Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
¹⁹F-NMR studies of trifluoroacetyl insulin derivatives
(USC Thesis Other)
¹⁹F-NMR studies of trifluoroacetyl insulin derivatives
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
9 f- n m r s t u d ie s o f t r if l u o r o a c e t y l in s u l in d e r iv a t iv e s Part I: Preparation and C haracterization o f Several Trifluoroacetyl insulin Derivatives Part II: ^ F -N M R Studies-the Effects o f Solvent, pH, Salts and Denaturants on the Conformation and Aggregation Properties o f Trifluoroacetylated Insulin Derivatives by Richard Alan Paselk A Dissertation Presented to the FACULTY O F THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial F u lfillm ent o f the Requirements for the Degree DOCTOR OF PHILOSOPHY (Biochemistry) January 1976 U N IV E R S IT Y O F S O U T H E R N C A L IF O R N IA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CALIFORNIA 9 0 0 0 7 This dissertation, written by Richard Alan Paselk under the direction of A L S . Dissertation Com mittee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of requirements of the degree of D O C T O R O F P H IL O S O P H Y Dean Date ........ ). 1 . J . .S L \.. . '1 DISSERTATION COMMITTEE j Chairman P T IR 8Io > V G P2.S I PREFACE The protein hormone insulin has been studied intensively for over 50 years using a wide variety o f physical and chemical techniques. Recently the detailed structure of crystalline zinc insulin has been elucidated by X -ray crysta llo - graphic techniques. However, the conformation o f insulin in solution and the relationship o f its structure to biological function are s till not fu lly elucidated. There is a considerable body o f evidence to indicate that the conformation o f insulin in solution is sim ilar to its conformation in the crystalline state. Most o f the tools presently available for solution studies do not possess su fficie nt re solution to determine the detailed structure o f insulin in solution. N uclear magnetic resonance (NMR) spectroscopy is a technique w ith a potential resolving power approaching that o f X -ray crystallography, at least for small molecules. The NMR spectra o f large molecules are generally too complex to obtain this inform ation. However, by the use of specific reporter groups, high resolution studies may be made o f specific regions o f a macro- 19 m olecule. In the research reported here fluorine ( F) has been used as a 19 reporter group attached to insulin, in conjunction w ith F-N M R , to gain in formation about three specific sites on the insulin m olecule. A comparison o f this inform ation w ith the properties o f these sites predicted by the crystalline structure o f insulin allows certain conclusions to be drawn regarding the solution structure o f insu lin. This dissertation is presented in two parts. Part I describes the preparation and p u rifica tio n o f a number of trifluoroacetylated insulin derivatives. These derivatives are characterized chem ically and b io lo g ica lly and the ir properties are compared w ith native insulin. Part II describes ^ F -N M R studies o f these derivatives, u tiliz in g the triflu o ro a ce tyl moiety as a ^ F -N M R reporter group. The effects of pH, solvent, divalent z in c , salts, quanidine hydrochloride and sodium dodecyl sulfate were observed in these studies. Lastly, the results of these studies are discussed in the lig h t o f the structure of crystalline zinc TABLE OF CONTENTS PREFACE LIST OF FIGURES LIST O F TABLES PART I Introduction - Part I M aterials Experimental Amino A cid Analysis Thin Layer Chromatography Column Chromatographic Procedures Reaction o f Insulin w ith Trifluoroacetate in W ater Reactions o f G lycin e M ethylester and Insulin w ith Ethyl T hioltrifluoroacetate in Dimethylformamide Reaction o f Insulin w ith Ethyl Thioltrifluoroacetate in Dimethylformamide: C orrelation of Kinetics w ith Product Formation Preparation and Isolation of T ri-triflu o ro a c e ty l- insulin Preparation o f M on o-, D i- and T ri-triflu o ro a c e ty l- insulin Derivatives Determ ination o f Free Amino Groups on Insulin by D initrophenylation Deamination o f T riflu o ro a ce tyl-in su lin Derivatives Sulfhydryl Group Analysis The Reaction o f Trypsin w ith T riflu oroa ce tyl-insu lin Derivatives Biological and Immunological Assays Isoelectric P recipitation C rystallization o f G ly c in e ^ - ^-T riflu o ro a ce tyl- insulin Results C haracterization of the Reaction o f Insulin HCI w ith Ethyl T hio ltrifluoroacetate in Dimethylformamide Chromatography o f T riflu oro a ce tyl-in su lin Characterization o f T ri-T riflu o ro a ce tyl-in su lin C haracterization o f the M on o-, and D i-T riflu o ro - a ce tyl-in su lin Derivatives Biological and Immunological A c tiv ity C rystallization o f G ly c in e ^ ” ^-T riflu o ro a ce tyl- insulin Reaction o f Insulin Hydrochloride w ith Ethyl T hiol trifluoroacetate in Aqueous Buffer Discussion The Trifluoroacetylation Reaction T r i- tr i flu o ro a ce tyl-in su lin M ono- and d i-triflu o ro a c e ty l-in s u lin Derivatives Biological and Immunological A c tiv ity PART Page II 55 Introduction - Part II 56 M aterials 65 T riflu o ro a ce tyl-p h e n yla la n ylg lycylg lycin e 66 N , € -ly s in e -triflu o ro a c e ty l-in s u lin -o c ta p e p tid e 66 G ly c in e ^ - ! -triflu o ro a c e ty l-S -s u lfo -in s u lin A Chain 66 G ly c in e ^ - ^ -ph enylalanine^- ^ -tr i flu o ro a ce tyl- 67 desoctapepti d e-i nsu I i n T rifluoroacetylation o f Insulin 67 Methods 68 N uclear M agnetic Resonance Measurements 68 Sedimentation V e lo c ity Studies 69 C ircu lar Dichroism Measurements 69 Results 71 ^ F -N M R Studies o f Z in c -fre e Insulin Derivatives 71 and Model Compounds as a Function o f pH and Solvent Chemical C haracterization o f Model Compounds 71 ^ F -N M R Spectroscopy o f Model Compounds 74 ^ F -N M R Spectroscopy o f Z in c -fre e Insulin 77 ^ F -N M R Studies o f Z in c T riflu o ro a ce tyl-in su lin 85 Derivatives at pH 6 .8 : The Effects o f Perturbants Sodium C itrate and Sodium Acetate 88 Potassium Thiocyanate 95 G uanidine Hydrochloride 96 vi Page Sodium Dodecyl Sulfate 102 C ircular Dichroism Spectra 109 Far UV Spectra 109 N ear UV Spectra 119 Sedimentation V e lo c ity Studies 128 Discussion 139 Model Trifluoroacetyl Derivatives 139 Z in c -fre e T riflu oro a ce tyl-in su lin Derivatives: 140 The Effects o f pH and A ce tic A cid Studies of T riflu oro a ce tyl-in su lin Derivatives 146 in the Presence of Z in c Comparison w ith Z in c -fre e Results 146 Effects of Low Concentrations o f Salts 147 The Effects o f Potassium Thiocyanate 149 The Effects o f G uanidine Hydrochloride 151 The Effects o f Sodium Dodecyl Sulfate 154 Conclusion 159 References 162 vii LIST OF FIGURES Scheme describing the synthesis o f m ono-, d i- and tri-triflu o ro a c e ty l insulin derivatives using phenyl- trifluoroacetate or ethyl th io ltriflu o ro a ce ta te . The kinetics o f the reaction o f ethyl thiol — tr i— fluoroacetate w ith amino groups in the presence o f triethylam ine or im idazole. The reaction o f ethyl thio ltriflu o ro a ce ta te and in sulin hydrochloride in the presence o f triethylam ine: a comparison o f the progress o f the reaction and the product spectra. DEAE-Sephadex chromatography w ith 0.01 M Tris buffer in 7M urea w ith a 0 .0 4 -0 .2 3 M NaCI grad ient (pH 7 .9 ) o f tri-triflu o ro a c e ty l-in s u lin and an insulin co n tro l. Sephadex G -5 0 chromatography w ith 0.01 M Tris buffer, 0 .04 M NaCI in 7M urea (pH 7.9 ) o f fr i- t ri f I uoroa cety I - i nsu I i n . DEAE-Sephadex chromatography w ith 0.01 M Tris buffer in 7M urea w ith NaCI gradient of triflu o ro - acetylated insulin derivatives. Crystals o f g ly c in e ^ ” ! -triflu o ro a c e ty l-in s u lin grown from citrate buffer in the presence o f Z n^+ . DEAE-Sephadex column chromatography w ith 0.01 M Tris buffer in 7M urea w ith a 0 .0 4 -0 .2 3 M NaCI gradient o f insulin hydrochloride treated w ith ethyl thiotrifluoro acetate in aqueous buffer. Sephadex G -5 0 column chromatography w ith 0.01 M Tris buffer, 0 .04 M NaCI in 7M urea (pH 7 .9 ) o f insulin hydrochloride treated w ith ethyl th io ltriflu o ro acetate in aqueous buffer. Figure 10 11 12 13 14 15 16 17 18 19 20 21 22 Page Structure o f the insulin monomer ind ica ting the 72 location o f the fluorine probes. 19F-ch emical shifts o f triflu o ro a ce tyl-a m i no acids, 75 peptides and insulin derivatives as a function o f solvent and pH. ^^F-N M R spectra o f triflu o ro a c e ty l-in s u lin 78 deri vati ves. ^ F -N M R spectra o f triflu o ro a c e ty l-in s u lin 83 deri vati ves. ^ F -N M R spectra o f triflu o ro a c e ty l-in s u lin 86 derivatives at pH 8 .7 . ^ F -N M R spectra o f g ly c in e ^ ” ^ -triflu o ro a c e ty l- 89 insulin at pH 6 .8 . The effect o f sodium acetate on the ^F-resonance 91 peak o f z in c -g ly c in e ^ ""^ -triflu o ro a c e ty l-in s u lin . The effect o f potassium thiocyanate (KSCN) on the 97 1 9 F-N M R resonance peak o f zin c-fre e g ly c in e A -1 - triflu o ro a ce tyl-in su lin at pH 6 .8 . The effect of guanidine hydrochloride on the 99 ^ 9 F-N M R spectra o f triflu o ro a c e ty l-in s u lin derivatives at pH 6 .8 . Comparison o f the effects of sodium dodecyl su l- 103 fate (1%) on the ^ F -N M R spectra of trifl uoro- a ce tyl-in su lin derivatives at pH 6 .8 . The effects o f increasing concentrations o f sodium 106 dodecyl sulfate on the '^F -N M R spectra o f zinc g ly c in e A -l - l y s i n e ^ ~ 2 9 _ t r i f | u o r o a c e t y | _ ] n s u | j n a j. p H 6 . 8 . Far UV circula r dichroism spectra o f zin c-fre e tr i- HO fluoroacetylated insulin derivatives at pH 6 .8 . The effects of zinc on the circu la r dichroism 114 spectra o f g ly c in e ^ H -triflu o ro a c e ty l-in s u lin at pH 6 .8 . IX Figure 23 24 25 26 27 28 29 30 The effects o f increasing concentrations o f sodium dodecyl sulfate on the far UV circula r dichroism spectra of zinc g ly c in e ^ -^ - ly s in e ^ '^ - tr iflu o r o - a ce tyl-in su lin at pH 6 .8 . A comparison o f the far UV circu la r dichroism spectra o f trifluoroace tylated insulin derivatives in sodium dodecyl sulfate solutions (1%) at pH 6.8. N ear UV circu la r dichroism spectra for triflu o ro a c e ty l-in s u lin . Near UV c ircu la r dichroism spectra for triflu o ro a c e tyl-in su lin derivatives at pH 6 .8 . The effect o f zinc ion on the near UV circu la r dichroism spectra o f g ly c in e ^ "^ -triflu o ro a c e ty l- insulin at pH 6 .8 . The effect of acetate ion on the near UV circu la r dichroism spectra o f z in c -g ly c in e ^ -^ -triflu o ro a c e ty l-in s u lin . The e ffe ct o f increasing concentrations o f sodium dodecyl sulfate on the near UV circula r dichroism spectra o f g ly c in e ^ -^ - ly s in e ^ ^ - tr iflu o r o a c e ty l- insulin. A comparison o f the near UV circular dichroism spectra o f triflu o ro a c e ty l-in s u lin derivatives in sodium dodecyl sulfate solution (1%). Page 116 120 122 124 126 129 131 133 x LIST OF TABLES Table I II III IV V VI V II V III IX X Page Amino acid analysis o f tri-triflu o ro a c e ty l-in s u lin 32 and deaminated tri-triflu o ro a c e ty l-in s u lin . Amino acid analysis of deaminated triflu o ro a c e ty l- 36 insulin derivatives. Immunoassays o f triflu o ro a c e ty l-in s u lin derivatives. 38 19F- N uclear magnetic resonance peak widths (Hz ) 80 for m o no-triflu o roa ce tyl-insu lin derivatives as a function o f p H . ^ F -N u c le a r magnetic resonance peak widths (Hz ) 81 for d i- and tri-triflu o ro a c e ty l-in s u lin derivatives. The effects o f citrate ion on 19p_fsJMR peak widths 93 (Hz ) o f the triflu o ro a ce tyl moiety o f g ly cin e ^ - ! - triflu o ro a c e ty l-in s u lin and glycineA-1 -phenylalanine ^ “ ^ -triflu o ro a c e ty l-in s u lin in the presence o f zinc ion . The effects o f acetate ion on ^ F -N M R peak widths 94 (Hz ) o f the triflu o ro a ce tyl moiety o f g ly c in e ^ - ^- triflu o ro a c e ty l-in s u lin and g ly c in e ^ - '-phen yla la n ine -triflu o ro a c e ty l insulin in the presence o f zinc ion . 1 9 F-NM R peak widths (Hz ) of triflu o ro a ce tyl-in su lin 101 derivatives as a function o f guanidine hydrochloride concentrations at pH 6 .8 . ^ F -N M R peak widths (Hz ) o f zinc triflu o ro a c e ty l- 108 insulin derivatives as a function o f sodium dodecyl sulfate (SDS) concentration at pH 6 .8 . C ircular dichroism ratios ( A2O8/ ^222^ f ° r z in c - 112 free insulin derivatives. xi Table XI XII X III C ircular dichroism ratios ( A 2O 8/ ^227) ^or z ' nc triflu o ro a ce tyl-in su lin derivatives in SDS solutions at pH 6 .8 . Svedberg constants (s20,w) o f insulin and triflu o ro a ce tyl-in su lin derivatives as a function o f pH. Svedberg constants (s20,w) o f zinc insulin and zinc g ly c in e ^ ” ^ -triflu o ro a c e ty l-in s u lin : the effects of sodium acetate at neutral pH. Page 118 136 137 xii PART I Preparation and C haracterization o f Several Trifluoroacetyl Insulin Derivatives 1 Introduction - Part I Relationship o f the structure o f crystalline insulin to the properties of in sulin in solution. Recently the m olecular structure o f crystalline zinc insulin has been resolved by X -ray crystallography (Adams et a L , 1969). The elucida tion o f this structure was the culm ination o f over 40 years o f study o f the insulin m olecule. During this time insulin had been studied by nearly every technique available to the biochem ist, in an attem pt to elucidate its structure and the re lationship of this structure to the biological a ctivitie s o f insulin. It is reassuring that many o f the known solution properties of insulin c la rifie d during this period may be explained in terms o f its solid state crystal line structure. Thus, for exam ple, the aggregation behavior o f insulin in solution is best explained by assuming the aggregation o f monomers into dimers w ith a subsequent aggregation o f the dimers into hexamers (Jeffrey and Coats, 1966a,b; Pekar and Frank, 1972). S im ilarly the crystalline form o f insulin is seen to be a hexamer composed o f three dimers (Blundell et a l. , 1972). As a second example it may be noted that the circu la r dichroism minima observed at 208 nm and 220 nm for insulin in solution may be correlated, respectively, to the regions o f C K-helix in the monomers and the (3 -sheet structure between monomers seen in crystalline insulin (Golman and Carpenter, 1974). Third, a series of im munological studies o f insulin allow ed the prediction o f an approximate chain folding for the insulin monomer (A rquilla et a l. , 1969) which is in fa irly close agreement w ith the chain folding observed in the crystal structure (Blundell et a l. , 1972). There is thus at least an overall relationship between crystalline and solution arrangements o f the insulin m olecule. Chemical studies o f insulin. In addition to correlations such as those men tioned above, it has also been possible to correlate some o f the specific chemical properties o f insulin w ith its crystalline structure and b iological a c tiv ity . There are two basic rationales for most o f the chemical m odification studies o f insulin: 1) the re a ctivities of various amino acid residues depend on their accessibility to the reagent used as w ell as the local environment o f the residue, and 2) changes in the biologica l a c tiv ity o f insulin w ith the m odification o f p a rticula r residues indicate that these residues are s p e cifica lly involved in the bio lo gica l function o f insulin (barring an overall conformation change caused by the m o d ifica tio n ). The elucidation o f the complete amino acid sequence o f insulin by Sanger and coworkers (Ryle et a l. , 1 955) provided the impetus for a variety o f chemical and enzym atic m odification studies o f insulin. Every reactive side chain o f insulin has since been m odified. Examples include: 1) the cystine disulfide bridges (Cecil and W ake, 1962; Markus, 1964; Zahn and Drechsel, 1968), 2) the phenyl ring o f tyrosine (Morris et a l. , 1970; Levy and Carpenter, 1970; Slobin and Carpenter, 1963, 1966), 3) the arginine side chain (Nakaya et a l. , 1967; Bungli and Bosshard, 1971), 4) the histidine ring (Weil e t a l . , 1965; Suzuki e t a l . , 1969; C ovelli and W o lff, 1967), 5) the alcohol o f serine and threonine (Maloney et a l. , 1964; Thomas, 1969), and 6) the © < -and & - amino groups (Lindsay and Shall, 1971; A rq u illa et a l. , 1969; Paselk and Levy, 1974b). 3 Unfortunately the interpretation o f the results o f these chemical studies was more d iffic u lt than had been hoped. This was due to the com plexity o f the in sulin molecule and to the accompaniment o f some m odifications by conformational changes. The occurrence o f conformational changes makes it d iffic u lt to deter mine whether any loss o f biologica l a c tiv ity is sp e cifica lly due to the m odifica tion o f a specific residue, or a different conform ation. However, recently some o f the results o f these chemical studies have been interpreted in the lig h t of the known crystalline structure of insulin (Blundell et a l. , 1972). For example, the relative reactivities o f the three disulfide bonds is correlated to the degree o f exposure of each bond to solvent (Blundell et a l. , 1972). Conversely, the lack o f se le ctivity in esterifying the carboxylate groups in a cid ic methanol is ex plained because a ll o f them are exposed in the dim er, w hich is the expected aggregate under these conditions (Blundell et a l. , 1972). As another example, it has been found that m odification o f the amino groups of the N -term inal phenyl alanine on the B chain o f insulin prevents the formation o f insulin crystals (Lindsay and Shall, 1971). The X -ray crystal structure o f zinc insulin shows that this residue lies in a tig h t pocket between the dimers o f the hexamer. An increase in bulk at this position thus prevents the formation of hexamers and therefore, o f crystals. The chemical studies have also indicated the residues which may be modified w ithout extensive loss o f b io lo gica l a c tiv ity (Blundell et a l. , 1972). O f p a rticu la r interest in this regard are the free amino groups (discussed below ). M o d ifica tio n o f the free amino groups o f insu lin. The inform ation presented 4 so far points to at least an overall relationship o f the structures o f insulin in the solution and crystalline states. However, much more detailed inform ation must be accumulated in order to relate the detailed structure of the m olecule under these two d ifferent conditions, a relationship of obvious importance in under standing the mechanism of action of insu lin. In an attem pt to further c la rify this relationship it was desired to synthesize ye t another group o f m odified insulin derivatives w ith a trifluoroacetyl group covalently linked to one or more o f the three free amino groups of insulin. The resultant derivatives could be studied chem ically and b io lo g ic a lly as w ell as by ^ F -N M R using the triflu o ro a ce tyl group as a reporter group. Trifluoroacetylation using ethyl th io ltriflu o ro a ce ta te in aqueous solvent was first described by Schallenberg and C alvin (1955), and used successfully by Goldberger and Anfinsen (1962) for the m odification o f the amino groups in ribonuclease. The enzym atic a c tiv ity was lost and regained only a fter appro priate rearrangements o f the disulfide bonds. The la b ility o f ethyl th io ltriflu o ro acetate in aqueous solutions leads to the formation o f large amounts o f ethane th io l, which leads to considerable disruption o f disulfide bonds at high pH. Since insulin is not readily regenerated from the disrupted A and B chains, the triflu o ro a ce tyla tio n o f the amino groups o f insulin in non-aqueous solvent sys tems, where hydrolysis o f the reagent is not a problem was desirable. The free amino groups o f insulin (theo<-am ino groups o f the N -term inal amino acids, glycine and phenylalanine, and their -am ino group o f the single lysine) have been more extensively studied than any o f the other chem ically 5 active groups of insulin. One of the advantages of working w ith the amino groups is the a b ility to control the degree o f reaction o f the three sites. The re lative re a ctivitie s o f the three amino groups have been found to depend on the solvent, the pH and the m odifying reagent employed. Thus, in organic and mixed organic/ aqueous solvent systems a number of reagents react most readily w ith the glycine A “ 1 amino group. Examples include: 1) t-butyIoxycarbonyl azide in dim e thyl- formamide/aqueous hydrogen-carbonate, 2) p-nitrophenylacetate in d im e thyl- sulfoxide in the presence o f triethylam ine (G eiger, 1971) and 3) S -e th y ltri- fluoroacetate or phenyl tri fluoroacetate in dimethylformamide in the presence o f triethylam ine or im idazole (Paselk and Levy, 1974). Conversely, most reagents react p re ferentially w ith the amino group o f phenylalanine in aqueous solutions. These reagents included: 1) fluorescein isothiocyanate (Bromer et a l. , 1967), 2) phenyl isothiocyanate (A frica and Carpenter, 1968; Brandenberg, 1969) and 3) diketene (Lindsay and Shall, 1969). The difference in reaction rates o f the g ly c in e ^ - ^ and phenylalanine^- ^ amino groups toward fluorescein isothio cyanate is suppressed in 7M urea solutions. This indicates that the d iffe re n tia l reaction rates o f these groups is due to differences in the environment o f these groups in the native protein. F in a lly, at high pH (10-13) in aqueous solutions certain reagents have been found to react preferen tially w ith the € “ amino group o f lysine (Evans and Saroff, 1957; L i, 1956). A t lower pH values this group is generally the least reactive o f the three amino groups due to its high pK . A second advantage o f w orking w ith the free amino groups in m odification studies of insulin is the a v a ila b ility o f techniques for separating the various 6 m ono-, d i- and tr i- derivatives. P articularly e ffe ctive is column chromatography on DEAE Sephadex u tiliz in g a 7M urea buffer w ith a salt gradient to elute the protein (Bromer and Chance, 1967). For exam ple, Lindsay and Shall (1971) have separated insulin and four o f its acetylated derivatives on a single colum n. More recently insulin and all seven o f its trifluoroacetyl a ted derivatives have been resolved using a m odification o f this chromatographic system (Paselk and Levy, 1974b). The biological and immunological a ctivitie s o f insulin derivatives modified at the free amino groups. The most important parameter characterizing any in sulin derivative is its biologica l a c tiv ity . Biological a c tiv itie s are most com monly measured in vivo by the mouse convulsion assay or the blood sugar de pression assay and in vitro by the rat epidydimal fat pad assay or the isolated fa t cell assay. M o d ifica tio n o f the amino groups o f insulin generally results in a decrease in b io lo g ica l and/or immunological a c tiv ity . The relative a c tiv itie s o f various derivatives depends on the particular groups attached to insulin, the location o f the attachment and the number (0-3) o f groups attached. As an illu stra tio n one may look at derivatives o f insulin involving the acetoacetyl group (Lindsay and Shall, 1971). T ri-a ce to a ce tyl- insulin is approxim ately 75% as a ctive as native insu lin. Another tri d e riva tive , invo lving a bulkier substituent (fluorescein thiocarbam ate), is less than 1% as active as native insulin by the mouse convulsion assay. The position o f the A 1 adduct is also im portant; glycine -a ce to a ce tyl-in su lin is only 85% a ctive , w hile both phenylalanine^” ^ - , and lysine^” ^^-a ce to a ce tyl-in su lin is about 7 75% active (an extensive tabulation to the properties o f the properties of various insulin derivatives m odified at the amino groups is found in Blundell et d . , 1972: Table IV ). In contrast to previous studies w ith other m odifying groups, acetylation w ith either p-nitrophenyl acetate (Brandenburg et a l. , 1972) or N -h yd ro xy- succinim ide acetate (Lindsay and Shall, 1971) resulted in essentially no loss o f b io lo g ica l a c tiv ity . These results could be explained by the small size o f the acetate unit (it simply was not interfering w ith the normal function of insulin); or, by postulating that the acetate units are cleaved o ff the insulin molecules by acetylases present in the test organisms, regenerating the native hormone. The addition o f the triflu o ro a ce tyl group to the free amino groups of in sulin was undertaken to distinguish between the above explanations for the behavior o f acetylated insulin, and to serve as environm ental probes for the three sites using the technique o f fluorine nuclear magnetic resonance (^ F -N M R ). The triflu o ro a ce tyl group is approxim ately the same size as the acetyl group, w hile the properties o f the two groups are d iffe re n t. Thus the same enzyme would not be expected to attack both groups. In a d d itio n , fluorine is a rare species in liv in g systems and enzymes specific to its compounds are not lik e ly to be present in the test organisms. 8 Materials Bovine zinc insulin was obtained from Eli L illy and Company (Lot N o. 493-88G P -017 ). L -l-T osyla m ido-2-phen ylethyIchlorom ethyl ketone-treated trypsin was purchased from W orthington Biochemical C orporation. This material was used in order to m inim ize the possibility o f n o n -tryptic cleavage o f peptide bonds by chymotrypsin, which is a frequent contaminant o f trypsin preparations. Dimethylformamide (M a ilinckro d t) was purified by d is tilla tio n under vacuum (10 mm Hg) a fter refluxing for two hours over calcium hydride. Phenyl tri fluoro acetate, ethyl th io ltriflu o ro a ce ta te and ninhydrin were purchased from Pierce Chemical Company, and were used w ithout further p u rific a tio n . L -p h e n yl- alanine and Ellman's reagent (5 ,5 -d ith io b is(2 -n itro b e n zo ic acid)) were purchased from Calbiochem . Bovine serum albumin was obtained from N u tritio n a l Bio chemical Company. The chromatography resins (DEAE Sephadex A -25; Sephadex G -5 0 , fine; and Sephadex G -2 5 , medium) were purchased from Pharmacia Fine Chemicals, Incorporated. M N Polygram SIL N-HR pre-coated plastic sheets were obtained from Brinkmann Instruments, Incorporated. N - triflu o ro a ce tyl-p h e n yla la n in e was prepared according to the procedure o f Schallenberg and C alvin (1955), reported mp 1 1 9 .4 -1 2 0 .6 °; found 120-121°. 9 Experimental U ltra vio le t absorption measurements were taken on a Zeiss spectrophoto meter (PMQ II). M e ltin g points were taken on a Fischer-Johns melting point apparatus. Removal o f salts and urea from solutions o f insulin and insulin derivatives was accomplished in an Amicon u ltra filtra tio n cell w ith U M -2 d iaflow membranes using 5mM ammonium bicarbonate (pH 7 .0 ). Amino acid analysis. Samples were hydrolyzed in evacuated sealed glass tubes w ith 6M HCI at 120° C for six hours. The hydrolysate was analyzed on a Technicon amino acid analyzer (Spackman et a l. , 1958). Thin layer chromatography. Thin layer chromatography o f amino acids and th e ir derivatives was carried out on M N Polygram precoated plastic sheets. Concentrated solutions o f derivatives to be chromatographed were spotted w ith drawn out c a p illa ry tubes and then allow ed to dry. The chromatograms were then developed in glass chambers containing the appropriate solvent systems at room temperature. Two solvent systems were employed: chloroform /m ethanol/ a cetic acid (95:5:1), to separate compounds soluble in non-polar solvents such as ether; and n-propanol/am m onium hydroxide (70:30), to separate compounds soluble in polar solvents such as dilu te a cid . When appropriate, the separated compounds were visualized on the developed chromatographic plates by one or more o f the follow ing methods: observation by u ltra v io le t lig h t; exposure of 10 the plate to iodine vapor; or treatment o f the plates w ith ninhydrin reagent. Column chromatographic procedures. A DEAE-Sephadex (A-25) column (2.5 cm x 45 cm) was used to isolate the various m ono-, d i- and tri-triflu o ro a c e ty l-in s u lin derivatives from the reaction m ixture by using the procedures o f Bromer and Chance (1967). Samples (5-40 m g/m l) were dissolved in 0.01 M T ris-0 .0 4 M NaCI in 7M urea (pH 7 .9 ) and applied to the column. W ith large concentrations o f the tri-triflu o ro a c e ty l-in s u lin derivative it was necessary to readjust the pH to 7 .9 w ith 1 N sodium hydroxide. The protein derivatives were eluted w ith a lin e a r gradient generated by running 0 .1 7 M or 0.23 M NaCI (1:1) into a stirred reservior containing 1:1 o f 0 .04 M N aC I, all in 7M urea- 0.01 M Tris (pH 7 .9 ). Flow rates were 40 m l/hour and fractions o f 7 .5 -8 .5 ml were co lle cte d . Protein concentrations were determined by absorbance at 278 nm. Resolution o f the various derivatives was greatly improved after the column had been exposed to the gradient (0 .0 4 -0 .2 3 M N aC I) several times. The appropriate fractions in each peak were pooled, concentrated and desalted by u ltra filtra tio n and ly o p h iliz e d . A Sephadex (G -5 0 , fine) column (2.5 cm x 90 cm) was used to compare the size o f insulin and triflu o ro a c e ty l-in s u lin in 7M urea. Samples (5 m g/m l)were applied to the column and eluted at a flow rate of 20 m l/ho ur using 0.01 M T ris-0.04 M NaCI in 7M urea (pH 7 .9 ). The reaction o f insulin w ith ethyl th io ltriflu o ro a ce ta te in w ater. Insulin hydrochloride (24 mg, 4 moles) was dissolved in 4 .0 ml o f a 0.2 M sodium bicarbonate buffer (pH 10.0 ). Ethyl th io ltriflu o ro a ce ta te (1 00 mg, dSO^moles) was then added to the insulin solution. The reaction was allow ed to proceed for 1 .5 hours w ith vigorous stirrin g . The isolated product was washed tw ice w ith 0 .2 M a ce tic acid and dried. This product was then chromatographed on Sephadex G -5 0 and on a DEAE Sephadex as described under chromatographic procedures. The degree of sulfhydryl group form ation was estimated using the procedure o f El I man (1959). The reactions o f g lycine methylester and insulin w ith ethyl th io ltriflu o ro - acetate in dim ethylform am ide. G lycin e methyl ester (1.33 mg, 1 2 .8 ^m o le s) was dissolved in dim ethylform am ide (5.10 ml) and a 0 .4 ml a liq u ot was removed for a tQ value. Triethylam ine (10^*1, 72yjkmoles) was added to the remaining solution and after 5 min ethyl th io ltriflu o roa ce ta te (4.74 mg, 3 0 ^*moles) was added. A t appropriate times a 0 .5 ml a liq u o t was removed and im m ediately mixed w ith 0 .5 ml o f 0.01 M HCI to stop the reaction. The progress o f the reaction was follow ed u tiliz in g the ninhydrin assay. To each sample one ml of ninhydrin reagent (pH 5) (C lark, 1964) was added. The reaction mix was then placed in a 100° C bath for twenty minutes, and then after co o lin g , 6 .4 ml o f 50% aqueous n-propanol was added to the solution. A fte r 10 min the optical density at 570 nm was recorded. For the tQ value, dim ethylform am ide (0.1 ml) containing ethyl th io ltriflu o ro a ce ta te (0.4 mg) was added prior to this assay. Insulin hydrochloride (13 mg, 2.23^m oles) was added to the remaining solution and, afte r 5 minutes, ethyl th io ltriflu o ro a ce ta te (4.74 mg, 30/*moles) was added. A t appropriate times, a 0 .5 ml a liq u o t was removed and analyzed as above for free amino groups. A dditional studies, identical to those described above w ith the exceptionthat im idazole (6 mg, 8 7 moles) was substituted for trie th ylam in e, were also conducted. In prelim inary experiments u tiliz in g sim ilar procedures, 2,5-1 utidiene or pyridine were used as the base. The reaction o f insulin w ith ethyl th io ltriflu o roa ce ta te in dim ethylform a mide: correlation of kinetics w ith product form ation. Insulin hydrochloride (100 mg, 17.2 jamoles) was dissolved in dimethylformamide (30 m l), trie th y l amine was added (80^ x l, 576y*moles) and a 0 .3 ml aliquot was removed and mixed w ith 3 ml ethyl ether for a tQ valu e. A fte r 5 minutes ethyl th io ltriflu o ro acetate (6 drops, approxim ately 240j*. moles) was added. A t 5 minute intervals 0 .3 ml samples were removed for ninhydrin assay and im m ediately precipitated from solution by the addition o f 3 ml o f ethyl ether. The mixture was then centrifuged, resuspended in ether two times, and the p e lle t was dried in vacuo ever phosphorus pentoxide. The p e lle t was then dissolved in 0 .4 ml o f w ater and 1 .2 ml o f ninhydrin reagent (C lark, 1964) (pH 6 .8 ) was added. The reaction m ix was then placed in a 100° C bath for twenty minutes, and then afte r coo ling, 6 .4 ml o f 50% aqueous n-propanol was rapidly mixed w ith the solution. A fte r 10 minutes the optical density at 570 nm was recorded. Simultaneous w ith the removal o f the samples for ninhydrin assay, 7 .5 ml samples were removed for analysis by chromatography on DEAE Sephadex. The samples were im m ediately precipitated by the addition o f excess anhydrous ether. The products were isolated by cen trifu g a tio n , washed tw ice w ith ether, and then dried in a desiccator over phosphorous pentoxide in vacuo. They were then dissolved in 7M urea buffer and chromatographed as described under chromatographic procedures. 13 Preparation and isolation o f tri-triflu o ro a c e ty l-in s u lin . Insulin hydro chloride was prepared by the method o f Carpenter (1958). A sample o f this m aterial (24 mg, 4 jxmoles) was dissolved in purified dimethylformamide (5.0 ml) and triethylam ine (IO jjJ , 72j*m oles) and stirred at 24° C for 5 minutes. Ethyl th io ltriflu o ro a ce ta te (18.4 mg, 120/Amoles) was dissolved in dimethylform am ide (1 .15 ml) and added d ire ctly to the insulin solution. The reaction was allowed to proceed for 60 minutes at w hich time the product was precipitated by the addition o f anhydrous ether (40 m l). The product was isolated by cen trifu g a tio n , washed w ith acetone, ether, and then dried in a desiccator over phosphorous pentoxide under high vacuum. Preparation o f m ono-, d i- , and tri-triflu o ro a c e ty l-in s u lin derivatives. Insulin hydrochloride (Carpenter, 1958) (200 mg, 33 j>moles) was dissolved in dim ethylformamide (40 m l). Im idazole (50 mg, 735j*m oles) was added and the solution stirred fo r 5 minutes. Phenyltrifluoroacetate (27.4 mg, 144/*moles) was dissolved in dim ethylform am ide (8.8 ml) and then added to the reaction flask and stirred for 4 hours at room temperature. The reaction was terminated by p re cip ita tio n o f the protein w ith ether and the products collected by ce n tri fugation. The m ixture o f products was washed tw ice w ith fresh ether to remove excess reagent and then dried under vacuum ever phosphorous pentoxide. Sim ilar procedures were used when ethyl th io ltriflu o ro a ce ta te was used as the m odifying reagent. Insulin hydrochloride (50 mg, 8.26y*m oles) was dissolved in dim ethylform am ide (10 ml) which contained im idazole (24 mg, 353 j a moles). Ethyl th io ltriflu o ro a ce ta te (50.5 mg, 320ja moles) dissolved in 14 dim ethylform am ide (2.3 ml) was added. A fte r 30 minutes the reaction was te r minated by the addition o f ether and the products isolated as above. Determ ination o f free amino groups on insulin by d in itro p h e n yla tio n . In sulin (I mg) or one o f its derivatives was mixed w ith a solution o f 10 mg sodium bicarbonate in 1 ml o f w ater in a drawn out combustion tube. Five percent (v /v ) dinitrofluorobenzene (2 ml) in ethanol was then added, and the tube was shaken for 21 hours at room tem perature. The m ixture was then a cid ifie d w ith 0.1 ml concentrated H C I, and then extracted three times w ith 7 ml portions o f anhydrous ethyl ether. Excess ether was removed by evaporation under vacuum. One ml o f concentrated HCI was then added to each tube, the contents frozen and degassed under vacuum, sealed, and hydrolysed for 6 hours at 120° C. The samples were then analysed by amino acid analysis or thin layer chromato graphy. In the la tte r case the hydrolysed sample was first separated into acid and ether soluble components by e xtra ctio n . The resultant ether and acid solutions were then concentrated and separated by thin layer chromatography using the appropriate solvent systems. Deamination o f triflu o ro a c e ty I-in s u lin derivatives. The triflu o ro a c e ty l- insulin derivatives (2.5 mg) were dissolved in 50% acetic acid (5 m l). Sodium n itrite (300 mg in 1 ml o f w ater) was added drop-wise to the solution over a period o f 35 minutes at room temperature w ith vigorous s tirrin g . A fte r an additional 5 minutes the solution was diluted w ith w ater (100 ml) and the material ly o p h iliz e d . The resulting residue was dissolved in 3 ml o f 0.01 M ammonium acetate (pH 8 ), and the solution adjusted to pH 8 w ith ammonium hydroxide. 15 This solution was then applied to a Sephadex G -2 5 colum n, eluting w ith 0.01 M ammonium acetate (pH 8 .0 ). The isolated material was then characterized by amino acid analysis. A duplicate deamination reaction was run on unmodified insu lin. Sulfhydryl group analysis. The possibility o f sulfhydryl group formation during the reaction o f insulin w ith ethyl th io ltriflu o ro a ce ta te in dim ethylform a mide was determined using the method o f Ellman (1959). Standard curves were constructed using both cysteine and bovine serum album in. The reaction o f trypsin w ith triflu o ro a c e ty l-in s u lin derivatives. G lycin e phenylalanine^” ^ -triflu o ro a c e ty l-in s u lin ; g ly c in e ^ ” ^ - , ly s in e ^ ^ - t r i- A _i B— 1 B— 29 flu o ro a ce tyl-in su lin and g lycin e , phenylalanine , lysine -triflu o ro - a c e tyl-in su lin were treated w ith trypsin. In each case the trifluo ro a ce tyla te d insulin d e rivative (2.6 mg, 0.43jxm oles) was dissolved in 1 .0 ml o f 0 .2 M sodium bicarbonate buffer (pH 8 .0 ). To this solution was added 0.22 ml o f an L -I-tosylam ide-2-phenylethylchlorom ethyl ketone-treated trypsin solution (pH 3 .0 , 0.001 M CaCI, 1 .2 m g/m l) and the reaction kept at 3 7 .5 ° C for 4 hours w ith occasional a g ita tio n . The reaction was then applied to a Sephadex G -5 0 column (2.5 cm x 40 cm) equilibrated w ith 0.1 M ammonium bicarbonate (pH 8 .0 ). For triflu o ro a c e ty l-in s u lin derivatives this column was developed w ith the same buffer to afford a desectapeptide insulin derivative and an o cta - peptide. When the reaction was carried out on native in su lin, desoctopeptide in su lin , a heptapeptide and alanine were isolated. The octapeptide and hepta- peptide were further p urified on a Sephadex G -1 0 column (2.5 cm x 40 cm), 16 elu tin g w ith 2 .5 M ace tic a c id . The various products were isolated, hydrolyzed, and subjected to amino acid analysis. Biological and im m unological assays. The tri-triflu o ro a c e ty l-in s u lin d e riva tive , p u rifie d by iso le ctric p re cip ita tio n , was characterized by the mouse convulsion and immunoassays performed by Eli L illy and Company. In a ddition , radioimmunoassays were performed on native insulin and several o f the tr i flu o ro a ce tyl-in su lin derivatives, including tri-triflu o ro a c e ty l-in s u lin , using the method o f Hales and Randale (1963). Isoelectric p re cip ita tio n o f tri-triflu o ro a c e ty l-in s u lin . The tri-triflu o ro a c e ty l-in s u lin d erivative was subject to two isoelectric precipitations. The dried residue (30 mg) was dissolved in 3 ml o f 0.06 M ammonium acetate (pH 8), containing zinc acetate (0.5 mg) and precipitated at pH 4 .5 w ith 5 M acetic a c id , washed w ith acetone, ether, and dried under high vacuum over phos phorous pentoxide. C rystallization o f g ly c in e ^ ~ ^ -triflu o ro a c e ty l-in s u lin . Crystals o f glycine -triflu o ro a c e ty l-in s u lin were grown from citra te buffer (pH 6 .0 ) in the pre sence of zinc acetate using the methods o f S chlichtkrulI (1956a, b ). The in sulin d e rivative (5 mg) was dissolved in 0.01 N HCI (0.85 ml) containing zinc acetate dihydrate (0.085 mg). Acetone (0.15 ml) was added and the pH adjusted to 6 .0 w ith 0 .5 M sodium c itra te . The turbid solution was le ft over night at room temperature and then kept at 4 ° C for several days. 17 Results Synthesis o f several triflu o ro a ce tyla te d insulin derivatives has been effected u tiliz in g ethyl th io ltriflu o ro a ce ta te or phenyl tri fluoroacetate in dim ethylform am ide. The reaction is shown schem atically in Figure 1. Methods for the p u rifica tio n and characterization o f these derivatives are also described. C haracterization o f the reaction o f insulin HCI w ith ethyl th io ltriflu o ro acetate in dim ethylform am ide. In order to synthesize the various triflu o ro - acetylated insulin derivatives w ithout extensive disulfide interchange taking p la ce , the reactions were carried out in dim ethylform am ide. A series of k in e tic studies were undertaken to characterize the reaction in this solvent using various organic bases to deprotenate the free amino groups o f in su lin. Parallel studies using g lycin e m ethylester as a model compound were also carried out. The progress o f the reactions were follow ed using the ninhydrin assays to detect unreacted amino groups. In early studies the protein was precipitated from the reaction m ixture w ith ether and the dried protein analysed as described under methods. This procedure was not possible for the model compound since the product is soluble in ether. P recipitation also led to inconsistent results under some reaction conditions w ith insu lin. A second assay system was thus developed in w hich the free amino groups were assayed w ithout p u rifica tio n from the reaction m ixture (Methods). 18 Figure 1. Scheme describing the synthesis o f m ono-, d i- and tri-triflu o ro a c e ty l- insulin derivatives using phenyl tri fluoroacetate or ethyl thiol triflu o ro - 9 acetate. R= CF3- C - or F I - 19 H2N-Gly - H2N -P h e • ■Asn-O H h 2n -I C F, - c - 0 - O RHN-Gly ^o r -■ > -A sn-O H L ys-A la-O H CFj-C-s-CHj-CHi R H N -P h e- Lys-Ala-OH RHN 20 Figure 2A shows the kinetics o f the reaction o f g lycin e methylester w ith e th y l-th io ltrifu lo ro a c e ta te using triethylam ine or im idazole as base. The two curves for the reaction are superposable: using either triethylam ine or im idazole gives the same reaction rate. The reaction o f insulin hydrochloride w ith e th y l- th io ltriflu o ro a c e ta te , however, is effected by the choice o f base (Figure 2B). The reaction o f insulin and e th y l-th io ltriflu o ro a c e ta te in the presence o f im ida zole instead o f triethylam ine leads to a slower reaction w hich does not go to com pletion under these conditions. Studies u tiliz in g 2 ,5 -lu tid in e or pyridine as base gave much slower reactions than those u tiliz in g im idazole w hile reaching a sim ilar plateau. In order to further characterize the reaction o f insulin HCI and ethyl th io ltriflu o ro a c e ta te , a series o f samples were w ithdraw n at d iffe re nt times during the course of the reaction. Analysis were performed both for free amino groups by the ninhydrin assay and for various triflu o ro a ce tyl a ted insulin deriva tives by chromatography on a DEAE Sephadex column in 7M urea b u ffe r. The results o f this study are seen in Figure 3. The various derivatives of this study were not isolated. However, the peaks may be assigned to specific derivatives by comparison to late r separations in which the ind ividu a l components were isolated and analysed (Figure 6, Table II). In this way the peaks may be assigned as follow s: peak A , unmodified insulin; peak D, g ly c in e ^ - ^-lysin e ^” 2 9 -triflu o ro a ce tyl insulin; peak E, tri-triflu o ro a c e ty l-in s u lin ; and peak B and C g ly c in e ^ - ^ -triflu o ro a c e ty l-in s u lin and g ly c in e ^ - !-p h e n yla la n in e ^"^ - triflu o ro a ce tyl insulin, respectively, w ith an admixture o f the remaining possible 21 Figure 2. The kinetics o f the reaction o f ethyl th io ltriflu o ro a ce ta te w ith amino groups in the presence of triethylam ine ( • ) or im idazole ( ▲ ). (A) Amino groups on glycin e m ethylester. (B) Amino groups on insulin hydrochloride. 22 NINRYDR.TN COLOR: A (STOnm) t j I I i i 0-8 0-6 I I ! 0-4 i I ! 0*2 j I i 200 \20 ISO R E A C T IO N X I 146 (IA IN } 40 23 Figure 3 . The reaction o f ethyi th io ltriflu o ro a ce ta te and insulin hydrochloride in the presence o f triethylam ine: a comparison o f the progress o f the reaction and the product spectra. Product spectra at: 5 minutes ( A )/ 10 minutes ( # ), and 20 minutes ( 0 ). Inset: re action progress (kinetics). 24 008 - ! 0*06 flO i*- d 004- O O fc r e a c t io v\ t-»w « (m»*> ro cn mono- and d i-triflu o ro a c e ty l-in s u lin derivatives. Chromatography o f tri-triflu o ro a c e ty l-in s u lin . T ri-triflu o ro a c e ty l-in s u lin was chromatographed on a DEAE Sephadex column in a urea containing bu ffe r, enabling the detection o f differences in charge between native insulin and this product. The reaction afforded a homogeneous product, w ith the results shown in Figure 4. The m odification o f the three amino groups in insulin results in the loss o f three positive charges causing this deriva tive to be eluted at a higher salt concentration than in su lin. The material present in this peak (Fractions 175-205) was pooled and isolated by desalting in an Am icon u ltra filtra tio n c e ll, follow ed by ly o p h iliz a tio n . The purified product was then chromatographed on Sephadex G -5 0 using a urea buffer in an e ffo rt to assess the apparent size o f the d e riva tive . The results, as shown in Figure 5 , indicate that this derivative behaves in a fashion identical to insulin w ith regard to its apparent m olecular w eig h t. Analysis o f the p urifie d derivative for free sulfhydryl groups by the Ellman technique indicated essentially no free sulfhydryl groups were present. C haracterization o f tri-triflu o ro a c e ty l-in s u lin . Synthesis o f the tr i— tri — fluoroacetylated deriva tive o f insulin should result in the m odification of the free amino groups on the m olecule. In order to assess whether the isolated product was tri-triflu o ro a c e ty l-in s u lin , samples were subjected to known conditions which would form dinitrophenyl derivatives o f any remaining free amino groups. When the treated derivative was then hydrolysed and compared w ith hydrolysed dinitrophenylated insulin by amino acid analysis the results indicated that the amino groups o f the tri-triflu o ro a c e ty la te d insulin were not 26 Figure 4. DEAE-Sephadex chromatography with 0.01 M Tris buffer in 7M urea with a 0.04-0.23 M NaCI gradient (pH 7.9) of tri-trifluoroacetyl- insulin (--------) and an insulin control (----- ): NaCI gradient (— • — ). 27 (278nm ) INSULIN 0 . 1 2 0 . 1 0 TRI-TRIFLUOROACETYL- INSULIN 0.08 0.06 0.04 0. 02 20 40 60 80 100 120 140 160 180 200 FRACTION NUMBER 28 NaCI, MOLARITY Figure 5. Sephadex G-50 chromatography with 0.01 M Tris buffer, 0.04 M NaCI in 7M urea (pH 7.9) of tri-trifluoroacetyl-insulin. 29 INSULIN CONTROL 0.24 TRI-TRIFLUOROACETYL- INSULIN -------------------------> 0.20 E c CD f- O J 0 .1 2 0.08 0.04 40 60 80 20 100 FRACTION NUMBER 30 fu lly blocked. This result was in disagreement w ith the ninhydrin assay w hich had given a negative test for amino groups in the purifie d d e riva tive . The major discrepancy w hich could not be explained by known errors in the methodology employed was the presence o f dinitroph en yl-phenylala nin e as assayed by thin layer chromatography, which was also seen as a loss o f phenyl alanine on amino acid analysis o f the hydrolysed d e riva tive . It thus appeared that triflu o ro a ce tyl-p h e n yla la n in e might be subject to hydrolysis and subsequent dinitrophenylation under the conditions employed. In order to test this p o ssibility pure triflu o ro a ce tyl-p h e n yla la n in e was prepared and then treated w ith l-flu o ro -2 , 4-dinitrobenzene under the same conditions used to prepare d in itro - phenylated insulin. Subsequent analysis o f the reaction m ixture by thin layer chromatography demonstrated the presence o f d in itroph enyl-phenylala nine . The dinitrophenyl derivatives were thus not used in further characterizations. As an a lte rn a tive to dinitro p h e n yla tion , deam ination w ith sodium n itrite in ace tic acid was used. Deamination o f amino acids w ith free amino groups results in th e ir conversion to the corresponding hydroxy acids and subsequent loss upon amino acid analysis (Levy and Carpenter, 1970). Thus, deamination o f native insulin resulted in the loss o f three amino acid residues: the N - term inal amino acids, phenylalanine and g ly c in e , and lysine-29 o f the B-chain (Table I). When deam ination was carried out on the triflu o ro a ce tyl d e riva tive , amino acid analysis (Table 1) showed no loss o f amino acid residues, indicating that the three free amino groups had indeed been m odified w ith the triflu o ro acetyl m oiety. 31 T A B L E I A M IN O A C ID A N A L Y S E S A m in o acid In s u lin T r itr iflu o r o - D e a m in a te d D eam in ated -----------------------------------a c etvlin su lin in s u lin tritriflu o ro - C alcd Obsd acetylinsu lin A s p a rtic acid 3 2 .8 5 2.96 3.06 2.99 T h re o n in e i 0 .8 8 0.9 8 1.00 0.8 5 S erin e 3 2.90 2.76 2.82 2.77 G lu ta m ic acid 7 6.85 6.8 4 7.0 4 6.8 7 P ro lin e i 0.7 0 1.01 0-95 1.02 G ly c in e 4 3-92 4.11 3 .1 0 4.1 0 A la n in e 3 3 .0 0 3.0 0 3.0 0 3-00 H a lf-c y s te in e 6 5 -4° 5-2 7 5.2 0 5-30 V a lin e 5 4.8 0 4.4 4 4 -4 1 4-43 Iso leu cin e l 0.72 0 .5 6 o -54 0.61 L e u c in e 6 5-85 5 -8 ? 6.01 6.02 T y ro s in e 4 3-75 3-65 1.8 7* 1.92 P h e n y la la n in e 3 2 9 5 2.81 2.0 0 2.91 L y s in e l 0 .9 8 0 .9 8 0 .1 0 1.05 H is tid in e 2 1.92 1.87 1.99 2.04 A rg in in e I 1.05 0 .9 7 1.03 0-95 * D e a m in a tio n o f in s u lin w ith n itro u s acid effects p a rtia l n itra tio n o f th e ty ro s in e residues re s u ltin g in a decreased y ie ld o f ty ro s in e upon a m in o acid analysis. 32 As an additional characterization o f the triflu o ro a c e ty la tio n , insulin and the m odified deriva tive were treated w ith trypsin. Trypsin has been shown to catalyze the hydrolysis o f the arginyl bond at position 22 o f the B-chain and the lysyl bond at position 29 o f the B-chain o f insulin (Young and Carpenter, 1961). Using Sephadex G -5 0 and G -1 0 , desoctapeptide-insulin, a heptapeptide (G ly - Phe-Phe-Tyr-Thr-Pro-Lys) and free alanine can be isolated. An finsen et a l. (1956) have shown that m odification of th e £ -am ino group o f lysine w ill prevent hydrolysis o f the lysine-alanine bond. When the tri-triflu o ro a c e ty l d erivative was treated w ith trypsin, free alanine was not released and tw o fragments were isolated whose amino acid compositions corresponded to desoctapeptide-insulin and the octapeptide (G ly-P he-P h e -T yr-T h r-P ro -L ys-A la ). These results thus indicate that the £ -am ino group in blocked w ith the triflu o ro a ce tyl m oiety. C haracterization o f the mono- and d i-triflu o ro a c e ty l-in s u lin der iv a tives. The chromatographic separation o f the various insulin derivatives formed by treatm ent o f insulin hydrochloride w ith ethyl th io ltriflu o ro a ce ta te and im idazole or phenyl tri fluoroacetate and im idazole is shown in Figures 6A and B. The various triflu o ro a c e ty l-in s u lin derivatives, w ith the exception o f those in peaks C andE, were pu rifie d on a DEAE-Sephadex chromatographic column and deaminated to establish the location o f the triflu o ro a ce tyl m oiety. Amino acid analysis o f the deaminated products (Table II) indicated that peak A was phenyl- B— 1 A 1 alanine -triflu o ro a c e ty l-in s u lin ; peak B, g l y c i n e 1-triflu o ro a c e ty l-in s u lin ; peak D, g lycin e ^""^, phenylalanine^- ^ -triflu o ro a c e ty l-in s u lin ; peak F, glycine , lysine^- ^ -triflu o ro a c e ty |-in s u lin . The identities o f the minor peaks (C 33 Figure 6 . DEAE-Sephadex chromatography w ith 0.01 M Tris buffer in 7M urea w ith an NaCI gradient (-----) of triflu o ro a ce tyla te d insulin derivatives (------- ). (A) T riflu o ro a ce tyl-in su lin derivatives derived from the reaction w ith ethyl th io ltriflu o ro a ce ta te and im idazole eluting w ith a 0 .0 1 -0 .2 3 M NaCI gradient. (B) T riflu o ro a ce tyl-in su lin derivatives derived from the reaction w ith phenyltrifluoroacetate elutin g w ith a 0 .0 4 -0 .1 7 M NaCI gradient. 34 - * 0.20 0.20 0.16 0.16 0.12 0.08 0.08 0.04 g 0.00 “ 0.00 0.32 0.16 0.24 0.12 0.16 0.08 0.08 0.00 0.00 100 120 140 FRACTION NUMBER 20 40 220 160 180 200 35 TABLE II A M INO ACID ANALYSES Amino acid Insulin Tri-trifluoro- Deaminated Deaminated trifluoroacetylinsulin acetylinsulin insulin derivatives Calcd Obsd _______________________________________ Peak A Peak B Peak D Peak F Peak G Aspartic acid 3 2.85 2.96 3.06 3.17 3.20 2.99 3.00 2.99 Threonine 1 0.88 0.98 1.00 1.00 1.05 0.97 0.92 0.85 Serine 3 2.90 2.76 2.82 3.03 2.71 2.66 2.71 2.77 Glutamic acid 7 6.85 6.84 7.04 6.70 7.04 6.80 6.67 6.87 Proline 1 0.70 1.01 0.95 0.74 1.06 1.11 0.96 1.02 Glycine 4 3.92 4.11 3.10 2.97 3.88 3.87 3.80 4.10 Alanine 3 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 Half-cysteine 6 5.40 5.27 5.20 4.45 4.74 4.99 4.99 5.30 Valine 5 4.80 4.44 4.41 5.01 4.36 4.29 4.32 4.43 Isoleucine 1 0.72 0.56 0.54 0.50 0.52 0.42 0.55 0.61 Leucine 6 5.85 5.87 6.01 5.79 5.82 5.68 5.67 6.02 Tyrosine 4 3.75 3.65 1.87* 1.35* 1.45* 1.60* 1.24* 1.92 Phenylalanine 3 2.95 2.81 2.00 2.90 2.09 2.72 2.09 2.91 Lysine 1 0.98 0.98 0.10 0.21 0.21 0.20 0.93 1.05 Histidine 2 1.92 1.87 1.99 1.86 1.96 1.84 1.83 2.04 Arginine 1 1.05 0.97 1.03 1.05 1.01 0.99 0.94 0.95 * Deamination of insulin with H N 0 2 effects partial nitration of the tyrosine residues resulting in a decreased yield of tyrosine upon amino acid analysis. 36 and E) w hich were not- isolated were assumed by th e ir elution positions to be lysine®“ 2 9 -triflu o ro a c e ty l-in s u lin ancj phenylalanine®- ^ ', lysine®“ ^ - t r if lu o r o - a c e ty l-in s u lin , w hich are the two remaining possible derivatives that could be formed. As w ith the tri-triflu o ro a c e ty l-in s u lin , trypsinization o f three o f the above derivatives (peaks B, D and F) was effected to ascertain the state o f the £ -am ino group o f lysine. These trypsin studies confirmed the results obtained from the deam ination experiments concerning the £ -am ino group o f lysine®- ^ . In a d d itio n , the homogeneity o f the various triflu o ro a c e ty l-in s u lin derivatives was established by ^ F -n u c le a r magnetic resonance spectroscopy techniques (Paselk and Levy, 1974). B iological and im m unological a c tiv ity . A sample o f the tri-triflu o ro a c e ty l d e riva tive was analysed for b io lo g ica l and im m unological a c tiv ity . The results o f the mouse convulsion assay on this deriva tive indicated a b io lo gica l a c tiv ity o f 1 6 .1 /2 7 units/m g, or approxim ately 70% o f the a c tiv ity o f crysta llin e zinc in s u lin . The immunoassay performed at this time indicated a re la tive immuno- re a c tiv ity o f 60% compared to native in su lin. The appropriate controls showed no loss o f a c tiv ity . Several o f the mono- and d i- triflu o ro a ce tyl insulin deriva tives as w e ll as the tri-triflu o ro a c e ty l d erivative were la te r subjected to radio immunoassays, the results o f w hich are shown in Table III. C rysta lliza tio n o f g ly c in e ^ - ^ -triflu o ro a c e ty l-in s u lin . Using the methods o f S ch lich tkrulI (1956a, b ), the crystals o f g ly c in e ^ - ^ -triflu o ro a c e ty l-in s u lin were grown from citra te buffer in the presence of zinc io n , as shown in Figure 7. Attempts to obtain rhombohedral crystals using sodium chloride in the pro cedure to slow the crysta lliza tio n process were unsuccessful. 07 TABLE III IM M UNOASSAYS O F TRIFLU O RO A CETY L INSULIN DERIVATIVES Derivative % immunoreactivity PhenylalanineB1-trifluoroacetylinsulin 60 GlycineA1-trifluoroacetylinsulin 86 Glycine'4' 1 , phenylaIanineB1-trifiuoroacetylinsulin 80 Glycine'4'1 , lysine8 2 9 -trifluoroacetyIinsulin 89 Glycine'4 ' 1 , phenylalanine81, lysine8 2 9 -trifluoroacetylinsulin 48 38 Figure 7. Crystals o f g ly c in e ^ ^ -triflu o ro a c e ty l-in s u lin grown from citra te O 1 buffer in the presence o f Zn . 39 I t • ; . . . . - « . 40 Reaction of insulin hydrochloride with ethyl thioltrifluoroacetate in aqueous bu ffe r. Insulin was reacted w ith ehtyl th io ltriflu o ro a ce ta te in an aqueous buffer by the method o f Stouffer and Watters (1965). Hydrolysis o f the triflu o ro a c e ty - lating reagent during the reaction released ethyl mercaptan (evidenced by ex trem ely strong odor o f thiols) and triflu o ro a ce tic a cid . The released acid stopped the reaction in less than one hour by lowering the pH of the solu tio n, resulting in the p re cip ita tio n o f the pro tein. The product of this reaction was analysed by chromatography on DEAE Sephadex and Sephadex G -5 0 in 7M urea buffer. On DEAE Sephadex most o f the protein eluted at a volume corresponding to insulin H C I, w ith approxim ately 10-15% eluting at a volume corresponding to tri-triflu o ro a c e ty la te d -in s u lin and about 5-10% eluting between these two peaks (Figure 8). Since the release o f ethyl mercaptan during the reaction could catalyze disulphide bond in te r change, the product was analysed on a Sephadex G -5 0 column in 7M urea to analyze for heterogeneity o f m olecular size. As seen in Figure 9 this product e xhibited extreme size heterogeneity. There is a sharp peak corresponding to peptides o f larger m olecular size than insu lin, probably polymers o f the insulin chains. This is follow ed by a peak corresponding to the m olecular size o f insulin (less than 25% o f the total area o f a ll peaks) as w ell as several peaks corresponding to smaller m olecular sizes, possibly the S-ethyl A and B chains o f in s u lin . Extensive disulphide interchange would be expected to leave some free sulfhydryl groups. Analysis o f the triflu o ro a ce tyla te d insulin product prepared 41 Figure 8. DEAE-Sephadex column chromatography w ith 0.01 M Tris buffer in 7M urea w ith a 0 .0 4 -0 .2 3 M NaCI gradient o f insulin hydrochloride treated w ith ethyl th io ltriflu o ro a ce ta te in aqueous buffer. 42 0*08 002 200 20 120 140 FRACTION NUMBER fe Figure 9 . Sephadex G -5 0 column chromatography w ith 0.01 M Tris buffer, 0 .0 4 M NaCI in 7M urea (pH 7 .9 ) of insulin hydrochloride treated w ith ethyl th io ltriflu o ro a ce ta te in aqueous buffer. 44 A (276 Y \rv \) 0*28 0*24 INSULIN CONTROL 0 0 4 ao 2 0 AO FRACTION NUM BER too in acqueous media for free sulfhydryl groups was carried out by Ellman's procedure (Ellm an, 1959). This analysis gave positive results, confirm ing the presence of free sulfhydryl groups and by inference, disulfide interchange. 46 Discussion The triflu o ro a ce tyla tio n reaction. The preparation, p u rific a tio n and some properties o f a series of triflu o ro a ce tyla te d -in su lin derivatives have been described herein. Previous reports in the literature have described the prepara tion o f tr ifl uoroacety I ated derivatives o f a number of proteins (G oldberger and Anfinsen, 1962; Fanger and Harbury, 1965; Stouffer and W atters, 1965). These derivatives have a ll been prepared u tiliz in g ethyl th io trifluo ro a ceta te in a lka line aqueous media. A sim ilar procedure was found to be unsuccessful for m odifying insulin in the present study. This was due to the rearrangement o f disulfide bonds in the insulin m olecule caused by ethyl mercaptan released during the hydrolysis o f ethyl th io ltriflu o ro a c e ta te . In order to avoid this hydrolysis and the consequent breakdown a n d/o r polym erization of insu lin , a reaction system u tiliz in g an organic solvent was em ployed. Reports in the literature have shown that insulin can be m odified by various reagents in dim ethylform am ide in the presence of an organic base (Levy and Carpenter, 1967). Insulin isolated from such a reaction m ixture is found to be fu lly a c tiv e . It was thus decided to examine the reaction o f insulin and ethyl th io ltriflu o ro a ce ta te in dim ethylform am ide in the presence of various organic bases. In in itia l studies the kinetics o f these reactions were studied and the synthesis o f the fu lly substituted insulin derivative was attem pted. 47 The reaction kinetics were follow ed by using ninhydrin to detect the re maining unmodified free amino groups on the insulin m olecule. For some o f the studies the ninhydrin assay was m odified for use in a medium containing dim ethyl formamide and the other components o f the reaction m ixture. It was found that the assay was not effected by dim ethylform am ide, and gave consistent results w ith the controls used under these conditions. For the kin e tic studies the reaction was stopped eith er by p re cip ita tin g w ith ether (normal ninhydrin assay) or by the addition o f one volume o f 0.01 N HCI to hydrolyse any remaining ethyl th io l tri — fluoroacetate. In itia l tria ls determined that the hydrolysis o f the excess reagent by added w ater was not instantaneous. Thus, a small quantity o f acid was added to protenate the free amino groups and any excess reagent could be hydrolyzed. In itia l kin e tic studies w ith a number of reagent and base concentrations were used to establish the concentrations used in the more detailed studies re ported here. These results in turn aided in establishing appropriate conditions and times for subsequent preparative synthesis. Comparative studies using sim ila r conditions were carried out w ith g lycine methyl ester as a model for insulin H C I. It is interesting to note that the reaction kinetics o f insulin HCI and g lycin e methyl ester w ith triethylam ine are q u a lita tiv e ly sim ilar (Figure 2A , B). In both cases an in itia lly rapid reaction levels o ff and reaches a plateau, in d ica tin g a slight amount o f remaining ninhydrin reactive compound. This plateau value was consistently reproduced w ith both compounds and is probably in d ic a tiv e o f the blank's not including a ll o f the interfering species present in the reaction m ixtures. When im idazole was substituted for triethylam ine in 48 the g lycine methyl ester reaction the kin e tic curves were superposable. Subs titu tio n o f im idazole for triethylam ine in the insulin HCI reaction however, y ie ld quite different kin e tic data. The in itia l shape o f the curves is not as steep, in d ica tin g a slower reaction rate. The fin a l plateau reached indicates a substantial amount o f ninhydrin reactive m aterial rem aining. Thus the reaction u tiliz in g im idazole does not seem to go to com pletion (Figure 2B). In a d d itio n , the final plateau reached was insensitive to a doubling o f im idazole concentration. These results would seem to indicate a degree o f s p e c ific ity in the reaction of insulin HCI and ethyl th io ltriflu o ro a ce ta te in the presence o f im idazole, and that only two o f the possible three free amino groups are reacting under these condi tions. An additional kin e tic experim ent was conducted w ith insulin HCI in the presence o f trie th yla m in e . The purpose o f the experim ent was to correlate the form ation of the various possible derivatives o f insulin w ith the k in e tic data (Figure 3 ). From a comparison o f the chromatographic patterns at 5 , 10 and 20 minutes it is possible to estimate the re lative re a ctivitie s o f the various amino groups in the dim ethyl formami d e /trie th y l amine system. Thus it would appear that the N -te rm in a l amino group on glycine^""^ is the most reactive since the three major products at 5 minutes are a ll m odified at this position. The re a c tiv itie s o f the free amino groups o f lysine^” 29 anc| phenylalanine^""^ would appear to be sim ilar since the peaks corresponding to the major products w ith these groups blocked, g ly c in e ^""^ -p h e n y la la n in e ^"^ -triflu o ro a c e ty l-in s u lin and g ly c in e ^ " ^ -ly s in e ^ "^ -triflu o ro a c e ty l-in s u lin , are o f sim ilar magnitude. 49 T ri-triflu o ro a c e ty l-in s u lin . In itia l attempts to prepare the tri-triflu o ro - a ce tyl-in s u lin de rivative u tiliz e d the same proportions o f reactants and solvent as in the kinetic studies, but in larger quantity. The reaction was then allow ed to proceed for one hour at w hich time the ninhydrin color had essentially reached the plateau value. Subsequently the concentrations o f the reactants were ad justed somewhat, to the values reported, in order to m inim ize the production of ethyl mercaptan and the consequent catalysis o f disulphide bond interchange. A fte r the synthesis and isolation o f essentially ninhydrin negative m aterial, it was necessary to demonstrate that it was indeed the tri-triflu o ro a c e ty l-in s u lin d e riva tive . When chromatographed in 7M urea buffer on a DEAE Sephadex column the deriva tive was w ell separated from unmodified in su lin . It eluted at a higher salt concentration than in su lin , as would be expected i f the free amino groups were blocked (Figure 4). The deriva tive eluted essentially as a single peak on this colum n, ind ica ting a nearly pure product. When the peak was isolated and rechromatographed in 7M urea on a Sephadex G -5 0 column a single sharp peak was eluted w ith the same shape and elution volumn as unmodified insulin (Figure 5). This indicated that the material had the same m olecular w eight as insulin and was thus not crosslinked by disulfide bridges, in contrast to the variety o f apparent m olecular weights observed when the de rivative was prepared in water (Figure 9). Analysis o f this purified derivative for free sulfhydryl groups w ith Ellman's reagent (Ellm an, 1959) indicated essentially no form ation o f sulfhydryl groups on the m olecule. For the product prepared in w ater, on the other hand, the Ellman procedure indicated a high degree o f 50 disulfide bond cleavage. In order to determine whether the amino groups were blocked, samples o f the p u rifie d d eriva tive were treated w ith fluorodinitrobenzene (Sanger's Reagent). However, the conditions used were found to cause hydrolysis of the triflu o ro a ce tyl group attached to phenylalanine. This method could thus not be used to ascertain whether the amino group o f the N -te rm in a l phenylalanine was blocked or free. It is also d iffic u lt to use this reagent to quantitate the degree o f blocking at the N -te rm in a l glycine due to the pa rtial descomposition o f d in itro p h e n yl-g lycin e during acid hydrolysis (Porter and Sanger, 1948). As an a lterna tive assay for the blocked amino groups deam ination w ith sodium nitrate was u tiliz e d . In this assay amino acids w ith free amino groups are converted to th e ir hydroxy analogues, w hile amino acids w ith th e ir amino groups blocked by trifluoroacetate are uneffected. When the protein is sub sequently hydrolysed the triflu o ro a ce tyl groups are removed by hydrolysis. Blocked amino groups are thus detected by th e ir presence and free amino groups by th e ir absence in amino acid analysis o f the d e riv a tiv e . By this assay the tri-triflu o ro a c e ty l-in s u lin d e rivative was indeed shown to have a ll three groups blocked (Table I). M ono- and d i-triflu o ro a c e ty l-in s u lin derivatives. A fte r successfully synthesizing and characterizing the tri-triflu o ro a c e ty l deriva tive an e ffo rt was made to prepare the various mono- and d i-triflu o ro a c e ty l-in s u lin derivatives. In the best preparations an alternate reagent (phenyltrifluoroacetate) and base (im idazole) were u tiliz e d . The com bination o f phenyl triflu oro acetate and 51 im idazole affords a slower reaction, resulting in the formation o f several p a rtia lly substituted products. For exam ple, this system gave a m ixture o f products afte r four hours w h ile the comparable trie th y la m in e /e th y l-th io ltriflu o ro a c e ta te system afforded complete reaction afte r only one hour. In a d ditio n , this system e lim in ates any possibility o f disulfide bridge interchange due to the ethane thiol which is produced during the reaction w ith ethyl th io ltriflu o ro a c e ta te . A comparison o f the product spectra obtained w ith the two reagents and im idazole as base at optim al reaction times may be seen in Figure 6 . The various deriva tives were separated from one another by chromatography on DEAE Sephadex in 7M urea buffer w ith elution by a sodium chloride concentration gradient. The fo llo w in g triflu o ro a c e ty l-in s u lin derivatives were isolated and B— 1 characterized by deamination and amino acid analysis: phenylalanine triflu o ro a c e ty l-in s u lin ; g ly c in e ^ ” ^ -triflu o ro a c e ty l-in s u l in; g ly c in e ^ phenylalanine^- ^ -triflu o ro a c e ty l-in s u lin ; and g ly c in e ^ - l, lysine®“ ^ - t r i - flu o ro a c e ty l-in s u lin . The remaining peaks, probably corresponding to lysine^- ^ d o o R—1 triflu o ro a c e ty l-in s u lin and lysine ” , phenylalanine -triflu o ro a c e ty l-in s u lin , were not isolated because of th e ir generally poor y ie ld and because they were often poorly resolved. The reaction products obtained using both tri fl uoroacety I- ating reagents in the presence o f im idazole indicated a preference for reaction at N -te rm inal amino groups. Flowever, the use o f a stronger base, trie th ylam in e, resulted in an increased rate o f reaction at the € -am ino group o f lysine^- ^ , re sulting p rim a rily in a fu lly substituted insulin d e riva tive . B iological and im m unological a c tiv ity . A sample of tri-triflu o ro a c e ty l- 52 insulin d e rivative was repurified by isoelectric p re cip ita tio n and analyzed for b io lo g ica l and im m unological a c tiv ity by Eli L illy and Company. This deriva tive was shown to have approxim ately 70% o f the b io lo g ica l and 60% o f the im m unological a c tiv ity o f crystalline zinc insulin. The loss o f b io lo g ica l a c tiv ity for the tri-triflu o ro a c e ty l d e riva tive is sim ilar to losses reported for other insulin derivatives w ith com pletely substituted amino groups blocked by a va rie ty o f d iffe rent groups d iffe rin g in both size and charge. Complete substitution w ith the acetyl moiety,, however, gives a d e riva tive w ith essentially no loss o f a c tiv ity in in vivo tests (Lindsay and Shall, 1971; G a ttn e r, 1972). The same deriva tive has decreased a c tiv ity in in v itro tests (Zahn et a l. , 1972). These results would seem to indicate an enzym atic deacetylation in in vivo systems. Normal acetylases would not be expected to hydrolyse the triflu o ro a ce tyl m oiety and in fact the triflu o ro a ce tyl derivatives show decreased a c tiv itie s comparable to the results seen w ith other stable mod ific a tio n s . However, studies on trica rb a m yl-in su lin , in w hich the carbamyl groups cannot be removed e n zym a tica lly, have shown that this de rivative possesses essentially fu ll b io lo g ica l a c tiv ity . A comparison o f these last results, both invo lving small m odifying groups, leaves the question of what is happening w ith the acetylated de riva tive unresolved. But it is clear that the b io lo g ica l a c tiv ity o f insulin is sensitive to rather subtle changes at the N -te rm in a l o f the insulin A -c h a in . Several o f the tri fl uoroacety I ated insulin derivatives were also subjected R— 1 to radioimmuno assays. The values for phenylalanine -triflu o ro a c e ty l-in s u lin , 53 g ly cin e A ” ^ -triflu o ro a c e ty l-in s u lin and g ly c in e ^ "^ -lysine® _^ -tr iflu o r o a c e ty l- insulin are sim ilar to the im m uno-reactivities o f several other insulin derivatives, w h ile the values obtained for g ly c in e ^ ” ^-phenylalanine® ” ^ -triflu o ro a c e ty l- insulin and the tri-triflu o ro a c e ty l d e rivative are somewhat higher than other m odified insulins based on comparative studies (Blundell et a l. , 1972). C rystalline insu lin. Crystals o f g ly c in e ^ ” !-triflu o ro a c e ty l-in s u lin were grown in citra te buffer in the presence o f zinc io n , as shown in Figure 7. The crysta lliza tio n o f insulin derivatives m odified at the g ly c in e ^ ^ residue has also been reported by several other groups (Brandenburg et a l. , 1972; Lindsay and S hall, 1969, 1970; W einert et a ]. , 1971). M o d ifica tio n o f the amino group o f the phenylalanine®” ^ residue has resisted crysta lliza tio n because o f its c ritic a l location in the crystal. Steric alterations in this area would probably interfere s ig n ific a n tly w ith the proper dim er association leading to the final hexamer u n it, as assessed by the X -ra y crystal log ra phi c results (Blundell et a l. , 1972). M inor m odifications (Krai I e t a l . , 1 971) or deletions ( Z a h n e ta l., 1972) . R— 1 . o f phenylalanine , however, have afforded crystalline m aterial. 54 PART II ^ F -N M R Studies-The Effects o f Solvent, pH, Salts and Denaturants on the Conformation and Aggregation Properties o f T rifluoroacetylated Insulin D erivatives 55 IntroducHon - Part II NMR spectroscopy is based on the absorption o f radiofrequency radiation by atom ic nuclei subjected to a strong magnetic fie ld . Populations o f identical nuclei give rise to bell shaped absorption curves in plots o f am plitude vs. fre quency. Three types o f data may be measured on these curves: the inte nsity, or area under the curve, w hich is proportional to the number of nuclei in a p a rticu lar chemical environm ent; the position o f the center o f the curve, or chemical sh ift, measured in frequency units re la tive to a standard, w hich is determined by the chem ical environment o f the nucle i; and the w idth o f the curve, mea sured in frequency units at h a lf peak h e ig h t, w hich is determined by the rate of motion o f the n u cle i. Thus NMR may be used to determine the number o f nuclei in a p a rticula r environm ent, draw inferences about the nature o f this environ ment and draw conclusions about the m o b ility o f the n u cle i. In itia l NMR studies o f proteins u tiliz e d proton magnetic resonance. Pro ton m agnetic resonance spectroscopy has the advantages that protons are ubi quitous in organic molecules and that the proton gives rise to the strongest magnetic resonance signals and thus offers the greatest sensitivity o f any nuclei for NMR spectroscopy. However, proton NMR o f proteins affords extrem ely complex spectra due to the large number o f nearly equivalent protons and the broadening o f the ind ividu a l peaks resulting from the incorporation o f the protons 56 into a m acrom olecule. Such spectra generally appear as broad, featureless and uninform ative envelopes composed o f the many overlapping resonance peaks. Nevertheless resonance peaks occurring at high or low frequency (to e ith er side o f the main envelope) are often resolved. By u tiliz in g these peaks detailed inform ation about the dyanmic behavior o f proteins in solution has been ob tained. For exam ple, the resonance peaks for the histidine protons of rib o - nuclease A are w ell resolved from the main envelope in proton NMR spectra o f this p ro te in . By u tiliz in g NMR in conjunction w ith titra tio n experiments it has been possible to assign specific pK values to each o f the histidine residues in the ribonuclease m olecule (Meadows et a l. , 1967; Meadows et a l. , 1968). This assignment could not have been made w ith other a va ila b le techniques. O ther proteins w hich have been studied extensively by proton magnetic reson ance include: lysozyme, staphylococcal nuclease and hemoglobin (Jardetsky and W ade-Jardetsky, 1971). A number o f strategies have been used in an attem pt to sim plify the nuclear magnetic resonance spectra o f proteins. One strategy involves the selective deuteration o f a ll but a few o f the amino acid residues o f a protein. This method yields greatly sim plified proton magnetic resonance spectra (Katz e t a l. , 1968; M arkley et a l. , 1968). Using this method it would be possible to assign specific resonances to each o f the amino acid side chains in a prote in. Such studies are lim ite d however, by methods for selective deuteration, such as growing simple organisms in deuterated m edia. A second approach involves the use o f nuclei other than protons as nuclear 57 magnetic resonance reporter groups. Reporter groups may be incorporated into small molecules w hich inte ract s p e c ific a lly w ith the proteins, such as enzyme substrates, or they may be incorporated into the protein its e lf. Reporter groups may be incorporated b io syn th e tica lly, or by specific chemical m od ifica tio n . A p a rtic u la rly good nucleus to serve as a reporter group is ^ F lu o rin e . 19 FI uorine has a number o f desirable properties, inclu d in g : 1) the ^ F lu o rin e nucleus is nearly as small as the proton, 2) ^ F lu o rin e chemical shifts are an order o f magnitude greater than the shifts o f corresponding proton analogues, 3) the chem ical shifts are very sensitive to alterations in the environm ent, and 4) 19Fluorine y i elds a vastly sim plified spectrum for the m odified protein due to the small number o f fluorine nuclei introduced. Recently ^ F lu o rin e has been ch em ically introduced into ribonuclease, hemoglobin and insulin as an NMR reporter group (Huestis and Raftery, 1971, 19 7 2 a ,b ,c; Paselk and Levy, 1974a, b ). In these reports ^ F was introduced as the triflu o ro a ce tyl (or in the case of hem oglobin, the triflu o ro a ce to n yl) group. The triflu o ro a ce tyl groups was chosen in these studies because: 1) it is sm all, and thus not lik e ly to greatly disturb the protein's conform ation, 2) it may be introduced at defined locations in the protein under chem ically m ild conditions, 3) it contains three ide n tica l ^ F n u c le i, thus givin g a stronger NMR signal, and 4) there is no s p littin g o f the ^ F -N M R signal from this group. In ribonuclease the triflu o ro a ce tyl group was chem ically introduced at lysine residues 1 and 7 by reacting ribonuclease S-peptide (ribonuclease residues 1-20) w ith ethyl th io ltriflu o ro a ce ta te in aqueous solution. The m odified 58 S-peptide was then combined w ith ribonuclease S-protein (residues 21-124) to give an active enzyme preparation. ^ F -N M R was then used to observe changes in the environments o f the triflu o ro a ce tyl groups on the lysine residues when various inhibitors were bound to the enzyme (Huestis and Raftery, 1971). The resultant resonance peaks were assigned to the ind ivid u a l m odified lysine residues by comparing spectra o f triflu o ro a ce tyla te d derivatives o f lysine and the S-peptide w ith the reconstituted ribonuclease. T rifluoroacetyl groups attached to the two lysine residues were found to have s ig n ific a n tly d ifferent chem ical environments, as reflected in th e ir d ifferent NMR chem ical shifts. The difference in shifts is consistent w ith the conform ation o f ribonuclease derived from X -ra y crystallography, in which lysin e -l is exposed to solvent, w h ile lysine-7 is in the in te rio r o f the m olecule. The addition o f various inhibitors o f ribonuclease resulted in alterations in the environm ent o f lysin e -7 , as observed in the form o f altered ^ F -N M R chemical shifts for this residue. A d d itio n a l ^ F -N M R studies o f this tri f I uoroacety lated p ro te in , as w e ll as o f tri fl uoroacety lated hemoglobin have also been reported (Huestis and Raftery, 1 9 7 2 a ,b ,c ). l^ F -N M R h a s also been used to study the conform ation o f angio tensin II analogs, incorporating fluorine as p -flu o ro -L -p h e n yla la n in e (Vine et a [ . , 1973. 19f_ n MR studies o f proteins involving fluorinated ligands or substrates have also been reported. For exam ple, Zeffren and Reavill (1968) and Zeffren (1970) have studied inhibitor-enzym e inte raction in chymotrypsin u tiliz in g triflu o ro a ce tylp h e n ylala n in e as the in h ib ito r and ^ F -N M R probe. ^ F -N M R 59 has also been used to study the binding o f triflu o ro a ce tyl glucosamine oligomers (substrate analogs) to lysozyme (M ille tt and Raftery, 1972). Sim ilar studies have also been made o f $ -la c to g lo b u lin A and aspartate transcarba my I ase (Robillard and W ishnia, 1972; Cheng and M a rtin e z-C a rrio n , 1972). 19F -N M R te ch n i ques have thus been useful to gain detailed inform ation about the conformations and behaviors o f m odified proteins in solution. NMR is also a technique o f extrem ely high potential resolution (Jardetsky and W ade- Jardetsky, 1971), and thus enables a comparison o f details o f protein structure in solution w ith the crysta llin e structure revealed by X -ra y crystallography. NMR spectra o f proteins may be influenced by a variety o f interactions o f the protein w ith its e lf and w ith its environm ent. In order to interpret the spectra it is thus necessary to have inform ation about the protein from a variety o f other techniques. This inform ation is also useful to confirm that the deriva tives are not markedly d iffe re nt in conform ation or aggregation properties from the native pro te in. A va riety o f studies have been concerned w ith the relationship o f structure and function for the hormone, in su lin . The com plex aggregation properties and conform ation o f this m olecule have been studied extensively. Sedimentation v e lo c ity and sedimentation equilibrium studies have shown that the aggregation state o f insulin is dependent upon concentration, the presence o f metal ions, pH and the io n ic m edia. A t low pH insulin does not aggregate extensively, existing predom inately as dimers near pH 2 .0 in 0 .0 5 -0 .1 0 M NaCI and at a protein concentrations o f 0 .1 -0 .2 5 % (Jeffrey and Coates, 1966a,b; Fredericq, 1956). The aggregation behavior that has been observed at low pH depends ^ on the p a rticu la r species o f salts present and th e ir concentrations (Fredericq and N eurath, 1950; Fredericq, 1956; Jeffrey and Coates, 1966a,b). Insulin does not bind zinc or related cations at low pH and its behavior at low pH is not affected by low concentrations o f these ions. The low pH aggregation behavior o f insulin may be described as follow s: monomers aggregate to dimers, and the dimers may then aggregate further to tetramers and possibly hexamers (Jeffrey and Coates, 1966a,b). This scheme is supported by studies under sim ilar conditions using the techniques o f u ltra v io le t spectroscopy (Rupley et a l. , 1967; Lord et a l. , 1973). Studies o f z in c -fre e insulin at neutral pH values demonstrate a marked dependence o f the aggregation state on insulin concentration. M olecular weights ranging from less than that o f the dim er to greater than that o f the octamer as a function o f insulin concentration have been reported (Pekar and Frank, 1972; Fredericq, 1956). Results from sedimentation equilibrium studies at neutral pH may be described in terms o f a model invo lving e q u ilib ria between monomers, dimers, hexamers and higher aggregates (Pekar and Frank, 1972). This model is sim ilar to the model for the low pH aggregation o f in su lin . Under m oderately a lka lin e conditions z in c -fre e insulin tends to disaggregate (Sloben and C arpenter, 1966; Goldman and Carpenter, 1974). A t moderate concen trations (less than 0 .2 5 % ), z in c -fre e insulin may even dissociate to monomers at pH 9 .1 -9 .5 (Fredericq, 1956; M arcker, 1950a,b). A t pH values near n e u tra lity the aggregation behavior o f insulin is strongly influenced by the presence o f zinc or related diva le nt cations. Insulin has two 61 tig h t binding sites for zinc per insulin hexamer. This tig h tly bound zinc cannot be removed by dialysis (Cunningham et a l. , 1955). In a d d itio n , insulin w ill bind at least seven additional zinc atoms per hexamer in a weaker complex (Cunningham et a l. , 1955). It has been found that at least two zinc atoms per hexamer are required for crysta lliza tio n (S c h lic h tk ru ll, 1958). C rystalline beef or porcine insulin exists predom inately as hexamers in neutral pH solutions, ex h ib itin g S 20 w values o f 3 .4 to 3 .5 (Cunningham et a l. , 1 955; Fredericq, 1956). The hexamer is the m ajor aggregation state for crysta llin e insulin over a re la tiv e ly w ide range o f concentrations: 0 .1 -2 .4 % (Blundell et a l. , 1972). However, increasing the zinc concentration (greater than two zinc atoms per hexamer) results in aggregation to larger polyers, w ith S 20 w va^ues U P 9.5 (Fredericq, 1956). A va rie ty o f insulin derivatives blocked at theirc<-am ino groups also form hexamers in zinc solutions. Phenyl isothiocyanate derivatives do not e x h ib it the zinc dependent m olecular w eight increase above the hexamer seen in native insulin (M arcker, 1960a,b). O p tic a l rotatory dispersion and circ u la r dichroism have also been used to study the aggregation properties and conform ation o f insulin. Insulin spectra derived from these techniques have major cotton effects centered near 192 nm, 208 nm and 222 nm, and a minor band near 273 nm (M ercola et a l. , 1967; Ettinger and Timasheff, 1971; Goldman and C arpenter, 1974). The cotton effects near 222 nm and 273 nm are sensitive to changes in aggregation, p a rtic u la rly to transitions between monomer and dimer states. The 273 nm cotton e ffe ct is due to assym etrically held arom atic residues in insulin dimers 62 (tyrosines and phenylalanines). The sensitivity o f this cotton e ffe ct to aggrega tion state has been interpreted in terms o f the aromatic residues being sym etrically exposed to solvent (and thus not contributing to c irc u la r dichroism spectra) in the monomer (Morris et a l. , 1968). This interpretation is in agreement w ith the location o f these residues in the dimer and hexamer o f crystalline insulin (Blundell et a l. , 1972). Goldman and Carpenter (1974) have made tentative assignments o f the various contributions o f this c irc u la r dichroism band to in d ivid u al residues of the insulin dim er. The 222 nm band can be largely assigned to the structure created by monomer-monomer interactions in the dimer (G old man and C arpenter, 1974; Ettinger and Timasheff, 1971). This band is strongly affected by monomer-dimer transitions. The addition o f zin c ion to insulin causes a strengthening o f the 273 nm band (Goldman and C arpenter, 1974). The cotton e ffe ct near 208 nm may be attributed to © ^-helical and dis torted h e lica l regions o f the insulin m olecule (Goldman and Carpenter, 1974; Ettinger and Timasheff, 1971). This band is sensitive to insulin conform ation, but not to aggregation. Z in c binding causes an attenuation o f the 208 nm band in c ircu la r dichroism spectra o f in su lin . This may be due to the zinc binding s p e c ific a lly to histidine B-10, a component o f the p rin cip le o c -h e lix of in sulin (residues B-10-19) (Goldman and Carpenter, 1974). A number o f studies have looked at the circu la r dichroism spectra o f m odified insulins (M ercola et a j. , 1967; Levy and C arpenter, 1970; Brugman and A rq u illa , 1973; Goldman and Carpenter, 1974). It has been found that the ratio o f the intensity o f the d ich roic band at 208 nm to the intensity o f the 63 band at 222 nm is correlated to the b io lo g ica l a c tiv ity (and by im p lic a tio n , the conform ation) o f insulin derivatives. Comparing an extensive series o f insulin derivatives Zahn and coworkers (Blundell et a l. , 1972, pp 345-346) have demon strated a very approximate inverse relationship between this ra tio and the b io lo g ica l a c tiv ity o f the derivatives. The ratios observed ran from approxim ately 1.12 for crystalline insulin (100% a ctive ) to approxim ately 1.6 for insulin derivatives w ith b io lo g ica l a c tiv itie s o f 20% and less. Recently reporter groups have been introduced into the insulin m olecule fo r use w ith spectroscopic techniques. Thus, in a prelim inary report, triflu o ro - a c e ty l-g ly c in e enriched w ith has been introduced onto the B-chain o f . 1 3 in su lin . This group was then observed w ith °C -N M R spectroscopy (Saunders and O ffo rd , 1972). M ercola et a l. , (1972) u tiliz e d the symmetric dye fluo re scein isothiocyanate bound to the amino groups o f g ly c in e ^ ”"^ , phenylalanine^- ^ B— 29 and lysine as reporter groups for absorption and circu la r dichroism spectro scopy between 400 and 550 nm. Their results indicated that the dye groups on g ly c in e ^ ” ^ and phenylalanine^- ^ were separated by 7 .6 A or less w h ile the dye groups on g ly c in e ^ ”" ^ and lysine^- ^ were separated by 10.5 R or less. U nfortunately the reporter group used in this study is large and the conformations o f the derivatives may d iffe r substantially from native insulin. 64 Materials Ethyl th io ltriflu o ro a ce ta te was the product o f Pierce Chemical Company. G ly c y lg y c y lg ly c in e and p h e n yla ia n ylg lycylg lycin e were obtained from Sigma Chemical Company. L -G ly c in e and L-phenylalanine were obtained from C albiochem . D, L-Lysine monohydrochloride was the product o f N u tritio n a l Biochemical Company. L -l-tosyla m id o -2 -p he n ylethylch loro m e th yl ketone (TPCK)-treated trypsin was purchased from W orthington Biochemical C orporation. The chromatography resins (Sephadex G -5 0 , fine; Sephadex G -1 0 , fine) were obtained from Pharmacia Fine Chem icals, Incorporation. Dowex 50W--X2 was obtained from Bio-Rad Laboratories. M N Polygram SIL N -H R precoated plastic sheets were obtained from Brinkmann Instruments, Incorporation. Insulin (Lot N o. 493-88G P D -017) was the product o f Eli L illy and Company. A ll other chem icals were o f a n a lytica l grade and were obtained from M a llin ckro d t Chem ica l Company. N -triflu o ro a c e ty I-g ly c in e , N -triflu o ro a c e ty l-p h e n y la la n in e , N , € - triflu o ro a c e ty I-ly s in e and N -triflu o ro a c e ty l-g ly c y lg ly c lg ly c in e were prepared according to the procedures o f Schallenberg and C alvin (1955). N -triflu o ro - a c e ty l-g ly c in e , reported mp 114-116°; found, 115-117°; N -triflu o ro a c e ty l- phenylalanine, reported mp 1 1 9 .4 -1 2 0 .6 °; found, 120-121°; N , € -triflu o ro - a c e ty l-ly s in e , reported mp 226-231°; found 224-230° (Schallenberg and C a lvin , 65 1955); N -triflu o ro a c e ty l-g ly c y lg ly c y lg ly c in e , reported mp 225-228°; found, 227-229° (Weygand and Ropsch, 1959). T riflu o ro a ce tyl-p h e n yla la n ylg ly c y lg ly c in e. P henyla lanylglycylglycin e was reacted w ith ethyl th io l tri fluoroacetate by the method o f Schallenberg and C alvin (1955), and recrystallized from e thyl-acetate-petroleum ether (bp 3 0 -6 0 °); mp 174-176°; y ie ld 50% . This product was shown to be ninhydrin negative and homogeneous on thin layer chromatography using two solvent systems (ammonium hydroxide: n-propanol 30:70; and chloroform : m ethanol: a ce tic a c id , 95:5:1). A n a l. Calcd for C]5 F3 O5 : C , 4 8.0 0; H, 4 .3 0 ; F, 15.19. Found: C , 4 8.15; H, 4 .3 3 ; F, 14.92. N , € -L y s in e ^ ^ ^ -triflu o ro a c e ty l-in s u lin octapeptide. T ri-triflu o ro - a c e ty l-in s u lin (Levy and Paselk, 1973) (170 mg, 28y*m ole) was dissolved in bicarbonate b u ffe r, pH 8 .0 . To this solution was added 14.2 ml o f TPCK- trypsin solution (pH 3 .0 , 1 mM C aC l, 1 .2 m g/m l) and the reaction stirred for 4 hours at 3 8 °. The reaction m ixture was chromatographed on Sephadex G -5 0 , follow ed by chromatography on Sephadex G -1 0 , as previously described (Levy and Paselk, 1973) to afford the N , £ -ly s in e ^ ” ^ -triflu o ro a c e ty l-o c ta p e p tid e deri va tiv e . G ly c in e ^ ” ^-triflu o ro a c e ty l-S -s u lp h o -in s u lin A cha in. Insulin hydro chloride (164 mg, 28 jAmole) was subjected to sulphotolysis according to the procedures of Dixon and W ardlaw (1960) to afford the S-sulpho derivatives of the A and B chains. The p u rifie d S-sulpho A chain (75 mg, 28^xm ole) was triflu o ro a ce tyla te d in dim ethylform am ide to increase the so lu b ility o f the 66 p e ptid e , using procedures previously described (Levy and Paselk, 1973). The reaction m ixture was lyo p h ilize d and the product p u rifie d on Sephadex G -2 5 , e lu tin g w ith 2 .5 N a cetic a c id , affording the ninhydrin negative peptide. G ly c i n e ^ - ^ -p h e n yla la n in e ^- ^ - tr i fl uoroacetyl -desoctapeptide-i nsul in . Insulin hydrochloride (100 mg, 172jAmole) was treated w ith TPCK-trypsin as previously described (Levy and Paselk, 1973) to afford desoctapeptide-insulin, w hich was p urified on Sephadex G -2 5 , eluting w ith 0.05 M ammonium acetate, pH 7 .5 . T riflu o ro a ce tyla tion o f the insulin d e riva tive was effected in d im e th yl- formamide w ith ethyl th io l tri fluoroacetate (Levy and Paselk, 1973). The product was p u rifie d on Sephadex G -5 0 , eluting w ith 0.05 M ammonium acetate, pH 7 .5 . The product was characterized by ninhydrin analysis and by deam ination follow ed by amino acid analysis (Levy and Paselk, 1973). T riflu oro a ce tyla tio n of in su lin . The various m ono-, d i- and tr i- triflu o ro - a c e ty l-in s u lin derivatives were prepared as described in part one o f this dissertation. For the studies invo lving zinc an a dditional step was performed: the various p u rifie d derivatives were further p u rifie d by dissolving in 30% a ce tic acid follow ed by chromatography on a Sephadex G -1 0 colum n. The protein peaks were then collected and ly o p h iliz e d . 67 Methods Amino acid analysis were performed on a Technicon amino acid analyzer (Spackman et a I . , 1958). Samples were hydrolyzed in 6 N HCI at 120° for 6 hours. U ltra v io le t absorption measurements were taken on a Zeiss spectrophoto meter (PMQ II). pH Measurements were made on a Radiometer Copenhagen model 26 or 4 pH meter. Elemental analysis was performed by Elek Laboratories, Los Angeles, C a lifo rn ia . M elting points were taken on a Fischer-Johns m elting point apparatus. Thin layer chromatography was performed on "polygram " SIL N -H R precoated plastic sheets. ^ F -N u c le a r magnetic resonance spectra were recorded on a Varian H A -100-15 spectrometer m odified to operate at 94.1 M H z. Spectra were accum ulated on a Fabritek model 1074 computer o f average transients. Sedimentation v e lo c ity measurements were made on a Beckman-Spinco Model E a n a lytica l ultracen trifuge equipped w ith a Schlieren o ptical system and rotor temperature controls. C ircu la r dichroism spectra were recorded using a m odified Beckman CD spectrometer described elsewhere (H orw itz and H e lle r, 1973). Spectra were averaged on a H ew lett-Packard 5480A signal analyzer. N ucle a r magnetic resonance measurements. The triflu o ro a ce tyl-a m in o acids, peptides and insulin derivatives were dissolved in w ater and the pH adjusted w ith HCI or N a O H . Am ino acid and peptide derivatives were run at a concentration o f 0.05 M . N -triflu o ro a c e ty l-g ly c in e was also run at a 68 concentration o f 1 .5 mM. Insulin derivatives were run at a concentration o f 10 m g /m l. Samples containing zin c were prepared by dissolving the appropriate insulin d e riva tive in 0.01 N N aC I. A 10y*l a liq u o t o f zinc solution was then added to give a final concentration o f e ith er 0.3% or 0 .6 % w /w zinc re la tive to in su lin . The pH was then adjusted accordingly using sodium hydroxide and hydrochloric a c id . The pH o f samples containing perturbants, such as salts or detergents, was checked and readjusted after each addition o f perturbant. A ll spectra were measured in 12 mm tubes (W ilmad Glass Company, N o . 514A-7PP) at a probe temperature o f 35 + 1 °, using a co n ce n trica lly held 4 mM ca p illa ry tube o f triflu o ro a c e tic acid as an external reference standard. A ll samples were allow ed to e q u ilib ra te to the temperature o f the probe before spectra were accum ulated. Spectra were accum ulated for 10-18 hours (approx im ately 230-420 scans). Peak widths were measured at h a lf maximum in te nsity. Sedimentation v e lo c ity studies. Samples o f triflu o ro a c e ty I-in s u lin deriva tives (10 m g/m l) were prepared in an ide ntical fashion to the solutions used in ^ F -m a g n e tic resonance studies. Samples were sedimented at 59,790 rpm. Sedimentation coefficients were corrected to S 2Q w using a partial sp ecific volume of 0.73 m l/g (Pekar and Frank, 1972) and solvent corrections for the proper salt concentrations. C ircu la r dichroism measurements: 240 nm - 360 nm region. Samples o f the triflu o ro a c e ty l-in s u lin derivatives (10 m g/m l) were prepared in an identical fashion to the solutions used in ^ F -n u c le a r magnetic resonance studies. Spectra were measured in a 1 .0 mm pathlength quartz c e ll. Spectra were recorded w ith 69 respect to the base-line signal obtained w ith solvent alone. Measurements from 190 nm - 250 nm: the various samples used above were dilu ted w ith appropriate solutions to give a concentration o f 1 mg/ml insulin d e riv a tiv e , a ll other concentrations remaining constant. Spectra were measured in a 0 .12 mm path- length quartz c e ll. Spectra were recorded w ith respect to the base-line signal obtained w ith solvent alone. 70 Results ^ F -N M R studies o f z in c-fre e insulin derivatives and model compounds as a function o f pH and solvent. The triflu o ro a ce tyl m oiety has been introduced at various points in the insulin m olecule (Figure 10; R=H or CF3 CO) as described e a rlie r. This group can now serve as an environm ental probe u tiliz in g ^ F -N M R techniques to elucidate the nature o f the complex aggregation behavior and con form ational changes o f insulin in solution. The triflu o ro a ce tyla te d insulin derivatives were studied at various pH values in the absence o f zin c ion. A series o f triflu o ro a ce tyla te d amino acids and peptides were also studied as model compounds under these conditions as an aid in interpreting the data. Chemical characterization o f model compounds. Lysine^"29_t-r|f|uoro_ a c e tyl-in su lin -o cta p e p tid e (G ly-P h e -P he -T yr-T h r-P ro -L ys-A la ) l c o c f 3 was obtained u tiliz in g trypsin cleavage of tri-triflu o ro a c e ty l-in s u lin . The id e n tity o f this product was established by amino acid analysis. No alanine was released on enzym atic hydrolysis indicating that the £ -am ino group o f lysine was blocked w ith the triflu o ro a c e ty l m oiety. It was also observed that the conditions used a ffe ct the cleavage o f the octapeptide resulted in a slight hydrolysis of the triflu o ro a c e tyl group on g ly c in e ^ ” ^, as assessed by ninhydrin analysis and amino 71 Figure 10. Structure o f the insulin monomer (redrawn from Blundell et a l. , 1972) in d ica tin g the location o f the fluorine probes (R = H or CF3C O ). A chain ( — ©— ); B chain ( m o m ); disulfide bridges ( «*m o h * ) . 72 B - 3 0 A - 1 B - 2 9 B -1 A - 21 acid analysis o f the deaminated product. The preparation o f g ly c in e ^ - ! -p h e n y l- a la n in e ^” ^-triflu o ro a ce tyl-d e so cta p e p tid e -in su lin was thus prepared d ire c tly from desoctapeptide-insulin using the same procedures as described for native in s u lin , to afford a homogeneous product w hich was ninhydrin negative. S- sulpho-insulin A chain was also triflu o ro a ce tyla te d on g ly c in e ^ - ^ as previously described, yie ld in g a ninhydrin negative product which was further characterized by amino acid analysis o f the deaminated d e riva tive . 19p_NM R spectroscopy o f model compounds. The ^ F -N M R spectra of the amino acids, peptides and insulin derivatives were taken in 5 N a ce tic acid and at pH 2 .0 + .1 , 6.8 + .1 and 8 .7 + .2 . The chemical shifts o f N -triflu o ro - a c e ty l-g ly c in e , N -triflu o ro a c e ty l-p h e n y la la n in e and N -triflu o ro a c e ty l-ly s in e in 5 N a ce tic a c id , pH 2 .0 , 6.8 and 8 .7 are shown in Figure 11 A , B and C. A pronounced dow nfield shift is observed when these compounds are transferred from 5 N a ce tic acid to w ater, pH 2 .0 ; however, the introduction of a negative charge on the carboxyl group when the pH was raised to 6.8 had only a minimal e ffe ct on the g lycin e and lysine d erivatives, and no e ffe ct on the phenylalanine d e riv a tiv e . There was also no e ffe ct o f concentration o f the ^ F -c h e m ic a l shift o f N -triflu o ro a c e ty l-g ly c in e in the range 1 .5 -5 0 mM a t pH 2 .0 . The e ffe ct o f m odifying the carboxyl group in these triflu o ro a ce tyl-m o d e l compounds by the addition o f amino acids to form N -triflu o ro a c e ty l-g ly c y lg ly - c y lg ly c in e , N -triflu o ro a c e ty l-p h e n y la la n y lg ly c y lg ly c in e and lysine^“ ^ ^ - tr i- flu o ro a ce tyl-in su lin -o cta p e p tid e resulted in dow nfield shifts in a ll solvents and pH values (Figure 11 A , B and C). A g a in , the dow nfield shifts observed when 74 Figure 11. ^ F -c h e m ic a l shifts o f triflu o ro a ce tyl-a m in o acids, peptides and insulin derivatives as a function o f solvent (5 N a ce tic acid (HAc) or w ater) and pH . (A) triflu o ro a ce tyl group bound to theo<-annino group g lycin e : 1) N -triflu o ro a c e ty l-g ly c in e ; 2) N -triflu o ro - a c e ty l-g ly c y lg ly c y lg ly c in e ; 3) g ly c in e ^ ” ^--trifluoroacetyl--S - sulpho-insulin A chain; 4) triflu o ro a c e ty l-g ly c in e -in s u lin deriva tives. (B) triflu o ro a ce tyl-g ro u p bound to the € -am ino group B— 29 o f lysine: 1) N ,6 -triflu o ro a c e ty I-ly s in e ; 2) lysine -triflu o ro - a c e ty l-in s u lin -o c ta p e p tid e ; 3) triflu o ro a c e ty l-ly s in e -in s u lin d e riv atives. (C) triflu o ro a ce tyl bound to th e **-a m in o group o f phenyl alanine: 1) N -triflu o ro a c e ty l-p h e n y la la n in e ; 2) N -triflu o ro - a c e ty l-p h e n y la la n y lg ly c y lg ly c in e ; 3) tri fluoroacety I-ph e nyla l a - n in e -in su lin derivatives; 4) g ly c in e ^ ” ^ , phenylalanine^” ^ — tri — f I uoroacety I -desoctapepti d e -i nsul i n . 75 A 8.7 B C -3 2 0 -310 -3 0 0 -2 8 0 76 the compounds were transferred from 5 N a ce tic acid to w ater were quite marked, but were re la tiv e ly insensitive to alterations in pH. Lysine^“ ^ -triflu o ro a c e ty l-in s u lin -o c ta p e p tid e could not be run at pH 6.8 because o f its low s o lu b ility . G ly c in e ^ - ^-triflu o ro a c e ty l-S -s u lp h o -in s u lin A chain was shifted further dow nfield than the sm aller model compounds, and again the chem ical shift was markedly altered on transferring the compound from 5 N a ce tic acid to w ater, pH 2 .0 , but was also insensitive to a pH alteration to 6.8 (Fig ure 11 A ). The lin e widths o f a ll o f these tri fluoroacety I-d e riva tive s were narrow (2-3 Hz) in both 5 N a ce tic acid and in w ater, and were essentially in va ria n t over the pH range investigated. The presence o f sodium dodecyl sul fate (1 % w /v ) did not a ffe ct e ith e r the chemical shifts or lin e widths o f the model compounds at neutral pH. 19F-N M R spectroscopy o f z in c -fre e in su lin . The ^ F -N M R spectra o f the various tri fluoroacety I-in s u lin derivatives were studied under the ide n tica l conditions used for the model compounds. The spectrum o f g ly c in e ^ - ^ , phenyl- n _ I n on ala n ine , lysine - triflu o ro a c e ty l-in s u lin in 5 N ace tic acid is shown in Figure 12A. The assignment o f each o f the three peaks was established by obtaining a spectrum o f the two m o n o -triflu o ro a ce tyl-in su lin derivatives and the two d i-triflu o ro a c e ty l-in s u lin derivatives. The ^F -resonance positions for the three tri fluoroacety I-groups w ent from low to high fie ld in the order: phenyl a la n in e , g ly c in e , lysine. The lin e widths in two m ono-, two d i- and one t r i— tri fluoroacety I-in su l in derivatives were a ll quite narrow (Table IV and Table V ). The chem ical shifts were located at low er fields than the various model compounds 77 Figure 12. ^ F -N M R spectra o f tri fluoroacety I-in s u lin derivatives; a) g ly c in e ^ ” ! , phenylalanine^” ^, lysine^” ^ -tr iflu o r o a c e ty l- insulin; b) composite from g lycin e A-1 , phenylalanine^” ^ - triflu o ro a c e ty l-in s u lin and g ly c in e ^ ” ^ , ly s in e ^ " " ^ -tri- flu o ro a ce tyI-in su lin ; c) g ly c in e ^ ” ^ -triflu o ro a c e ty l-in s u lin ; d) g ly c in e ^ - ! , phenylalanine^” ^ r lysine^- ^ ^ -triflu o ro a c e ty l- in su lin ; e) g ly c in e ^ ” ! , phenylalanine^” ^ -triflu o ro a c e ty I- desoctapeptide-insulin. P, phenylalanine; G , g lycin e ; L, lysine. DOP, desoctapeptide. 78 5 N A c e t i c A c i d p H 2 .0 pH 6 . 8 pH 6 . 8 G * L p H 8 . 7 pH 6 . 8 - i — i— i— i— i—h — i— i— l— i— I— < ~ •360 -340 -320 -300 -280 -260 Hz 79 TABLE IV GlyA -\° Gb PheB-Sa P 5 n acetic acid 3.1 2.7 pH 2.0 = f c 0.1 3.2 3.7 pH 3.0 = L 0.1 5.6 Insoluble pH 6.8 = b 0.1 12 ± 1 c “ -trifluoroacetyl-insulin. b Trifluoroacetyl peak position, Figure 3 .c No 1 9 F-nmr peak observed. TABLE V Gly^-Phe®-10 GlyA _ 1 -LysB '2 9 a G ly^-Phe^-Lys0'290 GlyA ‘1 -PheB -lc G6 P G L G P L G P 5 n acetic acid 2.9 2.8 2,7 3.3 3.0 3.0 3.1 3.5 3.3 pH 2.0 ± 0.1 3.3 3.2 3.2 3.2 Insoluble pH 6.8 ± 0.1 12 ± 1 12 ± 1 9 ± 1 3.7 4.6 pH 8.7 ± 0.2 4.1 4.8 AAd 4.3 a -trifluoroacetyl-insulin. 6 Trifluoroacetyl peak position, Figure 3. c -desoctapeptide-trifluoroacetyl-insulin. d Width of tri- fluoroacetylglycine + trifluoroacetyllysine peak. (Figure 11). As in the case o f the various model compounds, the transfer of the tr i flu o ro a ce tyl-in su lin derivatives from 5 N acetic acid to w ater, pH 2 .0 , resulted in dow nfield shifts o f 24 to 31 Hz (Figure 11). A composite spectrum derived from the two d i-triflu o ro a c e ty l-in s u lin d e riva tive spectra (Figure 13) is shown in Figure 12B. The spectra o f g ly c in e ^ ” !-triflu o ro a c e ty l-in s u lin and phenyl a la n in e ^- ^ -triflu o ro a c e ty l-in s u lin further corroborated the peak assignments. As shown in Table IV and Table V the peak widths again remained quite narrow. The tri-triflu o ro a c e ty l-in s u lin d e riva tive was insoluble at pH 2 .0 , and thus could not be studied at this pH . When the pH o f g ly c in e ^ - ^ -triflu o ro a c e ty l-in s u lin was adjusted to pH 3 .0 , the ^ F -N M R spectrum indicated a sign ifica nt broaden ing o f the resonance peak (Table IV ), suggesting the form ation o f higher aggregates in solution. A ll other derivatives were insoluble at this pH and thus could not be studied. The ^ F -N M R spectrum o f tri-triflu o ro a c e ty l-in s u lin at pH 6.8 afforded a broadened peak o f 9 .0 + 1 Hz (Figure 12D, Table V ), whose center corres ponded to the ^F -reso n a n ce peak o f g ly c in e ^ - !-triflu o ro a c e ty l-in s u lin at pH 2 .0 . When the m ono-derivative (g ly c in e ^ - ^ -triflu o ro a c e ty l-in s u lin ) was studied at pH 6 . 8, it afforded a broad peak o f 12 + 1 Hz (Figure 12C, Table IV ), as did the two d i-triflu o ro a c e ty l-in s u lin derivatives (Figure 12C, Table V ); however, when phenylalanine^- ^ -triflu o ro a c e ty l-in s u lin was studied at pH 6 . 8, no ^ F -m a g n e tic resonance peaks were observed. G ly c in e ^ ” ^, phenyl- a la n in e ^- !-triflu o ro a c e ty l-d e s o c ta p e p tid e -in s u lin , w hich w ould not be expected 82 Figure 13. ^ F -N M R spectra o f triflu o ro a c e ty l-in s u lin derivatives: (a) g ly c in e ^ ” ^-p h e n yla la n in e ^" ^ tri flu oroa cetyl-in su l in; (b) g ly c in e ^ "^ -ly s in e ^ "^ ^ triflu o ro a c e ty l-in s u lin ; (c) composite tri-triflu o ro a c e ty l-in s u lin derived from a and b. 83 lO O H z. to undergo extensive aggregation based on the behavior o f desoctapeptide-in- sulin (A rq u illa et a l. , 1969), afforded a tw o-peak spectrum (Figure 12F) w ith narrow peak widths at pH 6.8 (Table V ). In a ddition to the observed peak w idths, a dow nfield spectral shift o f 11 Hz was also observed for the triflu o ro a c e ty l- group bound to phenylalanine®""^, w h ile the triflu o ro a ce tyl-g ro u p on g ly c in e ^ - ^ indicated a n e g lig ib le shift (Figure 11). G ly c in e ^ - ^ , p henylalanine® ""^-trifluo roacetyl-insulin and g ly c in e ^""^, phenylalanine®""^, lysine®""^9-triflu o ro a c e ty l-in s u lin were also studied at pH 8 .7 . The spectrum o f the tri-triflu o ro a c e ty l-in s u lin d e riva tive is shown in Figure 12E. This spectrum indicates two narrow resonance peaks (Table V ), whose areas are in the ratio 1:2. Spectral assignments were based on the positions ob served for the g ly c in e ^ ” ^, p henylala n ine ® ""^-triflu o ro ace tyl-in sulin derivative at this pH (Figure 14). In addition to the narrow resonance peaks, dow nfield spectral shifts o f 7 and 10 Hz were observed for the triflu o ro a ce tyl m oiety on phenylalanine®""^ and lysine^- 29^ respectively, between pH 2 .0 and 8 .7 (Figure 11). 19F -N M R stu d ies o f zin c triflu o ro a c e ty l-in s u lin at pH 6 . 8; the effects of perturbants. The zin c triflu o ro a c e ty l-in s u lin derivatives were prepared by adding zin c io n , at a concentration o f 0.3% (or 0.6% where noted) by w eight re la tive to p ro te in , as described in methods. This concentration o f zinc ion gives a ratio o f two zin c atoms for every six insulin molecules and assures the predomin ance o f the hexamer in insulin solution at neutral pH (Fredericq, 1956). The presence o f zin c had no noticeable e ffe c t on the ^9 F-N M R spectra at pH 2 .0 ; 85 Figure 14. ^ F -N M R spectra o f triflu o ro a c e ty l-in s u lin derivatives at pH 8 .7 : (a) g ly c in e A “ l-p h e n y la la n in e ^ - ^ triflu o ro a c e ty l-in s u lin ; (b) tr i- tr i f I u o ro a c e ty l-in s u lin . 86 360 >340 -3X0 >300 -Z80 - £60 H Z the peak widths and positions were identical to those seen in the zin c free studies. The ^ F -N M R spectra for g ly c in e ^ ^ -triflu o ro a c e ty l-in s u lin at pH 6.8 w ith and w ith o u t zin c were also essentially the same: 12 + 1 Hz (Figure 15). As observed e a rlie r for the z in c -fre e triflu o ro a c e ty l-in s u lin derivatives, the ! ^F -N M R spectra for the zin c g ly c in e ^ - !-p h e n y la la n in e ^ - !-triflu o ro a c e ty l- insulin d e riva tive at pH 6 .8 was ide n tica l to that for the zinc g ly c in e ^ ” ! - d e riv a tiv e . N o resonance was observed for the zinc phenylalanine^- ! - t r i flu o ro - a c e ty l-in s u lin de riva tive under these conditions. The ! ^F -N M R spectra o f the z in c g ly c in e ^ - ! - lysine^“ 2 9 -fri f | UOro a ce tyl-in su lin d e riva tive gave a broad (17 Hz) flattened peak that w ould appear to be the sum o f two or more peaks (Figure 19A). Because o f its in s o lu b ility a t low pH , and thus the d iffic u lty o f assuring the com plexing o f zinc io n , the tri-triflu o ro a c e ty l-in s u lin d e rivative was not studied in the presence o f z in c . Sodium citra te and sodium a cetate. The !^ F -N M R resonance o f the t r i fluoroacetyl m oiety on g ly c in e ^ - ^ in the presence o f zinc at pH 6.8 was strongly effected by low concentrations o f sodium citra te or sodium acetate (Figure 16). In Table VI and Table V II the peak widths go from approxim ately 12 + 1 Hz (the peaks could not be distinguished cle a rly above the noise) to peak widths o f 3 -5 Hz upon the addition o f 0.01 M sodium c itra te or sodium acetate. N o further narrowing occurs upon the addition o f eith er salt at up to 0.05 M concentration. The e ffe ct was eq ually pronounced in both the zinc g ly c in e ^ - ! triflu o ro a c e ty l-in s u lin and the zin c g lycin e A - l-p h e n y la la n in e ^ - !-triflu o ro a c e ty l- insulin d e rivatives, but is sp e cific to the triflu o ro a ce tyl group on g ly c in e ^ - ! ; 8 8 Figure 15. ^ F -N M R spectra o f g ly c in e A -l -triflu o ro a c e ty l-in s u lin at pH 6 . 8; (a) z in c -fre e , (b) in the presence o f 0 .3 % z in c . 89 a 90 Figure 16. The effe ct o f sodium acetate on the ^F -resonan ce peak o f z in c - g ly c in e ^ ” ^ -triflu o ro a c e ty l-in s u lin ; (a) pH 2 .0 , no acetate; (b) pH 6 . 8, no acetate; (c) pH 6 . 8, 0.01 M sodium acetate added. 91 r~ c -340 -330 -300 H i -280 92 TABLE VI The Effects o f C itrate Ion on l^ F -N M R Peak W idths (Hz) o f the T rifluoroa ce tyl M oiety o f G ly c in e ^ -^ -triflu o ro a c e ty l-in s u lin and G ly c in e ^ ” ! -p h e n y la la n in e ^ "^ -triflu o ro a c e ty l-in s u lin in the presence o f Z in c Ion pH C itrate (M) Sodium C itrate (M) G I y j ^ a G ly f^ " 1, G Pheg-1a P 2.0 0 0.01 3 3 .3 3 .6 2.0 0.01 0.01 3 .7 ----- ----- 6.8 0 0.01 C 12° ___c 6.8 0.01 0.01 4 .5 3 ___c 6.8 0 0.02 ----- 12C ___c 6.8 0 0 .0 6 ----- 12C ___c 6.8 0.02 0 .06 ----- 3 .5 c a) triflu o ro a c e ty l-in s u lin ; b) triflu o ro a ce tyl peak position; c) peaks were not distinguishable from background noise under experim ental conditions. 93 TABLE V II The Effects o f A cetate Ion on ^ F -N M R Peak W idths (Hz) o f the T rifluoroacetyl M oie ty o f G ly c in e A “ l-triflu o ro a c e ty l-in s u lin and G ly c in e A “ ^-phenyl a Ian ine^-1 -triflu o ro a c e ty l-in s u lin in the presence o f Z in c Ion (0.3% w /w insulin) p H A cetate (M) G ly A - la G b G ly A -1 , G PheB" ia P 2.0 0 3 -4 4 .5 6 6.8 0.01 3 -4 6 _c 6.8 0.05 4 a) triflu o ro a c e ty l-in s u lin ; b) triflu o ro a ce tyl peak position; c) peak not dis tinguishable from background noise under experim ental conditions (peak widths o f 12 Hz or greater). 94 the resonance for the triflu o ro a c e ty l m oiety on phenylalanine®- ^ is not observed. When zin c phenylalanine®- ^ triflu o ro a c e ty l-in s u lin was treated w ith sodium acetate (0.02 and 0.05 M) at pH 6.8 no ^ F -N M R resonance peaks were observed. Low concentrations (0 .0 1 -0 .0 5 M) o f sodium citra te were found to catalyze the hydrolysis o f the triflu o ro a ce tyl group from the insulin derivatives at pH 6 . 8 . Prelim inary results ind ica te that citra te does not have a comparable e ffe ct on the l^ p -N M R spectra o f the z in c -fre e derivatives at pH 6 . 8 . In an e ffo rt to further characterize the effects o f sodium acetate on the 19p-N M R spectra o f the zin c d erivative s, the areas of the peaks o f ind ividu a l preparations were observed at pH 2 .0 , pH 6.8 and then again at pH 2 .0 . From these experiments the area o f the peak at pH 6.8 was found to be 74 + 12% o f the area o f the peak at pH 2 .0 . In order to ascertain w hether the narrowing o f the triflu o ro a ce tyl resonance o f the glycine^""^ substituent is specific for these ions, low concentrations o f sodium chloride and potassium isothiocyanate were substituted for citra te and a ce ta te . In Table V I it is seen that sodium chloride (0.01 and 0 .05 M) had no e ffe ct on the l^ F -N M R sPeci"ra o f zinc g ly c in e ^-^ -p h e n y la la n in e ® - ^ -triflu o ro a c e ty l-in s u lin . Low concentrations o f potassium isothiocyanate (0.01 to 0.05 M ) were also in e ffe ctive in causing this change. However, the a ddition o f sodium citra te (0.05 M) to samples already containing sodium chloride (0.05 M) was com pletely e ffe c tiv e in g ivin g the narrow resonance observed previously. Potassium isothiocyanate. Higher concentrations o f potassium isothio cyanate (0.5 M , 1 .0 M and 2 .0 M) were also found to be in e ffe c tiv e in causing 95 a reduction in the widths o f the resonance peaks o f eith er zin c glycineA -1 or zince g ly c in e A -1 -p h e n yla la n in e ^”" ^ -triflu o ro a c e ty l-in s u lin at pH 6 . 8. A p a ra lle l series o f experiments w ith potassium isothiocyanate (0.5 and 1 .0 M) was also conducted w ith the z in c -fre e g ly c in e ^ ” ^ -triflu o ro a c e ty l-in s u lin deriva tiv e . In these experiments the normal broad peak at pH 6.8 (12 + 1 Hz) was reduced in w idth (7 + 1 Hz) and apparent area w ith 0 .5 M potassium isothio cyanate. A t I M potassium isothiocyanate the peak area and w idth were approx im ately the same as in w ater at pH 6 . 8 . Increasing the concentration to 2 M resulted in a slig h t broadening o f the peak (Figure 17). G uanidine hydrochloride. A prelim inary series o f experiments were con ducted to determ ine the e ffe ct o f guanidine hydrochloride at pH 6.8 on the ^ F -N M R spectra o f the various insulin derivatives. Concentrations o f guanidine hydrochloride below 1 to 1 .5 molal tended to cause the various triflu o ro a c e ty l- ated insulin derivatives to pre cip ita te from solution. Thus the lowest concen trations used in the experiments was 1 .5 m o la l. ^ F -N M R spectra for zinc g ly c in e ^ ""^ -triflu o ro a c e ty l-in s u lin were obtained at guanidine hydrochloride concentrations o f 1 .5 , 3 .0 and 6 .0 m o la l. A t a concentration o f 1.5 molal a narrow but small peak was observed (Figure 18, Table V III). A t 3 molal guani dine hydrochloride, however, on ly a broad, fla t peak was apparent. The final concentration o f 6 molal gave a narrow and d istin ct peak o f greater area than th a t observed at 1.5 molal (Figure 18D), but o f approxim ately the same area observed for this sample at pH 2 .0 before the addition o f guanidine hydrochloride. Z in c g ly c in e A -1 -p h e n yla la n in e ^”^ -triflu o ro a c e ty l-in s u lin was also treated w ith 96 Figure 17. The e ffe ct o f potassium isothiocyanate (KSCN) on the ^ F -N M R resonance peak o f z in c -fre e g ly c in e ^ ” ^ -triflu o ro a c e ty l-in s u lin at pH 6 . 8; a) no KSCN; b) 0 .5 M KSCN; c) 1 .0 M KSCN; d) 2 .0 M KSC N . 97 6 -3 80 -3 C O -340 -320 -300 i -------- » — -280 H z. 98 1 o Figure 18. The e ffe ct o f guanidine hydrochloride on the F-N M R spectra o f triflu o ro a c e ty l-in s u lin derivatives at pH 6 . 8; a -d ) zin c g ly c in e ^ - ^ - triflu o ro a c e ty l-in s u lin ; e) zin c g ly c in e ^ ^-phe nylalanine^ ^ - t r i flu o ro a c e ty l-in s u lin . The concentrations o f guanidine hydro chloride are: a) none, b) 1 .5 m olal, c) 3 m olal, d) 6 m olal, and e) 6 m o la l. The ve rtica l bars represent the peak position for the model compound, N -triflu o ro a c e ty lg ly c in e , at the guanidine concentrations noted. The arrow shows the ^F -reso n a n ce position o f g ly c in e ^ ~ ^ -triflu o ro a c e ty l-in s u Iin at pH 2 .0 in w ater. 99 I T T T T T T - 400 -380 -360 -340 - 3*0 H Z . -300 -280 100 TABLE V II! ^ F -N M R Peak W idths (Hz) o f T riflu o ro a ce tyl-in su lin derivatives as a Function o f G uanidine H ydrochloride C oncentration at pH 6 .8 Deri vative Gu H CIa (molal) Peak W idths G b P G ly c in e A ” ^ 0 12 G ly c in eA”1 C 1 .5 4 .8 G ly c in e A ^ 3 .0 12-15 G ly c in e A _ ^C 6.0 4 G ly c in e A - ^, Phenylalanine^” ^ 6.0 5 5 .5 a) G uH Cl = G uanidine H ydrochloride; b) peak positions; c) triflu o ro a c e ty l- in s u lin . 1 0 1 6 .0 molal guanidine hydrochloride. As seen in Figure 18C tw o narrow peaks were observed for the triflu o ro a ce tyl group at the tw o d iffe re n t positions on insu lin. The resonance peaks for the insulin derivatives in guanidine hydrochloride solution were a ll shifted dow nfield re la tive to the same derivatives in w ater (compare Figures 11 and 17). When the model compound, N -triflu o ro a c e ty l- g ly c in e , was observed w ith ^ F -N M R under id e n tica l conditions, its resonance peak was found to be shifted by the same degree as the triflu o ro a ce tyl moiety on g lycin e ^"*! at 1.5 molal guanidine. This dow nfield sh ift therefore appears to be a general solvent e ffe ct on the chem ical shift o f ^ flu o r in e . A t 6 molal guanidine, however, the triflu o ro a ce tyl group on glycine^"*! is seen to have shifted re la tiv e ly further dow nfield than the corresponding model compound. n _ I The resonance for the triflu o ro a ce tyl group on phenylalanine in 6 molal guanidine has an even greater dow nfield shift than the group on g ly c in e ^ "! . This sh ift is evident in the re la tiv e ly greater separation o f the two peaks in 6 molal guanidine (24 Hz) than is observed under other conditions (16 Hz at pH 2 .0 ). A sim ilar separation (25 Hz) o f the corresponding peaks was observed w ith triflu o ro a c e ty l a ted desoctapeptide-insulin at pH 6.8 (Figure 12). Sodium dodecyl sulfate. The ^ F -N M R spectra o f various zin c triflu o ro a c e ty l-in s u lin derivatives were studied in the presence o f d iffe re nt concentrations o f sodium dodecyl sulfate (SDS) at pH 6 . 8 . Figure 19 shows a comparison o f the spectra at 1% SDS concentration o f the three derivatives: g ly c in e ^ - ^ -; g lycin e -p h e n yla la n in e ^” ^- ; and g ly c in e ^ ” ^ -ly s in e ^ ” ^ -*triflu o ro a c e ty l-in s u lin . A ll 102 Figure 19. Comparison o f the effects o f sodium dodecyl sulfate (1%) on the ^ F -N M R spectra o f triflu o ro a c e ty l-in s u lin derivatives at pH 6 . 8; a) z in c -g ly c in e ^ ""^ -triflu o ro a c e ty l-in s u lin ; b) z in c - g ly c in e ^ ” ^ -p h e n y la la n in e ^ "^ -triflu o ro a c e ty l-in s u lin ; c) z in c - g ly c in e ^ ” ^ -p h e n yla la n in e ^” ^ - t r if I u o ro a c e ty l-in s u lin . 103 1-------------1------------- 1 ' I---------- 1 t --\ T — ~ i W -340 -320 -300 -ZB0 Hz. 104 three derivatives show changes in the w idths and shapes o f the ^ F -N M R peaks re la tive to the corresponding detergent free solutions. In the g ly c in e ^ ” ^ -; and g ly c in e ^ - ^-p h e n yla la n in e ^” ^-d e rivative s (Figure 19A and B) the tr i fluoroacetyl m oiety at the g ly c in e ^ - ! position gives rise to narrow (4 .8 -5 .8 Hz) peaks, w h ile the corresponding group in the g ly c in e ^ "^ -ly s in e ^ ” ^ ^ -d e riv a tiv e (Figure 19C) gives rise to a broader peak (7 Hz) (Table IX ). Two o f these derivatives show pronounced differences in peak w idth for the triflu o ro a c e ty l- m oiety at d iffe re n t positions on the same m olecule. Thus in the g ly c in e ^ ” ^ - p h e n y l a l a n i n e ^ ” ^ -d e riv a tiv e the triflu o ro a ce tyl group affords a re la tiv e ly broad R — 1 resonance peak (8 Hz) when attached to phenylalanine and a narrow resonance peak (4.8 Hz) when attached to g ly c in e ^ ” ^, w h ile in the g ly c in e ^ - ^- l y s i n e ^ ~ 2 9 _ d e riva tive a re la tiv e ly broad resonance peak (7 Hz) is seen for the triflu o ro ace tyl group attached to g ly c in e A ” ! and a narrow resonance peak (5.5 Hz) when it is attached to lysine^” ^ (Figure 1 9 C , Table IX ). In Figure 20 the ^ F -N M R spectra o f zinc g ly c in e ^ -l -lysine^"29_ tr i flu o ro a c e ty l-in s u lin at a number o f d iffe re n t SDS concentrations are shown. In the absence of detergent a single broad peak,centered at approxim ately the same chem ical sh ift seen for the g ly c in e ^ ” ^ -d e riv a tiv e , is seen (Figure 20A ). The addition o f 0.5% SDS causes the peak to become shifted s lig h tly upfield as w e ll as becoming s lig h tly skewed and narrowed (Figure 19B). Increasing the concentration o f SDS to 1% gives rise to a spectrum consisting o f a fa irly sharp . • • D_00 and d istin ct peak corresponding to the triflu o ro a ce tyl group on lysine super posed onto a broader peak for this group on g ly c in e ^ ” ^ (Figure 20C ). SDS 105 Figure 20. The effects o f increasing concentrations o f sodium dodecyl sulfate (SDS) on the ^ F -N M R spectra of zin c g ly c in e ^ - !-ly s in e ^ ” ^ - triflu o ro a c e ty l-in s u lin at pH 6 . 8 . The concentrations o f SDS are: a) 0% , b) 0 .5 % , c) 1% , d) 2% , e) 4% . G , g lycin e ; L, lysine. 106 a -280 34-0 H i TABLE IX I^ F -N M R Peak W idths (Hz) o f Z in c T riflu o ro a c e ty l-in s u lin D erivatives as a Function o f Sodium Dodecyl Sulfate (SDS) C oncentration at pH 6.8 D eriva tive SDS (% w /v ) Peak W idths G a L P • A - l b G ly c in e ^ 1 0 12 • A - lb G ly c in e 1 5 .8 G ly c in e ^ ” ^- ^ Phenylalanine^” ^ 1 4 .8 8 (est.) G ly c in e ^ ” ! - . i • B-29b LysineD 0 0 .5 1 2 4 -1 6 - - 10- 7 (e st.) 5 .5 5 (est.) 4 4 .5 4 a) peak positions; b) triflu o ro a c e ty l-in s u lin . 1 0 8 concentrations of 2% and 4% lead to a slight narrowing o f the lysine peak, w h ile the g ly c in e ^ ” ^ peak sharpened considerably, becoming a separate and d istin ct peak (Figure 20D and E, Table IX ). The chem ical shifts o f the g ly c in e ^ - ^ peaks are approxim ately the same as those observed for this de riva tive at pH 2 .0 in w ate r. C ircu lar dichroism spectra. In an e ffo rt to compare the triflu o ro a c e ty l- 19 insulin derivatives to native in su lin , and to correlate the F-N M R spectra to possible conform ational changes in the p ro te in , c irc u la r dichroism spectra were obtained under various conditions in both the far (190-250 nm) and near (250- 350 nm) u ltra v io le t regions. Far UV spectra. The c irc u la r dichroism spectra o f insulin in the far u ltra v io le t, and in p a rtic u la r, the ra tio o f the rotational difference a t 208 nm to that at 222 nm, has been found to correlate closely to the conform ation o f the pro tein (Blundell et a l. , 1972, pp 345-346). Figure 21 presents a comparison o f spectra for several sine free triflu o ro a c e ty l-in s u lin derivatives at pH 6 . 8, a ll at a concentration o f 1 mg p ro te in /m l. It is apparent that the overall shapes o f the three spectra are quite sim ila r. The curve shown for g ly c in e ^ - ^ -p h e n y l- B— 1 alanine "" -triflu o ro a c e ty l-in s u lin is quite ty p ic a l, w ith only minor variations, o f the g ly c in e ^ ” ^ - and phenylalanine^ derivatives at pH 2 .0 and the B “ I A 1 phenylalanine -d e riv a tiv e a t pH 6 .8 . The g ly c in e A -1 - and the tri-triflu o ro - a c e ty I-d e riv a tiv e d iffe r m ainly in having a more pronounced minima near 208 nm. The close relationship o f these spectra are apparent from th e ir 208/222 nm ratios (Table X ). The curve shown for tri-triflu o ro a c e ty l-in s u lin is the most 109 Figure 21. FAR UV c irc u la r dichroism spectra o f z in c -fre e triflu o ro a ce tyla te d insulin derivatives at pH 6 . 8; (•) g ly c in e ^"” ^-phenylalanine^"” ^ - triflu o ro a c e ty l insulin; (® ) tri-triflu o ro a c e ty l-in s u lin ; ( A ) g ly c in e ^ "^ - tr i flu o ro a c e ty l-i nsulin . 110 TABLE X C ircu la r Dichroism Ratios ( A2Qg/ ^222^ ^°r ^ * nc“ ^ree Insulin D erivatives (1 mg p ro te in /m l) D eriva tive pH 2 .0 pH 6.8 Z in c Insulin 1.08 G ly c in e^ 1- 0 1.18 1.29 Z in c G ly c in e ^ ^ - a 1.16 P henylalanine^- ^ - a 1.19 1.17 G ly c in e ^ - ^ -p h e n yla la n in e ^” ^ - a 1 .21 1.23 T ri-a 1.42 a) triflu o ro a c e ty I-in s u lin 1 1 2 divergent o f the spectra. The 208/222 nm ratio o f this d e riva tive also diverged most s ig n ific a n tly from the other derivatives studied, as w ell as from z in c - insulin its e lf (Table X ). The far UV c ircu la r dichroism spectra at pH 6.8 for g ly c in e ^ - ^ -triflu o ro - a c e ty l-in s u lin in the presence and absence o f zinc and for zin c insulin are com pared in Figure 22. It is apparent that the derivatives in the presence and absence o f zinc are closely related to each other as w e ll as to zinc insulin. The small differences in the minima near 208 nm and the shoulder near 220 nm observed w ith the addition o f zin c to the d e riva tive are quite typ ica l o f the differences seen w ith zinc and z in c -fre e insulin (Goldman and C arpenter, 1974). The 208/222 nm ratio for zin c g ly c in e ^ ""^ -triflu o ro a c e ty l-in s u lin is also very close to the ratios for the various z in c -fre e derivatives (Table X ). The far UV c irc u la r dichroism spectra o f zinc g ly c in e ^ ” ^ -triflu o ro a c e ty l-in s u lin was also taken at pH 6.8 in the presence o f 0.05 M sodium acetate -0 .0 1 M sodium ch lo rid e . This spectra was found to be essentially id e n tica l to the spectra for the same d e riva tive taken in 0 .06 M sodium chloride at pH 6 .8 . In Figure 23 the far UV circu la r dichroism spectra for zin c g ly c in e ^ ” ^ - B-29 lysine -triflu o ro a c e ty l-in s u lin at 0% , 1% and 4% SDS concentrations are shown. Both concentrations o f SDS have sim ilar effects, causing a loss o f the shoulder at around 220 nm, w h ile the minimum at around 206 nm is more pro nounced and shifted s lig h tly to shorter w avelengths. In a ddition the peak at around 192 nm is substantially reduced by the presence o f SDS. The 208/222 nm ratios for these spectra are given in Table X I. A 113 Figure 22. The effects o f zinc on the c irc u la r dichroism spectra o f g ly c in e ^ ^ triflu o ro a c e ty l-in s u l in at pH 6 . 8; ( • ) no z in c; ( ■ ) 0 .3 % zin c present; ( a ) zinc in su lin . 114 5X i o -4 A A 2 2 0 WAVEL&NGiH ('**"') ZOO Figure 23. The effects o f increasing concentrations o f sodium dodecyl sulfate on the far UV c irc u la r dichroism spectra o f zin c g ly c in e ^ ” ^ - lysine ^“ ^ -triflu o ro a c e ty l-in s u lin at pH 6 . 8 . ( • ) no SDS present; ( A ) 1% SDS in solution; ( ■ ) 4% SDS in solution. 116 200 ' 220 ' 2« 4o WAVE.LENAH TABLE XI C ircu la r Dichroism Ratios ( ^ Q g / ^222^ ^°r ^ ' nc T riflu o ro a c e ty l- Insulin D erivatives in SDS Solution at pH 6.8 (1 mg p ro te in /m l) D eriva tive 0% SDS Cone. 1% 4% Z in c Insulin 1.08 G ly c in e ^ - ^ - a 1 .16b 1.84 G ly c in e ^ ^-p h e n yla la n in e ^” ^ - a 1.80 G ly c in e ^ - ^ -ly s in e ^ “ ^ - a 1.28 1.90 1.80 a) triflu o ro a c e ty l-in s u l in , b) in 0 .0 6 M N a C I. 118 comparison o f c irc u la r dichroism spectra for zinc g ly c in e ^ - ! - , zinc g ly c in e ^ - ! - p h e n yla la n ine^- ^ - and zin c g ly c in e ^""^ -ly s in e ^ ^ -triflu o ro a c e ty l-in s u lin in 1% SDS is shown in Figure 24. The curves for the zin c g ly c in e ^ - ^ - and the zin c g ly c in e ^ - !-p h e n y la la n in e ^ - ^-d e rivative s are nearly id e n tica l w h ile the curve for the zinc g ly c in e ^ - ^ -ly s in e ^ ""^ -d e riv a tiv e shows some differences in fine structure between 202 and 206 nm. The overall sim ila ritie s o f these curves are reflected in th e ir close 208/222 nm ratios (Table X I). N ear UV spectra. The c irc u la r dichroism spectra in the near UV (250-350 nm) were also studied. These spectra give inform ation about the local confor mation around the tyrpsine and phenylalanine residues o f in su lin . A specific advantage o f the near UV spectra is that they were taken at the same protein concentration as the ^ F -N M R spectra (10 m g/m l) w h ile the far UV spectra could not be observed at these concentrations, due to instrumental lim ita tio n s. In Figure 25 the near UV spectra o f the zinc free g ly c in e ^ - !-d e riv a tiv e at pH 2 .0 and 6.8 are compared to zinc insulin a t pH 6.8 as a reference. Again PH has little e ffe ct on the spectra. The near UV c irc u la r dichroism spectra for the various z in c -fre e derivatives were nearly id e n tica l except for tri-triflu o ro - a c e ty l-in s u lin . Representative spectra (differing only in fine structure near 270 nm) as w e ll as the spectra for tri-d e riv a tiv e are shown in Figure 26. The a dditio n o f zinc to g ly c in e ^ - ^ -triflu o ro a c e ty l-in s u lin causes the minimum near 275 nm to become more pronounced as w ell as a slig h t decrease in the intensity o f the maxima near 250 nm (Figure 27). N ote that the minima for the zin c g ly c in e ^ - ^-d e riv a tiv e closely approximates the minima for zinc 119 Figure 24. A comparison o f the far UV c irc u la r dichroism spectra o f triflu o ro acetylated insulin derivatives in sodium dodecyl sulfate solutions (1%) at pH 6 . 8; (H ) g ly c in e ^ ” ^ -triflu o ro a c e ty l-in s u lin ; (# ) g ly c in e ^ ” ! - , phenylalanine^” ^ -triflu o ro a c e ty l-in s u lin ; (A ) g ly c in e ^ ” ! l y s i n e ^ ” ^ ^ -triflu o ro a c e ty l-in s u lin . 1 2 0 VV *-01 X S 220 W A V E .LE N 6TH <m ^ ZOO 1 2 1 Figure 25. Near UV circular dichroism spectra of g lycin e ^- ^-triflu o ro- acefyl-insulin; ( # ) pH 2.0; ( • ) pH 6 .8 . 1 2 2 250 Z1S 300 3fcS 350 WAVE LE N & T H (wwi) 123 Figure 26. N ear UV c irc u la r dichroism spectra for triflu o ro a c e ty l-in s u l in derivatives at pH 6 . 8; ( A ) g ly c in e ^ ” ^ -triflu o ro a c e ty l-in s u lin ; ( • ) g lycin e ^""^-p h e n yla la n in e ^ ^ -triflu o ro a c e ty l-in s u lin ; ( ■ ) tri-triflu o ro a c e ty l-in s u lin . 124 o X i n 3 Z S 350 300 W A V E LeN G fTH U rn ) Figure 27. The effect of zinc ion on the near UV circular dichroism spectra of glycine^” !-trifluoroacetyl-insulin at pH 6.8; ( • ) zinc free; ( A ) 0 .3% zin c present; ( ■ ) zin c in su lin . 126 250 *7 5 300 325 3 5 0 WAV E L E N S TH (v\m) 12 7 in su lin . Again the differences in the minima for the zinc and z in c -fre e deriva tives are typ ica l o f differences in the 275 nm minima observed for zin c and z in c - free insulin (see, for instance, G oldm an and C arpenter, 1974). As described e a rlie r, the addition o f sodium acetate to a solution o f zinc g ly c in e ^ ” !-triflu o ro a c e ty l-in s u lin results in a substantial change in the ^ F - NMR peak w id th . However, the same concentration (0.05 M ) o f sodium acetate did not result in any sig n ifica n t differences in the CD spectrum o f this d e riva tive (Figure 28). The presence o f e ith e r 1% or 4% SDS nearly abolished the near UV m ini mum observed for zin c g ly c in e A ” l-|y s in e ^ ""^ -triflu o ro a c e ty l-in s u lin (Figure 29). Very little difference is observed between 1% and 4% SDS w ith this d e riv a tiv e . A sim ilar e ffe ct occurred w ith the glycineA-1 -and zin c g ly c in e ^ ” ^ - D _ 1 phenylalanine -d e rivatives in the presence o f 1% SDS (Figure 30). The loss o f the near UV minimum observed under these conditions is closely correlated w ith loss o f the CD shoulder at 220 nm. It is also interesting to note that w hile 1% SDS has essentially the same e ffe ct as 4% SDS on the CD spectra o f glycin e ^ ” ^-ly s in e ^ ^ - d e r iv a tiv e , the ^ F -N M R spectra indicate considerable d iffe r ences between 1% and 4% SDS, at least re la tive to the A chain amino te rm in a l. Sedimentation v e lo c ity studies. In an e ffo rt to correlate the observed ^ F -N M R spectral peak w idth changes w ith alterations in the aggregation state o f the triflu o ro a c e ty l-in s u l in d erivatives, S 20 w VQlues were obtained for several derivatives under a variety o f conditions. Sedimentation values (s2Q w ) 128 Figure 28. The e ffe ct o f acetate ion on the near UV c irc u la r dichroism spectra o f zinc g ly c in e ^ ” ^ -triflu o ro a c e ty l-in s u lin ; ( • ) no acetate; (A) 0.05 M ace ta te. 129 — I— 250 275 300 3Z5 350 WAV/5V.5N GitH 130 Figure 29. The e ffe c t o f increasing concentrations o f sodium dodecyl sulfate (SDS) on the near UV c irc u la r dichroism spectra o f g ly c in e ^ ” ^ - lysine^“'^ -triflu o ro a c e ty l-in s u lin . Jhe concentrations o f SDS are: ( # ) 0 % , ( * ) ! % , ( « ) 4 % . 131 srx 10-* A & 3Z0 350 300 WAVELENGTH (av*0 £ 5 0 Figure 30. A comparison o f the near UV c irc u la r dichroism spectra o f triflu o ro - a c e ty l-in s u lin derivatives in sodium dodecyl sulfate solution (1% ); (A) g ly c in e ^ ” ^ -triflu o ro a c e ty l-in s u lin ; ( # ) g ly c in e ^ - !-p h e n y l- a la n in e ^” ^ -triflu o ro a c e ty l-in s u lin ; ( ■ ) g ly c in e ^ - ^ - ly s in e ^ " ^ - tri fl uoroacety I - i nsul in . 133 V V +-01 X s 3 0 0 WAVEV-E P^GiTVi 320 250 134 for several z in c -fre e derivatives at pH 2 .0 + .1 and 6.8 + .1 are listed in Table X II. A t pH 2 .0 , sedim entation values indicated the presence o f insulin dim ers, w h ile at 6 .8 , the S 20 w vc,l ues indicated considerable aggregation. H ow ever, a t pH 6.8 desoctapeptide-insulin d e riva tive s till had a sedim entation value sim ila r to the insulin derivatives at pH 2.0. The s ig n ific a n tly low er s20,w vc,l ue f ° r the tri-triflu o ro a c e ty l-in s u lin d e riva tive is also reflected in 19 a narrower F-resonance peak w id th . Sedimentation v e lo c ity studies ind ica te that the z in c -fre e g ly c in e ^ - ^ - d e riva tive at 0.01 M sodium chloride and the zin c g ly c in e ^ - ^ -d e riv ia tiv e in the presence of 0 .06 M sodium chloride or 0 .05 M sodium aceta te -0.01 M sodium chloride share very sim ila r average aggregation states (Tables X II and X III). In a d d itio n , the S 2q w value observed for these derivatives is very close to the S 2q w value observed for zin c insulin at pH 7 .0 in the presence o f 0 .0 5 M sodium a cetate-0.01 M sodium c h lo rid e . Z in c insulin and in su lin -H C l were found not to be soluble at 10 m g/m l at pH 6 . 8, 0 .0 5 M sodium a cetate-0.01 M sodium ch lo rid e . Sedimentation v e lo c ity studies were also performed w ith g ly c in e ^ ” ^ -d e riv a tiv e in 1 .0 M KSCN in the presence and absence of zinc ion (0.3% re la tive to p ro te in ). The S 2q w values observed under those con di tions were 3.6s and 3.3s re spectively. The results indicates that the d e riva tive exists as a large aggregate, perhaps as the hexamer, in both the presence and absence o f zinc ion under these conditions. F in a lly , zinc g ly c in e ^ - ^ -p h e n yla la n in e ^- ^ - and zinc g lycin e — tri — flu o ro a c e ty l-in s u lin were compared by sedim entation v e lo c ity in the presence 135 TABLE XII Svedberg Constants (S 2 0 w) Insulin and Trifluoroacetyl-insul in D erivatives as a Function o f pH Compound pH 2.0 + .1 pH 6.8 + .1 Insulin 1 .4 3 .5 G lyA_1 -T F A -ln su lin 1.5 3 .5 PheB -1-TFA-lnsuM n 1.8 3 .6 G ly A - 1 , PheB -1-T F A -ln su lin 1 .6 3 .2 G ly A _ 1 , PheB_1, LysB-29-T F A -ln su lin - 2.8 G ly A _ 1 , PheB_1-T F A -D O P a-ln s u lin - 1 .4 a) DO P, Desoctapeptide 136 TABLE XIII Svedberg Constants (S 2 0 w) of Zinc Insulin and Zinc G lycine^""^- trifluoroacetyl-Insulin: the Effects of Sodium Acetate at Neutral pH Deri va tive pH A cetate (M) s Z in c Insulin 7 .0 0 .05 3 .4 Z in c G ly c in e A -1 - a 6.8 0.00 3 .5 6.8 0 .05 3 .5 a) triflu o ro a c e ty l-in s u lin . 137 o f 1% sodium dodecyl su lfate. Because neither the stoichiom etry o f SDS-protein binding (and thus the density) nor the hydrodynamic properties o f the SDS-protein com plex are known, it is impossible to correlate the S 2q values obtained w ith p a rtic u la r aggregation state. However, a comparison o f the S 2Q values o f the two derivatives at the same detergent concentration should give v a lid inform ation as to th e ir re la tive aggregation states. The S 2q values (uncorrected for deter gent concentration) for the derivatives are 1.5s and 1. 8s respectively. These values ind icate a sim ilar aggregation state for the two derivatives under these conditions. 138 Discussion Using the technique o f ^ F -N M R spectroscopy, these studies have been concerned w ith the effects o f pH and various perturbants on the conform ation and com plex aggregation behavior o f triflu o ro a c e ty l-in s u lin deriva tive s. As discussed e a rlie r (Part I) these derivatives possess high b io lo g ic a l and immuno lo g ica l a c tiv ity . The results o f c irc u la r dichroism and sedim entation v e lo c ity studies also indicate th e ir close relationship to native in s u lin . Several model compounds have also been studied in reference to the effects o f solvent, pH and various perturbants on ^F -reso n a n ce chem ical shifts as w ell as on peak w idths. The synthesis and isolation o f several triflu o ro a c e ty l-in s u lin derivatives has afforded the p o ssibility o f studying the environm ent at three d istin ct locations in the m olecule. Model triflu o ro a ce tyl deriva tives. A ll o f the triflu o ro a c e ty l-a m in o acids and peptide derivatives experienced a dow nfield shift o f the ^F -re so n a n ce peaks when transferred from 5 N a ce tic acid to w ater (pH 2 .0 ). Such deshielding effects have been observed when triflu o ro a ce tylp he n ylala n in e was transferred from nonpolar solvents to w ater (Zeffren and R ea vill, 1968; Z e ffre n , 1970). The ^F -reso n a n ce peaks were essentially insensitive to alterations in pH, suggesting that the presence or absence of a negative charge on the carboxyl group had little influence on the environm ent o f the N -triflu o ro a c e ty l m oiety. 139 These results are again consistent w ith the w ork o f Z effren and Reavil! (1968). A lterations in the environm ent o f the N -triflu o ro a c e ty l group on the amino acids and peptides were observed, however, when the «<-carboxyI group was m odified 19 by a d dition o f various numbers o f amino acids as shown in Figure 11 . The F- resonance peaks from these peptides, as in the case of the amino acid derivatives, were also insensitive to alterations in pH . For a ll the model compounds, the lin e widths remained quite narrow (2-3 H z), suggesting little re striction o f m otion. Concentrations o f zero to one percent o f the detergent, sodium dodecyl sulfate was found to have no e ffe c t on the chem ical shifts and peak widths o f the triflu o ro a ce tyla te d amino acids. The dow nfield sh ift o f the ^F -reso n a n ce peak o f the triflu o ro a ce tyl group on g lycin e in solutions o f guanidine hydrochloride is discussed under The effects o f guanidine hydrochloride. The other model compounds were not observed in guanidine solutions. Z in c -fre e triflu o ro a c e ty l-in s u lin derivatives: the effects of pH and a ce tic a c id . The ^ F -n u c le a r magnetic resonance spectra o f triflu o ro a c e ty l-in s u lin derivatives suggested several im portant facts concerning the effects o f pH on aggregation and the environments o f the introduced ^ F probes. The chemical 19 shifts o f the F-resonance peaks from the triflu o ro a ce tyl m oiety on g lycin e in the various insulin derivatives at pH 2 .0 , 6 . 8, and 8 .7 were essentially id e n tic a l; suggesting sim ilar environments at this position in the m olecule under these conditions. The chem ical sh ift is very close to that o f g ly c in e ^ ” ^ - t r i fl u o ro a ce tyl-S -su lfo -in su l in A cha in , in d ica ting a high degree o f exposure to the aqueous environm ent. The ^ F chem ical shifts o f the triflu o ro a c e ty l moiety 140 bound to phenylalanine and lysine in the insulin d e riva tive w ere, however, sensitive to changes in pH from 2 .0 to 8 .7 , resulting in dow nfield shifts. This deshielding e ffe ct could be caused by alterations in the magnetic environm ent around the probe resulting from a pH -induced conform ation change (M orris e t a l . , 1968), or titra tio n o f a specific group, independent o f or in conjunction w ith a state of insulin aggregation at pH 8 .7 . Based on the results o f Z effren and Reavill (1968), these dow nfield shifts could be caused by increased exposure to the aqueous environm ent o r, as mentioned above, exposure to a charged species in a re la tiv e ly nonpolar environ m ent. The chem ical shifts at pH 6.8 for the triflu o ro a c e ty l group bound to lysine and phenylalanine could not be determined because of the extreme w idth of the peaks. The triflu o ro a c e ty l m oiety on phenylalanine^""^ in the triflu o ro - a ce tyl-d e so cta p e ptid e -insu lin d e riv a tiv e , however, was observable, ind ica ting a sh ift to low er fie ld , suggesting an a lte ra tio n in its environm ent. Confor m ational alterations in desoctapeptide-insulin have been suggested by M ercola et a l. (1967) using circ u la r dichroism and by Carpenter (1966) using optical rotatory dispersion. The 19F -N M R spectra o f triflu o ro a c e ty I-in s u lin derivatives also exhibited considerable variations in peak widths as a function o f pH . An increased w idth can be interpreted in terms o f a restriction o f m olecular motion o f a specific group. A ll o f the triflu o ro a c e ty l-in s u lin derivatives had narrow resonance peaks at pH 2 .0 . G ly c in e ^ -l-triflu o ro a c e ty l-in s u lin e xhib ite d a sig n ifica n t increase in peak w idth at pH 3 .0 , suggesting that this group was now held more rig id ly as a result o f a conform ational change, or that aggregation was 141 beginning to restrict the motion o f this group. As the pH was raised to 6 . 8, further broadening was observed to 12 + 1 Hz. When phenylalanine^- ^ - t r i flu o roa ce tyl -in s u lin was studied at pH 6 . 8, no observable resonance peak was obtained, suggesting that this group was held even more rig id ly than the triflu o ro a c e ty l m oiety on g ly c in e . As shown in Table V , the two d i- tr ifl uoro- ace tyl derivatives also afforded a broad peak o f 12 + 1 H z, suggesting that the signal originated from the triflu o ro a c e ty l m oiety on g ly c in e , and th at no signal B— 1 was obtained from phenylalanine . The shape and intensity o f a peak from g l y c i n e ^ - l -ly s in e ^ “ ^ ^ -triflu o ro a c e ty l-in s u lin also suggested that the triflu o ro ace tyl group o f lysine®- ^ js also h ig h ly restricted and thus affords an extrem ely broad peak that cannot be observed above background. Because the ^ F - A — 1 • B— 29 resonance peaks from the triflu o ro a c e ty l groups on g lycin e and lysine occur closer together than the peak from triflu o ro a ce tylp h e n yla la n in e , we cannot exclude the po ssib ility that there may be some slig h t contribution to this broad peak at pH 6.8 from the triflu o ro a c e ty l lysine group. S ufficient lysine ^- ^ - triflu o ro a c e ty l-in s u lin was not a va ila b le to run a spectrum to absolutely estab lish the co n tribution by the triflu o ro a c e ty l lysine group. The tri - triflu o ro a c e ty l- insu lin d e riva tive also afforded a broad peak, although somewhat reduced in w idth (Table V , Figure 13), suggesting that the aggregation o f the tri-d e riv a tiv e was somewhat reduced under these conditions. Recent studies have suggested that m odifications o f the three amino groups o f insulin could a lte r the aggrega tio n behavior o f the m olecule (Massey and Smyth, 1972; A rq u illa e t a l. , 1969; Goldm an and Carpenter, 1974). 142 Several studies were performed to establish that the peak broadening re fle cte d the aggregation state o f in su lin . As mentioned above, desoctapeptide- insulin has been shown not to undergo extensive aggregation at pH 7 .4 -7 .5 (A rq u illa et a l. , 1969). The spectrum o f the g ly c in e ^ ” ^-phenylalanine® ” ^ - triflu o ro a c e ty l d e riva tive o f this compound at pH 6.8 supported this proposal, affording a w e ll-re so lve d tw o-peak spectrum (Table V , Figure 13). Several studies have suggested that insulin disaggregates at pH values above 8 (Fredericq, 1956; Slobin and C arpenter, 1966). A cco rd in g ly, g ly c in e ^ ” ^-phenylalanine® ” and g ly c in e ^ ” ^ -p h e n yla la n in e ^- ! -ly s in e ^ “ ^ ^ -triflu o ro a c e ty I-in s u lin were studied at pH 8 .7 . As seen in Table V and Figures 13 and 14, the peak widths have narrowed considerably, suggesting a disaggregation o f the triflu o ro a c e ty l- insulin d e rivative s. In an e ffo rt to further establish the relationship between aggregation states and ^ F -N M R peak w idths, sedim entation v e lo c ity studies were carried out on insulin and several o f the triflu o ro a c e ty l-in s u lin d eriva tive s. A t pH 2 . 0, g ly c in e ^ - ! -triflu o ro a c e ty l-in s u lin , phenylalanine® ” ^ -triflu o ro a c e ty l- in su lin , and g ly c in e7 ^ - !-p h e nyla la n in e® ” ^ -triflu o ro a c e ty l-in s u lin had S 20rw values quite close to z in c -fre e in su lin . Under sim ila r conditions insulin has been shown to e x h ib it a w eight-average m olecular w eigh t o f 12,000 daltons (Jeffrey and Coats, 1966a,b). The S20r w VQl ues obtained are s lig h tly low er than values reported in the lite ra tu re (Fredericq and N e u ra th , 1950; Cunningham et a l. , 1955; G u tfreun d, 1952), probably resulting from the low salt concen tra tio n (0.01 M N aC I) w hich was used to order du plica te the condition o f the 143 ^ F -N M R experim ents. A t pH 6 . 8, the broad ^ F -N M R peaks for the triflu o ro - acetyl derivatives are pa rallele d by a substantial increase in the S 2q w values, as shown in Table X II. The s lig h tly narrower peak w idth for the tri-triflu o ro - a c e ty l-in s u lin d e riva tive is again reflected in the S 20 w value w hich suggests a low er m olecular w eight than that o f native insulin or the other mono- and d i- triflu o ro a ce tyl derivative s. F in a lly , the suggestion that the triflu o ro a ce tyl d e riva tive o f desoctapeptide-insulin does not undergo extensive aggregation at pH 6 .8 is corroborated by the S 20 w value o f 1 .4 . The c irc u la r dichroism spectra for these derivatives at pH 2 .0 and 6.8 are consistent w ith aggregation phenomena causing the major differences in the ^ F -N M R spectra. The far UV c irc u la r dichroism spectra o f the phenyl ala n ine ^""^- and the g lycin e7 ^ - !-p h e n y la la n in e ^ ” ^ - derivatives are essentially unchanged between pH 2 .0 and 6 . 8. The spectra for the g ly c in e ^ ” ! - de riva tiv e a t pH 2 .0 is also sim ilar to the spectra o f these two derivatives, although the pH 6.8 spectra shows some in te nsifica tio n o f the 208 nm band. The strong 222 nm negative extreme in a ll o f these derivatives indicates that they a ll exist predom inantly as dimers and higher aggregates o f dimers under these conditions. This is because the 222 nm band is due largely to the presence o f -structure (Q u a d rifo g lio and U rry, 1968), and the ^ -structure in insulin is due to mono- mer-monomer interactions in the dimer (Blundell et a l. , 1972). S ensitivity o f the 222 nm band to dim er form ation has been reported by Goldm an and Carpenter (1974) in comparing the aggregation behavior o f insulin and m odified insulins w hich do not undergo normal aggregation. It m ight also be postulated 144 that the attenuated 222 nm band o f the tri-d e riv a tiv e indicates that the low er apparent m olecular w eight at pH 6.8 is due at least p a rtly to weakened monomer- monomer interactions in this d e riv a tiv e . The 208 nm band in insulin has been a ttrib u te d to o c -h e lic a l structure in insulin (Ettinger and Tim asheff, 1971). C rysta lline insulin is known to have c < -h e lic a l regions (Blundell et a l. , 1972) w hich probably contribute to this band. The s im ila rity o f this band in the various derivatives thus im plies a sim ila r amount o f o c -h e lix . This s im ila rity is expected since m odified deriva tives o f very low b io lo g ica l a c tiv ity , such as desoctapeptide in s u lin , have been found to retain this band w ith little change (Goldman and Carpenter, 1974). The c irc u la r dichroism spectra o f the triflu o ro a c e ty l-in s u lin derivatives also in d ica te that they m aintain conform ations sim ilar to that o f zinc in su lin . The ra tio o f the minima at 208 nm to that o f 222 nm has been shown to correlate w ith the b io lo g ic a l a c tiv ity , and by im p lic a tio n , the conform ation o f insulin (Blundell et a l . , 1972, pp 345-346). The ratios observed for the triflu o ro ace tyl derivatives are a ll w ith in the range o f ratios seen for a ctive insulin d e riva tive ( 40% o f the a c tiv ity o f native in s u lin ). If the tri-d e riv a tiv e is e xclu d e d , the rem aining derivatives are seen to be very closely related by this c rite ria . The somewhat more intense band at 208 nm for the z in c -fre e triflu o ro - a c e ty la te d -in s u lin compared to zin c insulin is not unexpected since the presence o f zin c causes an attenuation o f this band in native insulin (Goldman and C arpenter, 1974). 145 Studies of Trifluoroacetyl-insulin Derivatives in the Presence of Zinc. Comparison w ith z in c -fre e results. Z in c g ly c in e ^ ” ! triflu o ro a c e ty l-in s u lin was found to have a broad peak at pH 6.8 sim ilar to that o f the z in c -fre e de riva tiv e (Figure 15). As in the studies w ith the z in c -fre e derivatives the g ly c in e ^ ” ^- p hen yla la n ine^” ^ - d e riva tive was indistinguishable from the g ly c in e ^ - l-d e riv a - tiv e on the basis o f th e ir ^ F -N M R spectra. Phenylalanine^ ^ -triflu o ro a c e ty l- insulin gave no observable peak in the presence o f z in c . The ^ F -N M R data thus ind ica te that these three triflu o ro a c e ty l a ted derivatives m aintain sim ilar conform ations and aggregation states in the presence and absence o f z in c . The conclusion above is supported by the sedim entation v e lo c ity data (Tables X II and X III) w hich show g ly c in e ^ ” ! - t r i fl uoroacety I-in s u lin to have id e n tica l S 2Q w values (3.5s) in both the presence and the absence o f zinc ion at pH 6 . 8 . On the other hand, c irc u la r dichroism spectra for this d e riva tive in d ica te that some changes in conform ation take place upon zinc binding. The 208 nm minima is more intense in the presence o f zin c than in its absence. S im ilar observations are reported for unm odified insulin and have been explained by the binding o f zin c to histid ine B -IO , a p rin c ip le component o f the major ©<-h e lix o f insulin (Goldman and Carpenter, 1974). Z in c g ly c in e ^ - ^ - tr i flu o ro a c e ty l-in s u lin also exhibits a more intense 273 nm c irc u la r dichroism band. A sim ilar e ffe ct is again seen in zin c insulin and may be due to a tig h te r aggregation state. However, the overall c irc u la r dichroism spectra o f this d e riva tive in the presence and absence o f zin c compare very closely, again in d ica tin g the closely related conformations under these tw o conditions. 146 In contrast to the three derivatives already discussed, zin c g ly c in e ^ " ^ , lysine®“ ^ -triflu o ro a c e ty l-in s u lin e xh ib ite d a d iffe re n t ^ F -N M R spectra than the z in c -fre e d e riv a tiv e . The z in c d e riva tive gave a broader (17 Hz) flattened peak re la tive to the zin c free d e riv a tiv e . This peak appears as i f it m ight be the sum o f two broad peaks, although it is approxim ately centered on the chem ical shift observed for the g ly c in e ^ d e riv a tiv e . Effects o f low concentrations o f salts. ^ F -N M R spectra were obtained for the g ly c in e A - ! t h e phenylalanine® - ^ -; and the g ly c in e ^ - !-p h e n yla la n in e ® - ^ -triflu o ro a c e ty l-in s u lin derivatives in solutions containing low concentrations o f a va rie ty o f salts. Low concentrations (10-50 mM) o f sodium c itra te or sodium acetate resulted in a d istin ct narrowing o f the resonance peak o f the t r i fluoroacetyl substituent on g ly c in e ^ -l . In contrast, the resonance peak for the substituent on p h enylalanine^” ^ in these derivatives was not observed under these conditions. This w ould ind ica te that the aggregate size o f the protein was not reduced. However, the narrow peak for the substituent on g ly c in e ^ ” ^ indicates that this group has gained a considerable degree o f m otional freedom. In an e ffo rt to further characterize the effects o f these salts, zin c glycin e A “ 1 -triflu o ro a c e ty l-in s u lin was studied by sedim entation v e lo c ity and near UV c irc u la r dichroism . These studies were conducted w ith solutions containing e ith e r 0 .05 M sodium acetate or 0 .0 5 M sodium ch lo rid e . N e ith e r o f these techniques showed any e ffe ct o f sodium acetate on the insulin d e riv a tiv e . A series o f experiments was conducted comparing the to ta l area o f the ^ F - N M R peaks o f zin c g ly c in e ^ - !-triflu o ro a c e ty l-in s u lin in 0.01 M sodium 147 a ce ta te . In these experiments the w idths o f the peak obtained at pH 2 .0 and pH 6.8 were found to be com parable. However, the area o f the peak observed at pH 6.8 was only 74 + 12% o f the peak observed at pH 2 .0 . This difference in peak areas indicates that not as many triflu o ro a ce tyl-g ro u p s are contributing to the sharp peak at pH 6 . 8, the non-co n trib utin g groups probably e xh ib itin g broad and thus non-distinguished resonance contributions. Thus it appears that the triflu o ro a c e ty l groups on g ly c in e ^ ” ^ may exist in two (or more) d iffe re n t environments in the insulin aggregate. In this connection Blundell et a l. (1972) have found that the A - l g lycin e residues in the dimers o f the zinc insulin hexamer occupy two somewhat d iffe re n t environments in the observed X -ra y crystal structure. It is thus possible that sodium acetate is able to distinguish these two sites, p re fe re n tia lly causing a loosening o f one o f these A - l g lycin e residues. N e ith e r sodium citra te nor sodium acetate are usually considered to have strong effects on the conformations or aggregation states o f proteins. Indeed these salts gene rally have no specific observed effects on proteins in low con centrations (von Hippel and S chleich, 1969). Thus, to determ ine w hether the observed effects on the ^ F -N M R spectra were general salt effects, id e n tica l concentrations o f sodium chloride (10 mM - 50 mM) were u tiliz e d in sim ilar experim ents. The sodium chloride did not result in any narrowing o f the g lycin e ^ ""^ -triflu o ro a c e ty l group peak. The narrow peak was observed when sodium citra te (50 mM) was added to the sodium chloride (50 mM) solu tio n. F in a lly , z in c g ly c in e ^ " ! - ; and zin c g ly c in e ^1 ""^ -p h e n y la la n in e ^ "^ -triflu o ro a c e ty l- 148 Insulin were treated w ith low concentrations o f potassium thiocyanate (10 m M - 50 m M ). Potassium thiocyanate is known to be a very powerful chaotropic salt and it is often used to p a rtia lly disrupt protein structures at re la tiv e ly low ( ^ 1 M) concentrations (von H ippel and S chleich, 1969; Sawyer and Puckridge, 1973). The mode o f action o f this salt is apparently to destablize hydrophobic inte ractio n s, largely by a lte rin g the structure of the w ater in w hich the protein is dissolved. This salt did not cause a narrowing o f the ^ F -N M R resonance peaks at these concentrations. The results o f these low concentration salt studies ind ica te that citra te and acetate ions in te ra ct in a sp ecific way w ith the triflu o ro a c e ty l group or adjacent amino acid residues at the N -te rm in a l end o f the insulin A -c h a in . This inte ractio n results in a loosening o f the structure a llo w in g the triflu o ro acetyl group considerable freedom o f motion re la tive to the overall aggregate. The overall conform ation and aggregation states o f the insulin derivatives appear to be uneffected by this in te ra c tio n . These salts may also be able to discrim inate between the two A -c h a in term inal regions o f the insulin dim er. The e ffe ct o f potassium th io cya n a te . H igher concentrations o f potassium thiocyanate (0 .1 -2 M) were used in an attem pt to perturb the structure of zinc g ly c in e ^ ” ! a n d zin c g ly c in e ^ -^ -p h e n y la la n in e ^ '^ -triflu o ro a c e ty l-in s u lin . No narrowing o f the ^ F -N M R peaks was observed. This indicates that the size o f the aggregates was not reduced. Sedimentation v e lo c ity studies confirmed this interpretation since the S 20 w VQl ues obtained in 1 M potassium thiocyanate (3.6s) and 0 .0 5 M sodium chloride (3.5s) were nearly id e n tic a l. 149 The I^ F -N M R peak for z in c -fre e g ly c in e ^ - l-triflu o ro a c e ty l-in s u lin in 0 .5 M potassium thiocyanate was somewhat narrower and sm aller than the peak in w ate r. H ow ever, the ^ ^p-resoncnce peak in 1 M potassium thiocyanate was nearly id e n tica l to the peak in w ater at pH 6 . 8. This indicates that at I M the d e riva tive probably exists as a large aggregate, a proposal supported by the S 20 w vc|l ue in 1 M solution (3 .3 s ). The narrow peak in 0 .5 M potassium thiocyanate is probably due to a p artial opening o f the insulin m olecule, w hich then tightens up again at higher concentration. A sim ilar phenomena appears to occur in the zin c d e riva tive in guanidine hydrochloride solutions. A comparison o f these results for insulin (both in the presence and absence o f zin c) w ith some other proteins show insulin to be resistant to the effects o f th io cya n a te . Thus, for instance, Sawyer and Puckridge (1973) showed that thiocyanate (0 .5 M) caused ^ -la c to g lo b u lin to disaggregate from a tetram er to a dim er. In both cases the dissociations occurred w ith o u t detectable changes in the secondary or te rtia ry structures o f the proteins. Indeed, in the case o f £ -la c to g lo b in , concentrations o f 3 M thiocyanate or greater were required to n o ticea b ly perturb this protein's conform ation. The increased s ta b ility o f the insulin aggregate re la tive to these other proteins may be a re fle ctio n o f the re la tiv e ly high protein concentration used in the insulin experim ents. The dissociation o f proteins by chaotropic salts such as thiocyanate is expected to be sensitive to protein concentration (Sawyer and Puckridge, 1973). The tig h t binding o f zin c provides additional s ta b ility in the cases o f zin c insulin d erivatives. 150 The effects o f guanidine hyd roch loride. It was o f interest to investigate the effects o f denaturants on the properties o f the triflu o ro a c e ty l-in s u lin de riva tive s. G uanidine hydrochloride was chosen as one o f the most pow erful, w id e ly used and best understood o f protein denaturants (Tanford, 1968). W hile high concentrations o f guanidine hydrochloride have often been used to com plete ly denature proteins to random c o ils, it has recently become more interesting to use low concentrations o f this agent to gently perturb protein structure. Thus Roberts and Benz (1963) have combined the NMR technique and low concen trations o f guanidine hydrochloride to study the in itia l steps in ribonuclease u n fo ld in g . In the present study triflu o ro a c e ty la te d insulin derivatives were observed w ith ^ F -N M R spectroscopy a t a number o f guanidine hydrochloride concen trations in an attem pt to fo llo w th e ir d enaturation. G uanidine hydrochloride concentrations of 1 molal or less were found to p re cip ita te the insulin deriva tives (10 m g/m l) from solutions at pH 6 . 8 . The lowest concentration used was therefore 1 .5 m ola l. A t this concentration the g ly c in e ^ ” ^ -triflu o ro a c e ty l- insulin d e riva tive gave rise to a narrow ^ F -N M R peak o f s lig h tly greater than one h a lf the area o f the peak w hich occurred at 6 molal gu an idin e. This result is p a rtic u la rly interesting when compared to the sim ilar e ffe ct observed w ith low concentrations o f sodium acetate discussed previously. In both cases some, but not a ll, o f the g ly c in e ^ ” ! - t r if l uoroacetyl groups appear to achieve a considerable degree o f rotational freedom. This rotational freedom is probably not shared by the remainder o f the p ro te in , suggesting that the aggregate is 151 larg e ly in ta c t. The suggestion that the aggregate is not disrupted at 1 .5 molal guanidine is supported by the fa ct that the narrow peak at this concentration is replaced by a broad, fla t peak at 3 molal guanidine h yd rochloride. Such a broad peak is characteristic o f re la tiv e ly high m olecular weights and thus an aggregated insulin d e riv a tiv e . W hy the g ly c in e ^ -l-triflu o ro a c e ty l group should lose rotational freedom w ith increasing denaturant concentration is not clear. The opposite e ffe c t should be expected since denaturation is generally thought o f as a loosening o f the structure o f the prote in. How ever, the effects o f sodium acetate and sodium c itra te , as w e ll as o f 1 .5 molal guanidine hydrochloride seem to ind icate that some o f the g ly c in e A ” l-triflu o ro a c e ty l groups o f the zinc insulin d e riva tive in the aggregate are held in a p a rtic u la rly unstable and easily loosened w ay. It may be that p a rtia lly denaturing the protein results in these p a rticu la r groups being more tig h tly held due to an overall change in the protein conform ation. A t 6 molal guanidine hydrochloride a sharp peak is apparent in the ^ F - N M R spectra o f the g ly c in e ^ ” ^ d e riv a tiv e , sim ilar to the peak observed a t pH 2 .0 in w ate r. The g ly c in e ^ ” !-p h e nyla lan in e® ” ^ -triflu o ro a c e ty l-i nsul in d e riva tive was only studied in 6 molal guanidine hydrochloride. Under these conditions two re la tiv e ly sharp peaks were observed. These results indicate that these two derivatives are largely dissociated in 6 molal guanidine hydro- ch lo ri de. In ad ditio n to changes in peak w idth and area, changes in chem ical shifts were observed for the ^ F -N M R peaks in guanidine hydrochloride 152 solutions. The shifts o f the peaks were compared to the chem ical shifts o f the model compound, triflu o ro a c e ty l-g ly c in e . This was necessary because o f the se n sitivity o f l^F -ch e m ica l shifts to solvent com position. The difference in shifts for the model compound and the insulin d e riva tive in 1 .5 molal guanidine was approxim ately 15 Hz. This is the same as the difference seen in w ate r, im plying that the triflu o ro a ce tyl groups on these tw o compounds bear the same relationships to each o ther, enviro nm entally, under these two d iffe re n t co ndi tions. In contrast, the peaks for the model compound and g ly c in e ^ - ^ -triflu o ro a c e ty l-in s u lin in 6 molal guanidine hydrochloride e x h ib it a difference of 19 Hz in chem ical shifts. Comparison to model studies indicates that the triflu o ro acetyl group on the insulin d e riva tive is in a more polar environm ent re la tive to the model compound than is the case in w ater or 1 .5 molal guanidine. The shifts for the ^F -reso n a n ce peaks for g ly c in e ^ "!-p h e n y la la n in e ^ triflu o ro a c e ty I-in s u lin in 6 molal guanidine hydrochloride were also measured re la tiv e to triflu o ro a c e ty l-g ly c in e . The resonance peaks are both shifted down fie ld re la tive to the d e riva tive at pH 2 .0 in w ater, however, the peak for the phen yla lanine^” ^ group exhibits a re la tiv e ly greater sh ift than the g ly c in e ^ - ^ group. That is at pH 2 .0 in w ater these two peaks are separated by approx im ate ly 16 Hz w h ile in 6 molal guanidine hydrochloride they are separated by 24 H z. A sim ilar difference in shifts (26 Hz) for these two peaks is observed fo r the triflu o ro a ce tyla te d desoctopeptide d e riva tive at pH 6.8 in w ater. These data in d ica te that the phenylalanine ^- ^ m oiety is in both cases exposed to a more polar environm ent, probably going from a fa irly non-polar region in the 153 protein to a region o f greater exposure to the solvent. In this regard it is in te r esting to note that desoctapeptide insulin does not aggregate at neutral pH , apparently existing as monomers (A rq u illa et a l. , 1969; Goldm an and C arpenter, 1974). However, desoctapeptide insulin does m aintain a d e fin ite secondary and te tia ry structure, as evidenced by c ric u la r dichroism bands near 192 nm and 208 nm (Goldman and C arpenter, 1974). The effects o f sodium dodecyl sulfa te . As a further ch aracterization o f the solution properties o f insulin the effects o f the detergent sodium dodecyl sulfate (SDS) were studied. Detergents have been used both as s o lu b iliz in g agents and as denaturing agents in the study o f proteins. Detergents behave quite d iffe r e n tly than agents such as urea, guanidine hydrochloride or heat in denaturing proteins. Whereas the la tte r agents act quite generally to open up a proteins structure, resulting u ltim a te ly in random co ils; detergents act more s p e c ific a lly by binding to p a rticu la r regions o f a p ro te in , often causing the protein to re fold in a non-native but s till organized way (Tanford, 1968). Indeed the pro tein may e x h ib it d iffe re n t kinds o f fo lding at d iffe re n t detergent concentrations. These result from d iffe re n t numbers o f detergent molecules binding to the protein at d iffe re n t concentrations, depending on the detergent binding constants for the various regions o f the pro te in . In the present studies o f the effects o f SDS on triflu o ro a c e ty l-in s u lin derivatives two series o f experiments were conducted: 1) a single d e riva tive (g ly c in e ^ "^ -ly s in e ^ "2 9 -tri f j uoro a ce tyI-in su lin ) was compared by ^ F -N M R and c irc u la r dichroism spectra at various concentrations o f SDS (0 -4 % ), and 154 2) three d iffe re n t derivatives (g ly c in e ^ - ! g ly c in e ^ - ^-p h e n yla la n in e ^- ^- , and g ly c in e ^ - ^ -ly s in e ^ - ^ ^ -triflu o ro a c e ty l-in s u lin ) were compared by these same techniques at a single concentration o f SDS (1% ). In the first series o f experiments the detergent had a d iffe re n tia l e ffe ct in sharpening the two ^ F -N M R peaks. Thus the triflu o ro a c e ty l m oiety bound to lysine^“ 29 gave a sharp peak in 1% SDS, w h ile a 4% SDS concentration was required to give an equally sharp peak for the m oiety on g ly c in e ^ - ^ . Even at this fin a l concentration the peak for the triflu o ro a c e ty l m oiety on g ly c in e A-1 is not equal in height nor in area to the peak o f the m oiety on lysine^” ^ . The difference in widths for these tw o peaks a t low SDS concentrations indicate that the triflu o ro a ce tyl group on lysine^- ^ has a greater degree o f freedom than the group on g ly c in e ^ - ! . This in turn im plies a conform ational change o f the protein m olecule a llo w in g the region around lysin e^- 29 some motional freedom re la tive to the protein as a w hole. This observation is o f p a rticu la r interest since in other studies the g ly c in e ^ ^ m oiety g enerally exhibits the narrowest ^ F -N M R peaks and thus the greatest freedom o f movement. The g ly c in e ^ - ^ -ly s in e ^ “ ^ ^ -triflu o ro a c e ty l-in s u lin d e riva tive was also studied by c irc u la r dichroism spectroscopy at SDS concentrations o f 1% and 4% . A t both o f these concentrations the shoulder near 222 nm is lost, in d ica tin g a loss o f the $ -structure o f the insulin dim er. The loss o f (3 -structure in turn im plies a dissociation o f the dimers to monomers. The cotton e ffe ct at 276 nm is also g re atly reduced,an observation also associated w ith dim er dissociation (Morris et a l. , 1967). The c irc u la r dichroism spectra under these conditions are very sim ilar to spectra for desoctapeptide insulin and des-alanine, des- asparagine insulin reported in the lite ra tu re (Brugman and A rq u illa , 1973; Goldm an and C arpenter, 1974). Desoctapeptide insulin does not aggregate, w h ile in d e s-alanine, des-asparagine insulin aggregation is markedly attenuated (A rq u illa et a l. , 1969; Goldman and Carpenter, 1974). These derivatives also show very low b io lo g ica l a c tiv itie s , im plying conform ational changes in the m olecule (Bromer and Chance, 1967; Carpenter, 1966; A rq u illa et a l. , 1969). In the second series o f experiments a 1% concentration of SDS was found to have d iffe re n tia l effects on the !^ F -N M R spectra o f the triflu o ro a ce tyl group o f g ly c in e ^ ” ! on d iffe re nt insulin deriva tives. Thus the ^ F -N M R peak o f the g ly c in e ^ "!-triflu o ro a c e ty l group is re la tiv e ly sharp for the g ly c in e ^ ” ! - , and g ly c in e ^ ” !-p h e n y la la n in e ^ ” ! - , derivatives but is sig n ific a n tly broader on the g ly c in e ^ ” !-ly s in e ^ ” ^ - d e riv a tiv e . It is o f p a rtic u la r interest to compare the spectra o f zin c g ly c in e ^ ” ^ -ly s in e ^ “ ^ ^ -triflu o ro a c e ty l-in s u lin and zin c g ly c in e ^ ” ^ -p h e n y la la n in e ^ "!-triflu o ro a c e ty l-in s u lin . Both derivatives show broad, though somewhat d iffe re n t, ! ^F-resonance peaks in w ater at pH 6 . 8, as discussed previously. In addition each derivatives has one sharp and one broad peak in its ! ^F -N M R spectra in 1% SDS solutions. However, for g lycin e ^ ” !-ly s in e ^ “ ^ -triflu o ro a c e ty I-in s u lin the peak for g ly c in e ^ " ! is broad and the peak for lysine®” ^ is narrow , w h ile for g ly c in e ^ ” !-p h e n y la la n in e ^ "!- triflu o ro a c e ty l-in s u lin the peak for g ly c in e ^ ” ! is narrow and the peak for ph en yla la n ine^” ! is broad. Thus the data ind ica te that the g lycin e group has a re la tiv e ly greater degree of freedom in the second d e riva tive than in the The presence o f broad peaks in the spectra of both derivatives ind ica te th a t both are aggregated to some e xte n t. The sim ilar S20 values (uncorrected for SDS concentration) o f these derivatives also ind ica te that they are o f a sim ilar m olecular w e ig h t. The c irc u la r dichroism spectra o f these three derivatives are a ll very sim ilar to one another in 1% SDS. As discussed previously for the g ly c in e ^ - ^ -ly s in e ^ ” ^ - d e r iv a tiv e the shoulder near 222 nm is lost and the cotton e ffe ct at 276 nm is g reatly reduced. In summary, the broad peaks in the ^ F -N M R spectra o f the two d i derivatives ind ica te that these derivatives are aggregated to some e xte n t. As a comparison, the ^ p - N M R spectra o f triflu o ro a ce tyla te d ribonuclease (MW approxim ately 12,000) has a peak o f sim ilar w id th (approxim ately 8 Hz) for the triflu o ro a c e ty l group bound to a lysine residue in the proteins in te rio r (Hestis and Raftery, 1971). This residue probably shares the overall motion o f the p ro te in , since i t is on the in te rio r. Thus, the re la tiv e ly broad peaks o f the d i-d e riva tive s ind ica te that these derivatives are aggregated to at least the dim er (MW approxim ately 12,000). C onversely, the c irc u la r dichroism spectra ind icate the loss o f the normal monomer-monomer interactions involved in the insulin dim er. Taken together these results may mean: 1) that these derivatives in SDS solutions are aggregated in an abnormal w ay, not in vo lvin g the normal monomer-monomer interactions o f the insulin dim er; or 2) an SDS-protein complex is formed w hich is large enough and rig id enough for the monomers to give broad NMR peaks. The unexpected broadening o f the ^ F -N M R peak o f the g lycin e m oiety 157 in the g ly c in e ^ ” ! -p h e n yla la n in e ^” ^ -triflu o ro a c e ty l-in s u lin d e riva tive may be due to a close p ro xim ity o f g l y c i n e ^ - ! and l y s i n e ^ - 2 9 . These two residues are quite close in the crystal structure o f insulin (Blundell et a l. , 1 9 7 2 ) as w ell as in the trifluoro sce in -iso th iocya n a te d e riva tive in solution (M ercola et a l. , 1 9 7 2 ) . The difference in the NMR spectra could be due to the loss o f the local charge on lysine^- ^ in the triflu o ro a ce tyla te d d e riv a tiv e . A lte rn a tive ly, a close p ro xim ity o f the two triflu o ro a ce tyla te d residues may a lte r the binding characteristics o f SDS to the insulin m olecule. 158 Conclusion This study has been concerned w ith e lu cid a tin g the solution properties o f insulin u tiliz in g ^ F -N M R studies. T rifluoroacetylated insulin derivatives and 19p-N M R were u tiliz e d in order to obtain simple and re adily interpretable NMR spectra, as opposed to the com plex spectra obtained from proton NMR spectro scopy o f native insulin. The triflu o ro a c e ty la te d insulin derivatives were com pared to native insulin by a va rie ty o f techniques, including b io lo g ic a l assay, c irc u la r dichroism spectroscopy and sedim entation v e lo c ity studies. These comparisons showed the derivatives to be very close to native insulin in behavior and conform ation. The I^ F -N M R data yie ld e d a w ealth o f inform ation concerning the environ ments and re la tive m ob ilities o f the triflu o ro a c e ty l probes on the three amino groups o f insulin: g ly c in e ^ -l f phenylalanine^"" ^ and lysine^” ^ . Environmental 1 9 inform ation was obtained by comparing the chem ical shifts o f the F-N M R peaks for the insulin derivatives w ith the chem ical shifts for model compounds placed in d iffe re n t chem ical environm ents. Relative m ob ilities o f the triflu o ro a ce tyl probes at the three positions were derived from the measured ^ F -N M R , peak w idths. The ^ F -N M R spectra thus re fle c t the environments and m obilities o f the amino term inal regions o f the A and B chains and the lysine residue near the carboxyl term inal end o f the B chain. The conform ational and aggregational 159 dynamics o f these three regions o f the insulin m olecule in solution were in te r preted from the ^ F -N M R data in conjunction w ith c irc u la r dichroism and sedim entation v e lo c ity studies. The resultant picture o f insulin shows a ll three regions to be exposed to the solvent and quite m obile at pH 2 .0 , im plying that they are a ll loosely held on the surface o f the insulin dim er, the predominate aggregate species under these conditions. Upon raising the pH to 6 . 8, both zinc and z in c -fre e insulins aggregate to hexamers. The ^ F -N M R data show that the N -te rm in a l region o f the A chain is s till on the e xte rio r o f the aggregate, and retains a high degree o f m o b ility . The N -te rm in a l region o f the B-chain is more restricted and is tig h tly held in the aggregate. A comparison to the X -ra y picture o f Blundell et a l.(1 9 7 2 ) suggests that this region o f the m olecule is tig h tly held in a hydrophobic pocket between dimers in the zinc insulin hexamer. Raising the pH o f z in c -fre e insulin to pH 8 .7 causes a pa rtial disaggregation o f the hexamer. AM three regions studied are again mobile and appear to reside on the e xte rio r o f the aggregate. The environm ental changes observed for the N -te rm in a l region o f the B chain and the lysine region, re la tive to insulin at pH 2 .0 , may be due to the titra tio n o f nearby carboxyl groups in the insulin m o le cu le . A more de ta iled picture was obtained by observing the interaction o f zinc insulin derivatives at pH 6.8 w ith various perturbants o f protein conform ation a n d /o r aggregation. Low concentrations o f acetate or citra te ion appeared to in te ra ct very s p e c ific a lly w ith the A chain N -te rm in a l region o f in s u lin , leading 160 to a great deal o f m otional freedom independent o f the hexamer. There is no change in environm ent; the N -te rm in a l region o f the A chain remains in the solvent. It is interesting that the entire population o f A chain N -te rm in a l regions is not affected by these perturbants. Rather, the e ffe ct is p a rtia l, suggesting that there are perhaps two s lig h tly diffe re nt environments for this region o f the insulin monomer w ith in the zinc insulin hexamer. Low concen trations o f guanidine hydrochloride showed the same e ffe c t as acetate and c itra te ion. Studies in vo lvin g sodium dodecyl sulfate in d ica te a close re la tio n ship between the N -te rm in a l region o f the A chain and the lysine near the C - term inal o f the B chain. These results may re fle c t a close positioning o f these two regions in the insulin m olecule. These regions are in fa ct quite close to one another in crysta llin e zinc insulin (Blundell et a l. , 1972). The ^ F -N M R picture o f in su lin , w hich gives inform ation about the dynamic interactions o f the m olecule in solutio n, thus shows the N -term inal region o f the A chain to lie on the surface of the hexamer, probably in close n o g proxim ity to lysine , w h ile the N -te rm in a l region o f the B chain is tig h tly held on the in te rio r o f the hexamer. These results are in complete agreement w ith the results from X -ra y crystallography, w hich give the structure o f the insulin hexamer in great d etail under static conditions. 161 References Adams, M . J . , B lundell, T .L ., Dodson, E .J ., Dodson, G . G . , V ija y a n , M ., Baker, E .N ., H arding, M .M . , H odgkin, D .C ., Rimmer, B ., and Sheat, S. (1969) N ature (London) 224, 491. A fric a , B ., and Carpenter, F .H . (1970) Biochemistry 9 , 1962. A nfinsen, C .B ., Sela, M ., and T ritch , H. (1956) A rch . Biochem. Biophys. 65, 156. A rq u illa , E .R ., Bromer, W .W ., and M ercola, D . (1969) Diabetes 18, 193. B lu n d e ll, T ., Dodson, G . , H odgkin, D ., and M e rcola , D. (1972), in Advances in Protein Chem istry, I (Anfinsen, C .B ., Edsall, J .T . , and Richards, F .M ., eds.) p. 279, Academ ic Press, New Y o rk. Brandenberg, D . (1969) Hoppe-Seyler's Z . Physiol. Chem. 350, 741. Brandenberg, D ., G a ttn er, H .G . , and W ollm er, A . (1972) Z . Physiol. Chem. 353, 599. Bromer, W ., Sheehan, S ., Berns, A . , and A rq u illa , E.R. (1967) Biochemistry 6, 2578. Bromer, W ., and Chance, R.E. (1967) Biochim . Biophys. Acta 133, 219. Brugman, T .M ., and A rq u illa , E.R. (1973) Biochemistry 12, 727. B u n zli, H .F ., and Bosshard, H .R . (1971) Hoppe-Seyler's Z . Physiol. Chem. C arpenter, F .H . (1958) A rch . Biochem. Biophys. 78, 539. Carpenter, F .H . (1966) Am . J . M ed. 40, 740. C e c il, R ., and W ake, R .G . (1962) Biochem. J . 82, 401. Cheng, S ., and M a rtin e z -C a rrio n , M . (1972) J. B iol. Chem. 247, 6597. C la rk, J .M . (1964), Experimental Biochemistry , W .H . Freeman and C o ., San Francisco. C o v e lli, R .W ., and W o lff, J . (1967) J . B io l. Chem. 242, 881. Cunningham, L .W ., Fischer, R .L ., and V e stlin g , C .S . (1955) J. Am. Chem. Soc. 77, 5703. D ixon, G . H . , and W ardlaw , A .C . (1960) N ature (London) 188, 721. Ellm an, G .L . (1959) A rch . Biochem. Biophys. 82, 70. E ttinger, M . J . , and Timasheff, S .N . (1971) Biochemistry 10, 824. Evans, R .L ., and Saroff, H .A . (1957) J . B io l. Chem. 228, 296 Fanger, M .W ., and Harbury, H .A . (1965) Biochemistry 4 , 2541. Fredericq, E. (1956) A rch . Biochem. Biophys. 65, 218. Fredericq, E ., and N eurath, H. (1950) J . Am. Chem. Soc. 72, 2684. G a ttn e r, H .G . (1971) H oppe-Seyler's Z . Physiol. Chem. 352, 7. G a ttn e r, H .G . (1972), quoted in B lundell, T ., Dodson, G . , H odgkin, D ., and M e rco la , D ., Advances in Protein Chemistry (Anfinsen, C. B ., Edsall, J .T ., and Richards, F .M ., eds.) V o l. 1, p. 347, Academ ic Press, N ew Y o rk. G e ig e r, R. (1971) H oppe-Seyler's Z . Physiol. Chem. 352, 7. G oldberger, R .F ., and A nfinsen, C .B . (1962) Biochemistry 1, 401. 163 G oldm an, J . , and C arpenter, F .H . (1974) Biochemistry 13, 4566. G utfreund, H. (1952) Biochem. J . 50, 564. Hales, C . N . , and Randale, P .J. (1963) Biochem. J . 88, 137. Huestis, W .H ., and Raftery, M .A . (1971) Biochemistry 10, 1181. Huestis, W .H ., and Raftery, M .A . (1972a) Biochemistry 11, 1684. Huestis, W .H ., and Raftery, M .A . (1972b) Proc. N a t. A cad. Sci. U .S . 69, 1887. Huestis, W .H ., and Raftery, M .A . (1972c) Biochem. Biophys. Res. Commun. 49, 1358. H o rw itz, J . , and H e lle r, J. (1973) J . B io l. Chem. 248, 1051. Jardetsky, O . , and W ade-Jardetsky, N .G . (1971) A nn. Rev. Biochem. 40, 605. J e ffre y, P .D ., and C o a te s,J.H . (1966a) Biochemistry 5 , 489. J e ffre y , P .D ., and Coates, J .H . (1966b) Biochemistry 5 , 3820. K atz, J . J . , S train, H .H ., Lenssing, D .L ., and D oughtery, R.C. (1968) J . Am . Chem. Soc. 90, 784. K ra il, G . , Brandenburg, D ., Zahn, H ., and G eig er, R. (1971) H oppe-Seyler's Z . Physiol. Chem. 352, 1995. Levy, D ., and Carpenter, F .H . (1970) Biochemistry 9, 3215. L i, C .H . (1956) N ature (London) 178, 1403. Lindsay, D .G . , and S hall, S. (1969) Biochem. J . 115, 587. Lindsay, D .G , and S hall, S. (1970) Eur. J. Biochem. 15, 547. Lindsay, D . G . , and S hall, S. (1971) Biochem. J . 121, 737. Lord, R .S ., Gubensek, F ., and Rupley, J .A . (1973) Biochemistry 12 4385.164 M aloney, P .J ., A p rile , M .A . , and W ilson, S. (1964) J. N ew Drugs 4, 259. M arcker, K. (1960a) A cta Chem. Scand. 14, 194. M arcker, K. (1960b) A cta Chem. Scand. 14, 2071. M a rkle y, J . L . , Putter, I . , and Jardetsky, O . (1968) Science 161, 1249. Markus, G . (1964) J . B io l. Chem. 239, 4163. M assaglia, A . , Pennisi, F ., Rosa, U ., Ronce-Testoni, S ., and Rossi, C .A . (1968) Biochem. J . 108, 247. Massey, D .E ., and Smyth, D .G . (1972) Eur. J . Biochem. 31, 470. Meadows, D .H ., M a rkle y, J . L . , Cohen, J .S ., and Jardetsky, O . (1967) Proc. N a t. A cad. S ci. U .S . 58, 1307. Meadows, D .H ., Jardetsky, O . , Epand, R .M ., Ruterjans, H .H ., and Scheraga, H. (1968) Proc. N a t. A ca d . S ci. U .S . 60, 766. M e rco la , D .A ., M orris, J .W .S ., A rq u illa , E .R ., and Bromer, W .W . (1967) Biochim . Biophys. Acta 133, 224. M e rco la , D .A ., M orris, J .W .S ., and A rq u illa , E.R. (1972) Biochemistry 11, 3860. M ille t, F ., and Raftery, M .A . (1972) Biochemistry 11, 1629. M orris, J .W .S ., M ercola, D .A ., and A rq u illa , E.R. (1968) Biochim . Biophys. A cta 160, 145. N a ka ya , K ., H orinishi, H ., and Shibata, K. (1967) J . Biochem. (Tokyo) 61, 337. Paselk, R .A ., and Levy, D . (1974a) Biochemistry 13, 3340. Paselk, R .A ., and Levy, D. (1974b) Biochim . Biophys. Acta 359, 215. 165 Pekar, A . H . , and Frank, B .H . (1972) Biochemistry 11, 4013. Porter, R .R ., and Sanger, F. (1948) Biochem. J . 42, 287. Q u a d rifo g lio , F ., and U rry, P .W . (1968) J . Am. Chem. Soc. 90, 2760. Roberts, G . C . K . , and Benz, F .W . (1973) A nn. N .Y . A cad. Sci. 222, 130. R obillard, K .A ., and W ishnia, A . (1972) Biochemistry 11, 3841. Rupley, J . A . , Renthal, R .D ., and Praissman, M . (1967) Biochim . Biophys. A cta 140, 185. Ryle, A . P ., Sanger, F ., Smith, L .F ., and K ita i, R. (1955) Biochem. J . 60, 541. Saunders, D .J ., and O ffo rd , R.E. (1972) FEBS (Fed. Eur. Biochem. Soc.) L e tt. 26, 286. Sawyer, W .H ., and Puckridge, J . (1973) J. B io l. Chem. 248, 8429. Schallenberg, E .E ., and C a lv in , J. (1955) J. Am . Chem. Soc. 77 , 2779. S c h lic h tk ru ll, J . (1956a) Acta Chem. Scand. 10, 1455. S c h lic h tk ru lI, J . (1956b) Acta Chem. Scand. 10, 1459. Slobin, L . I . , and Carpenter, F .H . (1966) Biochemistry 5 , 499. Spackman, D .H ., Stein, W .H ., and M oore, S. (1958) A n a l. Chem. 30, 1190. Stouffer, J .E ., and W atters, J .A . (1965) Biochim . Biophys. Acta 104, 214. Suzuki, T ., Takenaka, O . , and Shibata, K. (1969) J . Biochem. (Tokyo) 66, 815. Tanford, C. (1968), in Advances in Protein Chemistry 23 (Anfinsen, C .B ., Anson, M .L ., Edsall, J .T ., and Richards, F .M ., eds.) p p 122-283, Academ ic Press, New Y o rk. 166 Thomas, J .H . (1969) Biochem. J . 115, 55. V in e , W .H ., Brueckner, D .A ., Needlem an, P ., and M arshal!, G .R . (1973) Biochemistry 12, 1630. Von H ip p e l, P .H ., and S chleich, T. (1969), in Structure and S ta b ility o f B iological Macrom olecules (Timasheff, S .N ., and Fasman, G . D . , eds.) p . 417, M arcel D ekker, In c ., New Y o rk. W e il, L ., Seibles, T.S., and Herskvits, T.T. (1965) A rch . Biochem. Biophys. I l l , 308. W e in e rt, M ., K ircher, K ., Brandenberg, D ., and Zahn, H. (1971) Hoppe- Seyler's Z . Physiol. Chem. 352, 719. W eygand, F ., and Ropsch, A . (1959) Chem. Ber. 92, 2095. Young, J .D ., and C arpenter, F.H. (1961) J. B io l. Chem. 236, 743. Zahn, H ., and Drechsel, E. (1968) H oppe-Seyler’s Z . Physiol. Chem. 349, 385. Z e ffre n , E. (1970) A rc h . Biochem. Biophys. 137, 291. Z e ffre n , E ., and R ea vill, R.E. (1968) Biochem. Biophys. Res. Commun. 32, 73. 167 UMI Number: DP21591 All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. UMI Dissertation Publishing UMI DP21591 Published by ProQuest LLC (2014). Copyright in the Dissertation held by the Author. Microform Edition © ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code ProQuest ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106- 1346
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
A new approach to the problem of measuring the properties of micelles
PDF
Computer modeling and free energy calculatons of biological processes
PDF
A study of the effect of glycogen on the oxidation of butyrate by rat liver slices
PDF
Chemical, physicochemical, and biological studies on the mucoproteins of plasma and serum
PDF
Connections of the dorsomedial hypothalamic nucleus in the rat
PDF
Adrenalectomy and fat absorption
PDF
A novel reaction of mismatched cytosine-cytosine pairs associated with Fragile X
PDF
The isolation and characterization of rat plasma albumins
PDF
Computer simulation of polar solvents with the SCAAS model
PDF
A seasonal study on the distribution of polycystine radiolarians from waters overlying the Catalina basin, southern California
PDF
Calculations of electrostatic interactions in proteins
PDF
A study of the ketone body metabolism of excised rat tissues with respect to labile phosphate content
PDF
A study of methods of cholesterol analysis
PDF
Disulfide exchange derivatives of human hair keratins
PDF
A study of protein separation using ion exchange resins
PDF
Steric effects in the acid, amine, and glyceraldehyde-3-phosphate dehydrogenase catalyzed hydrolysis of acyl phosphates
PDF
Building a model of organization acculturation: an interpretive study of organizational culture and stories
PDF
Chiroptical spectroscopy of nitrogenase
PDF
Colloid properties of some New Mexico clays
PDF
A critical analysis of ideology and discourse in two hotels
Asset Metadata
Creator
Paselk, Richard Alan
(author)
Core Title
¹⁹F-NMR studies of trifluoroacetyl insulin derivatives
School
Graduate School
Degree
Doctor of Philosophy
Degree Program
Biochemistry
Degree Conferral Date
1976-01
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
chemistry, analytical,OAI-PMH Harvest
Language
English
Contributor
Digitized by ProQuest
(provenance)
Advisor
[illegible] (
committee chair
), [illegible] (
committee member
)
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c17-2575
Unique identifier
UC11347972
Identifier
DP21591.pdf (filename),usctheses-c17-2575 (legacy record id)
Legacy Identifier
DP21591.pdf
Dmrecord
2575
Document Type
Dissertation
Rights
Paselk, Richard Alan
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA
Tags
chemistry, analytical