pegylation of brain-derived neurotrophic factor for preserved biological activity and enhanced...

PEGylation of brain-derived neurotrophic factorfor preserved biological activity and enhancedspinal cord distribution

Ryan G. Soderquist,1 Erin D. Milligan,2 Evan M. Sloane,2 Jacqueline A. Harrison,2

Klarika K. Douvas,1 Joseph M. Potter,2 Travis S. Hughes,3 Raymond A. Chavez,4

Kirk Johnson,4 Linda R. Watkins,2 Melissa J. Mahoney11Department of Chemical and Biological Engineering, University of Colorado at Boulder, 424 UCB, Boulder,Colorado 803092Department of Psychology and the Center for Neuroscience, University of Colorado at Boulder, 345 UCB,Boulder, Colorado 803093Department of Molecular, Cellular and Developmental Biology, University of Colorado at Boulder,347 UCB, Boulder, Colorado 803094Avigen, 1301 Harbor Bay Parkway, Alameda, California 94502

Received 3 June 2008; revised 25 July 2008; accepted 30 July 2008Published online 1 December 2008 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.32254

Abstract: Brain-derived neurotrophic factor (BDNF) wascovalently attached to polyethylene glycol (PEG) in orderto enhance delivery to the spinal cord via the cerebrospinalfluid (intrathecal administration). By varying reaction con-ditions, mixtures of BDNF covalently attached to one (pri-mary), two (secondary), three (tertiary), or more (higherorder) PEG molecules were produced. The biological activ-ity of each resulting conjugate mixture was assessed withthe goal of identifying a relationship between the numberof PEG molecules attached to BDNF and biological activ-ity. A high degree of in vitro biological activity was main-tained in mixtures enriched in primary and secondary con-jugate products, while a substantial reduction in biologicalactivity was observed in mixtures with tertiary and higherorder conjugates. When a biologically active mixture of

PEG-BDNF was administered intrathecally, it displayed asignificantly improved half-life in the cerebrospinal fluidand an enhanced penetration into spinal cord tissue rela-tive to native BDNF. Results from these studies suggest aPEGylation strategy that preserves the biological activity ofthe protein while also improving the half-life of the proteinin vivo. Furthermore, PEGylation may be a promisingapproach for enhancing intrathecal delivery of therapeuticproteins with potential for treating disease and injury inthe spinal cord. � 2008 Wiley Periodicals, Inc. J BiomedMater Res 91A: 719–729, 2009

Key words: intrathecal drug delivery; brain-derived neuro-trophic factor; PEGylation; biological activity; confocalmicroscopy

INTRODUCTION

Brain-derived neurotrophic factor (BDNF) is amember of the neurotrophin family of moleculesand exhibits therapeutic benefits for several neurode-generative diseases.1 BDNF can limit tissue damageafter spinal cord injury,2 augment the function ofspared neural systems,3 promote neural repair andregeneration,2 and promote cell survival and neuriteoutgrowth.4 BDNF also has potential as a therapeutic

agent for neuropathic pain, as low doses of BDNFcan suppress abnormal pain reactivity caused by pe-ripheral nerve injury.5,6

To treat disease and injury in the spinal cord,BDNF is commonly administered intrathecally viathe cerebrospinal fluid (CSF) surrounding the spinalcord. Protein in the CSF then diffuses into spinalcord tissue to reach and impact cellular targets.However, in most cases, protein half-life (typically1–3 h)7 due to turn over in the CSF limits the timescale over which biologically active levels of BDNFcan be maintained in the CSF. This in turn, limitsthe amount of BDNF that is capable of diffusingfrom the CSF into the parenchyma toward cellulartargets. Once in tissue, the protein is also subject toelimination via binding to cell surface receptors

Correspondence to: M. J. Mahoney; e-mail: [email protected]

� 2008 Wiley Periodicals, Inc.

and enzymatic degradation prior to reaching cellulartargets, which further limits its therapeutic availabil-ity. To improve the efficacy of intrathecal proteinadministration, delivery strategies designed toincrease the bioavailability of the protein in the CSFand in the parenchyma are necessary.

Covalent attachment of synthetic polymers to pro-teins has been shown to improve protein half-lifeand penetration into tissues.1,2,8–24 The polymerpolyethylene glycol (PEG) is the most widely usedfor this purpose.10,25–27 Although the attachment ofPEG to a protein can sterically hinder the protein’saccess to receptors and subsequent protein biologicalactivity, PEGylated proteins have an increased half-life in the bloodstream. The increased circulationtime of the protein compensates for the reduction inactivity and a therapeutic benefit is often observedwhen examined in vivo. Although improvements inthe half-life of PEGylated proteins in vivo often over-come in vitro biological activity losses,21,28 minimiz-ing the loss in activity by controlling the PEGylationreaction would be advantageous. The resultant ther-apeutic would not only improve patient compliance,as the number of necessary injections could bereduced due to the longer circulation time of theprotein, but it would be more cost-effective as lowerdosages would be necessary to achieve therapeuticeffects.

The focus of this work is to PEGylate BDNF inorder to enhance in vivo properties after intrathecaladministration while maintaining full in vitro biologi-cal activity, a feat that has only been reported withcarboxyl-directed BDNF PEGylation,21 but not previ-ously reported with amine-directed BDNF PEGyla-tion strategies.1,2 PEGylation of the N-terminus ofproteins tends to retain the biological activity of pro-teins when characterized in vitro,23,29 including epi-dermal growth factor (EGF).12 On the basis of thesuccess with this related growth factor, we pursueda PEGylation scheme that targets the N-terminus ofBDNF. Because there tends to be an inverse relation-ship between the number of PEG molecules attachedto a protein and its in vivo clearance rate,24 we alsoexamined the functionality of BDNF when attachedto two or more PEG molecules. Overall, ourapproach was to maximize the number of PEG mole-cules attached to the protein while maintaining thein vitro biological activity by minimizing the numberof PEG molecules that attach to residues locatedwithin BNDF functional sites. The effect of attachingone, two, three, or more PEG molecules to BDNF onin vitro biological activity was examined. The mix-ture with the highest level of biological activity wasidentified and the improvement in half-life and pen-etration into the spinal cord parenchyma relative tounmodified BDNF following intrathecal administra-tion was measured.

MATERIALS AND METHODS

BDNF PEGylation with aldehyde chemistry

PEG conjugation with BDNF using aldehyde chemistry[mPEG-ButyrALD, Fig. 1(a)] was conducted in reactionbuffer that consisted of 50 mM sodium phosphate with100 mM sodium chloride at a pH of 6.3. rhBDNF (1.780mg; Amgen) was added to 1 mL of the reaction buffer asthe protein mix. For a 60-fold PEG to BDNF dimer molarexcess and a 60-fold reducing agent to BDNF dimer molarexcess, 10 mg of mPEG-Butyrald-5000 (Nektar) was addedto 0.6575 mL of the reaction buffer as the polymer mix and1.72 mg of sodium cyanoborohydride (Sigma-Aldrich) wasadded to 10 mL of reaction buffer as the reducing agentmix. One hundred microliters of the protein mix, 200 lL ofthe polymer mix, and 175 lL of the reducing agent mixwere combined in a polypropylene tube and agitated atroom temperature for 24 h. Reactions were stopped bytransferring the products to 2808C until further analysis.Variations in the molar excess of constituents to BDNFwere conducted by holding the concentration of BDNFconstant.

BDNF PEGylation with NHS ester chemistry

NHS ester chemistry [mPEG-SPA, Fig. 1(b)] was alsoused to conjugate PEG with BDNF and this reaction wasconducted in phosphate buffered saline at a pH of 7.4.BDNF (0.89 mg) was added to 1 mL of PBS as the proteinmix. For a 15-fold PEG to BDNF dimer molar excess 2.4 mgof mPEG-SPA-5000 (Nektar) was added to 1 mL of the reac-tion buffer as the polymer mix. Two hundred and fiftymicroliters of the protein mix and 250 lL of the polymermix were combined in a polypropylene tube and agitated atroom temperature for 24 h. Reactions were stopped bytransferring the products to 2808C until further analysis.

Figure 1. Reaction diagrams of (a) conjugation withmPEG-ButyrALD and (b) conjugation with mPEG-SPA.

720 SODERQUIST ET AL.

Journal of Biomedical Materials Research Part A

Mass spec analysis

Matrix-assisted laser desorption ionization time of flight(MALDI-TOF) mass spectrometry (Voyager-DE STR, Per-kin–Elmer) was used to obtain mass information for PEG,BDNF, and representative PEG-BDNF conjugate mixtures.Samples (0.5 lL) were cocrystallized with matrix (sinapinicacid, Agilent) on gold-coated sample plates. Data weresummed over 100 acquisitions in delayed linear extractionmode with a 25 kV accelerating voltage, a 50 V guide wirevoltage, and a 300 ns delay.

Gel electrophoresis and immunoblotting

Reaction products were analyzed by SDS-PAGE using10% precast gels according to manufacturer recommendedreagents and protocols (Bio-Rad). Coomassie staining (Bio-Rad) was conducted with gels that had been loaded with8.9 lg of total protein. Immunoblotting with OPTI-4CNdetection was also conducted according to manufacturerrecommended reagents and protocols (Bio-Rad). Five hun-dred nanograms of total protein was loaded for detectionwith 1:2000 diluted rabbit-anti-BDNF polyclonal antibody(Chemicon Ab1779) as the primary antibody and 1:3000goat-anti-rabbit-HRP conjugate (Bio-Rad) as the secondaryantibody.

Band density analyses were conducted on coomassiestained gels with NIH ImageJ software. Image look uptables were inverted and the products of the mean andarea measurements were taken on a black and white scaleusing a polygonal fit around each observed band. Theproduct of the mean and area measurement was measuredas the intensity value for each species. The total intensityfor all bands in a conjugate mixture was determined andthe fractional intensity of the free BDNF band was multi-plied by the total protein concentration to estimate the re-sidual free BDNF concentration.

In vitro biological activity assay

The in vitro biological activity assay was conductedusing a rat pheochromocytoma cell line (PC12, passagenumber 13–18) that stably expresses the trkB receptor forBDNF. Cells were grown in RPMI medium supplementedwith 10% horse serum, 5% fetal bovine serum, and 1%penicillin–streptomycin (all from Invitrogen) in 24-well col-lagen coated plates (8 lg/cm2, Vitrogen-100, Angiotech).Twenty-four hours after seeding the plates at a density of1.0 3 106 cells/mL, the cells were incubated with 0.5 ng/mL of BDNF or PEG-BDNF conjugate mixtures for anadditional 24 h. The medium was removed by aspirationand the cells were fixed onto the plates with 4% parafor-maldehyde. The number of cells extending neurites longerthan two cell bodies was then assessed and error valueswere represented as standard error of the mean. For eachcondition 40 different groups consisting of 10 cells eachwere analyzed to determine the mean fraction of neuriteextension.

Intrathecal injections

All animal procedures were in accordance with theInstitutional Animal Care and Use Committee at the Uni-versity of Colorado at Boulder and NIH guidelines for thecare and use of laboratory animals (NIH Publication 85-23Rev. 1985) were also observed. Pathogen-free adult maleSprague-Dawley rats were used in all experiments. Rats(250–275 g at the time of arrival; Harlan Labs, Madison,WI) were housed in temperature (238C 6 38C) and light(12:12 light:dark; lights on at 0700 h) controlled roomswith standard rodent chow and water available ad libitum.The route of drug delivery for all experiments was intra-thecal (sub-dural, peri-spinal) and took 2–3 min to com-plete. An acute catheter application method under briefisofluorane anesthesia (3.0% volume in oxygen) wasemployed, as described previously,30 to inject BDNF orPEG-BDNF at the level of the lumbosacral enlargement.

In vivo half-life assessment

At predetermined time points after intrathecal injections(30, 90, 120, and 240 min), lumbosacral (lumbar) CSF sam-ples were collected under isoflurane anesthesia,31 afterwhich point the rats were immediately euthanized. CSFwas collected from six animals at each time point. CSFsamples were immediately transferred to dry ice and sam-ples were subsequently stored at 2808C for further analy-sis. An ELISA kit was used for the quantitative detectionof BDNF according to manufacturer protocols (Promega).

In vivo biodistribution assessment

Transcardial perfusions were performed as previouslydescribed32,33 with 0.9% saline (5 min) followed by chilled,fresh 2% paraformaldehyde in 0.1% PBS (5 min) on tworats for each condition at the 240-min time point. Cord sec-tions were collected and postfixed in 4% paraformalde-hyde overnight at 48C and then cryoprotected in a 30% su-crose solution. Sections were frozen while embedded inOCT Compound (Tissue-Tek), cryostat sectioned at 10 lmand thaw mounted onto Superfrost Plus Slides (Fisher).

For immunohistochemical analysis, the slide mountedsections were washed in PBS and incubated in a nonspe-cific protein block solution for 3 h. A polyclonal antibodyfor BDNF (Chemicon Ab1779) was diluted 1:1000 in block-ing solution and applied overnight at 48C. Slides wererinsed briefly in blocking solution and an AlexaFluor546secondary antibody (Molecular Probes) was diluted 1:200in blocking solution and applied for 4 h. For the detectionof cell nuclei, the slides were incubated with 1:1000 dilutedDAPI (Molecular Probes) for 10 min, washed, and incu-bated in PBS overnight. Cover-slips were then mountedonto the slides by applying 12–15 lL of Fluoromount-G(Fisher) to each tissue region. The sections were examinedwith confocal microscopy using a Zeiss Pascal LSM micro-scope with a 403 Plan NeoFluor (1.3) oil immersion objec-tive.

Line intensity profiles on acquired images were col-lected with NIH ImageJ software. Confocal images for

PEGYLATION OF BDNF FOR PRESERVED ACTIVITY AND ENHANCED SPINAL DISTRIBUTION 721


BDNF pAb staining of spinal cord cross sections were con-verted to black and white and the pixel length was cali-brated to cord length in lm. The plot profile tool was thenused to measure the fluorescence intensity as a function ofdistance from the edge of the cord. Fluorescence intensityvalues were averaged from eight different sections fromtwo different rats at each condition for a total of 80 differ-ent profiles per condition. Data were normalized againstthe fluorescence intensity at the edge of the cord for eachprofile. The 50% penetration distance was reported as thepoint where the average fluorescence intensity was 50% ofthe average fluorescence intensity at the edge of the cord.

RESULTS

Mass spec characterizationof PEG-BDNF conjugates

MALDI-TOF mass spectrometry was used todetermine the molecular weight of each species pres-ent in a given conjugate mixture. Here, we providerepresentative mass spec profiles for the PEG usedfor conjugation, BDNF, and a conjugate mixture pro-duced by reacting a 60-fold molar excess of PEGwith BDNF in the presence of 125-fold molar excessof reducing agent. Unreacted polymer (5 kDamPEG-ButyrALD) had a molecular weight of5.81 kDa, which is consistent with manufacturerclaims [Fig. 2(a)]. The molecular weight of mono-meric BDNF was 13.6 kDa [Fig. 2(b)], which is alsoconsistent with manufacturer claims. The peak at6.82 kDa was the doubly charged m/z species of theBDNF monomer.

Three additional peaks with molecular weights of20.0, 26.2, and 32.2 kDa were observed in the conju-gate mixture that were not seen in the profiles forthe unreacted polymer and BDNF [Fig. 2(c)]. Themolecular weight values obtained from mass spec-trometry are consistent with the attachment of one(primary conjugate), two (secondary conjugate), orthree (tertiary conjugate) 5.81 kDa PEG molecules toBDNF. The peak at 10.0 kDa was the doubly chargedm/z species of the primary conjugate. Although SDS-PAGE analysis cannot provide quantitative molecularweight information regarding PEGylated proteins,similar qualitative results were obtained. In additionto unreacted BDNF, three conjugate bands of increas-ing molecular weight were observed in SDS-PAGEcorresponding to primary, secondary, and tertiaryconjugate [Fig. 3(a), 603 PEG to BDNF molar excess].

Production of mixtures enriched in primaryand secondary conjugates

To generate PEG-BDNF mixtures with varyingamounts of primary, secondary, and higher order con-

jugates, the influence of reaction conditions on theamount of and type of conjugate generated wasexplored. Fixing the reducing agent concentration at125-fold molar excess and increasing the amount ofPEG present in the reaction buffer from 1-fold molarexcess to 10-fold increased the amount of primary andsecondary conjugate formed [Fig. 3(a)]. Mixtures com-posed of primary and secondary conjugate specieswere also produced when the reducing agent concen-tration was decreased to 10- to 60-fold excess and PEGlevels were held at 60-fold excess [Fig. 3(b)].

Mixtures containing primary, secondary, and terti-ary conjugate were formed at a high reducing agentconcentration (125-fold excess) in the presence ofhigh amounts of PEG (15-fold excess to 60-foldexcess) [Fig. 3(a)]. Specifically, the amount of pri-mary conjugate formed decreased and the amount ofsecondary and tertiary conjugate formed increasedas PEG excess increased from 15- to 60-fold [Fig. 3(a)].In the presence of high levels of PEG (60-fold excess),increasing the reducing agent concentration to 600-fold excess resulted in the production of primary, sec-ondary, tertiary, and higher-order conjugate species[Fig. 3(c)]. Using mPEG-SPA instead of mPEG-Butyr-ALD (i.e., NHS ester chemistry instead of aldehydechemistry) resultant mixtures were also highly in en-riched in tertiary and higher order conjugate products

Figure 2. MALDI-TOF mass spectrometry of (a) PEG spe-cies used for conjugation, (b) BDNF, and (c) PEG-BDNFconjugate mixture prepared with a 60-fold excess of PEGto BDNF and a 125-fold excess of reducing agent to BDNF.



at 15- and 60-fold PEG to BDNF molar excess values(data not shown).

On the basis of these findings, we identified thefollowing approaches to creating mixtures with dif-ferent fractions of primary, secondary, tertiary, orhigher order conjugate species. When 125-fold excessreducing agent is present in solution, conjugate mix-tures containing an abundance of primary conjugateand free BDNF can be formed in the presence of 1-or 2-fold excess PEG; a 10-fold excess of PEG resultsin the production of a mixture of primary, second-ary, and tertiary conjugates. Mixtures enriched inprimary and secondary conjugate are produced inthe presence of lower levels of reducing agent (60-fold excess) and 10- or 60-fold excess PEG. Mixturesenriched in primary, secondary, tertiary, and higherorder conjugate species are prepared in the presenceof higher levels of reducing agent (600-fold excess)and 60-fold excess PEG, or by PEGylating BDNFusing NHS-ester chemistry.

A comparison of coomassie staining and immuno-blotting to detect PEGylated BDNF conjugate mix-tures prepared with aldehyde chemistry were alsoconducted [Fig. 3(c)]. Primary and secondary conju-

gates were detected by antibodies for BDNF. How-ever, tertiary and greater than tertiary conjugate spe-cies were not detected by the antibody for BDNF.

Biological activity assessment

A neurite extension assay with the PC12-trkB cellline was conducted in order to assess the in vitro bio-logical activity of various conjugate mixtures. ThePC12-trkB cell line, which stably expresses the trkBreceptor, extends neurites from the cell body in thepresence of functional BDNF.34 The mean percent ofneurite extension was assessed by counting the frac-tion of cells extending neurites two times greaterthan the length of the cell body. As the concentrationof BDNF increased linearly from 0.1 to 0.5 ng/mL,the percentage of cells extending neurites increasedlinearly [Fig. 4(a)].

The presence of noncovalently attached PEG (noreducing agent) did not alter the in vitro biologicalactivity of BDNF [Fig. 4(b), 603 PEG]. When BDNFwas incubated with reducing agent in the absence ofPEG, slight decreases in biological activity occurred

Figure 3. SDS-PAGE analysis (10% Gels) of conjugate mixtures. (a) PEG-BDNF mixtures prepared with a 125-fold excessof reducing agent (RA) to BDNF with varied molar excesses of PEG to BDNF detected by coomassie staining. (b) PEG-BDNF mixtures prepared with a 60-fold excess of PEG to BDNF and either a 10-fold or a 60-fold molar excess of RA toBDNF detected by coomassie staining. (c) PEG-BDNF mixtures with a 60-fold molar excess of PEG to BDNF and a 600-fold molar excess of RA to BDNF detected by coomassie staining and immunoblotting.



with increasing amounts of reducing agent [Fig. 4(b)60–2403 RA]. At reducing agent to BDNF molarexcess values of 360-fold or greater, minor biologicalactivity reductions became more significant (p <0.05) reaching an ultimate reduction of 1.2-fold (p <0.05) with a 600-fold molar excess of reducing agentto BDNF [Fig. 4(b) 360–6003].

Conjugate mixtures enriched with different frac-tions of free BDNF, primary, secondary, tertiary, andhigher-order species were tested for biological activ-ity. Mixtures enriched in primary conjugate exhib-ited slightly reduced in vitro biological activities rela-tive to unmodified BDNF; however, the measuredreductions were not statistically significant [Fig. 5(a):1/125, 2/125]. Full bioactivity was preserved in mix-

tures composed of primary and secondary conju-gates [Fig. 5(b): 60/10, 60/60). A 1.5-fold (p < 0.001]reduction in activity was observed in the mixturecomposed of primary, secondary, and tertiary conju-gate [Fig. 5(c): 60/125]. The reduction in activity islikely due to the presence of higher levels of tertiaryconjugate. An 8.4-fold (p < 0.001) reduction in bio-logical activity was also observed in mixtures con-taining large amounts of tertiary and higher orderconjugates [Fig. 5(c): 60/600].

The in vitro biological activity was also assessedfor (PEG-SPA) conjugate mixtures produced by NHSester reaction chemistry [Fig. 5(d)] as this reactionwas able to generate extensive PEGylation withoutthe need for a reducing agent. A PEG to BDNF

Figure 4. PC12-trkB neurite extension results. (a) Application of BDNF at concentrations in the range of 0–0.5 ng/mLand (b) Application of 0.5 ng/mL of BDNF treated with PEG only or reducing agent (RA) only at the indicated molarexcess values. * indicates p-value < 0.05 relative to BDNF only condition (two-tailed t-test, error bars are standard error ofthe mean).

Figure 5. PC12-trkB neurite extension results after the application of BDNF and PEG-BDNF conjugate mixtures at a con-centration of 0.5 mg/mL (total BDNF for all species). ‘‘–" represents the expected degree of neurite extension from residualfree BDNF levels in each mixture [free BDNF level obtained from band density analysis of coomassie stained gels, andestimated degree of neurite extension obtained from the dose-response relationship in Fig. 4(a)]. (a) PEG-BDNF mixturesenriched in primary conjugate species. (b) PEG-BDNF mixtures containing primary and secondary conjugate species.(c) PEG-BDNF mixtures containing tertiary and higher order conjugate species. (d) PEG-BDNF mixtures prepared withmPEG-SPA instead of mPEG-ButyrALD. m/n indicates m-fold excess PEG to BDNF and n-fold excess reducing agent toBDNF respectively. * indicates p-value < 0.001 relative to control BDNF (two-tailed t-test, error bars are standard error ofthe mean).



molar excess of 15-fold resulted in the production ofa mixture of secondary, tertiary, and higher orderconjugates with a 3.6-fold (p < 0.001) reduction ofin vitro biological activity [Fig. 5(d): 15/0 (SPA)].Increasing the PEG to BDNF molar excess to 60-foldgenerated a mixture dominated by higher order con-jugates (>84%) with a 10.8-fold (p < 0.001) biologicalactivity reduction [Fig. 5(d): 60/0 (SPA)].

In vivo CSF half-life analysis

To determine whether an improvement in CSFhalf-life is observed when a PEG-BDNF mixtureenriched in primary and secondary conjugate prod-ucts is administered intrathecally, the conjugate mix-ture utilizing a 60-fold molar excess of PEG and a60-fold molar excess of reducing agent was adminis-tered in vivo. This conjugate mixture was selectedover the mixture with the 10-fold molar excess ofreducing agent and the same molar excess of PEGdue to the fact that it contained a lower amount offree BDNF even though both mixtures exhibited afully preserved in vitro biological activity.

The concentration of PEG-BDNF in the CSF after in-trathecal administration was compared with unmodi-fied BDNF over time. ELISA results demonstratedthat the PEG-BDNF mixture was detected with thesame avidity as the BDNF stock (data not shown). Atpredetermined time points after intrathecal injection,CSF samples collected at the lumbar region demon-strated that PEG-BDNF was much more abundantthan BDNF over the entire testing interval. The con-centration of PEG-BDNF relative to BDNF was 6-fold(p < 0.005) higher after 30 min, 6- to 7-fold higher (p <0.05) at 90 and 120 min, and 12.6-fold higher (p <0.005) at 240 min (Fig. 6). From this data, the half-lifeof PEG-BDNF in the CSF was calculated to be 167 minwhile the half-life of BDNF in the CSF was 62.7 min,values which suggest a 2.6-fold improvement in half-life for the fully active PEG-BDNF conjugate mixture.

Tissue penetration profiling

The fully bioactive PEG-BDNF conjugate mixtureexhibited a prolonged bioavailability in the CSF. Theprolonged stability of the protein conjugate mixture inthe CSF may be due to the increased size of the pro-tein, sterically hindering proteases from degradingBDNF. Although improvements in CSF availability ofthe protein were achieved, the ability of the conjugatemixture to penetrate into parenchymal tissue to reachcellular targets must also be assessed. Molecules thatare larger in molecular weight tend to diffuse moreslowly in tissue environments. Thus, it is possible thatthe availability of the conjugated protein to cellular

targets in parenchymal tissue would go down, due toa decrease in the rate at which the protein-conjugatesdiffuse through tissue. For this reason, the extent towhich the conjugate mixture was capable of penetrat-ing into the surrounding parenchymal tissue follow-ing intrathecal delivery was evaluated.

Lumbar region spinal cord tissues were collected4 h after intrathecal injections of BDNF and PEG-BDNF. Sections were fixed, immuno-stained forBDNF, and imaged by confocal microscopy. BDNFwas primarily detected on the periphery of spinalcord tissue [Fig. 7(a)]. PEG-BDNF was detected athigh concentrations in the cord periphery, but wasalso detected at elevated levels within the tissue[Fig. 7(b)]. Control tissue (no injection) did not exhibitbackground staining for human BDNF (data notshown). Line profiles of the relative fluorescence in-tensity vs. tissue depth confirmed that the PEG-BDNFconjugate mixture penetrated deeper into the spinalcord tissue than BDNF [Fig. 7(c)]. Comparisons of thefluorescence intensity values indicate that the PEG-BDNF conjugate mixture declined to 50% of its initialvalue at a depth of 34.57 lm compared with 8.02 lmfor the BDNF only injections. Consistent with theimprovement in CSF protein availability, the fully bio-active PEG-BDNF conjugate mixture also hasimproved availability in parenchymal tissue.

DISCUSSION

Many PEGylation strategies have focused on thepreservation of in vitro biological activity as an impor-

Figure 6. ELISA detection of BDNF in lumbar CSF sam-ples after intrathecal injections of BDNF or PEG-BDNF. *and ** indicate p-values < 0.05 and < 0.005, respectively,for time-matched PEG-BDNF compared with BDNF(unequal variance two-tailed t-test, error bars are standarderror of the mean).



tant criterion for therapeutic development.10,21–23,28,35,36

Site-specific PEGylation strategies to minimize disrup-tion of the binding site are often the most advantageousapproach for biological activity preservation.12,22,28,36–38

The functional sites for BDNF are the trkB and p75 re-ceptor binding regions. The aldehyde reaction chemis-try used in this work preferentially targets the N-termi-nus of a protein and the N-terminus of BDNF is notinvolved in p75 and trkB receptor binding.39 Five outof the 11 lysine residues (potentially reactive amines)on BDNF are within the trkB binding region, four ofwhich are also located within the p75 binding region.39

Preferential N-terminal PEGylation increases thepotential for bioactivity preservation as it decreases thepotential for attachment with these lysine residues in afunctional region. The full preservation of in vitro bio-logical activity with the attachment of one PEG toBDNF for mixtures enriched with PEG-BDNF primaryconjugates is likely due to the attachment of PEG at theN-terminus by nature of the reaction chemistry.

Although aldehyde chemistry is designed to pref-erentially attach a PEG molecule to the N-terminus,increasing the molar excess of PEG creates additional

covalent linkages between PEG molecules and sur-face lysine groups which subsequently decreases theyield of mono PEGylated species.12,29,40,41 In otherwords, even though the covalent attachment of PEGto a protein with aldehyde chemistry is preferentialfor the N-terminus, other amine containing groupson the protein are still reactive to PEG due to theiraccessibility and the presence of unreacted PEG. In aPEGylated mixture of interferon-b, peptide mappinghas demonstrated that primary conjugates consistalmost entirely (>90%) of PEG attached to the N-ter-minal peptide and that secondary covalent linkagespreferentially occur between PEG and a single acces-sible lysine group.42 Similar results demonstratingpreferential N-terminal attachment for the primaryconjugate species and subsequent attachment to ac-cessible lysine groups for the multi-PEGylatedspecies using site-directed aldehyde chemistry havealso been found with EGF,12 tumor necrosis factor,41

and GM-CSF.16 Therefore, even though the pre-ferential site of PEG attachment is at the N-terminus,the result that increasing the molar excess of PEGand/or reducing agent increased the formation of

Figure 7. Confocal imaging along the periphery of BDNF immuno-stained lumbar spinal cord tissue sections. (a) Spinalcord cross-section 4 h after intrathecally injected BDNF. (b) Spinal cord cross-section 4 h after intrathecally injected PEG-BDNF. (c) ImageJ analysis of fluorescence intensity values relative to distance from the spinal cord cross-section periphery.



multi-PEGylated species is consistent with previousfindings.

In general, the clearance rate of a protein isreduced as the number of PEG molecules attached toa protein is increased.23,24,43,44 However, as the func-tional sites for BDNF binding to neurotrophins con-tain lysine residues, increasing the number of PEGmolecules attached to lysine groups on the proteinwill increase the potential for binding disruptions.For this reason, a goal of this study was to identifythe maximum number of PEG molecules that couldbe attached to BDNF without compromising its bio-logical activity. The in vitro biological activity wasfully preserved in the mixtures enriched with pri-mary and secondary conjugates. This demonstratesthat conjugates in a mixture with large amounts ofBDNF bound to one or two PEG molecules werefunctional, or in other words that the preferentialattachment site for a second PEG molecule is notdisruptive to biological activity, suggesting that thesecond PEG molecule is not attached to a surface ly-sine group within the trkB or p75 receptor bindingregions of BDNF.

Increasing the fraction of tertiary and higher orderconjugates in a mixture, on the other hand, reducedthe in vitro biological activity as PEG-BDNF conju-gate mixtures with increasing levels of higher orderconjugates exhibited decreased biological activities.This indicates that attaching several PEG moleculesto BDNF reduces access to binding sites on the mole-cule. The polyclonal antibody used in this work neu-tralizes the bioactivity of BDNF applied to PC12-trkBcells in culture (data not shown) and manufacturerspecifications for the antibody indicate that it neu-tralizes the bioactivity of BDNF but not other neuro-trophins. Immunoblotting data using this antibodyfor BDNF demonstrated that it cannot detect tertiaryand higher order conjugates in the mixture withhigh avidity [Fig. 3(c)], indicating that antibodybinding was disrupted for these conjugate species.As the trkB receptor is more specific for BDNF thanthe other neurotrophins,39 this leads to the conclu-sion that tertiary and higher order PEG attachmentslikely occur on lysine residues within or near to thetrkB receptor binding regions of BDNF.

Elevated concentrations of PEG-BDNF in the CSFwere recognized over the course of 6 h, long enoughto match the CSF turnover rate in humans,45 eventhough the entire CSF volume turns over every 2–4 h for a rat. This turnover is essentially one-way,where CSF from the subarachnoid space is cleared tothe bloodstream or lymph nodes.45–48 The increasedability of PEG-BDNF to diffuse in and out of surfacetissue to avoid clearance, along with the shieldingeffects of PEGylation against proteolytic and enzy-matic degradation products in the CSF and surfacetissue, are likely responsible for its increased persist-

ence in the CSF over time. In the bloodstream,PEGylation can shield a protein from enzymatic deg-radation and antigenic determinants of the immunesystem.35,49 These agents are less abundant in theCSF than in the bloodstream, but there is increasingevidence for the presence of serine proteases and an-tigenic determinants in the CSF,31,50 in addition totheir presence in spinal cord tissue. PEGylation ofBDNF has been shown to reduce its rate of clearancefrom the bloodstream by nearly 10-fold.21 The 2.67-fold improvement in clearance from the CSF wasmore moderate than results in the bloodstream, butis consistent with the high turnover rate of productsfrom the CSF and reduced protease levels and com-ponents of the immune system in the CSF whencompared with the bloodstream.

To improve efficacy following intrathecal adminis-tration, the PEGylated protein must have improvedstability in the CSF and its penetration into the spi-nal cord tissue must be improved. PEG-BDNF conju-gate species in these studies exhibited enhanced pen-etration into spinal cord tissue [Fig. 7(c)]. Eventhough larger molecules typically exhibit a reducedability to diffuse through tissue, PEGylation createsa hydration layer around a protein which increasesits solubility,35,51 reduces its nonspecific electrostaticinteractions,1 and shields it from receptor mediateduptake by surface tissues thereby limiting its avail-ability to the interior tissue.2,26 Therefore, eventhough PEG-BDNF conjugate species are larger insize than BDNF, PEGylated species would beexpected to exhibit enhanced diffusion into spinalcord tissue. Prior work has shown that PEG-BDNFexhibits enhanced diffusion in ex vivo brain tissue sli-ces1 and in vivo penetration into the spinal columnand forebrain after prolonged exposure to continu-ous intrathecal infusions.2 Consistent with andimproving upon these findings, we have shown thatimproved diffusion of PEG-BDNF into spinal cordtissue in vivo also occurred after a single intrathecalinjection.

CONCLUSIONS

Ongoing work continues to validate the merits ofPEGylation for improving the overall efficacy oftherapeutic proteins. Directed and controlled PEGy-lation is a promising approach enabling a high pres-ervation of in vitro biological activity with animproved in vivo pharmacokinetic profile after intra-thecal delivery. PEGylation of BDNF using aldehydechemistry for control of primary and secondary con-jugate formation preserved the in vitro biological ac-tivity of the mixture while improving its penetrationinto spinal cord tissue and half-life in the CSF. The



half-life was improved following intrathecal admin-istration at a duration that is large enough in magni-tude to be effective in humans. The approach hereinused for the PEGylation of BDNF could also beextended to other therapeutic proteins that must bedelivered intrathecally to reduce dosage require-ments and prolong the therapeutic efficacy of treat-ments for a wide range of central nervous systemdisorders.

The authors thank John H. Mahoney of the Universityof Colorado for assistance with animal perfusions. Theauthors also thank Avigen (Alameda, CA) and Amgen(Thousand Oaks, CA) for the respective gifts of mPEG andBDNF used in this work.

References

1. Stroh M, Zipfel WR, Williams RM, Ma SC, Webb WW, Saltz-man WM. Multiphoton microscopy guides neurotrophinmodification with poly(ethylene glycol) to enhance interstitialdiffusion. Nat Mater 2004;3:489–494.

2. Ankeny DP, McTigue DM, Guan Z, Yan Q, Kinstler O, StokesBT, Jakeman LB. Pegylated brain-derived neurotrophic factorshows improved distribution into the spinal cord and stimu-lates locomotor activity and morphological changes afterinjury. Exp Neurol 2001;170:85–100.

3. Kishino A, Katayama N, Ishige Y, Yamamoto Y, Ogo H, Tat-suno T, Mine T, Noguchi H, Nakayama C. Analysis of effectsand pharmacokinetics of subcutaneously administered BDNF.Neuroreport 2001;12:1067–1072.

4. Encinas M, Iglesias M, Llecha N, Comella JX. Extracellular-regulated kinases and phosphatidylinositol 3-kinase areinvolved in brain-derived neurotrophic factor-mediated sur-vival and neuritogenesis of the neuroblastoma cell line SH-SY5Y. J Neurochem 1999;73:1409–1421.

5. Miki K, Fukuoka T, Tokunaga A, Kondo E, Dai Y, NoguchiK. Differential effect of brain-derived neurotrophic factor onhigh-threshold mechanosensitivity in a rat neuropathic painmodel. Neurosci Lett 2000;278:85–88.

6. Eaton MJ, Blits B, Ruitenberg MJ, Verhaagen J, Oudega M.Amelioration of chronic neuropathic pain after partial nerveinjury by adeno-associated viral (AAV) vector-mediated over-expression of BDNF in the rat spinal cord. Gene Ther 2002;9:1387–1395.

7. Bergman I, Burckart GJ, Pohl CR, Venkataramanan R, BarmadaMA, Griffin JA, Cheung NKV. Pharmacokinetics of IgG andIgM anti-ganglioside antibodies in rats and monkeys after intra-thecal administration. J Pharmacol Exp Ther 1998; 284:111–115.

8. Veronese FM, Caliceti P, Schiavon O, Sergi M. Polyethyleneglycol-superoxide dismutase, a conjugate in search of exploi-tation. Adv Drug Deliv Rev 2002;54:587–606.

9. Dang W, Colvin MO, Brem H, Saltzman WM. Covalent cou-pling of methotrexate to dextran enhances the penetration ofcytotoxicity into a tissue-like matrix. Cancer Res 1994;54:1729–1735.

10. Francis GE, Fisher D, Delgado C, Malik F, Gardiner A, NealeD. PEGylation of cytokines and other therapeutic proteinsand peptides: The importance of biological optimisation ofcoupling techniques. Int J Hematol 1998;68:1–18.

11. Belcheva N, Woodrow-Mumford K, Mahoney MJ, SaltzmanWM. Synthesis and biological activity of polyethylene glycol-mouse nerve growth factor conjugate. Bioconjug Chem 1999;10:932–937.

12. Lee H, Jang IH, Ryu SH, Park TG. N-terminal site-specificmono-PEGylation of epidermal growth factor. Pharm Res2003;20:818–825.

13. Krinner EM, Hepp J, Hoffmann P, Bruckmaier S, Petersen L,Petsch S, Parr L, Schuster I, Mangold S, Lorenczewski G, Lut-terbuse P, Buziol S, Hochheim I, Volkland J, Molhoj M, Sris-kandarajah M, Strasser M, Itin C, Wolf A, Basu A, Yang K,Filpula D, Sorensen P, Kufer P, Baeuerle P, Raum T. A highlystable polyethylene glycol-conjugated human single-chainantibody neutralizing granulocyte-macrophage colony stimu-lating factor at low nanomolar concentration. Protein EngDes Sel 2006;19:461–470.

14. Yun Q, Xing WC, Mal GG, Su ZG. Preparation and character-ization of mono-PEGylated consensus interferon by a novelpolyethylene glycol derivative. J Chem Technol Biotechnol2006;81:776–781.

15. Ramon J, Saez V, Baez R, Aldana R, Hardy E. PEGylatedinterferon-alpha 2b: A branched 40K polyethylene glycol de-rivative. Pharm Res 2005;22:1374–1386.

16. Kinstler OB, Brems DN, Lauren SL, Paige AG, HamburgerJB, Treuheit MJ. Characterization and stability of N-termi-nally PEGylated rhG-CSF. Pharm Res 1996;13:996–1002.

17. Tsutsami Y, Tsunoda S-I, Kamada H, Kihira T, Kaneda Y,Ohsugi Y, Mayumi T. PEGylation of interleukin-6 effectivelyincreases its thrombopoietic potency. Thromb Haemost1997;77:168–173.

18. Pardridge WM, Wu D, Sakane T. Combined use of carboxyl-directed protein pegylation and vector-mediated blood-brainbarrier drug delivery system optimizes brain uptake of brain-derived neurotrophic factor following intravenous adminis-tration. Pharm Res 1998;15:576–582.

19. Katre NV, Knauf MJ, Laird WL. Chemical modification ofrecombinant interleukin 2 by polyethylene glycol increases itspotency in the murine Meth A sacroma model. Proc NatlAcad Sci USA 1987;84:1487–1491.

20. Hudecz F, Clegg JA, Kajtar J, Embleton MJ, Pimm MV, Sze-kerke M, Baldwin RW. Influence of carrier on biodistributionand in vitro cytotoxicity of methotrexate-branched polypep-tide conjugates. Bioconjug Chem 1993;4:25–33.

21. Sakane T, Pardridge WM. Carboxyl-directed pegylation ofbrain-derived neurotrophic factor markedly reduces systemicclearance with minimal loss of biologic activity. Pharm Res1997;14:1085–1091.

22. Leong SR, DeForge L, Presta L, Gonzalez T, Fan A, ReichertM, Chuntharapai A, Kim KJ, Tumas DB, Lee WP, Gribling P,Snedecor B, Chen H, Hsei V, Schoenhoff M, Hale V, DeveneyJ, Koumenis I, Shahrokh Z, McKay P, Galan W, Wagner B,Narindray D, Hebert C, Zapata G. Adapting pharmacokineticproperties of a humanized anti-interleukin-8 antibody fortherapeutic applications using site-specific pegylation. Cyto-kine 2001;16:106–119.

23. Koumenis IL, Shahrokh Z, Leong S, Hsei V, Deforge L,Zapata G. Modulating pharmacokinetics of an anti-interleu-kin-8 F(ab0)(2) by amine-specific PEGylation with preservedbioactivity. Int J Pharm 2000;198:83–95.

24. Clark R, Olson K, Fuh G, Marian M, Mortensen D, TeshimaF, Chang S, Chu H, Mukku V, CanovaDavis E, Somer T, Cro-nin M, Winkler M, Wells JA. Long-acting growth hormonesproduced by conjugation with polyethylene glycol. J BiolChem 1996;271:21969–21977.

25. Bailon P, Berthold W. Polyethylene glycol-conjugated phar-maceutical proteins. Pharm Sci Technol Today 1998;1:352–356.

26. Roberts MJ, Bentley MD, Harris JM. Chemistry for peptideand protein PEGylation. Adv Drug Deliv Rev 2002;54:459–476.

27. Kinstler O, Molineux G, Treuheit M, Ladd D, Gegg C. Mono-N-terminal poly(ethylene glycol)-protein conjugates. AdvDrug Deliv Rev 2002;54:477–485.



28. Rosendahl MS, Doherty DH, Smith DJ, Carlson SJ, ChlipalaEA, Cox GN. A long-acting, highly potent interferon alpha-2conjugate created using site-specific PEGylation. BioconjugChem 2005;16:200–207.

29. Na DH, Lee KC, DeLuca PP. PEGylation of octreotide: II.Effect of N-terminal mono-PEGylation on biological activityand pharmacokinetics. Pharm Res 2005;22:743–749.

30. Milligan ED, Langer SJ, Sloane EM, He L, Wieseler-Frank J,O’Connor K, Martin D, Forsayeth JR, Maier SF, Johnson K, Cha-vez RA, Leinwand LA,Watkins LR. Controlling pathological painby adenovirally driven spinal production of the anti-inflamma-tory cytokine, interleukin-10. Eur J Neurosci 2005;21:2136–2148.

31. Milligan ED, O’Connor KA, Nguyen K, Armstrong CB, Twin-ing C, Gaykema RPA, Holguin A, Martin D, Maier SF, Wat-kins LR. Intrathecal HIV-1 envelope glycoprotein gp120 indu-ces enhanced pain states mediated by spinal cord proinflam-matory cytokines. J Neurosci 2001;21:2808–2819.

32. Mannes A, Caudle R, O’Connell B, Iadarola M. Adenoviralgene transfer to spinal cord neurons: Intrathecal vs. intrapar-enchymal administration. Brain Res 1998;793:1–6.

33. Milligan ED, Sloane EM, Langer SJ, Cruz PE, Chacur M, Spa-taro L, Wieseler-Frank J, Hammack SE, Maier SF, Flotte TR,Forsayeth JR, Leinwand LA, Chavez RA, Watkins LR. Con-trolling neuropathic pain by adeno-associated virus drivenproduction of the anti-inflammatory cytokine, interleukin-10.Mol Pain 2005;1:1–9.

34. Greene LA, Aletta JM, Rukenstein A, Green SH. PC12 pheochro-mocytoma cells: Culture, nerve growth factor treatment, and ex-perimental exploitation. Methods Enzymol 1987;147: 207–216.

35. Molineux G. Pegylation: Engineering improved biopharmaceut-icals for oncology. Pharmacotherapy 2003;23(8, Part 2): 3S–8S.

36. Gaertner H, Offord R. Site-specific attachment of functional-ized poly(ethylene glycol) to the amino terminus of proteins.Bioconjug Chem 1996;7:38–44.

37. Lee H, Park TG. Preparation and characterization of mono-PEGylated epidermal growth factor: Evaluation of in vitrobiologic activity. Pharm Res 2002;19:845–851.

38. Kubetzko S, Sarkar CA, Pluckthun A. Protein PEGylationdecreases observed target association rates via a dual block-ing mechanism. Mol Pharm 2005;68:1439–1454.

39. Robinson RC, Radziejewski C, Stuart DI, Jones EY. Structureof the brain-derived neurotrophic factor neurotrophin 3 het-erodimer. Biochemistry 1995;34:4139–4146.

40. Cindric M, Cepo T, Galic N, Bukvic-Krajacic M, Tomczyk N,Vissers JPC, Bindila L, Peter-Katalinic J. Structural characteri-zation of PEGylated rHuG-CSF and location of PEG attach-ment sites. J Pharm Biomed Anal 2007;44:388–395.

41. Edwards CK, Martin SW, Seely J, Kinstler O, Buckel S, Bend-ele AM, Cosenza ME, Feige U, Kohno T. Design of PEGy-lated soluble tumor necrosis factor receptor type I (PEGsTNF-RI) for chronic inflammatory diseases. Adv Drug DelivRev 2003;55:1315–1336.

42. Arduini RM, Li ZF, Rapoza A, Gronke R, Hess DM, Wen DY,Miatkowski K, Coots C, Kaffashan A, Viseux N, Delaney J,Domon B, Young CN, Boynton R, Chen LL, Chen LQ, Bet-zenhauser M, Miller S, Gill A, Pepinsky RB, Hochman PS,Baker DP. Expression, purification, and characterization ofrat interferon-beta, and preparation of an N-terminally PEGy-lated form with improved pharmacokinetic parameters. Pro-tein Expr Purif 2004;34:229–242.

43. Baker DE. Pegylated interferons. Rev Gastroenterol Disord2001;1:87–99.

44. Eliason JF. Pegylated cytokines: Potential application inimmunotherapy of cancer. BioDrugs 2001;15:705–711.

45. Kandel ER, Schwartz JH, Jessell TM. Principles of Neural Sci-ence. New York: McGraw-Hill; 2000.

46. Weller RO, Kida S, Zhang ET. Pathways of fluid drainagefrom the brain—Morphological aspects and immunologicalsignificance in rat and man. Brain Pathol 1992;2:277–284.

47. Johnston M. The importance of lymphatics in cerebrospinalfluid transport. Lymphat Res Biol 2003;1:41–45.

48. Abbott NJ. Evidence for bulk flow of brain interstitial fluid:Significance for physiology and pathology. Neurochem Int2004;45:545–552.

49. Greenwald RB, Yang K, Zhao H, Conover CD, Lee S, FilpulaD. Controlled release of proteins from their poly(ethyleneglycol) conjugates: Drug delivery systems employing 1,6-elimination. Bioconjug Chem 2003;14:395–403.

50. Scarisbrick IA, Towner MD, Isackson PJ. Nervous system-specific expression of a novel serine protease: Regulation inthe adult rat spinal cord by excitotoxic injury. J Neurosci1997;17:8156–8168.

51. Dhalluin C, Ross A, Leuthold LA, Foser S, Gsell B, Muller F,Senn H. Structural and biophysical characterization of the 40kDa PEG-interferon-alpha(2a) and its individual positionalisomers. Bioconjug Chem 2005;16:504–517.



pegylation of brain-derived neurotrophic factor for preserved biological activity and enhanced...

Documents