HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Ali, S. A. Articles by Steinkasserer, A. Search for Related Content PubMed PubMed Citation Articles by Ali, S. A. Articles by Steinkasserer, A. Social Bookmarking                      What's this?

J Biol Chem, Vol. 274, Issue 34, 24066-24073, August 20, 1999

Transferrin Trojan Horses as a Rational Approach for the Biological Delivery of Therapeutic Peptide Domains^*

Stuart A. Ali, Heidi C. Joao, Franz Hammerschmid, Jörg Eder^§, and Alexander Steinkasserer^¶

From the Novartis Research Institute, Brunnerstrasse 59, A-1230 Vienna, Austria

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

One novel approach for the biological delivery of peptide drugs is to incorporate the sequence of the peptide into the structure of a natural transport protein, such as human serum transferrin. To examine whether this is feasible, a peptide sequence cleavable by the human immunodeficiency virus type 1 protease (VSQNYPIVL) was inserted into various regions of human serum transferrin, and the resultant proteins were tested for function. Experimentally, molecular modeling was used to identify five candidate insertion sites in surface exposed loops of human serum transferrin that were distant from biologically active domains. These insertions were cloned using polymerase chain reaction mutagenesis, and the proteins were expressed using a baculovirus expression vector system. Analysis of the mutant proteins provided a number of important findings: (a) they retained native human serum transferrin function, (b) the inserted peptide sequence was surface exposed, and most importantly, (c) two of these mutants could be cleaved by human immunodeficiency virus-1 protease. In conclusion, this investigation has validated the use of human serum transferrin as a carrier protein for functional peptide domains introduced into its structure using protein engineering. These findings will be useful for developing a novel class of therapeutic agents for a broad spectrum of diseases.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Human serum transferrin (HST)¹ is a monomeric glycoprotein with a molecular mass of around 80 kDa and is able to bind tightly, but reversibly, two ferric irons together with two bicarbonate co-ions. It has two roles. First, it regulates the availability of free iron in the body fluids, preventing the production of potentially toxic free radicals, and providing bacteriostatic functions. Second, it mediates the transport and uptake of iron into cells.

The process of receptor-mediated endocytosis is responsible for the uptake of transferrin by cells. The general features of transferrin uptake by this process are now well understood (1-5). First, iron-saturated transferrin binds to the transferrin receptor at the cell surface and is then internalized by clatherin coated vesicles into early endosomes. Here, the pH is mildly acidic and causes the release of iron from the transferrin. The transferrin-receptor complex then recycles back to the plasma membrane, where the neutral pH results in the dissociation of the iron-deficient transferrin from the receptor and release into the plasma. This transferrin is then free to bind more iron, and the receptor is free to bind more iron-loaded transferrin.

Although all living cells require iron, only those cells that have a high requirement for iron express large numbers of transferrin receptors (6). Most normal resting human cells require little iron for healthy cell function and do not manifest detectable transferrin receptors on their plasma membranes (7). In contrast, erythrocyte precursors, together with all actively proliferating tissues, do show detectable levels of transferrin receptor expression (for examples, see Ref. 8). This reflects the increased requirement of iron for metabolism during growth and development.

Tumor cells also express high levels of transferrin receptors (see Refs. 9-12). This finding led to the proposal that the transferrin uptake pathway could be used for the targeted delivery of mimetic chemotherapeutic drugs (9). Trowbridge and Domingo (13) later reported that this was indeed possible by demonstrating that tumor cells could be killed using monoclonal anti-transferrin receptor antibody conjugated with fragments of the ricin and diphtheria toxin. These conjugates were effective at killing human tumor cells in vitro at concentrations of up to 10,000 lower than uncoupled antibody plus toxin. Raso and Basala (14) later showed that similar increases in cyctotoxicity could be achieved when transferrin, rather than an anti-transferrin receptor, is used (14). Since these studies, numerous other reports have demonstrated the use of transferrin to enhance the potency of a range of other toxic proteins and anti-neoplastic drugs, including Adriamycin (15-18), Doxorubicin (19), immunotoxins (20, 21), diphtheria toxin A chain (22), Pseudomonas exotoxin (23), cholera toxin (24), and RNase A (25).

In recent years, several studies have demonstrated that the use of the transferrin uptake pathway is highly effective in animal models and humans. Particularly, Laske et al. (26) showed that HST/diphtheria toxin conjugates are effective at eradicating human glioma tumors in mice. Treatment produced on average 95% regression in tumor volume after 30 days. In contrast, tumor volumes increased by about 300% in animals treated with free toxin and by 1000% in the controls. A clinical trial has demonstrated that this conjugate is also effective in humans (27). Patients with recurrent malignant brain tumors received the conjugate intracerebrally using high flow interstitial microinfusion. Two of the 15 treated patients showed complete remission, and 9 showed a 50% reduction in tumor volume, as determined by nuclear magnetic resonance imaging. In none of the patients were there any signs of systemic toxicity.

Whether transferrin or anti-transferrin receptor antibodies are used to target drugs via the transferrin uptake pathway, it is clear that this is an effective approach for drug delivery.

A novel way to use the transferrin uptake pathway for the cellular delivery of therapeutic agents would be to incorporate the drug into the structure of transferrin. One way to do this would be to modify the iron-binding site of transferrin such that instead of binding iron it is able to bind a drug molecule. A major problem with this is that because not all the structural determinants of iron and receptor binding are known, it would be very difficult to design a functional molecule. Alternatively, the drug could be incorporated into the structure of transferrin itself, using recombinant protein engineering techniques. This would be appropriate for peptide drugs. Peptide drugs often have short biological half-lives, and those that are water-soluble usually do not penetrate the cell membrane readily (for a review, see Ref. 28). A recombinant transferrin analogue that contains a therapeutically active peptide sequence would overcome these problems.

Surface exposed loops of globular proteins can frequently tolerate insertions of additional amino acids without altering the function of the protein (29). Therefore, it should be possible to introduce peptide sequences into the surface of transferrin without destroying function. As such, the peptides would "hide" in the surface of the transferrin analogue, be actively taken up into the cell, and then recognized (i.e. be functional) at the site of inhibitory activity. In this scenario, the function of the transferrin molecule would be analogous to that of a "Trojan horse" for the delivery of the peptide sequence into the cell. That there are no described toxic side effects of transferrin, that its use has already been approved from a number of clinical studies, its long circulatory half-life, and the fact that it has already been widely used for targeted delivery of drugs into cells that express high levels of transferrin receptors are all strengths that support its use for this purpose. The design and evaluation of recombinant transferrin analogues as a novel drug delivery system is the subject of this report.

To demonstrate proof of this principle, we have chosen to generate transferrin analogues targeted to the protease of HIV-1, the etiological agent of AIDS. A peptide sequence (VSQNYPIVL) that is cleavable by HIV-1 protease was selected as the foreign competitive "therapeutic" peptide. A functional transferrin analogue would maintain native transferrin function, yet be cleaved by HIV-1 protease. The HST analogue would therefore function as a competitive substrate for HIV-1 protease.

The idea of using a competing substrate to inhibit HIV-1 protease is not new. It is known that in vitro, the enzyme can be inhibited by competing synthetic peptides (28). However, these peptides are poorly soluble and are rapidly cleared from the general circulation. Recently, intracellular expression of a fusion protein containing the HIV-1 Vpr and the p17/p24 HIV-1 protease cleavage site from Gag was shown to inhibit the production of infectious HIV-1 particles (30). Additionally, because there are no known mammalian proteases that can cleave the SQNY cleavage motif in the p17/p24 cleavage site, such an inhibitor will be specific for HIV-1 protease. Thus, in this study, the p17/p24 cleavage site was chosen as the peptide sequence to be inserted into the HST analogues.

Importantly, HIV-1 replication induces cell activation, concomitantly up-regulating the levels of HST receptor expression (31). Therefore, HIV-1 infected cells would be ideal targets for HST-directed therapy, much in the same way as tumors are. It is not completely clear how HIV-1 causes cell activation, but it is known that transcription factors involved with activation, such as NFB and SP1, are up-regulated (31), and HIV-1 Rev protein can up-regulate the expression of HST receptors (32).

HST is itself an ideal transport molecule for an anti-HIV-1 drug because it is nontoxic, has a long circulatory half-life, and can penetrate all tissue compartments of the body that are known to harbor HIV-1 infected cells, including the brain. Thus, the idea of using an HST/HIV-1 protease cleavage site analogue to prevent HIV-1 infection and replication is a viable new therapeutic strategy.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Reagents-- Restriction and modification enzymes were purchased from New England Biolabs (Schwalbach, Taunus, Germany). Native HST was obtained from Sigma and Roche Molecular Biochemicals. Oligonucleotides and peptides were provided by Genosys (Cambridge, United Kingdom), which also synthesized the HST/HIV-1 protease cleavage site peptide conjugate. The BacPAK baculovirus expression system was purchased from CLONTECH (Palo Alto, CA). Antisera were produced in rabbits using standard techniques. HIV-1 protease was a gift from A. Billich (Novartis Research Institute, Vienna, Austria).

Design of Transferrin Analogues-- The three-dimensional structure of rabbit serum transferrin (RST), solved by x-ray crystallography, was used in the modeling study to investigate the effects of the insertion of the HIV-1 p17/p24 protease cleavage site sequence, VSQNYPIVL, on the tertiary structure of the native protein. The results of this investigation were assumed to be valid for HST (for which no tertiary structure is available), given the high degree of sequence identity of 81% between HST and RST.

The first step in this study was to select suitable insertion sites for the HIV-1 protease cleavage site sequence. For this step, careful consideration of the structural effects of the proposed mutations on the native protein structure and consequent alteration of function and/or activity must be made. Hence, surface exposed loop regions, distant from residues known to be essential for activity, present possible options for introducing mutagenic sequences that would have minimal effect on overall structure and function of the protein.

This was performed by inserting the nine amino acid residues of the HIV-1 protease cleavage site sequence into the molecular model of the N-lobe of the RST sequence. The atomic coordinates of the N-lobe were taken from the crystallographically solved structure obtained from the Brookhaven Protein Data Base (Brookhaven National Laboratory, Upton, NY). The molecular modeling was carried out on a Silicon Graphics Indigo II Workstation, using the Discover module from the Biosym software Insight II. An energy minimization was subsequently performed on the resultant mutant protein structure. The energy minimized structure shows that all residues more than 25 Å from the inserted residues were restrained. Energy minimizations using the steepest descent (2000 steps, derivative = 0.1), followed by dynamic runs (T = 1000 K, time step = 1 fs, steps = 1000), were used to give minimal energy model structures of the mutant proteins.

Cloning the Recombinant HST cDNAs-- The native HST gene used in this study was a cDNA subcloned into the baculovirus transfer vector pBacPAK8 as described by Ali and co-workers (33). This plasmid, termed p8T, contains the powerful polyhedrin promoter to drive expression of the HST gene, a polyadenylation site and sequences to allow homologous recombination with a baculoviral vector.

Cassette mutagenesis was used to construct the HST mutant M289, which contains the inserted protease cleavage site at codon 289 of the HST sequence. Essentially, a 251-base pair region of p8T was amplified by PCR using the primers 5'-tta att gaa ttc caa cta ttc agc tct cct gtt tct cag aac tac cct atc gtc ctc cat ggg aag gac ctg ctg ttt aag-3' and 5'-ccc tac act gtt aac act cca ctc-3'. These primers contain the EcoRI and HpaII restriction sites (italics) and the nucleotides coding for the inserted protease cleavage site (underlined). This PCR fragment was then cloned into p8T via the EcoRI and HpaII restriction sites and then checked using a commercially available sequencing kit (Sequenase, version 2, Amersham Pharmacia Biotech). The resulting construct, p8TM289, is identical to the wild type construct p8T.1 except that at codon 289 the nucleotides coding for the HIV-1 protease cleavage site are inserted. In a similar fashion, cassette mutagenesis was used to insert the nucleotide sequence coding for the HIV-1 protease cleavage site at codon 279, creating p8TM279. The primers used for the PCR were 5'-tta att gaa ttc ttt tga cag gac gat agg gta gtt ctg aga aac ttt gtc ttt gcc aaa atg ttc ctg g-3' and 5'-tgc agg cct cga gtt cga atc-3' and contain the XhoI and EcoRI restriction sites (italics) for cloning.

To construct mutants M33, M75, and M257, which contain the inserted HIV-1 protease cleavage site at codons 33, 75 and 257, respectively, the novel procedure PCR-ligation-PCR mutagenesis was used as described by Ali et al. (34-36). Essentially, this procedure involves two steps. In the primary PCR amplification, the two regions adjacent to the insertion site are amplified. The internal primers used in these reactions contain the sequence to be introduced. After a short phosphorylation step (15 min), the PCR products are ligated and amplified in a secondary PCR using the external primers used for the primary PCR. This results in the production of a fusion gene, which contains the inserted nucleotide sequence. The fusion product was then cloned into the wild type transfer plasmid p8T via the XhoI and EcoRI restriction sites. The respective primers used to produce the upstream and downstream PCR fragments were MUT1/MUT2 and MUT3/MUT10 for M33, MUT1/MUT4 and MUT5/MUT10 for M75, MUT1/MUT6 and MUT7/MUT10 for M257. The sequences of these primers are as follows: MUT1, 5'-tgc agg cct cga gtt cga atc-3'; MUT2, 5'-ttt aaa gcg gcc gct tat tgt gac gag ggg tcg ctg cc-3'; MUT3, 5'-tgc ttg ccc ctg gag gtt ctg cac-3'; MUT4, 5'-gta cat cag gcc ata tca cct aga ac-3'; MUT5, 5'-cct atc gtc ctg aat aac ctg aag cct gtg gtg gc-3'; MUT6, 5'-gta gtt ctg aga aac cat act tcg ggc cac gac gg-3'; MUT7, 5'-cct atc gtc ctg ggc ggc aag gag gac ttg atc-3'; MUT10, 5'-gga gag ctg aat agt tgg aat tc-3'. Sequences of all regions that were PCR-amplified were checked using a commercial DNA sequencing kit (Sequenase, version 2.0, Amersham Pharmacia Biotech).

Cell Culture and Production of Recombinant Virus-- Recombinant virus was generated using the BacPAK baculovirus expression system (CLONTECH), as described by Ali et al. (33) using the High Five insect cell line (BTI TN 5BI-4; Invitrogen Corp., San Diego, CA). Essentially, the transfer plasmid was co-transfected into the host cells using lipofection. The viral DNA is a linearized replication-deficient recombinant baculovirus DNA (BacPAK6) derived from Autographia californica nuclear polyhedrosis virus. Recombinant virus was isolated, expanded in suspension cultures, and titrated using standard methods (37). High Five cell monolayer cultures were grown at 27 °C in SF900II serum-free medium (Life Technologies GmbH, Berlin, Germany), supplemented with 1 mM glutamine, 100 units/ml streptomycin, and 100 µg/ml penicillin. Suspension cultures included an additional supplement of 10 units/ml heparin (sodium salt; grade 1-A from porcine intestinal mucosa; Sigma-Aldrich Handels GmbH, Vienna, Austria). These cultures were shaken at 80 rpm at a density of 0.5-2.0 × 10⁶ cells/ml.

Large Scale Protein Expression-- High Five cells grown in suspension culture were pelleted by centrifugation at 400 × g for 15 min in a preparative centrifuge and then resuspended in viral supernatant at a density of 4 × 10⁶ cells/ml and a multiplicity of infection of 1. The viral supernatant was prepared by dilution of the viral stock in medium and then adding fresh heparin, glutamine, and antibiotics to give the concentrations indicated above. The infection was allowed to proceed for 1 h at 27 °C in an Erlenmyer flask with shaking at 80 rpm. The cells were then pelleted as above, resuspended in fresh medium at 1 × 10⁶ cells/ml, and shaken for a further 48 h at 27 °C and 80 rpm. Cells were then removed from the HST-containing supernatants by centrifugation as described (33).

Purification of Recombinant Proteins-- Recombinant HST was purified from the expression culture supernatants using hydrophobic interaction chromatography with a phenyl-Sepharose® column (High Load^TM, 26/10, Amersham Pharmacia Biotech), as described by Ali et al. (33). Briefly, an equal volume of 2.4 M ammonium sulfate and 0.8 M tri-sodium citrate (pH 6) was mixed with the supernatant, and the precipitate was removed by centrifugation. The supernatant was then loaded onto the column and washed with 1.2 M ammonium sulfate, 0.4 M citrate (Buffer A), and the protein was eluted in a gradient of 0-50% water in Buffer A. The HST-containing fractions (identified using SDS-PAGE analysis) were collected and dialyzed against 20 mM Tris-HCl, pH 8 (Buffer B), for 2 h. As a final "polishing" step, the protein was chromatographed on a Q-Sepharose® Fast Flow ion exchange column. Briefly, after washing pure HST eluted in a gradient of 0-100% 1 M KCl in Buffer B. HST containing fractions were pooled and dialyzed against PBS (137 mM NaCl, 2.68 mM KCl, 1.47 mM KH₂PO₄, 8.09 mM Na₂HPO₄, pH 7.2). Purified proteins were analyzed by SDS-PAGE analysis and Western blot analysis using a rabbit anti-HST antibody or an immune serum raised against the HIV-1 protease cleavage site peptide.

Sequence Determination-- N-terminal sequence determination of the recombinant HST was made using an Applied Biosystems 470A automatic protein sequencer, according to the manufacturer's instructions.

Iron Loading of HST-- Saturation of HST with iron was carried out as follows. Apo-HST was mixed with ferric nitriloacetate (1:3 molar ratio of ferric chloride to nitriloacetic acid, disodium salt) in the presence of 0.5 M Tris-HCl, pH 8.5, and an excess of bicarbonate (45 mM). The value of 1.48 µg of iron/mg of HST for 100% saturation was used to calculate the amount of iron required for 10 or 100% saturation. For preferential iron loading of the C-lobe, the iron saturation reactions were carried out at pH 6.0 in 0.25 M morpholinoethanesulfonic acid (MOPS) buffer, instead of pH 8.5 in Tris-HCl buffer. To generate the iron-free isoforms, proteins were combined with an equal volume of 100 mM sodium acetate, 1 mM EDTA, 1 mM NTA, pH 4, for 30 min on ice. In both cases, salts were removed by extensive dialysis against Tris-HCl, pH 8.5, or PBS.

Multizone Urea-PAGE Analysis of HST Isoforms-- A novel multizone urea-PAGE technique developed by Ali et al. (38) to resolve the four HST isoforms generated during the iron binding process was used to determine whether the recombinant HST proteins could bind the correct amount of iron. Briefly, apo-HST was incubated with increasing concentrations of ferric-nitriloacetate in the presence of bicarbonate to generate partially and fully saturated isoforms. These isoforms were then separated on polyacrylamide gels containing 6 M urea, 0.1 M Tris, 0.01 M boric acid, and 0.05 M EDTA, with an electrode buffer of 0.88 M -alanine and 0.25 M Tris, pH 8.8. Proteins in the gels were visualized by staining with silver.

Cell Uptake Assay-- The cell uptake assay used in this study has already been described in our earlier publication (33). Essentially, CEM T cells (ATCC no. CCL119) normally cultured in RPMI 1640 medium supplemented with 2 g/liter NaHCO₃, 10% FCS (BioWhittaker UK Ltd., Berks, United Kingdom) were serum-starved in RPMI-MOPS medium (RPMI 1640 containing 165 mM MOPS, adjusted to pH 7 with NaOH, 0.1% bovine serum albumin, and glutamine plus antibiotics as indicated earlier) for 2 h prior to the assay. For the assay, 5 × 10⁶ cells were resuspended in 0.5 ml of RPMI-MOPS containing 300 nM recombinant or native HST in either the apo- or holoform and incubated in a 37 °C water bath for 30 min. Cells were then washed in ice-cold PBS to remove free HST.

For immunofluorescence, cells were bonded to the surface of adhesion slides (Bio-Rad) and then fixed in 3% formaldehyde in PBS. Cells were permeabilized with 0.1% triton in PBS for 5 min at 22 °C and then in methanol at 20 °C for 3 min. After blocking with bovine serum albumin, immunofluorescence staining was performed with a rabbit polyclonal anti-human transferrin antiserum (BioGenix Laboratories, San Ramon, CA) as primary antibody and a rhodamine-conjugated goat anti-rabbit antibody (Accurate Chemical and Scientific Co., Westbury, NY) as the secondary antibody. After mounting cells in Bacto FA mounting fluid (Difco), fluorescence was visualized using a fluorescence microscope (Axiovert 10, Zeiss GmbH, Jena, Germany).

Native Dot Blot Analysis-- Samples of 10 ng (2 µl) of each recombinant HST protein, HIV-1 p55Gag and the protease cleavage site peptide VSQNYPIVL were spotted onto nitrocellulose and then subjected to immunodetection analysis using polyclonal antibodies from rabbits immunized with either HST, M289 mutant, the protease cleavage site peptide, or p24Gag. Donkey anti-rabbit F'(ab₂) horseradish peroxidase conjugate (Amersham Pharmacia Biotech) was used as the secondary antibody, and the presence of horseradish peroxidase was detected using a commercially available chemiluminescence kit (ECL, Amersham Pharmacia Biotech).

Determination of the Specific Activity of HIV-1 Protease-- The specific unit activity of the recombinant HIV-1 protease used in these studies was experimentally determined as the minimum amount of the enzyme stock required to totally digest 1 µg of HIV-1 p55Gag in 1 h at 37 °C in 20 µl of HIV-1 protease assay buffer (2 M NaCl, 5 mM EDTA, 15% (w/v) glycerol, 0.01% (v/v) Triton X-100, 20 mM MOPS, pH 6.0).

HIV-1 Protease Cleavage Assay-- Recombinant HST or HIV-1 p55Gag (1 µg) were incubated with a dilution series of recombinant HIV-1 protease in 20-µl volumes of HIV-1 protease assay buffer (see above) at 37 °C for either 1 h or overnight (16 h). Reactants were then separated by SDS-PAGE on 10% gels, and the proteins were visualized by silver staining.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Design of Recombinant Proteins-- After careful inspection of the three-dimensional crystal structure of the N-lobe of RST, a number of exposed loop regions on the surface of the protein were identified as candidate regions for introduction of the HIV-1 p17/p24 protease cleavage site sequence, VSQNYPIVL (Fig. 1). The most favorable of these were loop regions that were most distant from all residues known to be important for protein functionality, including sites for iron and bicarbonate binding, hinge regions, and the putative receptor binding domain. With the above taken into consideration, the following five positions for insertion of the HIV-1 protease cleavage site sequence were further investigated: (a) 32-33, (b) 74-75, (c) 256-257, (d) 279-280, and (e) 288-289, as indicated in Fig. 1. These mutants were designated M33, M75, M257, M279, and M289, respectively. It is noted that the residues contained in loop regions of RST selected as insertion sites are highly homologous to those in HST, and in the case of HST, secondary structure prediction analysis indicates these residues to also be loop regions (39). As an example, the energy minimized molecular model of the mutant M289 is shown in Fig. 2. This figure shows that the inserted peptide cleavage site is surface exposed and does form a loop-like structure.

View larger version (49K): [in this window] [in a new window]   Fig. 1.   Candidate insertion sites for the HIV-1 protease cleavage site sequence. The sites, in the N-lobe of HST, are indicated by the arrows. N and C refer to the N and C termini of the polypeptide chain. The x-ray coordinates were obtained from the Brookhaven Protein Data Base.

View larger version (60K): [in this window] [in a new window]   Fig. 2.   Molecular model of HST mutant M289. The HIV-1 protease cleavage site sequence VSQNYPIVL was inserted after amino acid 288 in the HST sequence (indicated by the arrow).

Construction of Mutant cDNAs-- Five insertion sites for the HIV-1 protease cleavage sequence in HST were identified following analysis of the three-dimensional structure of RST (see above). These insertions were cloned into the HST gene using conventional PCR mutagenesis or a method that we developed to make insertions distant from convenient cloning sites (PCR-ligation-PCR mutagenesis).

Protein Expression and Purification-- The wild type and mutant HST analogues were expressed using a baculovirus expression system and purified using a combination of hydrophobic interaction chromatography and ion exchange chromatography as described previously (33). Typical expression levels were 10-20 mg/liter for each protein. The wild type protein is about 2 kDa smaller than the native protein. This is due to differences in glycosylation because deglycosylation using N-glycosidase-F yields protein with an identical size on SDS-PAGE gels (data not shown). Western blot analysis (Fig. 3B) shows that each of the proteins was immunoreactive with an anti-transferrin antibody. To confirm that the analogues do contain the introduced protease cleavage site sequence, the same proteins were subjected to Western blot analysis using an antiserum raised against the protease cleavage site peptide. Only the mutant HST analogues were immunoreactive (Fig. 3C).

View larger version (85K): [in this window] [in a new window]   Fig. 3.   Western blot analysis of purified recombinant HST. Lane b, native HST; lane c, wild type recombinant HST; lanes d-h, HST mutants 33, 75, 257, 279, and 289, respectively; lane a, 10-kDa molecular mass marker. Proteins were separated by denaturing SDS-PAGE and stained with silver (A) or were subjected to Western blot analysis using an anti-HST antibody (B) or an antiserum raised against the HIV-1 protease cleavage site peptide (C). Gels contained 200 ng of HST per lane.

The Protease Cleavage Sites Are Surface Located-- Immunodetection of recombinant HST proteins in a native dot blot assay using antibodies raised against the protease cleavage site peptide produced positive signals with each of the mutant analogues, but not the recombinant wild type HST, as shown in Fig. 4. If the inserted protease cleavage site domain was buried completely within the HST structure, then antibody binding to the native protein would not be expected. These analyses therefore provide further evidence that the engineered peptide domains were exposed on the surface of HST.

View larger version (42K): [in this window] [in a new window]   Fig. 4.   Detection of surface exposed peptide cleavage site domains in mutant proteins. Samples (1 µg) of the native and recombinant protein, HIV-1 p55 Gag, and the protease cleavage site peptide were spotted onto duplicate nitrocellulose squares (positions indicated in A) and then probed for immunoreactivity with antibodies raised against the protease cleavage site (B1 and B2). As controls, the nitrocellulose squares were probed with antibodies raised against M289 (C1 and C2) or HIV-1 p24 (D1 and D2). In this experiment, the serum or antibody used in each duplicate was derived from a different rabbit.

The Recombinant HST Proteins Correctly Bind Iron-- A novel multizone urea-PAGE technique was used to determine whether the recombinant HST proteins were able to bind the correct amount of iron, as described (38). The first experiment compared the ability of the native and recombinant wild type HST to bind iron. Incubation of the native protein in a 2-fold molar excess of iron at pH 8.5 saturates the protein, generating an isoform that migrates much faster through the gel than the slower moving apoform (data not shown). At partially saturating concentrations, two further isoforms are seen. The recombinant HST generates identical isoforms under the same conditions (data not shown).

Urea-PAGE analyses of iron binding by the mutant HST proteins is shown in Fig. 5. In each case, a faster moving protein species was generated when the proteins are incubated in a 2-fold molar excess of iron at pH 8.5 (compare the left and right lane in each panel). Under acidic conditions, the C-lobe was preferentially saturated with iron (middle lane in each panel); no N-lobe saturated isoform was seen. This demonstrates that like recombinant wild type and native HST, each mutant can bind two ferric ions. Moreover, the insertion of the peptide domain in the N-lobe of the mutants does not destroy the acid sensitivity of the HST molecule.

View larger version (21K): [in this window] [in a new window]   Fig. 5.   Urea-PAGE analysis of mutant HST proteins. In each panel, the left lane shows the apoprotein, and the right lane shows the holoprotein. Proteins incubated with iron under acidic conditions preferentially saturate the C-lobe (C-Fe-HST) and are shown in the middle lanes.

In a second experiment, the relative electrophoretic mobilities of the four HST isoforms generated by each of the mutants were compared using urea-PAGE. The apoproteins were incubated with 10% of the amount of iron required for complete saturation and then separated by electrophoresis (Fig. 6). For each mutant, all four HST isoforms were visible. Furthermore, the relative mobilities of each of the four isoforms in the urea gels were invariant between the mutants. These similarities in electrophoretic behavior provide further evidence that mutagenesis did not introduce gross conformational changes into the protein.

View larger version (34K): [in this window] [in a new window]   Fig. 6.   Urea-PAGE analysis of mutant proteins incubated with 10% of the iron required for complete saturation. Lanes a-f are the recombinant wild type, M33, M75, M257, M279, and M289 proteins, respectively. The positions of the apo- and holoisoforms are indicated, as are the HST isoforms containing iron bound by the C-lobe (C-Fe-HST) and N-lobe (N-Fe-HST).

Recombinant HST Is Biologically Active-- To analyze biological activity of the recombinant HST, immunofluorescence studies were carried out to observe cellular uptake. This approach has been previously used to demonstrate that the recombinant wild type HST is biologically identical to native HST (35). In this assay, cells were incubated with either recombinant wild type or native HST in its apo- or holoisoform at a 300 nM concentration (approximately of the normal serum concentration of HST in humans). The presence of HST was visualized using fluorescently labeled antibodies. Cells treated with holo-HST endocytosed HST, as demonstrated by a strong punctate intracellular immunofluoresence, whereas the apoproteins exhibit reduced extracellular binding and weaker immunofluoresence. These data concur with reports of Huebers et al. (40, 41).

Similarly, each of the mutant proteins was endocytosed and exhibited reduced cellular binding in the apoform (data not shown). This indicates that they, too, are biologically active. Therefore, insertion of the HIV-1 protease cleavage site domain in the N-lobe of HST disrupted neither iron binding nor recognition and cellular uptake. This is consistent with the molecular modeling predictions.

Mutants M279 and M289 are Cleaved by HIV-1 Protease-- The recombinant mutant HST proteins contain an inserted domain consisting of a peptide sequence that can be cleaved by HIV-1 protease. Treatment of these proteins with HIV-1 protease would demonstrate whether or not the introduced domains were accessible for cleavage by HIV-1 protease.

Experimentally, 1 µg of each protein was incubated for 3 h with 100 units of HIV-1 protease, (1 unit is the minimum quantity of HIV-1 protease required to cleave 1 µg of HIV-1 p55Gag to completion in 1 h). M289 was cleaved to completion, whereas approximately 10% of M279 was cleaved (data not shown). No detectable cleavage products were produced from the remaining three mutants, even after a prolonged incubation in 100-fold more enzyme for up to 4 days (data not shown).

N-terminal sequence analysis of the HIV-1 protease digestion products of M289 yielded two sequences (Fig. 7). The first sequence (derived from the 47-kDa fragment) corresponded to the N-terminal sequence of the correctly processed N-lobe. The second corresponded to the predicted N-terminal sequence that would be generated if the HIV-1 protease correctly cleaved the M289 protein. This proves conclusively that cleavage by HIV-1 protease was site-specific.

View larger version (16K): [in this window] [in a new window]   Fig. 7.   Cleavage of M289 by HIV-1 protease at the inserted HIV-1 protease cleavage site. M289 cleavage products resulting from digestion with HIV-1 protease were separated by gel electrophoresis and analyzed by N-terminal amino acid sequence analysis. The N-terminal sequence (VPDKTVRWCA) of the larger 47-kDa fragment (see lane M289 +) was identical to that of intact recombinant HST (see lane M289 ). The N-terminal sequence of the smaller 32-kDa fragment contained the digested HIV-1 protease cleavage site sequence (PIVL) followed by the expected adjacent amino acid sequence of the C-lobe (HGKDLL); see lane M289 +.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Proof of Principle: HST Analogues Are Biologically Active-- One novel approach for the biological delivery of peptide drugs is to incorporate the sequence of the peptide into the structure of a natural transport protein, such as HST. To examine whether this is feasible, a peptide sequence cleavable by HIV-1 protease was inserted into various regions of HST, and the resultant proteins were tested for function. Experimentally, molecular modeling was used to identify candidate insertion sites in surface exposed loops of HST that were distant from biologically active domains. Analysis of the recombinant proteins provided a number of important results: (a) they retained native HST function, (b) the inserted peptide sequence was surface exposed, and most importantly, (c) two of these mutants could be cleaved by HIV-1 protease. These findings validate the hypothesis that HST can be used as a recombinant transport system for functional protein domains that have been introduced into its structure using protein engineering.

Rational Approach Using Molecular Design-- In this study, a molecular modeling study was used to rationalize suitable insertion sites for the HIV-1 protease cleavage site. An alternative, but irrational, way to construct HST analogues would be to simply clone the HIV-1 protease cleavage site sequence into convenient restriction endonuclease sites within the HST gene. However, careful inspection of the HST structure indicated that all the unique restriction sites were buried within the protein or lay in domains thought to be necessary for correct protein folding, and would therefore disrupt HST function. Because this analysis was based upon the RST (and not HST) structure, it could be argued that for HST, some of these insertions could result in functional recombinant proteins. However, success would largely be a result of chance. Such an approach was used in an earlier publication that describes the production of recombinant -galactosidase proteins that contain the HIV-1 protease p6/PR cleavage site (42). The aim of this experiment was to develop a colorimetric assay for HIV-1 protease activity (proteolytic cleavage of -galactosidase destroys its activity). Eight positions for insertion of the protease cleavage sequence were chosen exclusively upon the availability of convenient restriction sites within the -galactosidase gene. The disappointing outcome of this approach was that only one of these mutants possessed -galactosidase activity, and furthermore, this was at a level of just 10% that of the wild type. Only this mutant was cleaved by HIV-1 protease.

In our approach, HST function was maintained in every one of the mutant analogues, as predicted using the molecular models. What was beyond the limits of our models was the ability to predict which of the analogues would be cleaved by HIV-1 protease. There are two possible reasons why some of the analogues were not cleaved. First, the inserted protease cleavage sites may not be in a correct steric orientation for binding within the active site of HIV-1 protease. If this is so, then it may be possible to convert a noncleaved analogue into a cleaved one by adding amino acids that introduce a partial helix either side of the peptide domain, thereby rotating its orientation within the loop. Cleavage of the analogue would also be prevented if the loop into which the peptide is inserted exerts structural constraints upon the peptide domain preventing correct binding within the active site of HIV-1 protease. However, molecular modeling analysis of these loops indicated that none were structurally constrained.

HST Analogues: A Rational Approach to HIV-1 Inhibition-- The use of peptide sequences cleaved by the HIV-1 protease is a rational approach to inhibit HIV-1. This is demonstrated in a recent publication that showed that transient expression of a recombinant protein consisting of HIV-1 Vpr and a C-terminally fused p17/p24 HIV-1 protease cleavage site could inhibit the production of infectious HIV-1 (30). Because Vpr is normally incorporated into HIV-1 virions and budding virions are the predominant site of GagPol polyprotein processing by HIV-1 protease, it was speculated that Vpr served to bring an excess of HIV-1 protease substrate to the site of protease activity, thereby inhibiting GagPol processing, preventing maturation of the virions. Although inhibition of HIV-1 production was convincingly demonstrated, the authors provided no evidence whatsoever that the fusion protein itself was cleaved by HIV-1 protease and that this cleavage event itself resulted in the inhibition of GagPol processing. However, if the production of infectious virus was a direct result of HIV-1 protease inhibition, then this is firm evidence to support the use of the HIV-1 protease cleavage site analogues. It would be interesting to examine whether an HSTM289/Vpr analogue would be similarly incorporated into virions and exhibit anti-HIV-1 activity.

Recently, Tritch et al. (43) described a peptide analogue of the p17/p24 cleavage site that could inhibit HIV-1 protease in vitro. This peptide differed from the native sequence by just two amino acids (MM instead of YP at the scissile bond) and was an absolute inhibitor of p55Gag cleavage. Interestingly, when the native p17/p24 cleavage site in p55Gag was replaced with this peptide sequence, the p55Gag was cleaved at a rate 10 times slower than the wild type protein. Therefore, we constructed an HST analogue containing this modified cleavage site (VSQNMMIVL instead of VSQNYPIVL) inserted at amino acid 289 and treated it with HIV-1 protease. However, this analogue was cleaved at a rate comparable to M289 (data not shown). This result emphasizes that it is not possible to assume that the cleavage rates of peptide substrates will be the equivalent to proteins containing the same cleavage site sequence. However, the fact that an inhibitory peptide analogue could be found does suggest that it may be possible to design similar inhibitory sequences that would be functional in an HST analogue.

Future Perspectives: Alternative HST Analogues-- It may be possible to generate a whole new range of HST analogues with different activities simply by using alternative foreign peptide domains that possess biological activity. For the therapy of HIV-1, as an example, candidate peptide sequences have been described. HIV-1 membrane fusion can be inhibited by peptides derived from the HIV-1 surface protein gp41, reducing virus infectivity (44, 45), and HIV-1 uptake via the CXCR4 co-receptor can be inhibited by peptides derived from stromal cell-derived factor 1 (46). Furthermore, peptides derived from the HIV-1 Tat transactivator protein can inhibit viral transcription, thereby inhibiting virus production (47, 48). It has also been shown that peptides derived from the transframe region of the GagPol polyprotein act as noncleavable inhibitors of HIV-1 protease (49), although, in this case, it remains to be tested whether these peptides can inhibit viral maturation. Finally, neurotoxicity of gp120 can be reduced by peptides derived from vasoactive intestinal peptide factor (50, 51), presenting a possible means for reducing intrauterine growth retardation and neurodevelopmental handicaps common among newborns from HIV-1 infected mothers. These are just a few of the many examples of peptides with therapeutic potential as HST analogues.

In conclusion, this investigation has demonstrated that it is possible to insert functional peptide domains into the structure of HST without disrupting HST function. This work has resulted in the development of protocols for mutagenesis and protein production that will permit the generation of a range of further HST analogues containing different peptide domains. As such, these HST Trojan horses will provide the ability to deliver defined peptide domains into cells. This not only provides a useful research tool for investigating protein function but also represents a rational approach for drug design, application of which will be useful for developing a novel class of therapeutic agents for a broad spectrum of diseases.

	ACKNOWLEDGEMENTS

We thank R. Datema, J. Hauber, and R. Newbold for helpful discussions and support of this project and E. Mlynar, S. Strommer, and R. Csonga for skilful technical assistance. For N-terminal protein sequencing, we thank A. C. Willis.

	FOOTNOTES

^* The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Present address: Dept. of Biochemistry, University of Oxford, South Parks Rd., Oxford OX1 3QU, Great Britain

^§ Present address: Novartis Pharma AG, S-386.943, CH-4002 Basel, Switzerland

^¶ To whom correspondence should be addressed: Dept. of Dermatology, University of Erlangen, Hartmannstrasse 14, D-91052 Erlangen, Germany. Tel.: 49-9131-85-36725; Fax: 49-9131-85-35799; E-mail: alexander.steinkasserer@derma.med.uni-erlangen.de.

	ABBREVIATIONS

The abbreviations used are: HST, human serum transferrin; HIV, human immunodeficiency virus; MOPS, 4-morpholinoethanesulfonic acid; PAGE, polyacrylamide gel electrophoresis; PCR, polymerase chain reaction; RST, rabbit serum transferrin; PBS, phosphate-buffered saline.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				Z. M. Qian, H. Li, H. Sun, and K. Ho Targeted Drug Delivery via the Transferrin Receptor-Mediated Endocytosis Pathway Pharmacol. Rev., December 1, 2002; 54(4): 561 - 587. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles