P

P1

DCCa

b

c

d

e

a

ARRAA

KCEIIRS

1

fhfa

n

O

h1

ARTICLE IN PRESSG ModelRBI-10466; No. of Pages 11

Process Biochemistry xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Process Biochemistry

jo ur nal home p age: www.elsev ier .com/ locate /procbio

roduction and characterisation of recombinant human chaperonin0 for treatment of inflammatory disease

ean J. Naylora, Ben Huntd, Angelo Guidolind, Allan W. Heyd,1, Stan Bastirasd,hristopher J. de Bakkera,b, David Y. Chinb,2, Christopher P. Marquisc, Daniel Lamberta,hristopher B. Howardb, Caroline A. Dobbina, Stephen M. Mahlerb,e,∗

Invion Ltd. (Formerly CBio Ltd.), Brisbane, QLD 4001, AustraliaAustralian Institute for Bioengineering and Nanotechnology, University of Queensland, QLD 4072, AustraliaSchool of Biotechnology and Biomolecular Sciences, University of New South Wales, Sydney, NSW 2052, AustraliaHospira Adelaide Pty Ltd., Thebarton, SA 5031, AustraliaSchool of Chemical Engineering, University of Queensland (UQ), QLD 4072, Australia

r t i c l e i n f o

rticle history:eceived 13 May 2015eceived in revised form 23 June 2015ccepted 24 June 2015vailable online xxx

eywords:pn10ndotoxinmmunomodulatornflammationAMPsystemic Lupus Erythematosus (SLE)

a b s t r a c t

Human chaperonin 10 (Cpn10) and chaperonin 60 (Cpn60) fulfil an essential role in mitochondrialprotein folding. Cpn10 is located in the mitochondrial matrix; however, it has also been detected inextra-mitochondrial compartments where it has demonstrated anti-inflammatory and immunoregula-tory activity, suggesting a potential therapeutic value for the treatment of inflammatory and autoimmunedisorders that corresponds with its recently proposed role as a resolution-associated molecular patterns(RAMPs) molecule. Ala-Cpn10, a recombinant, minimally modified Cpn10 synthesised in Escherichia coli,is formulated for use in clinical trials and pre-clinical studies with intravenous or subcutaneous adminis-tration. Herein, we report the development of a bioprocess for the production of ∼115 g of recombinanthuman Ala-Cpn10 from a 100 L E. coli fermentation with >99% purity (SDS-PAGE), <0.6 EU mg−1 endo-toxin, <18 pg mg−1 DNA and 3.2% molecular variants. This bioprocess was achieved through a carefuloptimisation of the gene construct, the promoter system and the fermentation process. RecombinantAla-Cpn10 produced in this process is active as a molecular chaperone indicating correct tertiary andquaternary structure and the stability profile indicates no significant changes with storage as a liquid

◦

for at least 3 years at 2–8 C. The results of validated characterisation assays demonstrate that purifiedAla-Cpn10 produced using the optimised process reported here is suitable for its intended purpose as aninvestigational drug product, and this material is currently being tested in a phase II study for efficacy,safety and impact on biomarkers in subjects with mildly active Systemic Lupus Erythematosus (SLE)under US IND 116156.

. Introduction

Chaperonin 10 (Cpn10) is a member of the stress or heat shockamily of proteins. This group of proteins is among the most

Please cite this article in press as: D.J. Naylor, et al., Production and chaof inflammatory disease, Process Biochem (2015), http://dx.doi.org/10

ighly conserved in nature, with respect to both structure andunction. Human Cpn10 has 100% homology with bovine, porcinend rabbit Cpn10 while rat Cpn10 differs from human by one

∗ Corresponding author at: Australian Institute for Bioengineering and Nanotech-ology, University of Queensland, QLD 4072, Australia.

E-mail address: s.mahler@eng.uq.edu.au (S.M. Mahler).1 Present address: Circassia Ltd., Northbrook House, Robert Robinson Avenue,xford Science Park, OX4 4GA, UK.2 Present address: Biosceptre International, North Ryde, NSW 2113, Australia.

ttp://dx.doi.org/10.1016/j.procbio.2015.06.022359-5113/© 2015 Elsevier Ltd. All rights reserved.

© 2015 Elsevier Ltd. All rights reserved.

amino acid [1]. Localised in various intracellular compartments,heat shock proteins (HSPs) have a well-defined role in proteinassembly, translocation and degradation. Cpn60 together with itsco-chaperone Cpn10, reside primarily within mitochondria wherethey fulfil an essential role in mitochondrial protein folding [2].Stress proteins are up-regulated in response to hypoxia, hypo-glycaemia, ionising radiation and heat-shock [3,4] and releasedfrom cells through as yet undefined mechanisms. Once in theextracellular environment, a number of stress proteins havedemonstrated autocrine, paracrine and endocrine immunologi-cal activity [5]. Cpn10 has demonstrated anti-inflammatory and

racterisation of recombinant human chaperonin 10 for treatment.1016/j.procbio.2015.06.022

immunoregulatory activity and has been proposed as a memberof the resolution-associated molecular patterns (RAMPs) familyof molecules [6]. As a RAMP, Cpn10 is hypothesised to regulatethe acute immune response with therapeutic potential to restore

ING ModelP

2 ioche

ha

oitCm(

flmsuwti[gts

bsts(iwoPe(tpl

tcmsembBiq[ap

cawaespacrc

s(

ARTICLERBI-10466; No. of Pages 11

D.J. Naylor et al. / Process B

omeostasis in a dysregulated setting such as chronic inflammationnd autoimmune disease.

Cpn10 members are dome-shaped, heptameric rings comprisedf identical 10 kDa subunits [7,8]. The test material described hereins Cpn10 produced in bacteria as a recombinant protein. An N-erminus comparison of native human Cpn10 with recombinantpn10 reveals the substitution of two hydrogen (H) groups for aethyl (CH3) and an amino (NH3

+) group in recombinant Cpn10Ala-Cpn10).

In vitro, recombinant Cpn10 inhibited TNF-� and IL-6 secretionrom human peripheral blood mononuclear cells (PBMC) stimu-ated with bacterial lipopolysaccharide (LPS) [9]. Data from animal

odels of inflammation show that Cpn10 pre-treatment of miceignificantly reduced serum concentrations of TNF-� and regulatedpon activation normal T cell expressed and secreted (RANTES)hile increasing serum IL-10 following sub-lethal LPS adminis-

ration [9]. Observations of potential immunomodulatory activityn animal models of human disease, including adjuvant arthritis10], experimental autoimmune encephalitis [11], allogeneic skinrafting [12], Graft-versus-host disease [9] and systemic lupus ery-hematosus [13] provide a rationale for clinical trials of Cpn10afety and efficacy in humans.

Nine clinical studies have been completed investigating recom-inant Cpn10 administration in various populations. In the clinicaltudies conducted to date there appear to be no hepatic, haema-ologic or other obvious effects of concern. The proof of concepttudies in subjects with rheumatoid arthritis (RA) and psoriasisPS), showed modest immunomodulatory effects and a reductionn the signs and symptoms of disease from pre-treatment levels

ith intravenous (i.v.) Cpn10 dosing. Cpn10 reduced productionf TNF-�, IL-1� and IL-6 in response to LPS stimulation of humanBMC from healthy volunteers and patients with autoimmune dis-ase [multiple sclerosis (MS), psoriasis (PS), rheumatoid arthritisRA)] [9,14,15]. Combined with the toxicokinetic profile generatedhus far, these data confirm the safety and tolerability of Cpn10 inatients and support further investigation in subjects with systemic

upus erythematosus or other autoimmune diseases.Currently the standard of care for subjects with mild-

o-moderate manifestations of SLE includes small moleculeompounds such as non-steroidal anti-inflammatory drugs, anti-alarial drugs such as hydroxychloroquine and corticosteroids

uch as prednisone. Patients with active disease and organ threat-ning lupus are usually treated with immunosuppressants (eitherethotrexate, leflunomide or azathioprine). Recently approved

iological agents (belimumab) target specific cytokines, such asLyS [B lymphocyte stimulator, also known as BAFF (B-cell activat-

ng factor)] a cytokine essential for the survival of B cells, althoughuestions remain regarding the durability of the clincial response16]. Thus opportunities still exist for the development of new, safend effective treatments for SLE, including other biological thera-eutics for this autoimmune disease.

The production of biologics (such as proteins) in an Escherichiaoli system incurs the risk of contamination with molecules knowns pathogen-associated molecular patterns (PAMPs), the mostell-known example being LPS. PAMPs describe the evolution-

rily conserved microbial components that are recognised by theukaryotic host through binding to pattern-recognition receptorsuch as CD14 or the Toll-like receptors (TLRs) [17]. Endotoxins arero-inflammatory and thus their presence in biologics may be toxicnd result in artefactual effects on eukaryotic cells. It is thereforeritical that recombinant protein preparations have minimal impu-ities such as endotoxin, DNA and other microorganism-derived

Please cite this article in press as: D.J. Naylor, et al., Production and chaof inflammatory disease, Process Biochem (2015), http://dx.doi.org/10

omponents.Regulatory bodies stipulate that the manufacturer of a biologic

hould assess and characterise both process-related impuritiesi.e. host cell proteins, endotoxin, DNA, etc.), and the levels of

PRESSmistry xxx (2015) xxx–xxx

product-related substances (i.e. molecular variants of the nativeproduct) (EMEA, CPMP/ICH/365/96). The product specification setsout an acceptance criteria for impurities and product-related sub-stances within numerical limits, and includes some detail of specifictests and analytical procedures used. Characterisation of the prod-uct is performed in the early stages of bioprocess development, andfollowing subsequent changes to the bioprocess. Apart from meet-ing specifications for process-related impurities, an important facetof bioprocess optimisation is minimising levels of product variant.

In compliance with regulatory requirements and in order tofacilitate the safe and effective testing of Cpn10 administration inpatients with autoimmune disease, it was essential that a rigor-ous purification methodology be devised to remove process-relatedimpurities from Cpn10.

This study describes the bioprocess development, productionand subsequent characterisation of recombinant human Cpn10produced in E. coli, the first chaperone protein to enter clinical trialsfor treatment of disease indications associated with inflammation.

2. Materials and methods

2.1. E. coli host strains

XL1-Blue (endA1 gyrA96 (nalR) thi-1 recA1 relA1 lac glnV44 hsdR1F′[::Tn10 proAB+ lacIq �(lacZ)M15]) (Stratagene, CA, USA). BR067(glnV44 (AS) rfbC1 endA1 spoT1 thi-1 hsdR17 creC510 �ompT) isa proprietary derivative of E. coli strain MM294 (ATCC #33625)developed at Hospira, Adelaide, Australia.

2.2. Cloning and expression of human Cpn10 in E. coli

Initial cloning and expression of Cpn10 was as described inSomodevilla–Torres et al. [18]. Ala-Cpn10 pPL550, a heat-inducibleexpression plasmid encoding human Cpn10 with an additionalN-terminal alanine was transformed into XL1-Blue and Cpn10expression examined in complex media. A master cell bank (MCB)of the XL1-Blue production clone was established at Acyte Pty Ltd.,Sydney, Australia which was subsequently expanded at Hospira,Adelaide, Australia. Synthetic genes with the appropriate flankingrestriction enzyme cleavage sites were obtained from GENEART AG,Regensberg, Germany.

Cpn10 expression from Hospira’s host strain BR067 andlac promoter-based expression vectors was examined followinggrowth and induction in defined media. A final production strain formanufacture of clinical trial material was selected on the basis ofhigh level Cpn10 expression and concomitant low level expressionof product variants. A MCB was established from a single selectedclone and a working cell bank (WCB) generated by expansion ofa single MCB vial. Both cell banks were characterised to ensuresuitability for use in manufacture of clinical grade material.

2.3. GMP fermentation: XL1-Blue cell line

A MCB of XL1-Blue with the Ala-Cpn10 pPL550 plasmid was‘pre-cultured’ overnight in Soya Peptone Broth with no antibioticsupplementation at 30 ◦C. An inoculum culture was subsequentlyprepared for the 150 L bioreactor maintaining the above mediaand growth temperature parameters. An aliquot of this inoculumwas dispensed into a 150 L bioreactor in a minimal media supple-mented with 1.5 g L−1 vegetable peptone and 1.0 g L−1 yeast extractand no antibiotic supplementation. The bioreactor cultivation didnot require nutrient feeding and the temperature was maintained

racterisation of recombinant human chaperonin 10 for treatment.1016/j.procbio.2015.06.022

throughout the growth phase at 30 ± 2 ◦C. The pH was maintainedat 7.0 ± 0.3 by the addition of an ammonia solution. Induction ofCpn10 was achieved by a temperature shift to 42 ◦C when the cul-ture reached an optical density at 600 nm (OD600 nm) of 10 and

ING ModelP

iochem

ft

2

oWktsnwbm6au

2

Mltbflo

2

c(eet

2a

resogw0

2

2

BSTlB1Ps

2

e

ARTICLERBI-10466; No. of Pages 11

D.J. Naylor et al. / Process B

urther incubation at 42 ± 0.5 ◦C for 3 h at which time the fermen-ation reached a cell density of between OD600 nm 20 and 25.

.4. GMP fermentation: BR067 cell line

In brief, 100 L fed batch-fermentations for Cpn10 were carriedut as follows. An inoculum culture, derived from a vial of theCB, was grown in Vegetable Peptone Broth supplemented with

anamycin. The inoculum was used to seed a 150 L fermenter con-aining 75 L chemically defined batch media (without antibioticelection) and following overnight batch growth, the glucose basedutrient feed was supplied on demand. The cultivation temperatureas maintained at 37 ± 2 ◦C throughout, pH maintained at 7 ± 0.3

y the addition of an ammonia solution and dissolved oxygen wasaintained at >30%. Induction was commenced by the addition of

.0 g isopropyl-�-d-thiogalactopyranoside (IPTG) when the culturechieved a cell density of OD600 nm ∼45 and allowed to proceed forp to 5 h with final culture cell densities in excess of OD600 nm 120.

.5. Cell lysis and preparation of a soluble lysate

Bacterial cells were collected with a flow-through WestfaliaSB-7 centrifuge and resuspended in 25 mM Tris–HCl pH 8.0. Fol-

owing overnight storage the cells were disrupted by 3 passageshrough a high pressure homogeniser (APV CD30) at 7000 psi. Theacterial homogenate was processed with a flow-through West-alia MSB-7 centrifuge and ∼90 L of clarified, Cpn10-containing,ysate was collected and either immediately processed or storedvernight at 4–8 ◦C.

.6. Purification of Cpn10

A three-step downstream process was developed for the purifi-ation of Cpn10, incorporating Big Bead Sulfopropyl-SepharoseSP-BB) cation exchange, Q-Sepharose Fast Flow (Q-FF) anionxchange and High Performance Sepharose SP (SP-HP) cationxchange chromatography. Chromatography was carried out usinghe K-Prime 40-II system (Millipore).

.7. De-pyrogenation of chromatography columns, containersnd buffers

All ion-exchange (IEX) chromatography columns were depy-ogenated by washing in 1 M NaOH with 100 mM NaCl andquilibrated with buffer until eluates returned to the pH of thepecific buffers. All containers used for buffer storage and receiptf column eluates in the various purification steps were depyro-enated using 1 M NaOH followed by extensive rinsing with MilliQater (Millipore) until washings were pH neutral. A pre-column

.2 � filter was used for the Q-FF and SP-HP steps.

.8. Big bead Sulfopropyl-Sepharose chromatography

.8.1. XL1-Blue cell lineThe filtered lysate was loaded over 20–40 min onto a SP-

B cation exchange column (BPG 300/500 with 8.6 L of SP-BBepharose resin, GE Biosciences) pre-equilibrated with 25 mMris–HCl, pH 8.0 (Buffer A), at 75 cm h−1 linear flow-rate and aoading rate of up to 17 g Cpn10 L−1 of resin. After washing withuffer A, the captured Cpn10 was eluted with Buffer A containing50 mM NaCl and a single fraction of 6–11 L was collected. SDS-AGE analysis showed the SP-BB pool to be >95% pure (data nothown).

Please cite this article in press as: D.J. Naylor, et al., Production and chaof inflammatory disease, Process Biochem (2015), http://dx.doi.org/10

.8.2. BR067 cell lineThe filtered lysate was loaded over ∼110 min onto a SP-BB cation

xchange column (BPG 300/500 with 18 L of SP-BB Sepharose resin,

PRESSistry xxx (2015) xxx–xxx 3

GE Biosciences) pre-equilibrated with Buffer A, at 75 cm h−1 lin-ear flow-rate and a loading rate of up to 31 g Cpn10 L−1 of resin.After washing with Buffer A, the captured Cpn10 was eluted withBuffer A containing 150 mM NaCl and a single fraction of 9–16 L wascollected. SDS-PAGE analysis showed the SP-BB pool to be >95%pure.

2.8.3. Q-Sepharose fast flow chromatographyThe SP-BB eluate was concentrated to half the starting volume

by tangential flow filtration (TFF) and diafiltered against 5 vol-umes of Buffer A at room temperature. The dialysed SP-BB poolwas loaded onto a BPG 200/500 column packed with 4.7 L of Q-FFanion exchange resin (GE Biosciences) pre-equilibrated in Buffer Aat a linear flow-rate of 75 cm h−1 and a loading rate of up to 35 gCpn10 L−1 of resin. The Q-FF anion exchange flowthrough contain-ing the Cpn10 was collected. Under the loading conditions, themajority of the E. coli cell proteins, nucleic acid and endotoxinremained bound to the matrix.

2.9. Sulfopropyl-Sepharose high performance chromatography

The Q-FF flowthrough from the XL1-Blue cell line was applied toa BPG 100/500 column packed with 1.67 L of SP-HP resin (GE Bio-sciences), pre-equilibrated with 50 mM Tris–HCl pH 7.6 (Buffer B),at ≤15 g Cpn10 L−1 of resin. The Q-FF flowthrough from the BR067cell line was applied to a BPG 140/500 column packed with 3.4 Lof SP-HP resin, pre-equilibrated with Buffer B, at ≤15 g Cpn10 perlitre of resin. The bound Cpn10 was eluted with a 13.3 mM/columnvolume NaCl gradient and 0.5 L fractions were collected. The frac-tions were pooled according to size exclusion chromatography(SEC), HPLC and SDS-PAGE analyses. The Cpn10 protein and NaClion concentrations were determined by UV Absorbance at 280 nmand the ion selective electrode (ISE) method (IDEXX Laboratories,Australia), respectively.

2.10. Formulation

The pooled SP-HP fractions were diluted with 50 mM Tris–HClpH 7.6–5 mg mL−1. Prior to dilution with 50 mM Tris–HCl pH 7.6,NaCl was dissolved in the buffer to achieve Na+ and Cl− concentra-tions of 152 mM and 190 mM respectively in the final formulation.The formulated Cpn10 was filtered through a 0.2 �m filter underaseptic conditions and 500 ml (2.5 g Cpn10) dispensed into 500 mLpyrogen-free plastic bottles.

2.11. Spectrophotometric and SDS-PAGE analysis of Cpn10

SDS PAGE of E. coli cell lysates and chromatography fractionswas performed using NuPAGE 4–12% Bis–Tris gradient gels (Invi-trogen). Gels were Coomassie Brilliant Blue stained and each gelincluded Cpn10 protein and molecular weight standards. Cpn10concentrations were determined by UV absorbance on a Cary 50Bio UV Spectrophotometer (Varian) at 280 nm using an extinctioncoefficient of 0.41 (mg mL−1)−1 cm−1.

2.12. Mass spectrometry

Mass spectrometry was performed at the University of Ade-laide Proteomics Centre, Adelaide, Australia. Whole moleculemasses were determined by direct injection into an API-100 massspectrometer (Applied Biosystems). Alternatively samples weredesalted prior to analysis by C8 reversed-phase HPLC using a

racterisation of recombinant human chaperonin 10 for treatment.1016/j.procbio.2015.06.022

10–60% acetonitrile gradient in the presence of 0.1% trifluoroaceticacid (TFA). Protein modifications were identified by a combina-tion of peptide mapping and mass spectrometry. Protein wasdigested with either Trypsin (Promega) or Endoproteinase LysC

ING ModelP

4 ioche

(tSwP

srsIsB

2

tGmN(cpm

2

csooNetw

2

7vbrttww

2

napaefcbtrKr

ARTICLERBI-10466; No. of Pages 11

D.J. Naylor et al. / Process B

Sigma–Aldrich). Digested samples were analysed with an acetoni-rile gradient on a Waters Symmetry 300 C18 column on a 2695eparation module (Waters) under the control of Millennium soft-are. UV peaks were manually collected and sent to the Adelaide

roteomics Centre for identification.Peptide masses were determined with a MALDI TOF/TOF mass

pectrometer (Ultraflex II, Bruker Daltonic GmbH) operating ineflectron mode under the control of flexControl software. Masspectra were analysed with flexAnalysis (Bruker Daltonic GmbH).ons were fragmented by laser-induced dissociation and theequence was assigned using the “Rapid De Novo” module ofioTools (Bruker Daltonic GmbH).

.13. Host cell DNA and endotoxin quantification

The concentration of E. coli DNA in purified Cpn10 prepara-ions was determined by Glycotope Biotechnology (Heidelberg,ermany) using the Threshold Total DNA Assay according to theanufacturer’s instructions (Molecular Devices, CA, USA; Producto. R8022). A quantitative chromogenic Limulus amebocyte lysate

LAL) test (BioWhittaker, QCL-1000) was used to estimate the con-entration of gram-negative bacterial endotoxin present in the finalreparation of Cpn10. Assays were performed according to theanufacturer’s instructions.

.14. Size exclusion chromatography

Soluble oligomer content was determined by size exclusionhromatography using a Superdex 200 HR 10/30 (GE Biosciences)ize exclusion column. Samples were analysed by injecting 100 �Lnto the column and separated by isocratic elution at a flow ratef 0.5 mL/min using 50 mM Sodium Phosphate pH 7.0 + 150 mMaCl as the elution buffer. The apparent molecular weight ofluting species was estimated by comparison of peak retentionimes with globular protein standards of a known moleculareight.

.15. Glutaraldehyde cross-linking of Cpn10

An amount of 580 �g of human recombinant Cpn10 in PBS (pH.4) was incubated with 0.01% (w/w) glutaraldehyde (APS) in a totalolume of 300 �L at 25 ◦C for 30 min. Reactions were quenchedy the addition of 15 �L of 2 M Tris–HCl pH 8.0. An aliquot of theeaction mixture was analysed by size exclusion chromatographyo separate the different oligomeric species and the peak eluting athe same retention time as the non-cross-linked Cpn10 oligomeras collected in two 0.5 mL fractions and analysed by SDS-PAGEith silver staining.

.16. Stability testing of purified and formulated Cpn10

Cpn10 batches 1, 2 and 3 (Table 1) were sterile filled underitrogen in glass vials at ∼75 mg mL−1 in 50 mM Tris–HCl pH 7.6nd 150 mM NaCl. Aliquots were placed at 2–8 ◦C and −80 ◦C for aeriod of 3 years. The integrity of the samples was tested at day 0nd at every six months up to 3 years. Stability tests included size-xclusion chromatography for heptamer verification, SDS-PAGEor purity and degradation, an A280 nm measurement for protein

Please cite this article in press as: D.J. Naylor, et al., Production and chaof inflammatory disease, Process Biochem (2015), http://dx.doi.org/10

oncentration, endotoxin analysis for sterility and a quantitativeioassay for activity. The bioassay constitutes the ability of Cpn10o work as a co-chaperone for GroEL (E. coli Cpn60) in the process ofhodanese refolding [19]. The refolding buffer was 20 mM MOPS-OH pH 7.5, 100 mM KCl and 10 mM MgCl2; bovine mitochondrialhodanese was purchased from Sigma.

PRESSmistry xxx (2015) xxx–xxx

2.17. Detection of proinflammatory E. coli contaminants by HEKcell assays

In order to detect trace levels of contaminants derived from bac-terial host, a bioassay was developed based on Toll-like receptor(TLR)-transfected HEK cells and a luciferase reporter system (i.e.pNifty2-Luc plasmid DNA, InvivoGen). HEK cells transfected withTLR2, TLR3, TLR4/MD2/CD14 and TLR9 were created for detectionof TLR ligands (agonists) Lipotechoic acid (LTA), PolyriboinosinicPolyribocytidylic Acid Poly (I:C), Lipopolysaccharide (LPS) and bac-terial CpG DNA (CpG).

HEK-TLR4-pNifty2-Luc and HEK-TLR9-pNifty2-Luc assays –HEK293XL or HEK293XL/hTLR9 cells (InvivoGen) and 10 �g mL−1

Blasticidin were grown in 75 cm2 flasks for 1 week after recov-ery from liquid nitrogen. Cells were briefly washed with PBSthen removed from the flasks using Trypsin–EDTA (0.05%). Cellswere split 1:3 in 75 cm2 flasks and cultured overnight. Cellswere transfected in flasks with Genejuice (Novagen) as follows:HEK293XL/TLR4/pNifty2-Luc Assay – 5 �g TLR4 plasmid DNA, 5 �gMD2 plasmid DNA, 5 �g CD14 plasmid DNA and 3 �g pNifty2-Luc plasmid DNA (all InvivoGen); HEK-TLR9-pNifty2-Luc – 3ugpNifty2-Luc plasmid DNA. DNA was mixed with Genejuice andincubated overnight (37 ◦C and 5% CO2) with the cells.

Cells were trypsinised from the flask and plated out into 24-well plates at 1 × 105 cells per well and left to incubate overnight.LPS from E. coli (Sigma L-6529) was dissolved in sterile distilledwater and stored at 4 ◦C at 1 mg mL−1 in glass vials. Immediatelybefore use vials were vigorously vortexed before dilutions weremade for assays. Human CpGB-ODN 2006 (InvivoGen) was recon-stituted to 500 �M with water before dilutions were made forassays. Cells were incubated with LPS (HEK293XL/TLR4/pNifty2-Luc), CpGB (HEK-TLR9-pNifty2-Luc), batches of Cpn10 or controlbuffers for 20 h. Medium was removed from the wells, and theadherent cells were briefly washed, before being processed forthe luciferase assay (Luciferase Assay System, Promega). Luciferaseactivity was read for 10 s on a Wallac Victor 2 (Perkin Elmer) Lumi-nometer.

HEK-TLR2-pNifty2-Luc and HEK-TLR3-pNifty2-Luc assays –HEK-TLR2-pNifty2-Luc and HEK-TLR3-pNifty2-Luc cell lines wereprepared in-house by stable transfections of HEK293XL cells withTLR2, TLR3 and pNifty2-Luc plasmids (InvivoGen). These cells weregrown in 75 cm2 flasks for 1 week after recovery from liquidnitrogen, in media containing 400 �g mL−1 Zeocin and 30 �g mL−1

Blasticidin. Cells were trypsinised from flask(s) and plated outinto 24-well plates at 1 × 105 cells per well and left to incubateovernight. Lipotechoic Acid (LTA) 5 mg (InvivoGen) was dissolvedin sterile distilled water to 10 mg mL−1 and stored at 4 ◦C in vials.Immediately before use vials were vigorously vortexed before dilu-tions were made for assays.

Poly (I:C) 10 mg (InvivoGen) was reconstituted to 2.5 mg mL−1

with water before dilutions were made for assays. Cells wereincubated with LTA (HEK-TLR2-pNifty2-Luc), Poly (I:C) (HEK-TLR3-pNifty2-Luc), batches of Cpn10 or control buffers for 20 h. Mediumwas removed from the wells, and the adherent cells were brieflywashed, before being processed for the luciferase assay (LuciferaseAssay System, Promega). Luciferase activity was read for 10 s on aWallac Victor 2 (Perkin Elmer) Luminometer.

3. Results and discussion

3.1. Production strain development

racterisation of recombinant human chaperonin 10 for treatment.1016/j.procbio.2015.06.022

Bioprocess development from the initial stages of proteinexpression through to generation of the final production strainincorporates a number of optimisation steps. The strain used inearly, small-scale bioprocess development may not be the optimal

ARTICLE IN PRESSG ModelPRBI-10466; No. of Pages 11

D.J. Naylor et al. / Process Biochemistry xxx (2015) xxx–xxx 5

Table 1Comparison of Cpn10 attributes during optimisation of production. (A) Upstream data for production of Cpn10 in host strains XL1-Blue and BR067, and downstream processingunit operations for purification of Cpn10 from BR067 production strain cell lysate, including Cpn10 purity and recovery for each unit operation. (B) Process contaminantlevels and product variant profile for Cpn10 produced in host strains XL1-Blue and BR067; other impurities include low levels of des (1–7) Cpn10; release testing; includesassays for Cpn10 concentration by absorbance at 280 nm (A280 nm); purity by SDS-PAGE; aggregation by size exclusion chromatography (SEC); chaperonin activity as anindication of biological activity and analysis of process related impurities, i.e. host cell DNA and endotoxin.

(A)

Upstream – fermentations Volumeprocessed (L)

Cpn10 amount (g) Cpn10 conc (g L−1) Cell density(OD600 nm)

Final yield (g)

XL1-Blue expression strain (n = 3)(wt Cpn10 in pPL550) 97 ± 3 64.1 ± 12.1 0.7 ± 0.1 25.5 ± 0.2 19.8 ± 4.8BR067 production strain (n = 3) (GRstop-opt Cpn10) 71 ± 2 442.1 ± 76.3 6.2 ± 0.9 126 ± 12 115.0 ± 9.3

Downstream processing (unitoperations) – BR067 production strain

Volume (L) Cpn10 amount (g) Cpn10 conc. (g L−1) Cpn10 purity (%) Step recoveries (%)

Lysate post filtration 95 529.2 5.57 ND NDSP-BB eluate 14 249.8 17.84 ND 47.2Diafiltered SP-BB 12.9 188.4 14.64 ND 75.4Q-FF eluate 22.6 162.4 7.20 >99 86.2SP-HP eluate 11.9 110.7 9.30 >99 68.3Formulated Cpn10 21.8 111.9 5.14 >99 100Filtered drug substance 21.7 111.4 5.14 >99 100

(B)

Cpn10 and variant profile (analysed by RP-HPLCa) wt GRstop-optn = 1 n = 3

Cpn10 87.9 96.8 ± 0.5G39E/Acetylated K40 3.6 0.7 ± 0.1R92K 8.1 0.0 ± 0.0G39D 0.0 1.9 ± 0.1Oxidised M31 0.0 0.4 ± 0.0Other impurities 0.4 0.2 ± 0.1

Total Impurities 12.1 3.2 ± 0.5

Characterisation test wt GRstop-optn = 3 n = 3

A280 nm Cpn10 (g L−1) 5.06 ± 0.06 5.09 ± 0.14Purity by SDS-PAGE (%) >99 >99SEC (%) 99.6 ± 0.2 99.0 ± 0.4Host cell DNA (pg/mgb) 4.6 ± 5.0 6.0 ± 1.7Endotoxin (EU mg−1) <0.4 ± 0.5 <0.6 ± 0.4Chaperone Activity (% of control) 107.7 ± 4.2 102.0 ± 2.4

a ectioner mg

soago

tphwtoiSow1suXpB

qw

The limit of quantitation for the RP-HPLC method was 2% while the level of detb The limit of quantitation for the Threshold Total DNA assay was 18 pg of DNA p

train for large-scale production, for reasons that include suitabilityf media formulation, promoter system, levels of product variantsnd choice of selectable marker. For example, codon optimisation isenerally required to maximise expression and minimise the levelf product variant(s) due to mistranslation at rare codons.

In the early stages of bioprocess development Cpn10 produc-ion was performed in XL1-Blue cells utilising a heat-inducibleromoter system (pPL550:cpn10). The XL1-Blue expression systemas been described previously in the literature [18–20] however, asill be shown, was found to be inadequate for large-scale produc-

ion. Bioprocess optimisation ultimately resulted in the productionf Cpn10 for Phase IIa clinical trials in strain BR067 from an IPTGnducible tac promoter-based expression construct. Fig. 1 showsDS-PAGE analyses of the soluble fraction from cell lysates through-ut the bioprocess development. Cpn10 is a heptamer of moleculareight (Mr) 70 kDa, with the monomeric form visualised at Mr

0 kDa by SDS-PAGE. Cpn10 yield (g L−1) in the optimised BR067train fermentate increased around 8- to 9-fold and improved prod-ct purity from ∼88% to 97% compared to product made in theL1-Blue strain (Table 1). Importantly, total product variants, as aercentage of purified Cpn10, were reduced from 12.1% in the XL1-

Please cite this article in press as: D.J. Naylor, et al., Production and chaof inflammatory disease, Process Biochem (2015), http://dx.doi.org/10

lue system to 3.2% in the final BR067 production strain (Table 1B).Although the XL1-Blue strain allowed production of sufficient

uantities of Cpn10 for early stage product development, thereere a number of drawbacks. The inadequacies included the

was ∼1%. of Cpn10 while the level of detection was 6 pg/mg.

requirement for complex media, use of �-lactam-based antibioticswhich can potentially cause allergic reactions, the difficult to con-trol heat shock inducible promoter at large fermentation volumes,inability to achieve high cell densities and generation of unac-ceptably high levels of mistranslation variants. The removal ofproduct variants was first attempted by total codon optimisationof the human Cpn10 cDNA for production in E. coli. Surprisingly,total codon optimisation in the BR067 strain significantly reducedexpression levels of Cpn10 (compare Fig. 1A, lane 5 and Fig. 1B, lane1); as a result partial codon optimisation was next attempted.

Investigation of XL1-Blue derived Cpn10 variant profile by RP-HPLC revealed one major contaminating peak; subsequent massspectrometry analysis identified a large amount of the produc-tion variant Cpn10-G39E and an additional variant where Cpn10was acetylated on residue K40 (Fig. 1C). For this reason, partialcodon optimisation first targeted the rare glycine codon encod-ing residue G39 in the human Cpn10 cDNA. Fig. 1A shows thatequivalent expression levels were obtained with the wild type(wt) Cpn10 cDNA and the G39 codon optimised (G39-opt) cDNAwhen produced from the XL1-Blue strain, but more importantly,the Cpn10-G39E variant peak visible by RP-HPLC was significantly

racterisation of recombinant human chaperonin 10 for treatment.1016/j.procbio.2015.06.022

or completely removed (data not shown).Given the inability of XL1-Blue to achieve high cell densities

(Table 1A), expression of the wt and G-opt cDNAs from the heat-inducible promoter system (pPL550:cpn10) was next attempted

ARTICLE IN PRESSG ModelPRBI-10466; No. of Pages 11

6 D.J. Naylor et al. / Process Biochemistry xxx (2015) xxx–xxx

Fig. 1. Analysis of Cpn10 expression levels and variant profiles when produced in E. coli. (A and B) SDS-PAGE analysis of clarified host strain lysates expressing Cpn10 fromthe wild-type (wt) cDNA and numerous codon optimised cDNAs. The band corresponding to Cpn10 is shown (Mr 10 kDa). (A) Strains XL1-Blue and BR067 with pPL5550 heatinducible promoter based Cpn10 expression constructs. (B) BR067 strain with pBR322, IPTG inducible tac promoter based Cpn10 expression constructs. (C and D) Cpn10purified from numerous expression constructs (indicated in panels A and B) was analysed by RP-HPLC to reveal levels of Cpn10 production variants. The constructs testedwere XL1-Blue cells expressing Cpn10 from the wt cDNA (Panel A, lane 3) and BR067 cells expressing Cpn10 from an all Glycine-codons optimised cDNA (G-opt; Panel B, lane3) and an all Glycine, Arginine and stop-codon optimised cDNA (GRstop-opt; Panel B, lane 4). (C) RP-HPLC analysis of purified Cpn10 using a 25–45% acetonitrile gradientover 40 min. Peak G39E+ acetylated K40 indicates co-elution of these Cpn10 variants; stop-1 and stop-2 indicate two stop-codon read-through Cpn10 variants. (D) RP-HPLCa in. ThX tion vr n10 p

iwce

nalysis of purified Cpn10 using a shallow 30–32% acetonitrile gradient over 40 mL1-Blue derived material, produced from the wt Cpn10 cDNA. The Cpn10 produceduced in GRstop-opt (i.e. BR067 production strain). In panels C and D the main Cp

Please cite this article in press as: D.J. Naylor, et al., Production and chaof inflammatory disease, Process Biochem (2015), http://dx.doi.org/10

n BR067 cells. Again equivalent expression levels were obtainedith the wt Cpn10 cDNA and the G39 codon optimised (G39-opt)

DNA when produced from BR067 cells (Fig. 1A). However asxpected, BR067 ferments typically reached a cell density of

e extended gradient revealed an arginine 92 to lysine (R92K) product variant inariants detected by RP-HPLC in panels C and D were undetectable or significantlyeak is indicated with an asterisk.

racterisation of recombinant human chaperonin 10 for treatment.1016/j.procbio.2015.06.022

OD600 nm ∼110 (data not shown) compared to XL1-Blue fermentswhich only reached a cell density of OD600 nm ∼25 (Table 1A). Theincrease in cell density achieved with the BR067, compared to theXL1-Blue, cell line was ∼4.6-fold.

ING ModelP

iochem

cssttos(

tbotss

tqtcmasgapooicaw

clbtTtltewtorwstv

gwscRvCs

otitp

ARTICLERBI-10466; No. of Pages 11

D.J. Naylor et al. / Process B

The use of the G39 codon optimised Cpn10 cDNA in BR067ells addressed numerous inadequacies of the original XL1-Bluetrain for large-scale production, however, the issues of the unde-irable �-lactam-based antibiotic selection marker and the difficulto control heat shock promoter still remained. In order to addresshese concerns and maximise Cpn10 expression levels, a seriesf pBR322-based expression constructs harbouring a Kanamycinelection marker, ColE1 origin of replication and lac repressor genelacI) were screened.

The expression constructs varied by their use of IPTG-induciblerc or tac promoter and transcriptional termination sequences. Theest Cpn10 (G39-opt) expression was achieved with a combinationf the IPTG-inducible tac promoter and the rrnBT1T2 transcriptionermination sequence in the BR067 cell line (Fig. 1B and data nothown). All subsequent work was completed in this productionystem.

The establishment of an optimal expression system and con-inual efforts to improve the RP-HPLC method, for detection anduantitation of Cpn10 production variants, uncovered yet addi-ional inadequacies in the wt and partially codon optimised Cpn10DNA. A combination of removal of Cpn10-G39E variant and opti-isation of the RP-HPLC gradient, revealed the presence of small

mounts (∼1%) of new production variant Cpn10-G39D (data nothown). A decision was made to attempt to optimise all rarelycine codons within Cpn10 to prevent the unforeseen appear-nce of further trace amounts of glycine variants as the programmerogressed. Fig. 1B shows that equivalent expression levels werebtained with the wt Cpn10 cDNA, the G39 codon optimised (G39-pt) cDNA and when all rare glycine codons were optimised (G-opt)n the Cpn10 cDNA. RP-HPLC analysis of Cpn10 from the all glycineodon optimised construct, revealed that while the major G39E+cetylated K40 peak had been minimised, two new additional peaksere present (Fig. 1C, G-opt).

Mass spectrometry analysis identified these peaks as two stopodon (TGA) aberrant translational read-through variants. Trans-ational read-through at TGA termination codons has previouslyeen documented during the overexpression of recombinant pro-eins [21] with read-through is a more common event at TGA thanAA termination codons [22]. At approximately the same time thathe two stop codon read-through variants were identified, a shal-ow gradient RP-HPLC analysis of purified Cpn10, expressed fromhe wt cDNA in the original XL1-Blue system, revealed the pres-nce of an additional Cpn10 variant where arginine residue 92as substituted for a Lysine (i.e. R92K) according to mass spec-

rometry analysis. The Cpn10-R92K variant was present at ∼8.1%f the total purified Cpn10 material (Table 1B). In prior RP-HPLCuns, utilising a steeper gradient, the Cpn10-R92K variant is hiddenithin the main Cpn10 peak (Fig. 1C). A shallow gradient analy-

is of the all glycine codon optimised construct, expressed fromhe BR067 cell line, also detected the presence of the Cpn10-R92Kariant.

A final codon optimised Cpn10 cDNA was designed where alllycine codons, all arginine codons and the stop codon (TGA to TAA)ere optimised (GRstop-opt). Fig. 1B shows that equivalent expres-

ion levels were obtained with the wt cDNA, G39-opt cDNA, G-optDNA and the GRstop-opt cDNA in BR067 cells. More importantly,P-HPLC analysis revealed that the two stop codon read-throughariants and the Cpn10-R92K variant were not detectable while thepn10-G39E and acetylated Cpn10-K40 were significantly minimi-ed (Fig. 1C and D, Table 1B).

Cpn10 production in the BR067 strain, utilising the GRstop-pt cDNA and optimised pBR322 expression construct (hereafter

Please cite this article in press as: D.J. Naylor, et al., Production and chaof inflammatory disease, Process Biochem (2015), http://dx.doi.org/10

ermed BR067 production strain), showed optimal characteristics,ncluding high cell density growth in defined media, suitable selec-ion marker and high expression yields with minimal levels ofroduct variants.

PRESSistry xxx (2015) xxx–xxx 7

3.2. Structure, function and stability

3.2.1. Confirmation of heptameric structureCpn10 is a heptamer, made up of seven identical monomers.

SDS-PAGE analysis of Cpn10 derived from BR067 production strainis shown in Fig. 2A; an overloading of the gel with Cpn10 quali-tatively shows >99% purity, with no other bands visible. In orderto confirm Cpn10’s heptameric structure, Cpn10 oligomers werepartially cross-linked with glutaraldehyde. Partial cross-linking ofCpn10 favoured intra-oligomer links to be formed (i.e. linkage ofmonomers within an oligomeric unit) and limited inter-oligomercross-links (i.e. artificial linking of several oligomeric units).

Subsequent separation by size exclusion chromatography andcollection of the material which eluted at the same position on thechromatogram as native Cpn10 oligomer, ensured that only intra-oligomer cross-linked species were isolated. SDS-PAGE analysisof this material revealed seven discrete protein bands, represent-ing Cpn10 monomer and cross-linked dimer, trimer, etc. throughto heptamer, indicating that Cpn10 adopts a heptameric struc-ture that is characteristic of all Cpn10 members (Fig. 2B). Theapparent molecular weight of heptameric Cpn10 was determinedto be 94 ± 3 kDa (Fig. 2C). The expected theoretical molecularweight of heptameric Cpn10 was estimated at 76 kDa using theExPasy Protein Parameter analysis tool (SIB; Swiss Institute ofBioinformatics). The slight discrepancy in apparent molecularweight could be attributed to the expanded packing of the sub-units within the Cpn10 heptamer resulting in an increased Stokesradius.

3.2.2. Activity and stability of purified, recombinant Cpn10The biological activity of Cpn10 (over time) was investigated

as a molecular co-chaperone in protein folding, and was used asan indicator for structural integrity. Cpn10 (>99% pure) expressedin E. coli was formulated in 50 mM Tris–HCl (pH 7.6) and 150 mMNaCl. When stored at ∼75 mg mL−1 and 2–8 ◦C for at least 3 years,Cpn10 retained full activity as a co-chaperone in the process ofE. coli GroEL-mediated rhodanese refolding (Fig. 2D). The bioassayrevealed that small differences in Cpn10 activity can be detected byvarying the amount of Cpn10, illustrating its usefulness for ensur-ing the integrity of Cpn10 production batches. In agreement withthe retention of full biological activity after 3 years stored at 2–8 ◦C,no significant changes to the product variant profile were observedby RP-HPLC and no changes were observed to the heptameric struc-ture by SEC (data not shown). Cpn10 is therefore a very stableprotein, which is probably not unexpected given that it is a stress-inducible protein and fulfils an essential role in the maintenance ofcellular homeostasis.

Methionine oxidation can have significant effects on the activityof a protein; for example, oxidation of the exposed, N-terminal-methionine of �-polypeptide in the light-harvesting complex fromRhodospirillum rubrum affects its ability to form the B820 subunitwith �-polypeptide and bacteriochlorophyll a [23]. Cpn10 con-tains a single Met residue at position 33 of the mobile loop (i.e.32-Ile-Met-Leu-34), which is crucial for its efficient interactionwith Cpn60. Therefore, potential oxidation of this Met residue wasinvestigated by mass spectrometry. Tandem mass spectra of Cpn10indicated the Met residue is readily oxidised by hypochlorite (datanot shown), however only 0.5% of the purified Cpn10 material wasobserved to be oxidised at the end of production (Table 1B). Moreimportantly, storage at ∼75 mg mL−1 under nitrogen for 3 years at2–8 ◦C resulted in <1.5% product oxidation, indicating that oxida-tion will not be a concern for biological activity.

racterisation of recombinant human chaperonin 10 for treatment.1016/j.procbio.2015.06.022

3.2.3. Mass spectrometryThe mass of purified Cpn10, determined by mass spectrome-

try was 10,872.2 ± 1 Da (data not shown), in agreement with the

ARTICLE IN PRESSG ModelPRBI-10466; No. of Pages 11

8 D.J. Naylor et al. / Process Biochemistry xxx (2015) xxx–xxx

Fig. 2. Characterisation of purified Cpn10. (A) SDS-PAGE analysis of purified Cpn10. Varying amounts of purified Cpn10 were loaded into adjacent lanes. Relative percentages(%) are indicated to illustrate a purity of >99%. (B) Glutaraldehyde cross-linking of human recombinant Cpn10. Silver stained 4–20% SDS-PAGE analysis of partially cross-linked Cpn10 oligomers reveals seven distinct species confirming a heptameric structure. Lane 1, molecular weight standards; Lane 2, cross-linked Cpn10; Lane 3, is a lighterexposure of lane 2 for better clarity of bands 5, 6 and 7. (C) Size exclusion chromatograms (Superdex-200 HR 10/30) of Cpn10. Cpn10 is a heptamer of identical subunits, andelutes with an apparent Mr of 94 ± 3 kDa. Cpn10 (cross-linked) is predominantly a heptamer containing intra-oligomer cross-links eluting with an apparent Mr of 93 ± 3 kDa,and some higher oligomers containing inter-oligomer cross-links. The molecular weight (MW) standards were thyroglobulin 669 kDa, catalase 232 kDa, albumin 67 kDa andchymotrypsinogen A 25 kDa. (D) Cpn10 is active as a molecular co-chaperone during the GroEL-dependent folding of denatured rhodanese. Rhodanese was denatured in 6 Mg , �, �

C ATP. Aa

toimaptdv

uanidine–HCl and diluted 100-fold (final concentration 400 nM) in the presence (�pn10 were added (as indicated) and rhodanese refolding was initiated with 5 mMctivity was measured.

heoretical mass of 10,871.5 Da. To confirm the primary sequencef Cpn10, a tryptic peptide map was developed and the 14 result-ng peptides were identified by mass spectrometry (Fig. 3). The

ass spectrometry-based identity of the peptides in peaks 1, 10, 13nd 14, was corroborated by their strong absorbance at A280 nm as

Please cite this article in press as: D.J. Naylor, et al., Production and chaof inflammatory disease, Process Biochem (2015), http://dx.doi.org/10

eptides containing tyrosine residues (data not shown). Mass spec-rometry was an invaluable tool for monitoring product integrityuring process development and for tracking the level of productariants.

, �) or absence (o) of 500 nM GroEL (i.e. spontaneous folding). Varying amounts oft the times indicated, GroEL action was stopped with 50 mM EDTA and rhodanese

3.3. Purified recombinant Cpn10 lacks proinflammatory E. colicontaminants

The observed effects of recombinant heat shock protein prepara-tions on the immune system have been quite controversial owing to

racterisation of recombinant human chaperonin 10 for treatment.1016/j.procbio.2015.06.022

the potential presence of pro-inflammatory E. coli components thatmay alone be responsible for activity [24]. To illustrate that recom-binant Cpn10 preparations lacked major E. coli pro-inflammatorycontaminants, highly sensitive HEK reporter cell lines transfected

ARTICLE IN PRESSG ModelPRBI-10466; No. of Pages 11

D.J. Naylor et al. / Process Biochemistry xxx (2015) xxx–xxx 9

Fig. 3. The identity of Cpn10 was confirmed with a tryptic peptide map resolvedby RP-HPLC (A) and analysed by mass spectrometry (B) Peptide peaks 1, 10, 13 and1t

wsaoit

3

sHattai

tccattr

Fig. 4. Lack of E. coli proinflammatory contaminants in preparations of recombinantCpn10. HEK cells transfected with TLR2 (Panel A), TLR3 (Panel B), TLR4/MD2/CD14(Panel C) and TLR9 (Panel D) were stimulated for 20 h with a range of the appropriateligand, three Cpn10 batches and the Tris buffer used to formulate Cpn10. The limit ofdetection for each assay is mark with an asterisk. Data represents the mean of tripli-cate wells ± S.D. LTA, Lipotechoic acid; Poly (I:C), Polyriboinosinic Polyribocytidylic

4 (marked with an asterisk) absorbed strongly at 280 nm due to the presence ofyrosine residues.

ith either TLR2, TLR3, TLR4/MD2/CD14 or TLR9 were used. Fig. 4hows that when TLR-transfected HEK cells were challenged withliquots of Cpn10 from three separate GMP batches of BR067, read-uts were below the level of detection for the appropriate ligand,ndicating a robust downstream process that yields Cpn10 prepara-ions that are essentially free of bacterially derived contaminants.

.4. GMP production of Cpn10

Human Cpn10 was produced in the E. coli BR067 productiontrain in 100 L fed-batch fermentations performed under cGMP atospira, Adelaide, Australia. Cpn10 protein was completely solublend purified from the cellular lysate after disruption and clarifica-ion. Compared to XL1-Blue at the same fermentation scale, use ofhe BR067 strain resulted in an ∼5-fold increase in cell density andn ∼9-fold increase in expression levels which ultimately resultedn a near 6-fold increase in final product.

Cpn10 was purified by a downstream process incorporatinghree sequential IEX chromatography steps: SP-BB, Q-FF and SP-HPhromatography (Table 1A). Throughout the downstream process,hromatographic fractions were analysed by RP-HPLC for puritynd yield of Cpn10 (Table 1A). The average pI for E. coli host pro-

Please cite this article in press as: D.J. Naylor, et al., Production and chaof inflammatory disease, Process Biochem (2015), http://dx.doi.org/10

eins is 5.7 [25]. As the pI of Cpn10 (8.91) is significantly greaterhan the majority of the E. coli proteins, the downstream processesulted in a final SDS-PAGE purity of >99%.

Acid; LPS, Lipopolysaccharide; CpG ODN 2006, unmethylated oligonucleotide 2006(human type B) containing CpG motifs.

SP-BB cation exchange captures facilitated concentration of theCpn10 while the majority of host cell protein contaminants wereremoved in the flow-through. SP-BB chromatography resulted in asignificant and variable loss of Cpn10 (44–83%), despite an excess

racterisation of recombinant human chaperonin 10 for treatment.1016/j.procbio.2015.06.022

of resin and conditions which should have facilitated bindingof Cpn10. Re-application of the non-binding fraction, contain-ing ∼50% of the Cpn10, resulted in a modest recovery (∼10%).

ING ModelP

1 ioche

Cstb>twt

tcHCvuCaep[

3

pprl1CtdbTBocc

uNFpaod

4

Ccewuicatbpolsb

[

[

[

[

[

[

[

[

[

276–287.[19] A. Brinker, G. Pfeifer, M.J. Kerner, D.J. Naylor, F.U. Hartl, M. Hayer-Hartl, Dual

ARTICLERBI-10466; No. of Pages 11

0 D.J. Naylor et al. / Process B

haracterisation of this non-binding material by RP-HPLC and masspectrometry indicated that it was identical to material retained onhe SP-BB resin. An attempt to dissociate Cpn10 from an unknownound entity using urea or organic alcohols improved binding to90%, however led to the formation of an apparent Cpn10 hep-amer dimer (<20% of total Cpn10) (data not shown). A decisionas made to accept the lower Cpn10 recovery on SP-BB to achieve

he maximum purity.Operation of Q-FF IEX chromatography at pH 8.0 ensured that

he Cpn10 was contained in the flowthrough, whilst residual hostell proteins, DNA and endotoxin bound to the column. The SP-P final polishing step resulted in the production of clinical gradepn10, with a purity >99% by SDS-PAGE. Table 1B shows the productariant profile for three cGMP Cpn10 batches and an average prod-ct variant level of 3.3%. Table 1B additionally shows numerouspn10 characterisation tests which ensure suitability for humandministration. Endotoxin (<0.6 EU mg−1), based on dosage lev-ls, is well below the required limit of 5 EU kg−1 body weighter hr for an intravenous application of a therapeutic protein26].

.5. GMP formulation of Cpn10

Cpn10 bulk drug substance (BDS) was manufactured by Hos-ira, a GMP facility in Adelaide, South Australia. Hospira hasroduced Cpn10 at a manufacturing scale (100 L) to supply mate-ial for preclinical and clinical studies, preparation of a referenceot, and stability studies. Cpn10 produced from the cell line XL-Blue was manufactured for phase I and phase II clinical trials,pn10 produced from the BR067 process was manufactured for GLPoxicology studies and phase II clinical trial material. Cpn10 pro-uced by both methods has demonstrated equivalent performancey biological, physicochemical and in vitro safety parameters.he two features identified as different were the reduction inR067 derived product of translational variants and the removalf an unidentified agonist, as detected by a macrophage reporterell line, both being desirable improvements on the old pro-ess.

The Sponsor transferred its bulk manufactured Ala-Cpn10 prod-ct to its GMP CMC partner, Gallus Biopharmaceuticals (Princeton,J), for formulation and release under GMP conditions in theirood and Drug Administration (FDA) approved facilities. Gallus Bio-harmaceuticals have since validated a suite of characterisationssays which have demonstrated suitability for the determinationf purity and identity of Cpn10 prior to formulation and release ofrug product according to defined, FDA approved, specifications.

. Conclusion

A bioprocess for the production of ∼115 g of clinical gradepn10 (representing an overall recovery of 26%) from a 100 L E.oli fermentation was developed incorporating cation and anionxchange chromatography purification. Cpn10 was characterisedith respect to purity by SDS-PAGE (>99%), identification of prod-ct variants (3.2%), in vitro biological activity and stability. Key

mprovements to process productivity were achieved throughhanging the host production cell line, the choice of promoternd termination sequences and through partial codon optimisa-ion. These changes resulted in elevated final cell density in theioreactor, higher specific Cpn10 productivity and reduced the pro-ortion of product variants, which was critical for the production

Please cite this article in press as: D.J. Naylor, et al., Production and chaof inflammatory disease, Process Biochem (2015), http://dx.doi.org/10

f clinical-grade protein. Biochemical stability testing revealed nooss in chaperone activity and no significant chemical change whentored at 75 mg mL−1 and 2–8 ◦C for at least 3 years. The finalioprocess yields Cpn10 of high purity which lacks detectable levels

[

PRESSmistry xxx (2015) xxx–xxx

of proinflammatory E. coli contaminants. The bulk material pro-duced by this process has been formulated and released in a GMP,FDA approved facility according to the acceptance specifications forvalidated characterisation assays for purity, identity and potency.Ala-Cpn10 is currently being trialled under a US IND116156 in ablinded, placebo-controlled phase II clinical study in patients withSLE.

Acknowledgment

Part of this work was supported by a Pharmaceuticals Partner-ship Programme (P3) from the Australian Federal Government andthe Australian Research Council.

References

[1] P.V. Viitanen, Mammalian Cpn10, in: M.J. Gething (Ed.), MolecularChaperones and Protein Folding Catalysts, Oxford University Press, New York,1997, pp. 199–201.

[2] F.U. Hartl, M. Hayer-Hartl, Molecular chaperones in the cytosol: from nascentchain to folded protein, Science 295 (2002) 1852–1858.

[3] U. Jakob, M. Gaestel, K. Engel, J. Buchner, Small heat shock proteins aremolecular chaperones, J. Biol. Chem. 25 (1993) 1517–1520.

[4] Z. Li, A. Menoret, P. Srivastava, Roles of heat-shock proteins in antigenpresentation and cross-presentation, Curr. Opin. Immunol. 14 (2002)45–51.

[5] W. van Eden, XToll: a recombinant chaperonin 10 as an anti-inflammatoryimmunomodulator, Curr. Opin. Investig. Drugs 9 (2008) 523–533.

[6] A.M. Shields, G.S. Panayi, V.M. Corrigall, Resolution-associated molecularpatterns (RAMP): RAMParts defending immunological homeostasis? Clin.Exp. Immunol. 165 (2011) 292–300.

[7] J.F. Hunt, A.J. Weaver, S.J. Landry, L. Gierasch, J. Deisenhofer, The crystal

structure of the GroES co-chaperonin at 2.8 ´̊A resolution, Nature 379 (1996)37–45.

[8] S.C. Mande, V. Mehra, B.R. Bloom, W.G. Hol, Structure of the heat shockprotein chaperonin-10 of Mycobacterium leprae, Science 271 (1996) 203–207.

[9] B.J. Johnson, T.T. Le, C.A. Dobbin, T. Banovic, C.B. Howard, F.de.M.L. Flores, D.Vanags, D.J. Naylor, G.R. Hill, A. Suhrbier, Heat shock protein 10 inhibitslipopolysaccharide-induced inflammatory mediator production, J. Biol. Chem.280 (2005) 4037–4047.

10] D.S.E. Agnello, G.M. Di, F. Leoni, D. Modena, P. Mascagni, M. Introna, P. Ghezzi,P. Villa, Preventive administration of Mycobacterium tuberculosis 10-kDa heatshock protein (hsp10) suppresses adjuvant arthritis in Lewis rats, Int.Immunopharmacol. 2 (2002) 463–474.

11] B. Zhang, M.D. Walsh, K.B. Nguyen, N.C. Hillyard, A.C. Cavanagh, P.A.McCombe, H. Morton, Early pregnancy factor treatment suppresses theinflammatory response and adhesion molecule expression in the spinal cordof SJL/J mice with experimental autoimmune encephalomyelitis and thedelayed-type hypersensitivity reaction to trinitrochlorobenzene in normalBALB/c mice, J. Neurosci. 212 (2003) 37–46.

12] H. Morton, D.A. McKay, R.M. Murphy, M.J. Somodevilla-Torres, C.E. Swanson,A.I. Cassady, K.M. Summers, A.C. Cavanagh, Production of a recombinant formof early pregnancy factor that can prolong allogeneic skin graft survival timein rats, Immunol. Cell Biol. 78 (2000) 603–607.

13] O.P. Kulkarni, M. Ryu, C. Kantner, M. Sárdy, D. Naylor, D. Lambert, R. Brown,H.J. Anders, Recombinant chaperonin 10 suppresses cutaneous lupus andlupus nephritis in MRL-(Fas)lpr mice, Nephrol. Dial. Transplant. 27 (2012)1358–1367.

14] D. Vanags, B. Williams, B. Johnson, S. Hall, P. Nash, A. Taylor, J. Weiss, D.Feeney, Therapeutic efficacy and safety of chaperonin 10 in patients withrheumatoid arthritis: a double-blind randomised trial, Lancet 368 (2006)855.

15] B. Williams, D. Vanags, B. Johnson, S. Hall, C. McCormack, P. Foley, J. Weiss, B.Johnson, E. Latz, D. Feeney, Efficacy and safety of chaperonin 10 in patientswith moderate to severe plaque psoriasis: evidence of utility beyond a singleindication, Arch. Dermatol. 144 (2008) 683–685.

16] N. Jacob, Cytokine disturbances in systemic lupus erythematosus, ArthritisRes. Ther. 13 (2011) 228.

17] R. Medzhitov, C. Janeway Jr., The Toll receptor family and microbialrecognition, Trends Microbiol. 10 (2000) 452–456.

18] M.J. Somodevilla-Torres, H. Morton, B. Zhang, S. Reid, A.C. Cavanagh,Purification and characterization of functional early pregnancy factorexpressed in Sf9 insect cells and in Escherichia coli, Prot. Exp. Purif. 32 (2003)

racterisation of recombinant human chaperonin 10 for treatment.1016/j.procbio.2015.06.022

function of protein confinement in chaperonin-assisted protein folding, Cell107 (2001) 223–233.

20] M.T. Ryan, D.J. Naylor, N.J. Hoogenraad, P.B. Hoj, Affinity purification,overexpression, and characterization of chaperonin 10 homologues with andwithout N-terminal acetylation, J. Biol. Chem. 270 (1995) 22037–22043.

ING ModelP

iochem

[

[

[

[

[25] A.W. Houry, D. Frishman, C. Eckerskorn, F. Lottspeich, U. Hartl, Identification

ARTICLERBI-10466; No. of Pages 11

D.J. Naylor et al. / Process B

21] K.V. Lu, M.F. Rohde, A.R. Thomason, W.C. Kenny, H.S. Lu, Mistranslation of aTGA termination codon as tryptophan in recombinant platelet-derivedgrowth factor expressed in Escherichia coli, Biochem. J. 309 (1995) 411–417.

22] J. Parker, Errors and alternatives in reading the universal genetic code,

Please cite this article in press as: D.J. Naylor, et al., Production and chaof inflammatory disease, Process Biochem (2015), http://dx.doi.org/10

Microbiol. Rev. 53 (1989) 273–298.23] Z. Wang, M. Shimonaga, Y. Muraoka, M. Kobayashi, T. Nozawa, Methionine

oxidation and its effect on the stability of a reconstituted subunit of thelight-harvesting complex from Rhodospirillum rubrum, Eur. J. Biochem. 268(2001) 3375–3382.

[

PRESSistry xxx (2015) xxx–xxx 11

24] B. Gao, M.F. Tsan, Induction of cytokines by heat shock proteins andendotoxin in murine macrophages, Biochem. Biophys. Res. Commun. 4 (2004)1149–1154.

racterisation of recombinant human chaperonin 10 for treatment.1016/j.procbio.2015.06.022

of the in vivo substrates of the chaperonin GroEL, Nature 402 (1999)147–154.

26] US Pharmacopoeia (USP), Stage 6 Harmonisation: Bacterial Endotoxins Test,vol. 85, 2012, pp. 1–6.