PT100580A - PRODUCTION AND RECOVERY OF RECOMBINANT NEUROTROPHINS - Google Patents

Description

74 061 6526-081-118 -2-

DESCRIPTIVE MEMORY

This patent application is a continuation-in-part of the co-pending patent application Serial No. 07 / 715,185, filed June 12, 1991, which is hereby incorporated by reference in its entirety.

1. INTRODUCTION

This invention relates to the fields of neurobiology, molecular biology and biochemistry of proteins and, in particular, neurotrophins and the processes of production and recovery thereof.

BACKGROUND OF THE INVENTION

Neurotrophins are a group of proteins involved in the functioning of the nervous system. They stimulate the growth of nerve cells and, during embryonic development, support their survival. To date, the researchers have identified four neurotrophins: nerve growth factor (NGF), Füllrich et al., 1983, Nature 303: 821-825). brain derived neurotrophic factor (BDNF) (Leibrock et al., 1989, Nature 341: 149-152). neurotrophic factor 3 (NT-3) (Maisonpierre et al., 1990, Science 247: 1446-14511 and neurotrophin-4 (Hallbook et al., 1991, Neuron 6: 845-858). essential for the development and survival of the peripheral and sympathetic nervous system and cholinergic neurons of the brain.It is now being tested for use in Alzheimer's disease.BDNF promotes the survival of sensory neurons in the central nervous system and represents a hope in therapy of NT-3 and NT-4 have recently been obtained, and their biological role is now being investigated. Like NGF, NT-3 appears to act on sensitive and sympathetic neurons. three subunits: α, Λ & X · Only β subunit has neurotrophic activity. NGF-β is 74 061 6526-081-118

produced as a pre-pro-protein which is secreted by the cell and processed in the mature form. The pre-propeptide contains 187 amino acid residues. During processing, the pre- and pro-sequences are removed since they are two amino acid residues at the C-terminus. This gives a mature protein of 118 amino acid residues. The mature protein has six cysteine residues that form three disulfide bonds. The protein also contains many basic amino acids resulting in a positive charge at neutral pH.

Structurally, neurotrophins form a family, apparently having developed from a common ancestral gene. (Hyman et al., 1991, WO 91/03568). For example, the amino acid sequence of the human N6F β-polypeptide (hNGF) is 90% homologous to that of mouse NGF and that of bovine NGF; and the three human neurotrophins, hNGF, hBDNF and hNT-3, share 60% homology at the amino acid level. The six cysteine residues are conserved in all known neurotrophins, implying that disulfide bonds are important for the function.

The natural sources of neurotrophins are limited. The mouse submaxillary gland is rich in NGF but the isolation is difficult and the amounts recovered are small (Hyman et al., 1991, WO 91/03568). Del la Valle et al. . (EP 0 333 574, 1989) have described the isolation of small amounts of hNGF from human placenta by chromatographic fractionation.

As a result, the researchers turned to molecular genetics to provide a rapid source of biologically active neurotrophins. The DNA sequences encoding NGF, BDNF and NT-3 were all isolated (Ullrich et al., Nature 303: 821-825; Hyman et al., WO 91/03568; Hohn et al., WO 91/03569; Kaisho et al., FEBS Letters 266: 187-191). The researchers transformed animal and non-animal hosts with these sequences in order to express neurotrophins.

Kanaya et al. (1989, Gene 83: 65-74) fused a sequence of 74 061 6526-081-118

DNA encoding the mature hNGF in one encoding the pre-pro sequence of the yeast-pairing factor. After expression, the culture supernatant contained a protein recognized by the hNGF-α antibodies. However, the partially purified protein exhibited low neurotrophic activity. In addition, the level of expression was low. Kanaya et al. Stated that these results may be due either to the inability of the yeast system to remove the two extra C-terminal amino acid residues during maturation or to a problem with hNGF protein folding.

Chan et al. (EP 0 370 171, 1990) produced mature hNGF in insect cells. They fused a gene for pre-pro-hNGF to the polyhedrin promoter and the baculoviral DNA leader sequence. They described the production of 6æg / ml hNGF, but did not physically characterize the product.

The researchers also expressed human NGF, BDNF and NT-3 in mammalian expression systems. Bruce and Heinrich (1989, Neurobiolocrv of Aaina 10: 89-94) expressed a DNA sequence encoding the complete precursor for hNGF in COS cells and detected the hNGF dimer in the conditioned medium. However, they were not able to determine the efficacy at which pre-pro-hNGF was converted to mature hNGF. Kakinuma et al. (EP 0 414 151, 1991) expressed active hNGF in CHO cells. Hyman et al. (WO 91/03568, 1991) expressed hBDNF in CHO cells. Nakahama et al. (EP 0 386 752, 1990) and Hohn et al. (WO 91/03569, 1991) expressed hNT-3 in COS cells.

Mammalian systems provide the most natural means for the production of mammalian proteins. However, the production of large amounts of proteins in these systems is very expensive. Consequently, there is a need to develop systems that are less costly and more productive. One such system is E. coli.

The researchers attempted to produce neurotrophins in E. coli. but without significant success. Gray and Ullrich (EP 0 121

74 061 6526-081-118 -5-338, 1984) constructed an expression vector encoding N-methionyl-hNGF and expressed the gene in E. coli. They described the identification of hNGF by immunodetection in Western blot, but did not isolate the protein nor demonstrate the biological activity.

Hu and Neet (1988, Gene: 57-65) attempted to express mouse NGF in E. coli. They cloned a DNA sequence encoding a mature mouse NGF in which they replaced the serine, the N-terminal amino acid in the native protein, with methionine. They inserted the DNA sequence into a plasmid having a temperature inducible lambda PL promoter. This system expresses other heterologous proteins at the rates of 10% -25% of the total cellular protein. They expressed the gene and isolated the NGF by precipitation with ammonium sulfate followed by dialysis against acetate buffer. However, as verified by bioassay, this system yielded only 0.0005% to 0.1% NGF. The authors speculated that the highly inconsistent and low yields were due to the NGF toxicity to the cells, instability or inefficacy in the translation of the AJRNm or to disaggregated disulfide bonds in the redoubled, oxidized protein.

Iwai et al. (1986, Chem. Pharm. Bull 34: 4724-4730) described the synthesis of a gene encoding hNGF with codons preferred in E. coli. They directly expressed the gene as N-methionyl-hNGF or as a human growth hormone fusion protein. Direct expression only has a quarter of the expression efficiency of the fusion protein. They examined the proteins by SDS-PAGE, but, on the other hand, did not isolate them.

Dicou et al. (1989, J. Neurosci. Res. 22: 13-19) expressed prepro-mouse NGF and a fragment of hNGF as β-galactosidase fusion proteins and found that the addition of protease inhibitors enhances yield.

In summary, previous attempts to express neurotrophins in E. coli have resulted in cell death, low levels of protein accumulation, and expression of proteins with virtually no biological activity.

SUMMARY OF THE INVENTION

J

This invention provides processes for the production and recovery of biologically active recombinant neurotrophins from non-animal host cells. The methods of producing recombinant neurotrophins comprise the step of culturing a host cell transformed with a recombinant DNA molecule comprising an expression control sequence operably linked to a DNA sequence encoding a neurotrophin. When neurotrophins are directly expressed in the cytoplasm, host cells are preferably protease deficient mutants and, to the best of our knowledge, heat shock regulator mutants. In bacterial hosts, the neurotrophin gene is preferably under the control of a controllable promoter and the expression is preferably non-induced or repressed until the cells reach the last log phase. Neurotrophins may also be secreted from the cell during production, providing a DNA sequence encoding a signal peptide upstream of the DNA sequence encoding the neurotrophin.

Recovery processes of biologically active recombinant neurotrophins produced by host cell cultures comprise the step of solubilizing neurotrophin in a solution comprising a strong denaturing agent, the solution essentially free of reducing agents. According to one embodiment of the invention the solution further comprises a protease inhibitor in an amount effective to inhibit the degradation of neurotrophin by proteases. Further embodiments of the invention further comprise some or all of the following steps: exchanging the strong denaturing agent with a weak denaturing agent; adjusting the solution to comprise a basic amino acid or one equivalent to a concentration effective to maintain the solubility of the neurotrophin in a non-denaturing medium; purifying the neurotrophin of other 74 061 6526-081-118 7-

molecules in solution; removing the weak denaturing agent; and complete the purification in the absence of the denaturing agent.

In another embodiment of the invention, the neurotrophin molecule expressed in E. coli and recovered in a biologically active form is neurotrophin-4. In a preferred embodiment of the invention directed to the expression of biologically active NT-4 in E. coli. the NT-4 is encoded by a nucleic acid molecule comprising a sequence substantially as depicted for hNT-4 in Figure 11 (SEQ ID Na22) or may comprise a sequence that is at least about 70 percent homologous to such a sequence.

4. DESCRIPTION OF THE DRAWINGS Figure 1 is a schematic representation of plasmid PRPN133. The solid line represents the DNA sequences derived from pBR322 with the origin of replication (ORI) and the β-lactamase (ampicillin resistance) gene (Ap) indicated. The identifying characteristics of the plasmid are indicated by regions framed with arrows indicating the direction of transcription or replication. PL indicates the lambda PL promoter. rbs1 is the wild type promoter and the T7 phage ribosomal binding site Φ1.1. The hNGF gene encodes a mature polypeptide in which the second amino acid residue, the serine, is replaced by threonine. cI857 indicates the heat-inactivating λ repressor gene Figure 2 shows dose-response curves for recombinant human NGF produced in E. coli by (A) E8 and (B) DRG dorsal root ganglion (DRG) explants of chicken embryo E8 dissociated. Figure 3 depicts the nucleotide and amino acid sequences of the wild-type LamB (Fig 3A, SEQ ID Nal) and the synthetic LamB signal sequences, LamB1 (Fig. 3B, SEQ ID Na2), LamB2 (Fig. 3C , SEQ ID NO 3), LamB 3 (Fig. 3D, SEQ ID No 4) and

6526-081-118 -8

LamB4 (Figure 3E, SEQ ID Ns5). Figure 4 depicts the signal kinetics of the signal sequence of the modified LamB signal sequences, LamB1 and LamB3. The BL21 / DE3 strain containing the appropriate LamB-hBDNF plasmid was grown. In this strain, the gene encoding the T7 RNA polymerase has been inserted into the chromosome and is under transcriptional control of the lac promoter (Studier and Moffatt, J. Mol. Biol. 189: 113-30). The addition of IPTG to the growing cells for 10 minutes allows the synthesis of T7 RNA polymerase. Subsequent addition of rifampicin, 0.2 mg / ml, blocks transcription by E. coli RNA polymerase but allows transcription by T7 RNA polymerase. This results in the selective transcription of the LamB-hBDNF gene which is immediately placed downstream of the last T7 01.1 promoter and the ribosome binding site of rbs2. Cells were pulsed with 35 S-methionine for 30 seconds and then captured with an excess of cold methionine. The cultures were sampled at the times indicated after capture and the labeled proteins were analyzed by 15% SDS-PAGE and fluorography. Processing was determined by densitometric scanning of the precursor and mature forms of LamB-hBDNF. Figure 5 is a schematic representation of pRPN121. The solid line represents the DNA sequences derived from pBR322 with the origin of replication (ORI) and the 0-lactamase (Ap) gene indicated. The distinguishing characteristics of the plasmid are indicated by regions framed with arrows indicating the direction of transcription or replication (ORI). LacUV5 is the promoter. rbs2 indicates the T7 promoter 01.1 and the T7 ribosome binding site §1.1 with minor wild-type nucleotide substitutions designed to create suitable restriction sites. LamB2 is the signal sequence. hBDNFmyc is the structural gene. Figure 6 depicts a Fast S-SEPHAROSE 1 fractionation protein profile of hBDNFmyc. Extract W3110 I (ÂμF / pRPN121 after DEAE chromatography was fractionated 74 061 6526-081-118

W*. As described on a 1.6 cm X 6.5 cm Fast S-SEPHAROsA column) in 7 M urea, 50 mM histidine, pH 5.0, 1 mM EDTA. Fractions indicated were analyzed by 15% SDS-PAGE and proteins visualized by Coomassie dye. Fractions 26-30 were pooled. Figure 7 is a summary of the purification procedure. Strain W3110 I <1> / pRPN121 was allowed to grow, induced and extracted as described above. The pooled column fractions were analyzed by SDS-PAGE on a 15% acrylamide gel and stained with Coomassie blue. Track 1, DEAE; lane 2, Fast S-SEPHAROSI® 1; Lane 3, Fast S-SEPHAROSI 2; lane 4, C4 reverse phase HPLC. Molecular weight standards are indicated. Figure 8 depicts a dose response curve for purified hBDNFmyc stimulation of the development of neurite in E8 chicken embryo DRG. The activity of hBDNFmyc is compared to that of recombinant hBDNF purified from a Chinese hamster ovary cell line. The two proteins were purified to more than 95% by C4 reverse phase HPLC. Figure 9 depicts a dose-response curve for stimulating the development of neurite development in E8 chicken embryo DRG by recombinant hBDNF purified from E. coli. The hBDNF was purified to greater than 95% by C4 reverse phase HPLC. Figure 10 is a schematic representation of pRPN149. The solid line represents the DNA sequences derived from pBR322 with the origin of replication (ORI) and the β-lactamase (Ap) gene indicated. The distinguishing characteristics of the plasmid are indicated by regions framed with arrows indicating the direction of transcription or replication (ORI). LacUV5 is the promoter. rbs2 indicates the T7 promoter §1.1 and the T7 ribosome binding site §1.1 with minor wild-type nucleotide substitutions designed to create suitable restriction sites. LamB1 is the signal sequence. hBDNF is the gene 74 061

Structural properties. Figure 11 is the DNA sequence of a portion of the isolated human genomic phage clone 7-2 encoding human NT-4 (SEQ ID Na22; Access Nail to ATCC 75070). The predicted hNT-4 protein encoded by genomic clone 7-2 is represented by the letter amino acid symbols (SEQ ID No 23). The framed region represents the predicted cleavage site of the hNT-4 preprotein. The arrows indicate the residues conserved in the pre-sequence. The underlined region (N-R-S) represents a consensus sequence for N-glycosylation. The region within a circle represents the starting methionine. The linker acceptor site is located on base pair 461-462 (AG), representing the 3 'end of the intron. Figure 12 is a schematic representation of pRG91. The solid line represents the DNA sequences derived from pBR322 with the origin of replication (ORI) and the β-lactamase gene indicated. The distinctive features of the vector are shown by framed regions. LacUV5 is the promoter. rbs2 is the phage T7 promoter $ 1.1 and the T7 ribosome binding site §1.1. LamB2 is the signal sequence. Plasmid pRG91 (Regeneron Pharmaceuticals) is a vector based on pBR322 designed for the expression of recombinant proteins and their secretion into the periplasmic space of Escherichia coli. The vector consists of the tight, regulated lacUV5 promoter followed by the phage T7 promoter ¢ 1.1 and the ribosome binding site inserted between the restriction sites EcoRI and NruI in pBR322. These control elements direct the expression of the LamB2 signal sequence to which the recombinant protein gene sequences can fuse. DNA sequences between the unique NruI and PvuII restriction sites were deleted, resulting in a larger number of plasmid copies. This plasmid confers resistance to ampicillin (Ap). Figure 13 is a dose-response curve for hNT-4 74 061 6526-081-118 -11

expressed in E. coli. The soluble protein extract of an induced culture of the strain RFJ26 containing the plasmid pRG173 was assayed in dorsal root ganglia E8 explants. The background activity of this assay was 0.2 units. Figure 14 depicts the effect of hNT-4 on CAT activity. Treatment of cultures enriched in motor neurons with a partially purified extract of an induced culture of the strain RFJ26 containing the plasmid pRG173 resulted in a 3.6 fold increase (at 1:20 dilution) in CAT activity after 48 hours, when compared to untreated controls (C-NT) and buffer (C-buffer). The E. coli extract was passed through a Sepharose-S column as described below prior to treatment of cultures enriched in motor neurons.

DETAILED DESCRIPTION OF THE INVENTION

As used herein, the term &quot; neurotrophin &quot; refers to any naturally occurring member of the neurotrophin family. This includes naturally occurring proteins that share amino acid sequence homology with any known neurotrophin and which retain the six characteristic cysteine residues. (Some of these proteins may be structurally included in the neurotrophin family, however they may exhibit a different biological activity from neurotrophic activity). the term &quot; neurotrophin &quot; refers also to constructed neurotrophins whose amino acid sequences are derived or modeled according to naturally occurring neurotrophins. For example, chimeric neurotrophins comprising amino acid sequences from different neurotrophins (e.g., NGF and NT-3) or from the same neurotrophin from different species (for example, hBNDF and porcine BDNF) are included; neurotrophins whose genetic sequences have point substitutions, mutations by addition or deletion; neurotrophins derived from the pre-pro-sequence (e.g., neurotrophins having the two carboxy-terminal amino acid residues encoded by the codons immediately preceding the codon of?

stop of the native gene (&quot; full-length neurotrophin &quot;); and includes fragments of a neurotrophin which exhibit neurotrophic activity (for example, those whose first six amino acids are altered or deleted). These examples are descriptive and are not intended to limit the definition.

This invention provides processes for the production and recovery of recombinant biologically active neurotrophins. It further provides the homogenous, purified recombinant neurotrophins prepared by these processes. A recombinant protein, as used in this specification, is a protein expressed from a recombinant DNA molecule in a host cell transformed therewith. A recombinant DNA molecule is a hybrid molecule comprising DNA sequences from different sources that have been linked together. Sam-brook et al. (1989, Molecular Clonidine: A Laboratorv Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor) describe many conventional techniques for recombinant DNA technology.

According to this invention, the production of recombinant neurotrophins involves culturing a non-animal host cell transformed with a recombinant DNA molecule having an expression control sequence operably linked to a DNA sequence encoding a neurotrophin. An expression control sequence is operably linked to a DNA sequence encoding a polypeptide when the expression control sequence directs and promotes the transcription and translation of the DNA sequence. Culture of a transformed host cell involves incubation of the cell under culture conditions suitable for cell growth and expression of the DNA sequence.

DNA sequences encoding neurotrophins are available from various sources. The literature describes DNA sequences for the four known human neurotrophins (ie, hNGF (Gray et al., EP 0 121 338, 1984), hBDNF (Hyman et al .. WO 91/03568), hNT-3 (Hohn et al. WO 91/03569) and 74 061 6526-081-118

NT-4 from Xenopus (Hallbook et al., 1991, Neuron 6: 845-858)) and for many non-human neurotrophins. Preferably, these sequences are directed in the construction of oligonucleotide primers suitable for amplification. Genomic libraries are adequate sources of DNA for PCR matrices. Also useful are cDNA libraries from cells known to express neurotrophin mRNA. For example, human fetal brain mRNA cDNA libraries contain sequences for hBDNF. The cDNA and genomic libraries may also be analyzed using any of the methods known in the art to identify and isolate DNA sequences encoding neurotrophins. Alternatively, synthetic or semi-synthetic genes may be constructed using standard DNA synthesizers. No doubt researchers will find DNA sequences encoding new neurotrophins. The processes of this invention will also be useful in the production and recovery of these neurotrophins.

Specifically, the invention generally relates to the production of NT-4 or a derivative or fragment thereof by growing a recombinant bacterium containing a nucleic acid encoding NT-4 or a derivative or fragment thereof under certain conditions, so that the NT-4 or a derivative or fragment thereof are expressed by the bacterium, and to the recovery of the produced NT-4 or derivative or NT-4 fragment. In one preferred embodiment, the NT-4 is human NT-4. In another embodiment, an NT-4 derivative which is a chimeric or fusion protein is produced. In a more preferred embodiment, the produced NT-4, or the NT-4 derivative or fragment, is biologically active, ie, capable of exhibiting one or more of the known NT-4 functional activities, as assayed by any of the known methods in the art or disclosed herein (e.g., in vitro assays of the ability to promote the development in E8 DRG explants of the ability to stimulate CAT activity in motor neuron cultures; see Section 9, infra).

In a specific embodiment of the invention, the sequences encoding human NT-4 are expressed in an expression system 74 061

6526-081-118 -14-

in E. coli and a purification scheme is used as described below to yield useful amounts of human NT-4. The nucleic acid encoding human NT-4 which is thus expressed may be that contained in nucleic acid pRG173 (ATCC Accession Number 75131) or HG7-2 (Accession Number to ATCC 75070) or shown in Figure 11 (SEQ ID NO: NB22) or isolated by any of the methods known in the art, or as follows: The mixtures of 5 'and 3' oligonucleotides representing all possible codons corresponding to the known NT-4 sequences or conserved amino acid sequences of known neurotrophins are used as primers in the polymerase chain reaction (PCR). Primary and secondary PCR amplification reactions of human or other mammalian (or other mammalian) cDNA or genomic libraries result in the isolation of a PCR product that can be used as a 32 P labeled probe to isolate a cDNA or genomic clone from total length encoding NT-4. The term &quot; human neurotrophin-4 &quot; as used herein shall be understood to mean any human Xenopus NT-4 homologue (Hallbook et al., 1991, Neuron 6: 845-858), including a molecule yet distinct homologous neurotrophin (e.g., at least about seventy percent homology). The literature describes a variety of expression control sequences useful for expressing DNA sequences in non-transformed animal hosts. These include, among others, in bacteria, the lac system, the trp system. the TAÇ system, the TRC system. the lambda P2 promoter, the last T7 promoters and the control regions of the coat protein fd; and in yeast, the phosphoglycerate kinase promoter, the Gal 4 promoter, the His promoter, the alcohol dehydrogenase promoter, the alkaline phosphatase promoter and the? -coupling factor promoter. Controllable expression control sequences are preferred, and among these, a temperature-inducible lambda PL promoter, the lacOV5 promoter and the T7 promoter Φ1.1 for expression in E. coli are further preferred. The literature also describes a variety of

, or expression vector / host suitable for bacterial and fungal hosts, including plasmids, bacteriophage, cosmids, and derivatives thereof. Examples of such systems are, for bacteria, col Fl, pCR1, pBR322, pMB9, RP4, lambda phage, M13 and filamentous viruses; and, for yeasts, 2 / i. Preferred plasmids in bacteria are stable in the host cell and have an average copy number of 20-200 copies per cell. Plasmids derived from pBR322, such as those described by Panayotatos (1987, Enoineerinq Efficient Expression System, in Plasmids-A Practical Approach, ed .; Herdy, K., IRL Press, Oxford / Washington, D.C.) were used. In particular, the RPN class plasmids, developed by Regeneron Pharmaceuticals, Inc. and most prominently described below, are preferred. The art is also familiar with many non-animal hosts useful for expressing heterologous proteins, including E. coli. Bacillus. Streptomvces. Saccharomvces and Pichia pastoris. E. coli is preferred. Culture of transformed host cells results in the expression of neurotrophins. Bacterially expressed neurotrophins have been found to be subject to degradation by intracellular proteases, especially those induced as part of the &quot; heat shock &quot; response. to foreign proteins (Goff et al., 1984, Proc Natl Acad Sci USA 81, 6647-6651). Accordingly, it is preferable to use protease-deficient mutants as host cells. Expression of the genes by heat shock is regulated by heat shock regulator genes, such as the E. coli htpR gene. Mutants of the HtpR gene are deficient in the expression of protease genes by heat shock, as well as in other genes that contribute to the response to heat shock. Expression of recombinant neurotrophins in heat shock regulator mutants, especially htpR &quot; mutants, has been found to significantly increase yield. The double mutants HtpR &quot; lon. The strain LC137

This strain is available from Prof. Alfred Goldberg, Harvard University, and U.S. Patent 4,123,826, filed Oct. 31, 1996, which is incorporated by reference herein. 758 512 (Goldberg et al.) Describes other suitable strains. Another preferred E. coli strain is FRY26.

Neurotrophins are toxic to the bacterial cells that express them. Accordingly, and in order to maximize yield, it is preferred to induce the expression of neurotrophins only in dense bacterial cultures, for example, cultures in the latter log phase. An inducible system based on the lambda PL promoter and the temperature sensitive repressor cI857 was used. The production of neurotrophins was typically induced by changing the growth temperature to 42øC for 30 minutes and continuing the incubation for 3-20 hours at 380-42øC. Induction of expression for more than 16 hours in actively growing cells eventually causes cell death. The inability of E. coli to accumulate neurotrophins may result from one or more neurotrophin gene or protein properties. The structure of the neurotrophin mRNA, particularly the structure close to the translation start point, may prevent efficient translation. Alternatively, neurotrophin may prevent its own synthesis by direct interaction with its mRNA, or may interact directly or indirectly with some component of the DNA replication / transcription / translation mechanism of E. coli.

In accordance with a further embodiment of the present invention, neurotrophin was introduced into the periplasmic space of E. coli instead of expressing it intracellularly. This was accomplished by fusing a signal sequence to a mature neurotrophin. A signal sequence gene fused to the 5 'end of the neurotrophin gene can provide a nucleotide sequence near the translation start point that most contributes to efficient translation, resulting in high levels of neurotrophin accumulation. In addition, sequestration of neurotrophin into the periplasmic space prevents it from interfering with any cytosolic component required to

synthesis of proteins. It also protects it from attack by cytosolic proteases. It is also possible that the secretion of a mature neurotrophin in the perlplasmic space may provide a medium that most contributes to the adequate folding of the protein.

A signal sequence is provided by the construction of a recombinant DNA molecule in which the neurotrophin-encoding DNA sequence comprises, from 5 'to 3', a fused gene encoding a signal or control sequence suitable for the host cell which is in-frame with a DNA sequence for neurotrophin. The literature describes various signal sequences useful in such constructs. For example, LamB, OmpA and PhoA are useful in E. coli. (Denèfle et al., 1985, Gene 85: 499-510; Wong et al., 1988, Gene 68: 193-203). LamB and, in particular, the modified LamB signal sequences are preferred which increase the translation efficiency of LamB mRNA. We have constructed genes for the modified LamB signal sequences as follows. Degenerate substitutions on the third nucleotide of several LamB codons were prepared by substituting G or C for A or T. These substitutions do not alter the amino acid sequence of the LamB signal peptide but decrease the potential number of hydrogen bonds in any structure secondary education. This reduces the stability of possible secondary structures involving this LamB mRNA region. Codon alterations were also introduced based on codon usage models to further approximate the codons most frequently used by E. coli.

The LamB signal sequence was also modified to enhance the processing efficiency of the LamB precursor protein in mature protein. The native LamB has a hydrophobic core of 10 amino acid residues. Mutational analysis of various E. coli signal sequences suggests that the length of the hydrophobic core region may have a strong effect on the activity of the signal sequence. It has been found that increasing the length of the hydrophobic region by the addition of up to ten hydrophobic amino acid residues enhances the processing efficiency of the LamB fusion precursor polypeptides. less than six is preferred, and still more preferably four. The choice of the added hydrophobic amlno acids is not critical, nor is the critical location to which they are added to the hydrophobic core region critical. However, it is preferred to add the Leu-Ala-Val-Leu tetrapeptide (&quot; LAVL '&quot;) (SEQ ID No 6) at a suitable restriction site near the N-terminal end of the hydrophobic region. Particular genes for the modified LamB signal sequences are described in Example 7.

According to the invention, the recombinant neurotrophins are released from the host cell culture by collection of the cells, lysis, lysate centrifugation and collection of the lysate pellet. Cells are typically harvested by centrifugation. To inhibit the degradation of neurotrophin upon collection of the cells, these are preferably resuspended in 50 mM EDTA. The cells are then used directly or frozen for further use. The art is familiar with many cell lysis techniques including enzymic (e.g., lysozyme), chemistry (e.g., alkali, SDS) and mechanics (e.g., French Press, hydrodynamic cut). Mechanical means are preferred because there is less risk of injury to neurotrophins. It is preferred to lysate the cells by passing them through a French press or a STANSTED® cell crusher at 68,948 kPa. The lysate pellet is obtained by centrifugation at about 15,000 x g.

This invention further provides methods for recovering recombinant neurotrophins produced in cell cultures. These processes overcome the problems associated with conventional recombinant protein recovery processes that have been applied to neurotrophins.

Native neurotrophins are soluble in neutral buffers. However, recombinant neurotrophins from E. coli behave as insoluble proteins. Recombinant neurotrophins were detected in the cytosolic fraction and, when exported, have been recovered from the periplasmic space using standard techniques such as osmotic shock, spheroplasty or

saw?. (Bochner et al. U.S. Patent 4,680,262). However, they are only isolable as a small fraction of the total neurotrophins in the cell.

Many of the mammalian proteins expressed in bacteria are insoluble. When produced in a foreign ionic medium, they do not bend properly and expose normally concealed hydrophobic regions. Accordingly, the proteins form aggregates and precipitate as inclusion bodies. The literature suggests that the cysteine residues of these proteins are at least partially reduced or mismatched (Tsuji et al., 1987, Biochemistry 26: 3129-3134).

The literature describes standard techniques for purifying recombinant proteins from inclusion bodies (see, e.g., Builder et al., U.S. Patent No. 4,620,948; Hershenson et al., U.S. Patent 4,961,969; and Hung et al. 734-362). These techniques involve the dissolution of the protein in a solution comprising a strong denaturant and a reducing agent. After exchanging the strong denaturant for a weak denaturant, the proteins are renatured in a neutral solution and oxidized to form the correct disulfide bonds. Most of the recombinant neurotrophins produced by these processes are biologically inactive. This is thought to be due, in part, to the complexity of the tertiary structure of neurotrophins. After denaturation of the protein and reduction of the sulfhydryl groups, the polypeptide does not fold back into the configuration suitable for the correct disulfide bond. Furthermore, these results suggest that they are present in the recombinant neurotrophin aggregates which are not typical inclusion bodies.

A process for recovering recombinant biologically active neurotrophins from the insoluble fraction of host cells has been found. The process essentially involves the solubilization of the neurotrophin in a strong denaturing agent while maintaining the correct state of 74 061 6526-081-118-Ttar of. Oxidation of the protein by avoiding the use of reducing agents. Reduction of the disulfide bonds during purification destroys the activity of purified neurotrophin polypeptides. This is an unexpected result because the literature (for example, Tsuji et al., 1987, Biochemistrv 26: 3129-3134) teaches that recombinant proteins insoluble in inclusion bodies should have incorrectly formed or reduced disulfide bonds.

For the neurotrophins to be still recovered, the strong denaturing agent is replaced with a weak one, however, it has also been found that the addition of a basic amino acid, such as histidine, or an equivalent thereof, helps to keep the neurotrophin soluble during the process of renaturation accompanying the removal of the weak denaturant (Prior et al., 1990, WO 90/06764).

These findings suggest a new model for the behavior of recombinant neurotrophins. It is believed that most of the recombinant neurotrophin polypeptides produced by the host cell are folded properly and form the correct disulfide bonds. Furthermore, since neurotrophins have a high pH and are positively charged at neutral pH, they are thought to form associations with negatively charged molecules in the cell, for example DNA, RNA and other proteins. Consequently, they become insoluble at neutral pH. Basic amino acids help in their dissociation.

More specifically, the present process for the recovery of recombinant neurotrophins produced by a host cell culture comprises the denaturing step of neurotrophin by dissolving it in a solution comprising a strong denaturing agent, which solution is also essentially free of reducing agents. The term &quot; denaturing agents &quot; as used herein refers to compounds which, in aqueous solution, reversibly deplete dissolved proteins by at least partially eliminating the tertiary and secondary structure through the disruption of hydrogen or the like.

alteration of the thermodynamic medium of the protein. Strong denaturation solutions include guanidinium salts (for example, guanidinium hydrochloride) and alkali metal thiocyanates (for example, sodium thiocyanate) at concentrations of 4 M - 9 M and 7 M - 9 M urea. Preferred denaturation values are 7 M-9 M guanidinium HCl. Preferred conditions are pH 7.0-pH 9.0 and room temperature. The guanidinium 8 M HCl, pH 8.0 is preferred.

The correct oxidation state of the sulfur atoms in the cysteine residues should also be maintained during this step or we risk losing the ability to recover any biologically active neurotrophin. Consequently, the solution must be essentially free of reducing agents. A solution essentially free of reducing agents is one in which the disulfide bonds of the protein are maintained. Even the addition of small amounts of disulfide reducing agents, such as β-mercaptoethanol or dithiothreitol, during the practice of this invention will destroy the activity of the purified neurotrophin.

Recombinant neurotrophins are apparently subjected to degradation by metalloproteinases. Accordingly, neurotrophins are preferably recovered in solutions comprising metalloproteinase inhibitors. Heavy metal chelators, such as EDTA, are preferred. The concentration of EDTA may reach at least at least 1 mM and at most about 200 mM. Concentrations of 5 mM-80 mM are preferred and more preferred than 50 mM. After the chelated heavy metal ions have been dialyzed from the solution, the EDTA can be reduced or eliminated. Langley et al., 1990, EP 0 398 753 describes peptides useful as inhibitors of metalloproteinase. The recovery process may also comprise one or more of the following steps. Upon solubilization of the neurotrophin, the strong denaturing agent is exchanged in the solution for a weak denaturing agent. Weak denaturation solutions include 4 M-9 M urea, and preferably 6 M-8 M, pH 7.0

74 061 6526-081-118 - ρ 9.0, at room temperature. The most preferable weak denaturation solution is that of 7 M urea, 50 mM Tris-HCl, 10 mM NaCl, 5 mM EDTA, pH 8.0. Dialysis is the preferred process of exchange. If the strong denaturing agent is 7.0 M urea -9.0 Μ, the weak denaturing agent of this step is preferably lower concentration urea.

The weak denaturing agent is removed from the solution so as to renature the neurotrophin. Dialysis or diafiltration against a non-denaturing solution is the preferred method of removal. Preferred is 0 M NaCl at 1M as non-denaturing solution.

Another step in the recovery process is the purification of the recombinant neurotrophin from other contaminants in the solution. Any of the typical protein isolation techniques known in the art may be used. A three-step isolation involving the cation exchange in S-SEPHAROSE ™, the anion exchange in DEAE-SEPHAROSE ™ and reverse phase high pressure liquid chromatography (HPLC) is preferred. The isolation step may commence at any stage after solubilization of the neurotrophin and may continue through the denaturing and removal agent exchange steps. Purification is preferably initiated in the urea phase because this agent does not interfere with ion exchange chromatography and because partial purification makes the dissolution of neurotrophin in the non-denaturing solution easier.

It has been found that, unless the neurotrophin is relatively purified, it will precipitate out of the solution when the weak denaturing agent is removed. The solubility depends on the degree of purity of the neurotrophin as well as the character of the other contaminants in the solution. For example, negatively charged molecules such as DNA will interfere with the solubility even of very pure neurotrophins. Accordingly, the solubility of the neurotrophin in the non-denaturing solution is maintained by adjusting the solution to comprise a basic amino acid or an equivalent, such as 74 061 6526-081-118

Indole acetic acid. The concentration of this species should be effective in keeping the neurotrophin in solution. Basic amino acid solutions include histidine, lysine and arginine or their salts at concentrations above 10 mM. Histidine is useful at concentrations of 20 mM - 500 mM. The histidine is more preferred in concentrations of 20 mM-10 mM. If the neurotrophin in the denaturation solution has been sufficiently purified prior to the removal of the denaturant, this step can be eliminated.

The results indicate that the present process produces uniform neurotrophin molecules having the same N-terminal amino acid. In contrast, the expression of neurotrophins in the mammalian CHO cell system has been found to produce a mixture of neurotrophin molecules with several N-terminal amino acids. Accordingly, the recombinant neurotrophins produced by the instant process appear to be unique.

The processes of this invention are also useful in the recovery of other proteins having biochemical properties similar to neurotrophins. That is, the processes are useful in recovering proteins that dock correctly in bacteria and which form the appropriate disulfide bonds, but, once denatured and reduced in conventional recovery techniques, are very difficult to renature with suitable disulfide bonds. This includes proteins having a pH greater than about 9.0 and having at least two disulfide bonds. (Applicant molecules include the secretory leukocyte protease inhibitor (Miller et al., 1989, J. Bacterioloq. 171: 2166-2172) and full length recombinant CD4 (Fisher et al., 1989 WO 89101940)).

As described hereinbelow in an example section infra. the present invention describes the expression of biologically active human neurotrophin-4. The human NT-4 DNA sequence was subcloned into the plasmid vector of pRG91 DNA, resulting in pRG173. This plasmid containing hNT-4 was transformed into E. coli strain RFJ26. and the processes described in the present 74 061 6526-081-118

-24-

4 specification were used to retrieve

biologically active, from the culture system. However, applicants should not be limited to such a specific embodiment. For example, any nucleic acid sequence substantially homologous to the HG7-2 region encoding human NT-4 may be used to construct any number of DNA plasmid expression vectors as described in the specification or known to the person skilled in the art, which on the other hand can be used to transform any number of E. coli bacterial strains so as to produce useful amounts of biologically active NT-4.

The following examples are presented so that this invention can be better understood. These examples are for the purpose of illustration only and are not construed to limit in any way the scope of the invention.

r. (Ibanez et al., 1990, EMBO J. 9: 1477-1483). In addition, PAN-20 was designed to decrease the G + C content of this region of the NGF gene so as not to affect the sequence of the resulting protein. The amplified DNA fragment was digested with the restriction endonucleases SalI and EaqI and ligated between the SalI and EagI sites of pRPN50 (Regeneron Pharmaceuticals, Inc.) resulting in the plasmid pRPN102. PRPN50 is derived from the expression plasmid pNKS97 (Panayotatos, 1987, Engineering an Efficient Expression System, in: Plasmids-A Praticai Approach, Herdy, K., IRL Press, Oxford / Washington, DC) by insertion of the lambda PL promoter into the promoter insertion site. The lambda repressor gene P? CI857 was incorporated in pRPN102 to facilitate repression at 30 ° C and to allow thermal shock derepression at 42 ° C. This DNA sequence was amplified by PCR using the primers EVD-26 (57-CCATTATCGC GACCAGAGGT-37) (SEQ ID NB9) and EVD-27 (57-TCTTGCTCGC GAGTTATCAG CTATGCG-37) (SEQ ID No 10) which generate a restriction site NruI at each end. This 821 bp fragment extends 70 bp upstream of the cI857 coding sequence in a region containing the constitutive promoter PRM as well as 38 bp beyond the cI857 termination codon. The cI857 PCR amplified fragment was digested with NruI and ligated into the NruI site of pRPN102. Candidates for this ligation were analyzed to find a plasmid with the cl857 insert that has the same transcription direction as the hNGF gene, resulting in the plasmid pRPN133 (Fig. 1). Plasmid pRPN133 was deposited with the ATCC and was assigned access NA 75029.

The E. coli mutant strain was transformed. LC137, with RPN133 to give pRPN133 / LC137.

400 ml of LB medium supplemented with 100 mg / l ampicillin was seeded with pRPN133 / LCl37 and incubated at 28 ° C with shaking in a 2 liter flask. In the last phase of growth

(OD590 = 1.0-2.0 under these conditions), the synthesis of hNGF was inhibited by addition to the overnight culture of 400 ml of pre-Lb medium -heated at 55 ° C and incubation was continued with stirring at 42 ° C for 5 hours.

Cells were harvested by centrifugation and cell pellets stored at -20 ° C. Cell pellets (1.0-2.5 g) were thawed, resuspended in 20.0 ml of buffer A (100 mM Tris-HCl, 50 mM EDTA, pH 8.0), passed through a cell crusher STANSTED® at 68 948 kPa and centrifuged at 23,000 xg for 15 minutes at 4 ° C. The pellet was washed twice with 10.0 ml of 2M guanidinium-HCl, 5 mM EDTA, pH 8.0.

The pellet was resuspended in 40 ml of 8 M guanidinium-HCl, 5 mM EDTA, pH 8.0, using a Potter homogenizer, to solubilize the recombinant hNGF. The suspension was then centrifuged at 23,000 xg for 15 minutes. The supernatant was dialyzed against 2 liters of buffer B (7 M urea, 100 mM histidine, 0.1 mM EDTA, pH 6.0) with two changes of buffer, at 25 ° C for 20 hours. The dialysate was centrifuged at 23,000 xg for 15 minutes at 4 ° C. The supernatant was loaded onto a S-SEPHAROSE® column (2.5 cm x 6.0 cm d x h) equilibrated in buffer B and washed to baseline absorbance at 280 nm with the same buffer. Proteins were eluted with a gradient of 300 ml of 0.0 M NaCl 1.0 M in buffer B. The fractions containing hNGF were pooled.

The pooled fractions were dialyzed against a 100 fold excess volume of buffer C (7 M urea, 50 mM Tris-HCl, 0.1 mM EDTA, pH 8.5) at 25 ° C for 20 hours.

The dialysate was loaded onto a DEAE-SEPHAROSE column (2.5 x 7.5 cm d x h) equilibrated in buffer C at a flow rate of 2.0 ml per minute. Fractions passed through the hNGF-containing column were pooled and dialyzed twice against 40 volumes of 100 mM histidine, 0.1 mM EDTA, pH 6.0 at 4 ° C overnight. O 74 061 6526-081-118 -27-

&amp; &amp; The dialysate was centrifuged at 23,000 x g for 15 minutes at 4 ° C. The dialysate was then applied to a TOYOPEARL CM 6505 (1.0 x 5.0 cm) (weak cation-exchange) column equilibrated with 100 mM histidine, 0.1 mM EDTA, pH 6.0 and eluted with a gradient of NaCl from 0.0 M to 1.0 M. The fractions containing the hNGF were pooled and filtered through a MILLIPORE GV® filter and stored at 4 ° C. The amino-terminal sequence of the purified protein was confirmed by direct sequencing. The biological activity of the purified protein was tested for the development of neuritis in explanted E8 and dissociated dorsal root ganglia (Fig 2A and 2B) (Lindsay et al., 1985, Dev.Biol., 112: 319-328). Following these criteria, recombinant hNGF purified from E. coli was shown to be as active as NGF purified from the mouse salivary gland by this process.

7. EXAMPLE: PRODUCTION AND RECOVERY OF RECOMBINANT hBDNF AND E. coli hBDNFmvc

7.1. LAMB BASED SIGNAL SEQUENCES

LamB-based signal sequences were constructed to facilitate both the synthesis and secretion of neurotrophins in E. coli. The LamB signal sequence (Fig. 3A, SEQ ID Nal) is a naturally occurring E. coli signal sequence which has been selected for the construction of a series of secretory vectors based on the pRPN series of developed expression vectors in Regeneron Pharmaceuticals, Inc. Secretion vectors were constructed by cloning synthetic DNA fragments encoding the LamB signal sequence into pRPN09 or pRPN16. These plasmids are derived from the expression vector pNKS97 (Panayotatos, 1987, Engineering an Efficient Expression System, in: Plasmids-A Practical Approach, ed. Herdy, K., IRL Press, Oxford / Washington, DC) in which a lacUV5 promoter . LamB was inserted into the insertion site of the structural gene so that the expression of LamB is under the control of the lacUV5 or T7 promoter §1.1. (I.e.

The synthetic DNA fragment, LamB1 (Fig. 3B, SEQ ID NO: 2), encodes 25 amino acids identical to the wild-type LamB signal sequence. However, the nucleotide sequence was altered relative to the wild-type nucleotide sequence in order to construct unique restriction enzyme sites and to maximize translational efficiency. LamB1 also has seven degenerate nucleotide substitutions, C or G replacing A or T. This reduces the stability of the possible secondary structure of the mRNA. LamB1 also has several novel restriction sites.

LamB2 (Fig. 3C, SEQ ID No 3) was also created as LamB1 modification. LamB2 3 'nucleotides were made to prepare LamB2 by PCR amplification. These changes introduced an Eaq1 restriction site which facilitates the cloning of blunt-ended DNA fragments. The fusion of the mature hBDNF to LamB1 results in the efficient synthesis of the fusion protein in E. coli. The authenticity of the synthesized product was confirmed by selective synthesis in a DNA-dependent coupled transcription-translation cell-free protein synthesis system, by the selective synthesis of the product using a T7 RNA polymerase expression system in E. coli. and by synthesizing the product with a mvc marker at the C-terminus that allows the chimera to be identified with a myc-specific monoclonal antibody. Any of these fusion proteins synthesized in E. coli was processed to mature hBDNF as evidenced by their mobility on SDS-PAGE. The level of expression obtained with LamB1 or LamB2 results in an accumulation of hBDNF of about 1% to 10% of the total cellular protein.

LamBl was modified to increase processing efficiency. Four amino acids (Leu-Ala-Val-Leu or LAVL) were inserted into the Nhel site of LamB1 to give LamB3 (Fig. 3D, SEQ ID Na4). The same insert in LamB2 was performed to obtain LamB4 (Fig. 3E, SEQ ID Ns5). This insertion results in the extension of the hydrophobic core region of LamB1 and LamB2 from 10 to 14

Amino acids. It results in a 5-fold increase in the processing speed of the hBDNF fusion protein in mature hBDNF (Fig. 4).

The LamB3- and LamB4-hBDNF molecules that are exported to the periplasm are processed into mature hBDNF faster than the wild-type LamB fusions. However, fewer molecules are exported, so the net amount of mature hBDNF, in this case, is not increased. In any case the increased translocation efficiency of LamB3 and LamB4 should result in increased yields of other proteins. LamBl The signal sequence fragment was constructed as two complementary synthetic oligonucleotides (LamB80, 5'-ATGATCACAC TGCGTAAGCT TCCGCCTAGCT GTAGCAGTAG CAGCAGGTGT AATGTCTGCA CAGGCCATGG CCCGGGATCC-3 '(SEQ ID NB11) and LamB88, 5'-CTAGGGATCC CGGGCCATGG CCTGTGCAGA CATTACACCT GCTGCTACTG CTACAGCTAG CGGAAGCTTA CGCAGTGTGA TCATCATG-3 '(SEQ ID NO: 12)), designed to produce, after denaturation / renaturation, protruding ends corresponding to those of the Spel and SphI restriction sites. This DNA fragment was ligated into the SphI and Spel restriction sites of pRPN16 (Regeneron Pharmaceuticals, Inc.) resulting in the plasmid pRPN52.

This plasmid was subsequently modified, in order to create additional restriction sites, by insertion of a synthetic DNA fragment, the denatured / renaturated product consisting of the LamB2A oligonucleotides (5'-CATGGCCAGT CGGCCGAG-3 '(SEQ ID N813)) and LamB2B (5'-GATCCTCGGC cgactggc-3 '(SEQ ID N814)), between the unique NcoI and BamHI restriction sites in the LamB1 signal sequence in pRPN52. The resulting plasmid, pRPN88, contains the LamB2 signal sequence with unique restriction sites in the signal peptidase recognition sequence to facilitate the fusion of the signal sequence to any other DNA sequence. The LamB2 signal sequence in pRPN88 is under transcriptional control of the lacUV5 promoter or the T7 promoter Φ1.1 and translation control of the T7 Φ1.1 ribosome binding site. 74 061 / Λ-k; '6526-081-118 μ ·. -30-,

7.2. PRODUCTION AND RECOVERY OF RECOMBINANT hBDNFmvc Human BDNFmyc is a protein comprising, from the N-terminus to the C-terminus, mature hBDNF fused to an antigenic marker. The label is a peptide having the amino acid sequence EQKLISEEDL (SEQ ID NB15). These ten amino acid residues are derived from the human c-myc proto-oncogene. Antibodies recognizing this marker are useful in identifying hBDNFmyc in a sample (Evan et al., Mol. Cell Biol., 5, 3610-3616; see also, S. Squinto et al. &Quot; Assay Systems for Detecting Neurotrophic Activity &quot;; U.S. Patent Application 07 / 5321,283, incorporated herein by reference). The hBDNFmyc DNA sequence was amplified by PCR from pCDM8-hBDNFmyc (Regeneron Pharmaceuticals, Inc.) using the oligonucleotide primers N8-hBDNF (5'-CCCACTCTGA CCCTGCCCGC CGAGGG-3 * (SEQ ID No 16)) and C2-hBDNFmyc (5'-GCTATGCGGC CGCTACAGAT CCTCCTC-3 '(SEQ ID No 17)) to construct a fusion protein comprising the LamB2 signal sequence and the hBDNFmyc protein. The amplified DNA fragment was digested with EaqI and cloned into the Bali and Eacr sites of the LamB2 sequence in pRPN88, resulting in the plasmid pRPN98.

One can also produce a DNA sequence encoding hBDNFmyc by fusing a DNA sequence encoding mature hBDNF to another encoding the ten amino acid myc tag. DNA sequences encoding mature hBDNF are isolated by PCR amplification as described. (Hyman et al., 1991 WO 091/01568). A double-stranded DNA sequence encoding the ten amino acid peptide myc tag can be chemically synthesized. For the purpose of this Example, the hybrid hBDNFmyc DNA sequence is then provided with the restriction sites Bali and Eaql at the 5 'and 3' ends, respectively.

It should be noted that there are two Bali sites in pRPN88 but the site in the sequences derived from pBR322 is approximately 50-fold less sensitive to Bali cleavage than the LamB2 signal sequence site to the demethylation of dcm (New England Biolabs). In pRPN98 the LamB2 signal sequence is fused to the mature part of the hBDNFmyc protein such that cleavage in the signal peptidase recognition sequence should yield the hBDNFmyc protein which leaves the histidine residue at +1 with respect to the processing site of the -protein (Lei-brock, 1989, Nature 341: 149-152).

DNA sequences between the unique NruI and PvuII sites in pRPN98 were deleted to remove sequences that control the copy number of the plasmid (Twigg and Sherratt, 1980, Nature 283: 216-218). The resulting plasmid, pRPN121 (Fig. 5), has a number of copies about 3 times greater than the upstream plasmid. Plasmid pRPN121 was deposited with the ATCC and assigned accession NB 75028.

It has been transformed into W3110 I ^ F &quot; of E. coli with pRPN121 to produce W3110 1SF / PRPN121.

500 ml of LB medium was inoculated with W3110 I <3F_ / pRPN121 and allowed to grow to the last log phase with stirring at 37Â ° C (OD590 of approximately 1.0), whereupon lactose was added to a final concentration of 1% (w / v) and the culture was aerated overnight at 37 ° C. Cells were harvested by centrifugation, washed once in 0.2 M Tris-HCl, pH 8.0 and pellet frozen at -70øC. The cell pellet (approximately 5 g) was thawed and resuspended in 20 ml of 0.2 M Tris-HCl, pH 8.0, 1 mM CaCl 2 and 25 units of micrococcal nuclease (Boehringer Mannheim). The cell suspension was passed through a French pressure cell at 55 158 kPa and then centrifuged at 15,000 rpm for 15 minutes in a SA600 rotor at 4 ° C. The pellet was resuspended in 30 ml of 0.2 M Tris-HCl, pH 8.0, 10 mM EDTA, 2% Triton X-100 and gently stirred at room temperature for one hour and then centrifuged at 15,000 rpm for 15 minutes. minutes in an SA600R rotor at 4 ° C. The pellet was washed twice in 20 ml guanidinium-HCl 2 Μ. The ball was resuspended in 10

8 mM HCl, 10 mM Tris-HCl, pH 8.5, 10 mM NaCl, 1 mM EDTA using a Potter homogenizer. The extract was raised to 20 ml with the same buffer.

The extract was dialyzed overnight at room temperature against 100 X 7 M urea volume, 50 mM Tris-HCl, pH 8.5, 10 mM NaCl, 1 mM EDTA.

The dialysate was loaded onto a DEAE ZETA-PREP ™ (Cuno, Inc., Meriden, CT) disc equilibrated in 7 M urea, 50 mM Tris-HCl, pH 8.5, 1 mM EDTA, at a flow rate of 3 ml / min and washed until it reaches the baseline with the same buffer.

Fractions passed through the disk were raised to 50 mM histidine, pH 5.0, and then dialyzed overnight at 4 ° C against 100 X 7 M urea volume, 50 mM histidine, pH 5.0, 1 mM EDTA. The dialysate was loaded onto a 1.6 cm x 6.5 cm column of S-SEPHAROSE ™ equilibrated with 7 M urea, 50 mM histidine, pH 5.0, 1 mM EDTA, at a flow rate of 1 ml / minute, and washed until reaching the baseline with the same buffer. Proteins were eluted with a gradient of 0.0 M - 1.0 M NaCl in 200 ml. Fractions containing hBDNFmyc (Fig. 6) were collected.

Fractions containing hBDNFmyc were dialyzed against 200 X 50 mM histidine volume, pH 5.0, 50 mM NaCl, 1 mM EDTA. The dialysate was loaded onto a 1.6 cm x 1.5 cm S-SEPHAROS® column equilibrated with 50 mM histidine, pH 5.0, 50 mM NaCl, 1 mM EDTA, at a flow rate of 1 ml / minute and washed until it reaches the baseline with the same buffer. Proteins were eluted with a gradient of 0.0 M - 1.0 M NaCl in 200 ml. Fractions containing hBDNFmyc were collected and pooled. The hBDNFmyc of this sample (approximately 85% pure) was directly applied to a 0.45 cm x 5.0 cm C4 reverse phase colony (VYDAC ®) and eluted with a 0-67% acetonitrile gradient in 0, 1% in 80 ml at a flow rate of 0.75 ml / minute. The peak fraction had a purity greater than 95% (Fig. 7). 74 061 6526-081-118 -33- '

The N-terminal amino acid sequence analysis confirmed that the purified protein is the authentic hBDNFmyc with a homogeneous N-terminus (Leibrock, 1989, Nature 341: 149-1521. The purified hBDNFmyc was biologically active in promoting development of neuritis in dorsal root ganglia (DRG) explants and in nodal ganglion explants of E8 chicken embryos, as well as in promoting the survival and dendritic development of dissociated E8 chicken DRG neurons (Fig. is comparable to that of recombinant human BDNF purified from mammalian cell extracts in DRG explant assays. The tested material corresponds to Fig. 7, lane 4. 7.3 PRODUCTION AND RECOVERY OF RECOMBINANT hBDNF The process described for the production and recovery of hBDNFmyc was also used to produce and recover mature, recombinant total length hBDNF, and gave a protein with similar bioactivity (Fi The construction of the plasmid pRPN149 (Fig. 10), which expresses a LamB1-hBDNF fusion protein, is analogous to the construction of pRPN121. The synthetic LamBl DNA fragment described above was cloned into the SphI and Spel restriction sites of pRPN09 (Regeneron Pharmaceuticals, Xnc) resulting in the plasmid pRPN31. The mature hBDNF DNA sequence was amplified by PCR from pCDM8-hBDNF (Regeneron Pharmaceuticals, Inc.) using oligonucleotide primers N1-hBDNF (5'-ACTCTGACCC TGCCCGCCGA GGGGAGCTG-3 ') (SEQ ID No 18) and C1-hBDNF (5'-GCGCGGATCC CTATCTTCCC CTTTTAATGG TCAATGTAC-3 ') (SEQ ID Ns19). This DNA fragment was cloned into SmaI and BamHI sites of pRPN31 resulting in the plasmid pRPN34. The HindIII-BamHI fragment including the LamB1-hBDNF fusion gene was subsequently cloned into the HindIII-BamHI restriction sites of pRPN52 (described above) resulting in the plasmid pRPN66. The NruI-PvulI deletion of the pBR322 sequences resulting in a larger number of copies (described above) was performed on pRPN66 resulting in the plasmid pRPN149. In this plasmid, the expression of LamBl-hBDNF is under the control of 74 061 6526-081-118 34 and

lacUVS and the T7 promoter §1.1 and the ribosome binding site. Plasmid pRPN149 was deposited with the ATCC and assigned accession NA 75027.

E. coli i3f "W3110 was transformed with pRPN149. Recombinant hBDNF was produced and recovered by the procedure described in Example 7.

8. EXAMPLE: PRODUCTION AND RECOVERY OF RECOMBINANT hNT-3 Human NT-3 is produced and recovered by procedures similar to those described for hNGF and hBDNF.

A DNA sequence encoding mature full-length hNT-3 is amplified by PCR from a cDNA library of human brain cells (Hohn et al., 1991, WO 91/03569, incorporated herein by reference), use oligonucleotide primers EVD-45 (5'-CCTATGCAGA GCATAAGAGT CACCGAGGA-3 ') (SEQ ID Na20) and EVD-7 (57-GTAAGGGCGG CCGAAGTTTA ATAAATAAAG GTC-3') (SEQ ID Na21). These primers are derived from the rat NT-3 DNA sequence (Maisonpierre et al., Science 247: 1446-1451). The sense primer is almost identical to the human NT-3 sequence. The antisense primer hybridizes to approximately one hundred base pairs downstream of the stop codon of the human gene.

This DNA fragment has a C-terminal EagI restriction site suitable for insertion into Bali and Eagl sites of pRPN88, where it will replace the hBDNF DNA sequence. The resulting plasmid is used to transform E. coli 3110. Recombinant hNT-3 is then produced and recovered as in Example 7. It is expected that recombinant hNT-3 purified from E. coli exhibits neurotrophin activity similar to that described by Hohn et al., 1991, ( WO 91/03569).

9. EXAMPLE: PRODUCTION AND RECOVERY OF RECOMBINANT HNT-4

E. COLI 9.1. The DNA sequence encoding the putative mature region of the human NT-4 gene (hNT-4) was amplified by PCR from the DNA HG7-2 of NT-4, using the oligonucleotides N1-NT4 (5 7-CCGGGGTCTCTGAAACTGCACCAGCGAGTCG-37) [SEQ ID NO: 23] and C1-NT4 (5 7-GGTGCAGTTTCAGAGCATACGCCGGCTGCGGTTGGC-37) [SEQ ID NO: 24] as primers. The C1-NT4 oligonucleotide produces an EagI restriction site at 37 relative to the NT-4 gene. The fragment produced by PCR was digested with EagI and cloned into Mscl-Eagl digested pRG91. The resulting plasmid, pRG173, consists of the LamB signal sequence fused to the mature region of hNT-4 (the glycine at amino acid residue 81 in the translated sequence HG7-2) under transcriptional control of the lacUV5 and T7 01.1 promoters and translation control of the T7 ribosome binding site 01.1. This plasmid also has the ropl deletion which increases the number of copies of the plasmid. Construction was confirmed by restriction enzyme analysis of purified plasmid DNA, purified plasmid DNA sequence analysis, in vitro synthesis of a protein of approximate size that is expected to be encoded by pRG173 and in vivo synthesis of a protein of the size which is expected to be encoded by pRG173 and which has neurite development stimulating activity (see discussion, supra) and which has the appropriate N-terminus as determined by amino acid sequencing (GVSETAPAE [SEQ ID Na25]).

9.2. PURIFICATION OF HUMAN NT-4 EXPRESSED IN E. COLI'S RFJ26 / PRG173

It was allowed to grow overnight in LB medium, supplemented with 100 mg / l ampicillin at 37øC, with aeration, a 5 ml culture of pRG173 transformed into the E. coli strain RFJ26. The culture overnight was diluted in 500 ml of LB and allowed to grow to OD590 = 5.3 at which point the expression of hNT-4 was induced by the addition of 1% (w / v) lactose and the culture was allowed to grow overnight with aeration. Cells were then collected by centrifugation and frozen at -70 ° C. The cell pellet (7.2 grams) was thawed and resuspended in 73 ml of 200 mM Tris-HCl, pH 8.0, 50 mM EDTA and lysed by 3 sequential passes through the STANSTEE® cell crusher. 74 061 6526-081-118 36

The lysate was then centrifuged at 26,000 x g for 10 minutes, yielding 83 ml of supernatant containing the soluble fraction. The soluble fraction was dialyzed against 50 mM Tris-HCl, pH 8.5 at 4øC for 5 h and then diluted 10-fold with 20 mM MES, pH 6.0, and loaded onto a Fast S-Sepharose column equilibrated with 20 mM MES , pH 6.0 and eluted with 1 M NaCl in 20 mM MES, pH 6.0. Recombinant human NT-4 protein which stimulated the development of E8 DRG was recovered in the wash with 1 M NaCl. The insoluble fraction was resuspended and homogenized in 83 ml of 8 M guanidinium HCl, 50 mM Tris-HCl, pH 8.5 , 10 mM NaCl, 1 mM EDTA. The insoluble fraction was dialyzed against 7 M urea, 50 mM Tris-HCl, pH 8.5, and loaded onto a Fast DEAE-Sepharose column equilibrated in 7 M urea, 50 mM Tris-HCl pH 8.5. Separated fractions were collected and dialyzed against 7 M urea, 100 mM histidine, pH 5.5, 1 mM EDTA and loaded onto a Fast S-Sepharose column equilibrated in 7 M urea, 100 mM histidine, pH 5.5, 1 mM EDTA , and hNT-4 eluted with a 0-1M NaCl gradient in the same buffer. Fractions containing hNT-4 were pooled and dialyzed against 100 mM histidine, pH 5.5, 1 mM EDTA. Approximately 95% of the hNT-4 protein was fractionated with the insoluble material.

9.3. PREPARATION OF ENRICHED CROPS IN NEURON MOTORS

Embryos (E14) from Sprague-Dawley rats (HSD or Zivic-Miller) were used in these experiments. Pregnant rats were sacrificed by carbon dioxide asphyxiation and the embryos were quickly removed and placed in an ice-cold medium for further dissection. Aseptically the spinal cord was removed from rat embryos at 14 days' gestation. The spinal cord was cut from the tail to the bulb (at the level of the first dorsal root ganglion), free of sensitive ganglia and adherent meninges. The medulla was then sub-divided into ventral and mid-dorsal segments for separate cultures. The tissues of the ventral spinal cord were cut into small pieces and incubated in 0.1% trypsin (GIBCO) and

(Sigma) Type 1 deoxyribonuclease (Sigma) in PBS at 37 ° C for 20 minutes. The trypsin solution was then removed, washed and replaced with medium consisting of 45% minimum essential Eagle medium (MEM), nutrient mixture of 45% HamF12 (F12), 5% heat inactivated calf fetal serum, (GIBCO), 5% heat-inactivated horse serum (GIBCO), glutamine (2 μM), penicillin G (0.5 μg / ml) and streptomycin (0.5 μg / ml). The tissue was then mechanically dissociated by gentle milling through a Pasteur pipette and the supernatants were pooled and filtered through a nylon (Nitex, Tetko; 40 μl) fiber. The filtered cell suspension was then subjected to a modification of the fractionation procedure described by Schnaar and Schaffner (1981, J. Neurosci. 1: 204-217). All steps were performed at 4 ° C. Metrizamide was dissolved in F12: MEM medium (1: 1) and a discontinuous gradient was set up consisting of a pad of 18% metrizamide (0.5 ml), 3 ml of 17% metrizamide, 3 ml of metrizamide a 12% and 3 ml of 8% metrizamide. The filtered ventral spinal cord cell suspension (2.5 ml) obtained as described above was stratified in a discontinuous gradient and the tube was centrifuged at 2500 g for 15 minutes using a swing rotor (Sorvall HB4). Centrifugation resulted in three cell layers: fraction I (at the interface of 0-8%), fraction II (at the interface of 8-12%) and fraction III (at the interface of 12-17%). Cells from each interface were removed in a small volume (about 1 ml), washed twice with defined serum-free medium consisting of 50% F12 and 50% MEM, supplemented with glutamine (2 mM), insulin (100 μg / ml), transferrin (100 μg / ml), progesterone (20 nM), putrescine (100 μg) and sodium selenite (30 nM) (Bottenstein and Sato, 1979, PNAS 7: 514-517). The viable cell count was obtained by hemocytometer counting in the presence of blue trypane. Fractionated ventral spinal cord cells (enriched in motor neurons) were then plated at a density of 100,000 cells / cm 2 in 6 mm wells precoated with poly-L-ornithine (Sigma: 10 μg / ml) and laminin (GIBCO: 10 μg / ml). NT-4 treatment was given on the day of placement. Cultures were maintained in defined serum-free medium at 37øC in a 95% air / 5% CO 2 atmosphere at a relative humidity

were collected on the day 2 (48 hours), the cells collected for the measurement of choline acetyltransferase Fonnum, 1975, J. Neurochem, 24: 407-409 ).

9.4. RESULTS The soluble fraction of the purification scheme of a culture of pRG173 / RFJ26 induced overnight was assayed in E8 DRG explants. Figure 13 shows a dose dependent stimulation of the development of neuritis in E8 DRG explants. The insoluble fraction was assayed on E8 dorsal root ganglia (DRG) explants and showed no bioactivity. Recombinant human NT-4 protein recovered from the 1M NaCl wash of the soluble fraction was assayed for activity in dissociated motor neuron cultures prepared as described supra in the Section of Example 9.3. and other tests. Addition of recombinant human NT-4 at a 1:20 dilution resulted in a 3.6 fold increase in choline acetyltransferase activity in the culture enriched in motor neurons 48 hours post-treatment. The 3.6-fold increase was measured relative to untreated controls (C-NT) and buffer (C-Buffer) (Figure 14).

9.5. DISCUSSION Subcloning of the NT-4 coding region of bacteriophage HG7-2 downstream of the UV5 and T701.1 LAC promoters, the T701.1 ribosome binding site and the LamB signal sequence resulted in the plasmid pRG173. The transformation of pRG173 into the E. coli strain RFJ26 and the induction of large-scale cultures of pRG173 / RFJ26 resulted in the expression of a biologically active form of recombinant human NT-4. The ability to express a biologically active form of human NT-4 in a prokaryotic expression system in vitro substantially increases the ease with which the production of recombinant human NT-4, peptides or derivatives thereof can be staggered either for therapeutic application or for application in

Diagnosis.

Thus, the present example demonstrates that a DNA sequence encoding a human NT-4 is susceptible to transcription and translation in a prokaryotic system so that human NT-4 is expressed, and the biologically active form is susceptible to the purification schemes of so that the activity remains subsequent to the purification. 10. DEPOSITION OF MICROORGANISMS Recombinant DNA molecules pRPN133 (hNGF), pRPN121 (hBDNFmyc) and pRPN149 (hBDNF) were deposited on June 7, 1991, pRG173 (NT-4) on October 28, 1991 and the recombinant bacteriophage HG7 -2 on August 22, 1991 at the American Type Culture Collection, 12301 Parklawn Drive, Rockville, Maryland 20852, under the Budapest Treaty on the International Recognition of the Deposit of Microorganisms in patents, and with the accession numbers indicated.

RECOMBINANT DNA MOLECULE ATCC ACCESSION NUMBER PRPN133 (hNGF) 75029 pRPNl21 (hBDNFmyc) 75028 pRPN149 (hBDNF) 75027 pRG173 (hNT-4) 75131 HG7-2 (human NT-4 genomic clone) 75070 The present invention does not is limited in scope by the deposited microorganisms or the specific embodiments described herein. Indeed, it will be apparent to those skilled in the art from the foregoing description and accompanying figures, various modifications of the invention beyond those described herein, it is envisaged that such modifications are included within the scope of the appended claims.

The various publications cited herein are incorporated by reference in their entirety.

While a number of embodiments have been described

It is evident that these basic embodiments may be altered to provide other embodiments utilizing the processes and compositions of this invention. Accordingly, it will be appreciated that the scope of this invention includes all of the alternative embodiments and variations which are defined in the foregoing description and the claims appended thereto; and the invention is not limited by the specific embodiments which have been given by way of example herein.

Claims (86)

A method of recovering a recombinant neurotrophin produced in a non-animal host cell culture comprising the step of solubilizing the neurotrophin in a solution comprising a strong denaturing agent, essentially free of reducing agents. A process according to claim 1, characterized in that the strong denaturing agent is a 4 M - 9 M guanidinium salt, a 4 M - 9 M alkali metal thiocyanate or 7 M - 9 M urea. A process as claimed in claim 2, characterized in that the strong denaturing agent is guanidinium 7 M-9 M HCI. A process according to claim 3, characterized in that the strong denaturing agent is 8 M HCl guanidinium HCl. A process as claimed in claim 1, further comprising the step of exchanging the strong denaturing agent with a weak denaturing agent. A process as claimed in claim 5, wherein the weak denaturing agent is 4 M-9 M urea. Process according to claim 6, characterized in that the strong denaturing agent is guanidinium HCl 7 M - 9 M and the weak denaturing agent is 6 M - 8 M urea. A process according to claim 7, characterized in that the strong denaturing agent is guanidinium HCl 8 M and the weak denaturing agent is 7 M urea. A method according to claim 1, characterized in that the solution further comprises a metalloproteinase inhibitor in an amount effective to inhibit the degradation of neurotrophin 74 061 6526-081-118 by proteases. 10. A process according to claim 2, wherein the solution further comprises a metalloproteinase inhibitor in an amount effective to inhibit the degradation of neurotrophin by proteases. 11. The process of claim 10, wherein the solution further comprises 1 mM EDTA-200 mM. 12. The process of claim 3, wherein the solution further comprises 5 mM EDTA-80 mM. 13. A process according to claim 4, characterized in that the solution further comprises 50 mM EDTA. 14. The process of claim 7, wherein the solution further comprises 5 mM EDTA-80 mM. 15. A process according to claim 8, wherein the solution further comprises 50 mM EDTA. 16. A process according to claim 5, further comprising the steps of (1) adjusting the solution to comprise a basic amino acid or indole-acetic acid at a concentration effective to maintain the solubility of the neurotrophin in a non- denaturant and (2) removing the weak denaturing agent. 17. A process according to claim 6, further comprising the steps of (1) adjusting the solution to comprise a basic amino acid or indole-acetic acid at a concentration effective to maintain the solubility of the neurotrophin in a non- denaturant and (2) removing the weak denaturing agent. 18. A process according to claim 6, further comprising the steps of (1) adjusting the solution of to comprise 20 mM histidine - 500 mM and (2) remove the weak denaturing agent. A process according to claim 7, further comprising the steps of (1) adjusting the solution to comprise 20 mM - 100 mM histidine and (2) removing the weak denaturing agent. A process according to claim 8, further comprising the steps of (1) adjusting the solution to comprise 20 mM-100 mM histidine and (2) removing the weak denaturing agent. A process according to claim 14, further comprising the steps of (1) adjusting the solution to comprise 20 mM - 100 mM histidine and (2) removing the weak denaturing agent. A process according to claim 15, further comprising the steps of (1) adjusting the solution to comprise 20 mM-100 mM histidine and (2) removing the weak denaturing agent. The process of claim 3, 4, 7, 8, 12, 13, 14, 15, 19, 20, 21 or 22, characterized by recovering recombinant hNGF, hBDNF or hNT-3. A method according to claim 23, which comprises recovering recombinant hNGF having threonine substituted serine as the second amino acid residue, full length hBDNF and full length hNT-3. A process for the production of a recombinant neurotrophin characterized in that it comprises the step of culturing a non-animal host cell transformed with a recombinant DNA molecule comprising a controllable expression control sequence operably linked to a DNA sequence encoding a neurotrophin, said cell The present invention relates to a heat exchange buffer gene mutant host. 26. The method of claim 25, wherein the host cell is a mutant strain of E. coli HtpRTS lon. 27. The method of claim 26, wherein the host cell is a mutant strain of E. coli HtoRts lon. 28. The method of claim 27, wherein the host cell is the E.coli strain LC137. 29. The method of claim 26, wherein the expression control sequence comprises a? PL promoter. 30. The method of claim 27, wherein the expression control sequence comprises a &quot; X PL promoter. 31. A process according to claim 30, characterized in that hNGF is produced. A process according to claim 31, characterized in that an hNGF is produced wherein the threonine substitutes the serine as the second amino acid residue. 33. A process according to claim 25, characterized in that it comprises the step of culturing E. coli strain E. coli transformed with pRPN133 (hNGF). A method according to any of claims 26-30, characterized in that hNGF, hBDNF or hNT-3 is produced. A process for the production of a recombinant neurotrophin characterized in that it comprises the step of culturing a non-animal host cell transformed with a DNA sequence, &lt; comprising an expression control sequence operably linked to a DNA sequence comprising a fused gene encoding a 5 'to 3' in-frame signal sequence with the neurotrophin. 36. The method of claim 35, wherein the host cell is E. coli and the signal sequence is LamB or a modified LamB. A method according to claim 36, characterized in that the expression control sequence comprises a lacUV5 promoter and a T7 ribosome binding site Φ1.1. 38. A method according to claim 37, characterized in that the signal sequence is LamBl: SEQ ID NO: 2 -> Bcll | &gt; Hind3 I &gt; Nhel I &gt; BspMl 1 1 1 20 1 1 | 140 1 * ★ 1 * 1 * C ATG ATG ATC ACA CTG CGT AAG CTT CCG CTA GCT GTA GCA GTA GCA GCA GGT G TAC TAC TAG TGT GAC GCA TTC GAA GGC GAT CGA CAT CGT CAT CGT CGT CCA Met Met lie Thr Leu Arg Lys Leu Pro Leu Ala Goes Ala Goes Ala Ala Gly &gt; a a a a a a LAMBI aaa aaaa &gt; &gt; BamHl &gt; Smal I &gt; I I II 60 | ITA 80 * I * III * GTA ATG TCT GCA CAG GCC ATG GCC CGGGATCCCTAG CAT TAC AGA CGT GTC CGG TAC CGG GCCCTAGGGATC Vai Met Ser Ala Gin Ala Met Ala &gt; a a LAMBI a a &gt; 74 061 6526-081-118 -46 LamB2: SEQ ID NO3> Bcll> Hind3> Nhel> BspMl I III I 20 | | 40 | * * | * | * * C ATG ATG ATC ACA CTG CGT AAG CTT CCG CTA GCT GTA GCA GTA GCA GCA GGT G TAC TAC TAG TGT GAC GCA TTC GAA GGC GAT CGA CAT CGT CAT CGT CGT CCA Met Met Ile Thr Leu Arg Lys Leu Pro Leu Ala Go Ala Vai Ala Ala Gly &gt; a a a a a a LAMB2 aaaaaaa &gt; > BamHl 1> Ncol> Ball II> Eagl 1 1 1 1 60 1 1 1 1 80 I 1 * 1 * 1 ★ 1 * GTA ATG TCT GCA CAG GCC ATG GCC AGT CGGCCGAGGATCCCTAG CAT TAC AGA CGT GTC CGG TAC CGG TCA GCCGGCTCCTAGGGATC Will Be Met Ala Gin Ala Met Ala Ser &gt; a 'to LAMB2 a a &gt; LamB3: SEQ ID NO: 4 &gt; &gt; Bcll I &gt; Hind3 &gt; Nhel &gt; Scal &gt; BspMl * 20 ★ | 40 * | + * C ATG ATG ATC ACA CTG CGT AAG CTT CCG CTA GCA GTA CTG CTA GCT GTA GCA G TAC TAC TAG TGT GAC GCA TTC GAA GGC GAT CGT CAT GAC CAT CAT CGT GTA GCA GCA GGT GTA ATG TCT GCA CAG GCC ATG GCC AGTCGGCCGAGGATCCCTAG CG CGT CGT CCA CAT AGA CGT CGTC CGG CGC TAC TCAGCCGGCTCCTAGGGATC Go Ala Ala Gly Go Met Met Ala Gin Ala Met Ala &gt; a a a a a a a a a Met Met Ile Thr Leu Arg Lys Leu Pro Leu Ala Vai Leu Leu Ala Vai Ala &gt; a a a a a a a A a a a a a> Eagl | &gt; Ncol &gt; Ball | &gt; BamHl 1 1 1 1 60 801 1 1 100 * + * 1 I * 1 1 * Or LamB4: SEQ ID NO: 5 &gt; Nhel &gt; Bcll &gt;Hind3; &gt; Nhel &gt; Scal | I &gt; BspMl I 1 1 1 4 1 20 * 1 1 1 1 | 40 1 * 1 1 * ATG ATG ATC ACA CTG CGT AAG CTT CCG TAC TAC TAG TGT GAC GCA TCA GAA GGC Met Met Ile Thr Leu Arg Lys Leu Pro aaaaaaa LAMB4 CTA GCA GTA CTG GAT CGT CAT GAC Leu Ala Goa Lea aaaa CTA GCT GTA GCA GAT CGA CAT CGT Leu Ala Vai Ala &gt; a a &gt; &gt; Eagl 1 &gt; Ncol &gt; Ball 1 1 I> BamHl 60 * * 801 *! 1 1 1 1 i * 1 | 100 1 * GTA GCA GCA GGT GTA ATG TCT GCA CAG GCC ATG GCC AGTCGGCCGAGGATCCCTAG CAT CGT CGT CAT CAT AGA CAT CGT CGTC CGG CAT CGG TCAGCCGGCTCCTAGGGATC Go Ala Ala Gly Ala Met Ser Ala Ala Metall Ala Ala &gt; a a a a a a a a 39. A method according to claim 38, characterized in that full-length hBDNF is produced. 40. The method of claim 39, wherein the signal sequence is LamB1 (SEQ ID NO2). 41. A process according to claim 38, characterized in that hBDNFmyc is produced. 42. The method of claim 41, wherein the signal sequence is LamB2 (SEQ ID No3) or LamB4 (SEQ ID Na5). A process according to claim 35, characterized in that it comprises the step of culturing the E. coli strain i3f-W1330 transformed with pRPN149 (hBDNF). A method according to any of claims 35-38, characterized in that hNGF, hBDNF or hNT-3 is produced. 74 061 6526-081-118 A method of recovering a recombinant protein produced in a culture of non-animal host cells, said protein having a pH of at least 9.0 and having at least two disulfide bonds, comprising the step of solubilizing the protein in a solution comprising a strong denaturing agent, said solution being essentially free of a reducing agent. A plasmid characterized by being the plasmid pRPN133 (hNGF). A plasmid characterized by being the plasmid pRPN121 (hBDNFmyc). A plasmid characterized by being the plasmid pRPN149 (hBDNF). A homogenous, purified neurotrophin characterized in that it is recovered by the process of claim 1. A homogenous, purified neurotrophin characterized in that it is recovered by the process of claim 5. A homogenous, purified neurotrophin characterized in that it is recovered by the process of claim 9. A homogenous, purified neurotrophin characterized in that it is recovered by the process of claim 10. A homogenous, purified neurotrophin characterized in that it is recovered by the process of claim 16. A homogenous, purified neurotrophin characterized in that it is recovered by the process of claim 17. A homogeneous, purified neurotrophin according to any of claims 49 to 54 characterized in that it is hNGF, hBDNF or hNT-3. 74 061 6526-081-118 -49- A homogeneous, purified neurotrophin according to claim 55 characterized in that it is an hNGF in which the threonine substitutes serine as the second amino acid residue. A homogeneous, purified neurotrophin according to claim 55 characterized in that it is a full-length neurotrophin. Gene for a modified LamB signal sequence gene which comprises a DNA sequence encoding a LamB signal sequence having at least one degenerate G or C substitution for A or T. The gene of claim 58, having 7 degenerate substitutions. The gene according to claim 59, characterized in that it is LamBl (SEQ ID Ns2) or LamB2 (SEQ ID NB3). A gene for a modified LamB signal sequence gene which comprises a DNA sequence encoding a LamB signal sequence having a hydrophobic core region of 11-20 amino acid residues. The gene of claim 61 which comprises a core region of 14 amino acid residues. The gene of claim 62 wherein the core region comprises the peptide LAVL (SEQ ID NO: 6). A gene according to claim 63, characterized in that it is LamB3 (SEQ ID No 4) or LamB4 (SEQ ID NO: 5). A process according to claim 3, characterized in that recombinant neurotrophin-4 is recovered. A method according to claim 65, 74 061 6526-081-118 was synthesized for recovering human neurotrophin-4. A method according to claim 66 for recovering human neurotrophin-4 encoded by a DNA sequence as contained in bacteriophage HG7-2 as deposited in ATCC accession number 75070 and shown below: SEQ ID N222 ( Following Sequence) 74 061 6526-081-118 10 20 30 40 50 60 70 80 CTTGTCACCCAGGTGGCACCCGAGTGGTGCACTCTCTGCTCACTGCAACCTCGGCCTCCTGGGTTCGAGTGATTCrcCTACCTCA 90 100 110 120 130 140 150 160 170 GCCTACTGAGTAGCTGGGATTACACGCGTGCAGCACTATGCCCGGrrAATTrTGGTATTTTTGGTAGAGATGAGGTTTCACCATG 180 190 200 210 220 230 240 250 TTGACCAGCTGCTCTGGAACTCCTCACCTCAAGTCATCCACCTGCCTCAGCCTCCCAGAGTGCTGGGATTAGAGGTGTGGGGCAC 260 270 280 290 300 310 320 330 340 AGTGCCTGGCCTGTAGTAGTTGAATATTTATTATTAATCTACAAGXTGCGCATTACGCAAGCCCTAGATATAGGGTCCCCCAAAC 350 360 370 380 390 400 410 420 ttctagaacaagggcttccccacaatcctggcaggcaagcctcccctggggttcccaacttctttccccactgaagtttttaccc intron - 430 440 450 460 470 T 480 490 CCXICXCIAAXCCCAGCCTCCCTCTTTCTGTCTC CAG GTG CTC CGA GAG / ATGSCXC CCT CTC CCC TCA TGC QVLREVLPLP 5 c &gt; 500 510 520 530 540 550 * * * * * * * XCC CTC CCC ATC CTC CTC CTT TTC CTC CTC CCC AGT GTG CCA ATT GAG TCC CAA CCC CCA CCC S L P L L 560 570 580 590 600 I 1610l 6201 * * * * * 1 * * 1 TCA ACA TTG CCC CCT TTT CTG GCC CCT GG TGG GAC CTT CTC TCC CC X CGA GTA GTC CTG TCT S T L P P W 630 640 I 650 660 670 680 * • ★ * * AGG GGX GCC CCT GCT GGG CCC CCT CTG CTC TTC CTG CTG GAG GCT GGG GCC TTT CGG GAG TCA R G A P A L E A G A F R E 5> 690 700 710 720 730 740 * * * * * * GCA GGT GCC CCG GCC AAC CGC AGC CGG CGT CGG GTG AGC GAA ACT GCA CCA GCG AGT CGT CGG A G A P A N R S R R G V S 750 760 770 780 790 800 810 is * * * * * * GGX GAG CTG GCT GTG TGC GAT GCA GTC AGT GGC TGG GTG ACA GAC CGC CGG ACC GCT GTG GAC G E L A V C D A V 5 G W V T D F R T V D> 820 830 - 840 850 860 870 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 8B0 B90 900 910 920 930 * * * * * * CAG TAC TTC TTT GAA ACC CGC TGC AAG GCT GAT AAC GCT GAG GAA GGT GGC CCG GGG GCA GGT 940 950 960 970 980 990 * * * * * ★ GGA GGG GGC TGC CGG GGA GTG GAC AGG AGG CAC TGG GTA TCT GAG TGC AAG GCC AAG CAG TCC G G G C R G V D R R H W S S C K A K 0 s &gt; 1000 1010 1020 1030 1040 1050 1060 * * * * * * * * TAT GTG CGG GCA TTG ACC GCT GAT GCC CAG GGC CGT GTG GGC TGG CGA TGG ATT CGA ATT GAC AND V R A L T A G G R V G K R W 1 R ϊ D &gt; 1070 1080 1090 1100 1110 1120 1130 * * * * * * ACT GCC TGC GTC TGC ACA CTC CTC AGC CGG ACT GGC CGG GCC TGAGACCCATGCCCAGGAAAATAACAGAG TAC V C T L S R T G R A &gt; 1140 1150 1160 1170 1180 1190 1200 1210 CTGGATGCTCAGAGACCTCAGGGATGGCCCAGCTCATCTAAGGACCCCAGTTTGGGAACTCATCAAATAATCACAAAATCACAAT 1220 1230 1240 1250 1260 1270 1280 1290 1300 TCTCTGATTTTGAGCTCAATCTCTGCAGGATGGGTGAAACCACATGGGGTTTTGGAGGTTGAATAGGAGTTCTCCTGGAGCAACT 1310 1320 1330 1340 1350 1360 1370 1380 1390 1400 tcagggtaataatgatgatgatataataataatagccactatttactgagtgtttactgtxtcttatccctaatacaxaactccx CAGATCAACICXCATG 74061 -52- 6526-081-118 A method according to claim 66 for the recovery of human neurotrophin-4 encoded by a DNA sequence as contained in DNA plasmid pRG173, as deposited with the ATCC under accession number 75131 and shown below:, locUV5 rbS2 69. A process according to claim 25, characterized in that neurotrophin-4 is produced. 70. A process according to claim 69, characterized in that human neurotrophin-4 is produced. A method according to claim 70 wherein the human neurotrophin-4 is encoded by a DNA sequence as contained in bacteriophage HG7-2, as deposited with ATCC accession number 75070 and set forth in claim 67 (SEQ ID Na22). 72. A process according to claim 70, The human neurotrophin-4 is encoded by a DNA sequence as contained in the E. coli plasmid vector. pRG173, as deposited with ATCC accession number 75131 and set forth in claim 68. 73. A method according to claim 25, characterized in that it comprises the step of culturing the E. coli strain RFJ26 transformed with the E. colif plasmid vector pRG173, as deposited with ATCC accession number 75131 and presented in claim 68. A process according to claim 35, characterized in that neurotrophin-4 is produced. A process according to claim 74, characterized in that human neurotrophin-4 is produced. 76. The method of claim 75, wherein the human neurotrophin-4 is encoded by a DNA sequence as contained in bacteriophage HG7-2 as deposited with ATCC accession number 75070 and set forth in claim 68 . A method according to claim 75, characterized in that human neurotrophin-4 is encoded by a DNA sequence as contained in the E. coli plasmid vector. pRG173, as deposited with the ATCC accession number 75131, as set forth in claim 68. 78. The method of claim 35 comprising the step of culturing the E. coli strain RFJ26 transformed with the E. coli plasmid vector. pRG173, as deposited with the ATCC under accession number 75131 and having the sequence set forth in claim 68. A vector plasmid of DNA characterized as pRPN91. 74 061 6526-081-118 -54- The DNA plasmid, characterized in that it is the pRG173 DNA plasmid, as deposited in the ATCC under accession number 75131 and having the sequence set forth in claim 68. A homogenous, purified, neurotrophin-4, characterized in that it is recovered by the process of claim 49. A homogeneous, purified neurotrophin-4 according to claim 81 characterized in that it is human neurotrophin-4. The homogeneous, purified Neurotrophin-4 according to claim 82 characterized in that it is encoded by a DNA sequence as contained in bacteriophage HG7-2 as deposited in ATCC accession number 75070 and set forth in claim 67. The purified homogeneous Neurotrophin-4 according to claim 82, characterized in that it is encoded by a DNA sequence as contained in the E. coli plasmid vector. pRG173, as deposited with the ATCC accession number 75131, as set forth in claim 68. A method of producing human NT-4, comprising culturing a bacterium, containing a recombinant nucleic acid encoding human NT-4, under conditions such that human NT-4 is expressed by the bacterium, and recovering the NT-4 produced. The method of claim 85, wherein the human NT-4 is biologically active. Lisbon 9. JttL 1992 By REGENERON PHARMACEUTICALS, INC = OFFICIAL AGENT-