`
`FEBS Letters 528 (2002) 257^260
`
`Increased stability of human growth hormone with
`reduced lactogenic potency
`Alexey A. Schulgaa, Alexander A. Makarovb; , Iliya V. Levichkina, Yuliya V. Belousovac,
`Vladimir M. Lobachovb, Irina I. Protasevichb, C. Nick Paced, Mikhail P. Kirpichnikova
`aShemyakin-Ovchinnikov Institute of Bioorganic Chemistry, Russian Academy of Sciences, Miklukho-Maklaya St. 17/10, Moscow 117997, Russia
`bEngelhardt Institute of Molecular Biology, Russian Academy of Sciences, Vavilov St. 32, Moscow 119991, Russia
`cResearch Center of Molecular Diagnostics and Therapy, Simpheropolsky Blvd. 8, Moscow 113149, Russia
`dDepartment of Medical Biochemistry and Genetics, Texas ApM University, College Station, TX 77843, USA
`
`Received 10 July 2002; revised 22 August 2002; accepted 25 August 2002
`
`First published online 5 September 2002
`
`Edited by Gunnar von Heijne
`
`Abstract Human growth hormone (hGH), whose main func-
`tion is the somatic growth stimulation, induces diverse e¡ects
`including lactation. We examined the possibility of hGH stabi-
`lization by elimination of its lactogenic activity. Chimeric GHs
`were constructed by replacement of di¡erent segments of hGH
`with sequences derived from non-lactogenic porcine GH. As was
`observed in the rat Nb2-11C lymphoma cell test,
`lactogenic
`activity of some chimeric hormones was seriously destroyed.
`This kind of hormones displayed the substantial
`increase in
`thermal and guanidine hydrochloride stability. The more stable
`hGH variants were found to be more soluble in Escherichia coli
`cells. ß 2002 Published by Elsevier Science B.V. on behalf of
`the Federation of European Biochemical Societies.
`
`Key words: Growth hormone; Inclusion bodies;
`Lactogenic activity; Mutagenesis; Stability
`
`1. Introduction
`
`Human growth hormone (hGH, somatotropin) is a single
`chain polypeptide hormone consisting of 191 residues with
`two disul¢de bonds. Its main function is the stimulation of
`somatic and bone growth, as well as an increase in the size
`and mass of organs and tissues. In addition,
`it in£uences
`protein, carbohydrate, and lipid metabolism. A unique pecu-
`liarity of human and primate growth hormones (GHs) is the
`broad species speci¢city and the capability of a direct e¡ect on
`the mammary gland cells by increasing RNA and protein syn-
`thesis (lactogenic activity) [1]. Practically all of the recombi-
`nant hGH produced by bacteria at high temperature (37‡C) is
`localized in large insoluble aggregates, the so-called inclusion
`bodies [2], which pose a serious obstacle to e⁄cient hGH
`production for biomedical use. One of the ways of solving
`the problem of hGH aggregation in vivo is to increase its
`intrinsic stability so as to compensate for the lack of disul¢de
`bonds in bacterial cytoplasm.
`
`Numerous e¡ects caused by hGH are based on its ability to
`bind to speci¢c receptors on the target cell surfaces. Evidently,
`the hGH binding interfaces have evolved to make favorable
`intermolecular contacts with the receptors. This may be at the
`expense of intramolecular interactions and burying hydropho-
`bic residues, and so lower the hGH conformational stability
`[3]. Indeed, there are many examples showing that the replace-
`ment of the ligand binding residues results in increased pro-
`tein stability [4,5]. Then, attenuation of one of the hGH minor
`activities might lead to a stabilization of its structure. To
`check such a possibility, we have constructed a series of chi-
`meric GHs with homologous replacements in hGH based on
`the structure of porcine growth hormone (pGH). Since the
`latter is devoid of lactogenic activity, we expect that the ac-
`tivity will be lost in some of the chimeric hormones. The
`primary structures of human and porcine GHs are highly
`homologous (67%) and their tertiary structures are similar.
`Therefore,
`it may be anticipated that the mutations intro-
`duced into the hGH molecule will not signi¢cantly change
`its conformation. This work reports a study of lactogenic
`activity and stability of chimeric hormones ‘14^33’, ‘41^73’,
`‘14^95’, and ‘78^95’. Figures designating chimeric GHs corre-
`spond to the residues from the pGH sequence that are in-
`serted into hGH; the numbering follows the hGH sequence.
`In accordance with this, variant ‘14^33’ is a hGH in which the
`region £anked by residues 14 and 33 is replaced with a similar
`region of pGH. We show that the elimination of lactogenic
`activity results in stabilization of hGH molecule and that in-
`tracellular hormone solubility is determined by its stability.
`
`2. Material and methods
`
`2.1. Gene expression and protein puri¢cation
`Genes for chimeric GHs were constructed by the ‘homolog recom-
`bination’ method [6]. The nucleotide sequences of GHs genes were
`determined in both directions. Expression plasmids for GHs were
`constructed by replacement of gene 10 in pGEMEX1 (Promega)
`with a GH gene [2]. Proteins were puri¢ed from the Escherichia coli
`strain BL21(DE3) harboring the corresponding expression plasmid [7].
`
`*Corresponding author. Fax: +7-095-135 14 05.
`E-mail address: aamakarov@genome.eimb.relarn.ru (A.A. Makarov).
`
`Abbreviations: GH, growth hormone; hGH, human growth hor-
`mone; pGH, porcine growth hormone; DSC, di¡erential scanning
`calorimetry; CD, circular dichroism; GdnHCl, guanidine hydrochlo-
`ride
`
`2.2. Nb2-11C cell proliferation assay
`Synchronization of Nb2-11C cells in the G0/G1 phase and monitor-
`ing of cell proliferation were performed as described earlier [8]. The
`doubling rates were calculated after 72 h from the following equation:
`No. of doublings = log [(no. of cells in the presence of a hormone)/(no.
`of cells in the absence of a hormone)]/log2. The initial number of cells/
`ml was 250 000.
`
`0014-5793 / 02 / $22.00 ß 2002 Published by Elsevier Science B.V. on behalf of the Federation of European Biochemical Societies.
`PII: S 0 0 1 4 - 5 7 9 3 ( 0 2 ) 0 3 3 2 5 - 2
`
`FEBS 26533 11-9-02
`
`MYLAN INST. EXHIBIT 1097 PAGE 1
`
`MYLAN INST. EXHIBIT 1097 PAGE 1
`
`
`
`258
`
`A.A. Schulga et al./FEBS Letters 528 (2002) 257^260
`
`2.3. Di¡erential scanning calorimetry (DSC)
`Microcalorimetric measurements were carried out on a DASM-4
`microcalorimeter (NPO Biopribor, Pushchino, Russia) in 0.48 ml cells
`at a heating rate of 1 K/min on 0.4^0.8 mg/ml proteins in 10 mM
`Na2HPO4, pH 7.5. The partial molar heat capacity of the protein
`(Cp), denaturation temperature (Td) and calorimetric denaturation
`enthalpy (vHcal) were determined as described elsewhere [9]. The ac-
`curacy of the calorimetric enthalpy was 6%, that of Td within 0.2‡C.
`
`2.4. Circular dichroism (CD)
`Far-UV CD spectra were recorded on a Jasco J-715 spectropo-
`larimeter equipped with Neslab RTE-111 water bath in a 0.02 cm
`cell on 0.3 mg/ml proteins in 10 mM Na2HPO4, pH 7.5. Protein
`melting was carried out at the same heating rate as in microcalorim-
`etry (1 K/min). Denaturation temperatures were determined from the
`peaks in the ¢rst derivatives of the melting pro¢les (accuracy 0.3‡C).
`The results were expressed as molar ellipticity, [3] (deg cm2 dmol31),
`based on a mean amino acid residue weight of 115.7 for hGH, 113.1
`for variant ‘14^95’, 115.1 for variant ‘14^33’, 114.3 for variant ‘41^73’
`and 115.5 for variant ‘78^95’.
`
`2.5. Equilibrium guanidine hydrochloride (GdnHCl)-induced
`denaturation
`Denaturation experiments were performed at 25‡C in 10 mM
`Na2HPO4, pH 7.5, at a ¢nal protein concentration of 0.2 mg/ml.
`The GdnHCl concentrations were determined by refractometry. The
`unfolding transition was monitored by CD at 222 nm.
`
`3. Results and discussion
`
`3.1. hGH and chimeric GHs lactogenic activity
`Lactogenic activity of GHs was determined by an increase
`in the amount of preliminarily synchronized cells of the rat
`lymphoma line Nb2-11C in response to the preparation under
`study. Hormones devoid of lactogenic activity are not able to
`stimulate proliferation of this cell
`line [10]. The results of
`measurements of proliferative activity are shown in Fig. 1.
`It is seen that activity of GHs ‘14^33’ and ‘78^95’ do not
`di¡er from that of hGH. The ‘14^95’ GH is devoid of ability
`to stimulate cell proliferation in the whole interval of concen-
`trations from 0 to 500 pM. The ‘41^73’ GH activity is mark-
`edly decreased. Thus, cells begin to proliferate only at the
`hormone concentration of 8 pM (such a ‘threshold’ concen-
`
`Fig. 1. The e¡ect of various concentrations of hGH (E) or chimeric
`GHs ‘14^95’ (a), ‘14^33’ (O), ‘41^73’ (S) and ‘78^95’ (8) on the
`proliferation rate of Nb2-11C lymphoma cells.
`
`Fig. 2. CD spectra of hGH (double chain line ^ 20‡C, triple chain
`line ^ 90‡C) and chimeric GHs ‘14^33’ (solid line), ‘14^95’ (dotted
`line), ‘41^73’ (dashed line) and ‘78^95’ (chain line) at pH 7.5, 20‡C.
`Inset: Temperature dependence of CD at 222 nm for hGH (1) and
`chimeric GH ‘14^33’ (2).
`
`tration for hGH is 0.2 pM), whereas cell doubling in 72 h is
`observed at the ‘41^73’ GH concentration of 440 pM (and at
`2 pM hGH). It is quite obvious that such a drastic decrease in
`the hormone activity is due to mutations in the 41^73 region.
`The ‘14^33’ variant is equipotent to hGH in Nb2-11C bio-
`assay (Fig. 1). Moreover, substitution of the segment 8^19 in
`hGH for homologous segment derived from pGH a¡ects nei-
`ther the proliferative activity in Nb2-11C bioassay, nor the
`binding a⁄nity for the Nb2 prolactin receptors [8]. Thus,
`the homologous substitutions in the hGH segment 8^33 do
`not in£uence the hGH binding to Nb2 rat prolactin receptors.
`In contrast, according to [11], the hGH mutant containing
`segment 11^33 of pGH causes tremendous disruption in bind-
`ing a⁄nity for the human prolactin receptor. Similarly, intro-
`duction of Lys168Ala/Glu174Ala mutations into hGH caused
`more than 6000 times decrease in its activity in the test of
`FDC-P1 cells transfected with the full-length human prolactin
`receptor and only 12 times decrease in the Nb2-11C test if the
`half-maximal concentrations for stimulation of cell prolifera-
`tion (EC50) were compared [12]. Altogether, this suggests that
`the functional determinants for the hGH binding to the hu-
`man and rat prolactin receptors do not coincide. The com-
`plete loss of the ‘14^95’ GH activity and preservation of that
`of ‘14^33’ and ‘78^95’ GHs in the rat lymphoma Nb2 test
`mean that the region 34^77 contains amino acid residues
`that are critical for the interaction of hGH with the rat pro-
`lactin receptors.
`
`3.2. Secondary structure and thermostability of hGH and
`chimeric hormones
`To study conformational changes in chimeric proteins, CD
`in the far-UV region was used. The CD spectrum of hGH
`(Fig. 2) is typical of a protein with a high content of K-helices
`and coincides with the previously published data [13]. All
`spectra of hybrid proteins are practically identical in shape
`
`FEBS 26533 11-9-02
`
`MYLAN INST. EXHIBIT 1097 PAGE 2
`
`MYLAN INST. EXHIBIT 1097 PAGE 2
`
`
`
`A.A. Schulga et al./FEBS Letters 528 (2002) 257^260
`
`259
`
`for hGH at 90‡C made up about 50% of initial level (Fig. 2).
`Similar results were obtained in [14] at acid pH. This is in-
`dicative of preservation of a signi¢cant part of secondary
`structure in the denatured protein and supports calorimetric
`data also pointing to a partial unfolding of hGH during heat
`denaturation due to aggregation.
`
`3.3. Equilibrium denaturation of hGH and chimeric hormones
`by GdnHCl
`The GdnHCl denaturation of GHs was extensively studied
`[15^17]. Non-human species of GH were characterized by the
`presence of the unfolding intermediates having a tendency to
`self-associate. At low protein concentrations ( 9 0.2 mg/ml),
`the hGH denaturation, studied by various techniques, was
`shown to be completely reversible and consistent with the
`two-state mechanism. However, at higher protein concentra-
`tions there is the e¡ect of protein concentration on the free
`energy of unfolding due to the presence of self-associating
`unfolding intermediates [18].
`The GdnHCl denaturation data for hGH and chimeric hor-
`mones are presented in Fig. 4. The S-shaped dependences of
`CD intensities on GdnHCl at 222 nm are indicative of a
`highly cooperative melting process [9]. Mutations in hGH
`resulted in an increase of the GdnHCl concentration at 50%
`protein unfolding (D50%) for GHs ‘14^33’ and ‘14^95’ (Table
`1).
`
`3.4. Relationship between inclusion bodies formation in E. coli
`cells and GHs stability
`It is well known that the reducing milieu of bacterial cyto-
`plasm prevents disul¢de bond formation in proteins [19]. At
`the same time, reduction of disul¢de bridges in hGH signi¢-
`cantly lowers its stability and results in an increased tendency
`for self-association as compared with the disul¢de-containing
`hGH molecules [20]. This may have relevance for the forma-
`tion of inclusion bodies during the heterologous expression of
`hGH gene in E. coli.
`
`Fig. 4. The GdnHCl-induced equilibrium denaturation of hGH
`(F, solid line) and chimeric GHs ‘78^95’ (8, dotted line), ‘41^73’
`(S, dashed line),
`‘14^33’ (R, chain line) and ‘14^95’ (b, double
`chain line) at 25‡C, pH 7.5, as detected by CD at 222 nm. Protein
`concentration was 0.2 mg/ml.
`
`Fig. 3. Temperature dependence of the excess heat capacity of hGH
`(1) and chimeric GHs ‘78^95’
`(2),
`‘41^73’
`(3),
`‘14^95’
`(4) and
`‘14^33’ (5) at pH 7.5.
`
`to that of hGH. However, the spectral amplitudes in charac-
`teristic minima and maximum are di¡erent, i.e. chimeric hor-
`mones and hGH are characterized by a di¡erent helical con-
`tent (Table 1).
`The e¡ect of mutations on the parameters characterizing
`the thermal denaturation of the hormone was studied using
`DSC. Fig. 3 shows the temperature dependence of the excess
`heat capacity of hGH and chimeric hormones at pH 7.5. In all
`cases denaturation was irreversible. Partial heat capacities of
`hGH and hybrids at 25‡C are the same and equal
`to
`(0.34 þ 0.02) cal/KWg. This value is typical of compact globular
`proteins [9]. The calorimetric enthalpy of hGH denaturation
`(Fig. 3, curve 1) is 48 kcal/mol that is more than two times
`lower than mean values of denaturation enthalpy correspond-
`ing to full unfolding of the near-size proteins [9]. A similar
`result was obtained for hGH in the acid pH interval [14]. The
`midpoint of the hGH thermal denaturation was equal to
`81‡C. For GHs ‘41^73’,
`‘14^95’ and ‘14^33’, denaturation
`temperature increases for 2.7, 4.0 and 8.3‡C, respectively,
`whereas for ‘78^95’ it decreases for 6‡C (Table 1).
`To collate the thermodynamic and structural data for hGH
`and ‘14^95’, the temperature dependence of the CD at 222 nm
`was measured. hGH undergoes a cooperative transition at
`81.2‡C, whereas it occurs at 86.6‡C for ‘14^95’ (Fig. 2, inset).
`These values are very close to the corresponding Td values
`obtained using scanning microcalorimetry. The CD amplitude
`
`Table 1
`Content of K-helices, temperature of heat denaturation and GdnHCl
`concentration at D50% for hGH and chimeric hormones at pH 7.5
`a
`Td (‡C)b
`D50% (M)c
`Hormone
`fH
`‘14^33’
`0.66
`89.3
`4.77
`‘14^95’
`0.60
`85.0
`4.85
`‘41^73’
`0.58
`83.7
`4.54
`hGH
`0.56
`81.0
`4.54
`‘78^95’
`0.52
`75.0
`4.33
`aCalculation of K-helix content (fH) was based on equation fH =
`3([3]222+2340)/30 300 [21]. The estimated uncertainty is 3%.
`bThe estimated uncertainty is 0.2‡C.
`cThe estimated uncertainty is 1%.
`
`FEBS 26533 11-9-02
`
`MYLAN INST. EXHIBIT 1097 PAGE 3
`
`MYLAN INST. EXHIBIT 1097 PAGE 3
`
`
`
`260
`
`A.A. Schulga et al./FEBS Letters 528 (2002) 257^260
`
`Table 2
`The proportion (%) of soluble GHs at di¡erent growth tempera-
`tures, determined from SDS^PAGE of native lysis supernatants and
`total cell protein
`24‡C
`Hormone
`100
`‘14^33’
`100
`‘14^95’
`100
`‘41^73’
`100
`hGH
`30
`‘78^95’
`nd ^ not determined.
`
`30‡C
`100
`100
`100
`25
`20
`
`37‡C
`7
`100
`30
`5
`0
`
`42‡C
`nd
`20
`5
`nd
`nd
`
`is an exception to the rule. Being most stable to the thermal
`denaturation it is equipotent to hGH in bioactivity deter-
`mined in Nb2 test. But it appears to be the exception that
`proves the rule. Really, as was discussed above, the functional
`hGH determinants for binding to the human and rat prolactin
`receptors do not coincide completely. The homologous sub-
`stitution of the segment 11^33 in hGH for porcine segment
`causes dramatic decrease in binding a⁄nity for the human
`prolactin receptor [11]. Thus, although the variant ‘14^33’ is
`fully active in the Nb2-lymphoma bioassay,
`it most likely
`lacks the lactogenic potency in humans.
`
`Acknowledgements: We thank P.M. Rubtsov for providing us with
`the plasmid pPGHsyn/rec. This work was supported by RFBR Grant
`01-04-48660 and NIH FIRCA Grant TW01058-01.
`
`References
`
`[1] Wells, J.A., Cunningham, B.C., Fuh, G., Lowman, H.B., Bass,
`S.H., Mulkerrin, M.G., Ultsch, M. and De Vos, A.M. (1993)
`Recent Prog. Horm. Res. 48, 253^275.
`[2] Levichkin, I.V., Arsen’eva, D.A. and Schulga, A.A. (1998) Mol.
`Biol. (Moscow) 32, 280^284.
`[3] Meiering, E.M., Serrano, L. and Fersht, A.R. (1992) J. Mol.
`Biol. 225, 585^589.
`[4] Schreiber, G., Buckle, A.M. and Fersht, A.R. (1994) Structure 2,
`945^951.
`[5] Shoichet, B.K., Baase, W.A., Kuroki, R. and Matthews, B.W.
`(1995) Proc. Natl. Acad. Sci. USA 92, 452^456.
`[6] Schulga, A.A., Levichkin, I.V., Kurbanov, F.T., Okorokov, A.L.,
`Pozmogova, G.E. and Kirpichnikov, M.P. (1994) Nucleic Acids
`Res. 22, 3808^3810.
`[7] Anisimova, M.V., Shul’ga, A.A., Levichkin, I.V., Kirpichnikov,
`M.P., Poliakov, K.M., Skriabin, K.G., Vijayalakshmi, M.A. and
`Varlamov, V.P. (2001) Bioorg. Khim. (Moscow) 27, 27^31.
`[8] Belousova, Y.V., Schulga, A.A., Gabrielyan, A.E., Levichkin,
`I.V., Lopatin, S.A., Chernov, B.K., Zakharova, I.V., Kuzina,
`N.V., Mikhailova, L.I., Moskaleva, E.Y., Kondratenko, T.Y.,
`Severin, E.S., Skryabin, K.G. and Kirpichnikov, M.P. (1996)
`Mol. Biol. (Moscow) 30, 673^680.
`[9] Privalov, P.L. (1979) Adv. Protein Chem. 33, 167^241.
`[10] Tanaka, T., Shiu, R.P.C., Gout, P.W., Beer, C.T., Nobie, R.L.
`and Friesen, H.G. (1980) J. Clin. Endocrinol. Metab. 51, 1058^
`1063.
`[11] Cunningham, B.C. and Wells, J.A. (1991) Proc. Natl. Acad. Sci.
`USA 88, 3407^3411.
`[12] Fu, G., Colosi, P., Wood, W.I. and Wells, J.A. (1993) J. Biol.
`Chem. 268, 5376^5381.
`[13] Zhukovsky, E.A., Mulkerrin, M.G. and Presta, L.G. (1994) Bio-
`chemistry 33, 9856^9864.
`[14] Gomez-Orellana, I., Variano, B., Miura-Fraboni, J., Milstein, S.
`and Paton, D.R. (1998) Protein Sci. 7, 1352^1358.
`[15] Brems, D.N., Plaisted, S.M., Havel, H.A., Kau¡man, E.W., Sto-
`dola, J.D., Eaton, L.C. and White, R.D. (1985) Biochemistry 24,
`7662^7668.
`[16] Brems, D.N., Brown, P.L. and Becker, G.W. (1990) J. Biol.
`Chem. 265, 5504^5511.
`[17] Bastiras, S. and Wallace, J.C. (1992) Biochemistry 31, 9304^9309.
`[18] De Felippis, M.R., Alter, L.A., Pekar, A.H., Havel, H.A. and
`Brems, D.N. (1993) Biochemistry 32, 1555^1562.
`[19] Martineau, P. and Betton, J.-M. (1999) J. Mol. Biol. 292, 921^
`929.
`[20] Youngman, K., Spencer, D.B., Brems, D.N. and De Felippis,
`M.R. (1995) J. Biol. Chem. 270, 19816^19822.
`[21] Chen, Y.H., Yang, J.T. and Martinez, H.M. (1972) Biochemistry
`11, 4120^4131.
`
`We compared the tendency of the wild-type and chimeric
`GHs to form inclusion bodies in bacteria. Soluble and total
`cellular fractions were prepared and analyzed by SDS^PAGE
`(Table 2). All the proteins were produced at the same level
`(about 300 mg/l). At 37‡C about 95% hGH is localized within
`a cell as insoluble aggregates. As growth temperature de-
`creases, a soluble protein fraction is increased, and at 24‡C
`hGH is mainly accumulated in a soluble form in the cyto-
`plasm. On the contrary, pGH is found in inclusion bodies
`in the whole temperature interval from 37 to 9‡C (data not
`shown). Substitutions in the region 14^95 as a rule improve
`intracellular solubility of hGH. Thus, already at 30‡C (vs.
`24‡C for hGH) chimeric GHs ‘14^33’ and ‘41^73’ are detect-
`able mainly in a soluble fraction. ‘14^95’ GH is fully soluble
`even at 37‡C. Moreover, about 20% of this protein is soluble
`even at 42‡C. This markedly distinguishes it both from hGH
`and pGH. ‘78^95’ GH resembles hGH by its properties.
`The comparison of intracellular solubility of chimeric hor-
`mones and corresponding values of D50% reveals a regularity ^
`the more resistant the protein to GdnHCl denaturation, the
`higher its solubility. The most stable variant,
`‘14^95’ GH
`(Table 1), has the highest solubility in the series of the hor-
`mones studied (Table 2), and the least stable hormones, ‘78^
`95’ GH and hGH, have the lowest solubility (only 20% of ‘78^
`95’ GH and 25% of hGH are soluble at 30‡C). ‘14^33’ GH
`with intermediate value of D50% was found entirely in the
`soluble fraction at 30‡C. A similar relationship between the
`protein solubility and its stability was found for a single chain
`antibody fragment [19].
`The cytoplasmic expression is preferable for protein pro-
`duction because it provides maximal product yield. The
`main obstacle to this route is the protein deposition in inclu-
`sion bodies as an insoluble and inactive material. The data
`described in this work demonstrate the possibility to increase
`intracellular solubility of a protein by increasing its intrinsic
`stability.
`
`3.5. Elimination of lactogenic activity results in stabilization of
`hGH
`Our results indicate an inverse relationship between the
`lactogenic activity of hGH and its stability. Demolishing lac-
`togenic activity made it possible to construct the hGH var-
`iants (‘41^73’ and ‘14^95’) exhibiting higher conformational
`stability and lower propensity to form inclusion bodies in
`bacterial cytoplasm. At the ¢rst glance, the mutant ‘14^33’
`
`FEBS 26533 11-9-02
`
`MYLAN INST. EXHIBIT 1097 PAGE 4
`
`MYLAN INST. EXHIBIT 1097 PAGE 4
`
`