences are observed. For example, the C-S stretching line near 668 cm-l is much weaker in Figure 4(a) and (b). Aooording to Lord & Yu (1970a), the weakening of the C-S stretching frequency is related to B change in the C-S-S bond angle. On the basis of the model oompoun& studies of Lord & Yu (197&), it appears that the C-S - S angles of the three disulfide links in these latter examples of dena$ured insuk~ are about 114” and those in native insulin about 103”.

CONFORMATION

OF

INSULIN

AND

127

PROINSULIN

Examination of Figure l(a) and (b) and Figure 4(a) and (b) in the 500 cm-l region shows a complementary relationship between the 495 cm-l and the 532 cm-l lines. We also note there are small changes in the amide V region around 750 cm-l. It is also of interest to note that the new line at 1020 cm-l in Figure l(b) may appear as either the shoulder of the 1032 cm-l line or that of the 1004 cm-l line (see Fig. 4(a) and (b)), depending on the conditions of denaturation. The amide III (at 1227 cm-l) appears to be resolved in the high-concentration denatured sample. With these minor exceptions, the three spectra of denatured insulin are similar. The agreement, found in the amide I and amide III regions suggests that all denatured samples oi insulin have the same chain conformation @-conformation). All the foregoing Raman spectra of denatured insulin have been obtained in the solid state. In order to determine the effect of water molecules on the conformation of denatured insulin, we have obtained the spectrum of an aqueous gel of denatured insulin (Fig. 5). The gel was prepared by seeding at 25°C. Difficulties usually exist in

I

I

I 600

I

I 000

I

I 1000

I

I

1200 Roman shift A cm“

FIQ. 6. Raman spectrum of wet gel of denatured r, 5cm-1 /min; d, 1%; p, 250 mW. X, spurious.

I

I

1400

I

I

1600

insulin (by seeding). A o, 6 am-l;

I

s, 500 cts/sec;

obtaining good Reman spectra of turbid, gel-like solutions of biological macromolecules (Lord $ Yu, 1970a; Lord & Mendelsohn, 1972). Such turbid solutions (optically heterogeneous) prevent free entrance of the exciting radiation into the solution and free egress of the Raman scattering therefrom (Mendelsohn, 1972). As is evident, the speotrum shown in Figure 5 is somewhat inferior in quality. The amide I and III regions, however, agree closely with the spectra of insulin fibrils in the solid state, indicating that the /3-structure of insulin fib& remains in aqueous environment. This may be understandable if the cross -/3 structure model (Ambrose & Elliott, 1951; Burke & Rougvie, 1972) (non-polar residues on the surface of the fibrils serve to protect the interior hydrogen-bonding from disruptive effects of polar solvents) is correct. The rest of the spectrum is somewhat diffuse, but shows notable differences in the 600 to 800 cm-l region, as compared to that of Figure l(b). The lines at 624 and 644 cm-l, known as conformation-independent, are weakened (with reference to the

N-T.

128

YU,

C. S. LIU

AND

D.

C. O’SHEA

S-S stretching at 513 cm-l). The origin of this intensity decrease is not clear at present. In summary, the small differences observed between the spectra of various denatured insulin samples may be due to a slight difference in the chain folding and the packing of the side groups under various conditions. (d) Comparison of solution and solid-state spectra of native insulin In the absence of methods for determining the structures of proteins in solution, biochemists have had to use X-ray diffraction studies with protein crystals and assumed that the protein structures are the same in crystals and in solution. Since we have demonstrated that many Raman lines in the spectrum of insulin depend on its structure and conformation, it is of interest to compare the Raman spectra of insulin in crystals and in solution and determine the effect of solvent molecules on the insulin conformation. The spectra of aqueous insulin at pH 2.40 and 8.30 are shown in Figures 6(b) and 7. Under comparison, it shows that there is a small shift in the line-

(b)

I zoo

Aqueous

insulin HCI

I

I

I

400

600

800

I

1000

! 1200

1400

I

1600

I

1800

J

Roman shift Ll cm-t

FIG. 6. Comparison between the Raman spectra of insulin in the solid state and in solution. The aqueous solution had a concentration of 100 mg insulin/ml. at pH 2.40. The insulin HCI solid d, 1%; was obtained by air-drying the solution. (a) A o, 6 cm-'; 8, 2500 cts/sec; r, 10 cm-l/min; d, 1%; p, 180 mW. p, 198 mW. (b) A u, 4 cm-‘. , 8, 2500 cts/sec; r, 10 cm-‘/min;

shape in the amide III region between insulin HCl (solid) and aqueous insulin HCl (Fig. 6(a) and (b)). This may indicate that on dissolution the relaxation of the backbone chain and side groups results in a slight increase in random-coil structure. Slight changes in line-shape also occur near 950 cm- l. Except for these small differences, the rest of the spectra shows good agreement. When the pH value was changed from 2.40 to 8.30, no significant changes were observed in the amide I and III regions.

CONFORMATION

OF

INSULIN

AND

PROINSULIN

129

r

,

I

I

I

,

600

400

I

800

--AL_,

1:

1000 1200 Raman shift A cm-'

FIG. 7. Reman spectrum of aqueous insulin at pH 8.30. Concentration, d, 1%; p, 180 mW. 8, 2500 cts/sec; r, 10 om-l/min;

1600

1460

L-

100 mg/mI. A o, 8 cm-‘;

t‘ 1 3

i

I

/

I

400

I

600

FIG. 8. R&man speotof proinsulin (port+) in the solid state. The sample was powder lyophilized from acetic acid. A CT,4 cm-' ,* 8, 1000 cts/seo; r, 10 cm-l/min; d, 1%; p, 200 mW. 9

N-T.

130

YU,

C. S.

LIU

AND

D.

C. O’SHEA

This indicates, so far as one can tell from present Raman results, that there are no conformational changes. The spectral changes noted in the 900 to 1000 cm-l region might be explained in terms of the ionization of the side-chain carboxyl groups and the imidazonium rings. The intensity increase at 950 cm-l might be due to the C-C stretching of the ionized glutamic side chains (Lord & Yu, 1970b) and the line at 990 cm-l in the spectrum of aqueous insulin HCl due to the imidazonium rings of the histicline residues (Yu, 1969). (e) Rawmn spectra of proinsulin;

conformation of the insulin peptide in proinsdin

moiety and the connecting

Does the insulin moiety in proinsulin exist in a conformation different from the free insulin ‘2What is the conformation and function of the connecting peptide ‘2We have obtained a Raman spectrum of a 0*2-mg sample of porcine proinsulin, which is shown in Figure 8 and tabulated in Table 3. This spectrum, after being redrawn to remove the noise, is compared with that of porcine insulin crystals shown in Figure 9. Comparison reveals that more than eleven conformation-dependent lines (st,arred in Table 3) agree in both spectra. This indicates that the insulin moiety exists in a conformation very nearly the same as insulin itself. The sharp line at 267 cm-l has shown up consistently in proinsulin spectra. It may be due to a grating ghost, The next question to be answered is one concerning the conformation of the connecting peptide consisting of 33 amino acid residues. If the conclusion about the conformation of the insulin moiety in proinsulin is correct, then the differences between the two TABLE 3

Raman spectrum of porcine proinsulin Frequencies

in cm-l

612 560 624 644

(l.OB) (0.7) (1.4)* (1.0) (3.2B)’

Skeletal v(S-S) J

680 666 (O&h)* (1*6)*

1

(1.75) (0.7sh)* (0.7)* (0.7sh)* (2.Osh)

853 831 (MD) (4.4D) 898 (8.05) 945 (3.6)* 961 919 (2.6)* (2.6)

bending

I

(0.3sh) (O+)* (2.15) (3.75)

741 750 769 726 814

Frequencies

Spurious ( ?)

267 (3.85) 406 445 495 308 515

Assignments

(lyophilized powder) (200 to 1800 cm-‘)

Skeletal

bending

Phe TY~ .(C-S) Skeletal

bending

I

1

TY~ v(C-C)

I

in cm-l

Assignments

1004 (10.05) 1019 (1.0s)

Phe v(C -N)

1032 (4.0s)

Phe

1106 1128 1160 1081 1177 1206 1246 1270 1298 1314 1341 1420

(1.3)* (2.6) (1.6) (0.6)* I (3.0) (50s) (5.5) (6.6) (3.25) (I.Osh) (6.5) (1.5sh)

1450 1636 1587 1607 1615 1663 1680

(9.5) (0.4) (1.3) (3.6D) (3.6D) (9.0) (6.0)

of C-peptide

v(C-N)

Tyr

Tyr and Phe Amide III Amide III CH deformation

Symmetrical CO; Stretching CHa deformation Phe Phe and Tyr Tyr Amide I Amide I

See legend to Table 1 for symbols. * Indicates crystals.

conformation-dependent

lines, which

are also present

in the spectrum

of insulin

CONFORMATION

OF

INSULIN

AND

131

PROINSULIN

-Prqinsulin -----Insulin

\

I

I

1

I

400

I

I

600

,

1

800

/

I

IO00

/

I

1200

1

I

1400

I I-1-1600

Roman shlff A cm-’

Fm. 9. Comparison between Ramen spectra of proinsulin 8nd insulin in the solid state. Both are of porcine origin. The spectrum of proinsulin was redrawn from Fig. 8. The insert shows our graphical subtraction between the two spectra in the amide I region.

spectra is the contribution from the C-peptide, characteristic of its conformation. Because the C-peptide does not contain any Phe residues and the 624 cm-l line is considered to be conformation-independent (Lord & Yu 1970a), the spectra have been normalized by recording them with the 624 cm-l line intensity adjusted to the same value. A graphical subtraction in the amide I region shows that the C-peptide contributes an amide I line at 1663 cm-l with a shoulder near 1685 cm-l. Since these two frequencies are characteristic of a-helical structure and random-coil form, respectively, we conclude that the C-peptide moiety in proinsulin contains a considerable fraction of a-helical structure. In fact, X-ray crystallographic studies (Fullerton, Potter & Low, 1970) have indicated the presence of two short segments of u-helices in the C-peptide of proinsulin. Frank & Veros (1968) have studied circular dichroism spectra of proinsulin and insulin. They concluded that the insulin moiety existed in the same conformation as in insulin itself and that the connecting peptide took on a randomcoil conformation. Several attempts have been made to combine the reduced A and B chains of insulin by oxidation and resulted in very poor yields (Dixon & Wardlaw, 1960; Du, Zhang, Lu & Tsou, 1961). On the other hand, it has been demonstrated that the reduced proinsulin undergoes spontaneous reoxidation with oxygen to form biologically active insulin (Steiner & Clark, 1968). These facts and the present results imply that the function of C-peptide is to bring the necessary cysteine residues into jux.taposition for the correct formation of disulfide links and that once the three disulfide links are formed, the insulin moiety becomes highly stabilized and the cleavage of the

132

N-T.

YU,

C. S. LIU

AND

D.

C. O’SHEA

C-peptide from proinsulin will not appreciably change the existing conformation of the insulin moiety. The C-peptide, however, may undergo conformational changes when it is released. Raman studies of the free C-peptide are now in progress. This work was supported by grant GM 18894-01 of the U. S. National Institute of General Medical Soiences and a Frederick Gardner Cottrell grant from Research Corporation. We thank Dr R. E. Chance of the Lilly Research Laboratory proinsulin of high purity.

for the 2-mg sample of porcine

REFERENCES

Ambrose, E. J. & Elliott, A. (1951). Proc. Roy. Sot. A., 208, 75. Beaven, G. H., Gratzer, W. B. & Davies, H. G. (1969). Ewwp. J. Biochem. 11, 37. Bellocg, A. M., Lord, R. C. & Mendelsohn, R. (1972). B&him. biophy8. Acta, 257, 280. Bhmdell, T. L., Cutfield, J. F., Cutfield, S. M., Dodson, E. J., Dodson, G. G., Hodgkin, D. C., Mercola, D. A. t Vijayan, M. (1971). Nawre, 281, 506. Bromer, W. W., Sinn, L. G. & Behrens, 0. K. (1957). J. Amer. Chem. Sot. 79, 2807. Burke, M. J. & Rougvie, M. A. (1972). Biochemistry, 11, 2435. Dixon, G. H. & Wardlaw, A. C. (1960). Nature, 188, 721. Du, Y.-C., Zhang, Y.-S., Lu, Z.-X. & Tsou, C.-L. (1961). Sci. Xi&a, 10, 84. Fanconi, B., Tomlinson, B., Nafie, L. A., Small, W. t Petiaolas, W. L. (1969). J. Chem. Phye. 51, 3993. Fullerton, W. W., Potter, R. & Low, B. W. (1970). Proc. Nat. Ad. Sci., Wash. 66, 1213. Frank, B. H. & Veros, A. J. (1968). Biochem. Biophya. Rm. Corm. 32, 155. Haugen, W. P. & Lipscomb, W. N. (1969). Actu Cry&. A-25 (suppl.), 5185 (XV-25). Koenig, J. L. (1972). J. PoZy. Sci. part D. In the Press. Koltun, W. L., Waugh, D. F. & Bear, R. S. (1954). J. Amer. Chem. Sot. 76, 413. Lord, R. C. (1971). 23rd Int. Gong. Pure Appl. Chem., Supplement to Pure and Appl. Chem. 7, 179. Lord, R. C. & Mendelsohn, R. (1972). J. Amer. Chem. Sot. 94, 2133. Lord, R C. & Yu, N.-T. (1970a). J. Mol. Biol. 50, 509. Lord, R. C. & Yu, N.-T. (1970b). J. Mol. Biol. 51, 203. Mendelsohn, R. (1972). Ph.D. Thesis, Massachusetts Institute of Technology. Miyazawa, T. (1967). In Poly-or-Amino Acids, ed. by G. D. Fasman, p. 69, New York: Marcel Dekker, Inc. Miyazawa, T., Shimanouchi, T. & Mizushima, S. (1958). J. Chem. Phys. 29, 611. Reithel, F. J. (1963). Adw. in Protein Chem. 18, 159. Richards, R. E. & Thompson, H. W. (1947). J. Chem. Sot. p. 1248. Schiffer, M. & Edmundson, A. B. (1970). Biophye. J. 10, 293. Schlichtkrull, J. (1956). Actu Chem. Stand. 10, 7. Small, E. W., Fanconi, B. & Peticolas, W. L. (1970). J. Chem. Phys. 52, 4369. Steiner, D. F. & Clark, J. L. (1968). Proc. Nut. Acud. Sci., Wash. 60, 622. Wallach, D. F. H., Graham, J. M. & Oseroff, A. R. (1970). FEBS Letters, 7, 330. Waugh, D. F. (1944). J. Amer. Chem. Sot. 66, 663. Waugh, D. F. (1954). A&. in Protein Chem. 9, 326. Yu, N.-T. (1969). Ph.D. Thesis, Massachusetts Institute of Technology. Yu, N.-T., Liu, C. S., Culver, J. & O’Shea, D. C. (1972). Biochim. biophye. Acta, 263, 1. Yu, N.-T. St Liu, C. S. (1972). J. Amer. Chem. Sot. 94, 3250.