Purification Development

Tech-transfer run of VLP 1 performed UCL/iQur with 3P

KM71H pHe7 HA2.3,(M2e) 3 VLP prep Gel of final Product HA2.3,(M2e) 3 1:2 HA2.3,(M2e) 3 1:4 HA2.3,(M2e) 3 1:8 rHBc (50ng) Marker HA2.3,(M2e) 3 Anti-core Blot 1 sec exposure HA2.3,(M2e) 3 1:2 HA2.3,(M2e) 3 1:4 HA2.3,(M2e) 3 1:8 rHBc (50ng) Marker HA2.3,(M2e) 3 1 min exposure VLP1, made during technical transfer visit, is truncated

Degradation products can be partially prevented using more protease inhibitors Marker S20S19S18S17S16S15S14S13S12 S10 S11 KDa KDa 59 KDa ~ 130 KDa ~ 95 KDa Aggregates??? 32X3=96 32X4=128 45X3=135 45X2=90 For S13, S14, S19 and S20 the 37 and 45 Kda bands are not present suggesting less degradation and inhibition of the cleavage at the MIR 2 (before and after the M2e sequence). Also the 32 KDa band (fragment derived from linker cleavage together with the 27 KDa band) is not present in these samples. However the top bands (130 and 95 KDa) are stronger compare with other samples. This suggests that the top bands may be aggregates of 32 and/or 45 KDa bands)

Purification Development Tech-transfer run of VLP 2 Fermentation at 3P Purifications at 3P and iQur

Purification of VLP2 3P g Anti-core Blot 10E11 14E11 Anti-M2 Blot 14C2 In-House Anti-LAH H3 % Full length product (%)62.4 Truncated product (%)20.7 Total core related protein (%)83.1 Contaminant (%)16.9 Yield (mg)3.0 All 1 sec exposure iQur purification 3P purification Yields from triplicate preps 3.0mg; 3.8mg; 3.7mg

KM71H pHe7 LAH H3,K1VLP 2 100g prep: Discreet pass homogenisation 20K SN 20K Pellet Diluted 20K SN CL4B Input 1:10 Permeate 1Permeate 2 S1000 Input CL4B Input TFF rHBc (50ng) Marker Crude lysate + Triton X100 1 sec exposure1 min exposure CL4B 100g of VLP2 cell pellet can be processed using Lab60, TFF and CL4B Lab60 in discrete pass mode TFF using 1600cm 2 cassette (115cm 2 for 25g scale) CL4B loaded at 20% CV (5% when processing 25g)

KM71H pHe7 LAH H3,K1VLP 2 100g prep: Discreet pass homogenisation: CL4B Void on XK 50/92 S1000 #5 #8 #12 #29 #18#20 #32 #26 #23 rHBc (50ng) MarkerInput #38 #41 #44 #59 #47#50 #65 #56 #53 rHBc (50ng) Marker#35 Fractions pooled as product After concentration total yield = 20mg (from 25g cell pellet ~3mg) 1 sec exposure S1000 loaded at 5% CV (2.5% when processing 25g)

LAH H3,K1 1:4 Marker LAH H3,K1 Anti-core Blot 1 sec exposure 1 min exposure In house 1 sec exposure 1 min 14E11 Anti-M2 Blot 1 sec exposure 14C2 1 min 1 sec exposure 1 min AX-LAH H3 KM71H pHe7 LAH H3,K1VLP 2 100g prep discreet pass homogenisation: Final Product 100g scale prep carried out x4 x2 with pellet from 3P x2 with pellet from UCL Yields in range 16 – 20mg

Why did 3P DSP fail? 20K s/n 20K pellet Dil s/n 0.1 filtrate PermetateTFF CL4B Input 1: filter SDS wash TFF 0.2 filtrate rHBc (50ng) Marker Crude lysate 1 sec exposure KM71H pHe7 LAH H3,K1 3P g VLP prep Lysis through to CL4B XK26/92- as expected

KM71H pHe7 LAH H3,K1 3P g VLP 2: CL4B Void immediately run on XK 50/92 S1000 #7 #13 #17 #31 #20#23 #34 #29 #26 rHBc (50ng) Marker #5 #40 #43 #46 #71 #49#51 #73 #65 #54 rHBc (50ng) Marker#37 Protein concentration by Bradford Assay Input 0.28mg/ml VLP pool (29 – 51) yield 3.1mg 1 sec exposure Is a hold between CL4B and S1000 to blame?

KM71H pHe7 LAH H3,K1 3P g VLP 2. CL4B Void stored at 4 o C overnight, filtered (0.2µm) then run on XK 50/92 S1000 #7 #13 #17 #31 #20#23 #34 #29 #26 rHBc (50ng) Marker#5 #40 #43 #46 #71 #49#51 #73 #65 #54 rHBc (50ng) Marker#37 Protein concentration by Bradford Assay Input 0.32mg/ml VLP pool (29 – 51) yield 3.0mg 1 sec exposure 12hr hold-step has no effect on final product

Run #1 Run #2 Run #3 Runs #5-7 Pre-loading a new S1000 column is required before reproducible separation is achieved VLP2 was lysed, filtered and passed over CL4B The CL4B void volume was then passed over an XL16/40 column filled with new S1000 matrix sequentially Samples were collected from each run 100ml of beads will be saturated by 3.2mg of CL4B void protein UV-vis traces show that four runs are necessary to “pre- saturate” a column before reproducible separation is achieved Run #4

Purification Development Further optimisation of growth/induction

Time course samples from 3P VLP2 fermentation CL4B profiles (western) 0hrs 4hrs 16hrs 32hrs 48hrs VLP expected in CL4B void NB: extraneous bands circled are related to the markers used and should be ignored 0hrs 4hrs 16hrs 32hrs 48hrs Arrows indicate where full length protein would be detected Gels show full length is only in CL4B void volume as expected Data suggests optimal expression is at 32hrs not 48 (confirmed verbally by 3P)

Effects of feed strategy on protein production -UCL mini-fermenters K1:K1 MeOH only induction VLP1 MeOH only induction VLP1 Mixed feed (Methanol/Glycerol) K1:K1 clearly has optimal expression at 40hrs. This differs from other tandem core VLP VLP1 has improved expression when grown under mixed-feed conditions Expression of VLP1 is decreased after 40hrs in either growth mode The loss of expression after 24hrs in MeOH reflects the iQur experience using shake flasks Data suggests optimal expression for VLP1 is less than 30hrs for either MeOH or a mixed-feed strategy

Method of fermentation affects the production of full-length protein even in shake flasks VLP1 mixed-feed VLP1 61S mutant mixed-feed VLP1 MeOH onlyVLP1 61S MeOH only Red arrows indicate full length protein and blue show truncated tandem core These gels are taken from initial lysis samples from shake flasks Truncated protein is not normally seen since it is rapidly purified out There is a marked increase in truncated protein in mixed-feed samples These data agree with larger scale fermentation studies which suggest Mixed-feed leads to more protein truncation

KM71H pHe7 HA2.3,(M2e) 3 VLP 1 25g UCL fermentation 24hr mixed-feed induction 20K s/n 20K pellet Dil s/n 0.1 filtrate PermetateTFF CL4B Input 1: filter SDS wash TFF 0.2 filtrate rHBc (50ng) Marker Crude lysate Anti-core Blot 1 sec exposure 1 min exposure

HA2.3,(M2e) 3 1:4 Marker HA2.3,(M2e) 3 Anti-core Blot 1 sec exposure 1 min exposure 10E11 1 sec exposure 1 min 14E11 Anti-M2 Blot 1 sec exposure 14C2 1 min 1 sec exposure 1 min In-House Anti-LAH H3 AEC KM71H pHe7 HA2.3,(M2e) 3 VLP 1 25g UCL fermentation 24hrs induction, mixed feed VLP I. Concentration of final product TFF using Spectrum Labs 0.75MDa hollow fibre unit Pump speed 200ml/min throughout run Concentration phase, 200ml, TMP 3psi, P permeate 0.4psi Final concentration phase, volume reduced to 10ml Protein concentration determined by Bradford Assay = 0.63mg/ml Quality of product improved

Purification Development Additional/alternative purification procedures

KM71H pHe7 LAH H3,K1 VLP2 Cim-DEAE AEX S1000 pure VLP loaded onto Cim-DEAE (1ml) in 20mM Tris, pH 8.4, 5mM EDTA, 1M 2M Urea, 1M NaCl – this should not bind DEAE Total input Via by- pass Load column at 1M NaCl VLP all bound or stuck in loop Wash loop 2M Urea + NaCl 2 NaOH VLP not in loop Wash columnNaOH Wash column GuHCL Wash column 0.1% Trition X100 20mM Tris pH 8.4 5mM EDTA OD due to Triton not elution of VLP 30% IsoPropanol Elution with IPA indicates hydrophobic interaction TE 30% IsoPropanol TE 1M Acetic Acid, 1.5M NaCl 100mM NaCitrate pH 6.0, 3M NaCL Conclusion – VLP are binding non-specifically to column and not to charge group (at this pH) Expected to elute here

Total input Via by- pass Load column at 0 NaCl Step elute at 10%, 15% 20%, 50% and 100% NaCl (1M) Hold in 100% NaClovernight TE 2M Urea 1M NaOH, 2M NaCl TE GuHCl TE 30% IsoPropanol TE 0.1% Trition X100 TE 30% IsoPropanol TE 30% IsoPropanol – Hold 40 min TE 100mM NaCitrate pH 6.0, 3M NaCL 1M Acetic Acid, 1.5M NaCl S1000 pure VLP loaded onto Cim-DEAE (1ml) in 20mM Tris pH 8.4, 5mM EDTA, 1M 2M Urea,– this should bind DEAE Conclusion – VLP are not binding specifically to the charge group KM71H pHe7 LAH H3,K1 VLP2 Cim-DEAE AEX

Optimisation of ammonium sulphate precipitation Most VLP is precipitated by 40% ammonium sulphate rHBc is recovered at 50% efficiency but tandem core is only recovered at the 20% rate This is not improved by the addition of Triton X100 or 2M urea

20% AmSO 4 30% AmSO 4 40% AmSO 4 1 sec exposure 5 min exposure rHBc (50 ng) Marker pHe7 LAH H3,K1 S/N 100mM Citrate buffer pH mM Citrate buffer pH mM Citrate buffer pH mM HEPES buffer pH mM HEPES buffer pH mM HEPES buffer pH 7.5 TE buffer pH 8.4 rHBc (50 ng) Marker pHe7 LAH H3,K1 S/N 100mM Citrate buffer pH mM Citrate buffer pH mM Citrate buffer pH mM HEPES buffer pH mM HEPES buffer pH mM HEPES buffer pH 7.5 TE buffer pH 8.4 rHBc (50 ng) Marker pHe7 LAH H3,K1 S/N 100mM Citrate buffer pH mM Citrate buffer pH mM Citrate buffer pH mM HEPES buffer pH mM HEPES buffer pH mM HEPES buffer pH 7.5 TE buffer pH 8.4 Precipitation and recovery of VLP2 using ammonium sulphate and pH change VLP (post CL4B & S1000) was precipitated using increasing percentages of saturated ammonium sulphate The pellets were resuspended in buffer at varying pH levels Tandem core was recovered at both pH 8.4 and 6.0 It is possible that the change on pH6 tandem core may be more permissive to AEX Differential precipitation itself may be a viable final polishing step

rHBc (50 ng) Marker VLP1 S/N 100mM Citrate buffer pH mM Citrate buffer pH mM Citrate buffer pH mM HEPES buffer pH mM HEPES buffer pH mM HEPES buffer pH 7.5 S/N TE buffer pH mM HEPES buffer pH 8.0 rHBc (50 ng) Marker VLP1 S/N 100mM Citrate buffer pH mM Citrate buffer pH mM Citrate buffer pH mM HEPES buffer pH mM HEPES buffer pH mM HEPES buffer pH 7.5 S/N TE buffer pH mM HEPES buffer pH 8.0 rHBc (50 ng) Marker VLP1 S/N 100mM Citrate buffer pH mM Citrate buffer pH mM Citrate buffer pH mM HEPES buffer pH mM HEPES buffer pH mM HEPES buffer pH 7.5 S/N TE buffer pH mM HEPES buffer pH 8.0 Precipitation and recovery of VLP1 using ammonium sulphate and pH change VLP1 (post CL4B & S1000) was precipitated using increasing percentages of saturated ammonium sulphate The pellets were resuspended in buffer at varying pH levels No recovery of VLP1 was seen at all using ammonium sulphate This differs from VLP2 when at least some material was recovered These data confirm that VLP1 is not compatible with ammonium sulphate precipitation

AEX does not work at pH 8.4 but may be possible at pH 6.0. Therefore, a pH change is required pH change by dilution VLP 2 final product from 100g Prep 1ml of VLP final product at 0.8mg/ml (in 20mM Tris pH 8.4, 5mM EDTA) diluted with 9ml pH 6.0 buffer (100mM Na-Citrate) Filtered through 0.2 µm filter – Note approximately 90% of protein was removed by filtration Filtrate applied to Cim-DEAE (1ml) 0.6 µm pore pH change by diafiltration VLP final product from 100g Prep TFF using Spectrum Labs 0.75MDa hollow fibre cassette pre-washed in normal VLP storage buffer (20mM Tris pH 8.4, 5mM EDTA) 2ml of VLP final product at 0.8mg/ml applied to TFF and circulated with permeate closed to mix. Resulting in dilution to 9ml total (0.17mg/ml) Diafiltration inlet of TFF connected to reservoir of pH 6.0 buffer (100mM Na-Citrate) Diafiltration with 10 volumes of buffer – theoretical result 99.9% buffer exchange At a point equivalent to 1 volume exchange visible precipitation is observed accompanied by corresponding increases in measured pressure This corresponds to a measured pH transition through pH 7.0 – 6.5 At the end of total diafiltration period the precipitate was removable by a 0.8 µm filter

KM71H pHe7 LAH H3,K1 S1000 purified VLP buffer exchanged from pH 8.4 to pH 6.0 by dilution 10ml S1000 product diluted (1:10) to bring to pH 60 loaded onto Cim- DEAE (1ml) 0.6µm pore Buffer A = 100mM Na-Citrate pH 6.0 Buffer B = 100mM Na-Citrate pH 6.0, 1M NaCl Buffer C = 1M NaOH, 2M NaCl 1.Step to 20%, 60%, 80% and 100% Buffer B 2.Wash Buffer A 3.Wash Buffer C NaOH

SEC-HPLC analysis of AEX purified S1000 material VLP2 prep Prep after pH change by dilution Fr 15 collected from AEX 70% NaCl material eluted from AEX appears to be VLP (zoomed)

KM71H pHe7 LAH H3,K1 UCL fermentation VLP 2 prep at pH 6.0 and Cim-DEAE AEX Crude lysate + Triton X100 20K SN 20K Pellet CL4B Input 1:10 Diluted 20K SNPermeate CL4B Void CL4B Input TFF rHBc (50ng) Marker Crude lysate – Triton X100 ECL 1 sec exposure AEC stain Lysis at pH6 improves the quality of CL4B void material

KM71H pHe7 LAH H3,K1 UCL fermentation VLP prep at pH 6.0 and Cim-DEAE AEX: CL4B XK26/40 #3 #4 #5 #13 #7#9 #15 #11 #10 rHBc (50ng) Marker #2 ECL 1 sec exposure AEC stain Fractions 2 – 5 pooled as CL4B void. This then used as input for further studies with S1000 and Cim-DEAE

KM71H pHe7 LAH H3,K1 UCL fermentation VLP prep at pH 6.0 and Cim-DEAE AEX: S1000 XK26/55 #5 #7 #9 #19 #11#13 #21 #17 #15 rHBc (50ng) Marker #3 10ml CL4B Void loaded onto S1000 XK26/55 in 20mM Tris pH 8.4, 5mM EDTA ECL 1 min exposure ECL 5 min exposure S1000 SEC-HPLC

KM71H pHe7 LAH H3,K1 UCL fermentation VLP prep at pH 6.0 and Cim-DEAE AEX: Cim-DEAE-1 10ml CL4B Void loaded onto Cim-DEAE (1ml) Buffer A = 100mM Na-Citrate pH 6.0 Buffer B = 100mM Na-Citrate pH 6.0, 1M NaCl Buffer C = 1M NaOH, 2M NaCl Note: 1 st NaOH wash conductivity did not reach correct value – probable bubble in pump – 2 nd wash correct

KM71H pHe7 HA2.3,(M2e) 3 UCL fermentation VLP prep at pH 6.0 and Cim- DEAE AEX Crude lysate + Triton X100 20K SN 20K Pellet CL4B Input 1:10 Diluted 20K SNPermeate CL4B Void CL4B Input TFF rHBc (50ng) Marker Crude lysate – Triton X100 ECL 1 sec exposure 1 min exposure VLP1 can also be lysed at pH6 but the yield is lower

KM71H pHe7 HA2.3,(M2e) 3 UCL fermentation VLP prep at pH 6.0 and Cim-DEAE AEX: CL4B XK26/40 #3 #4 #5 #13 #7#9 #15 #11 #10 rHBc (50ng) Marker #2 Fractions 2 – 5 pooled as CL4B void. This was used as input for further studies with S1000 and CIM-DEAE ECL 1 sec exposure 1 min exposure

KM71H pHe7 HA2.3,(M2e) 3 UCL fermentation VLP prep at pH 6.0 and Cim-DEAE AEX: S1000 XK26/55 10ml CL4B Void loaded onto S1000 XK26/55 in 100mM Na-Citrate pH ml CL4B Void loaded onto S1000 XK26/55 in 20mM Tris pH 8.4, 5mM EDTA VLP1 needs further buffer optimisation: National Physical Laboratory

Cim-DEAE-0.2µm Over Max allowed pre-column pressure Cim-DEAE- 0.6µm No pressure issues Effect of pore size of Cim-AEX column Both columns are identical except for the variation of pore size CL4B (pH6) material was passed over both. Yields are very low after salt elution In both cases, a sizeable amount of material remains on the column and can only be liberated using NaOH This is suggestive of aggregation and precipitation on the column NaCl ElutionNaOHPost NaOH In SEC, 12g pellet VLP2 yielded 2mg. Therefore, 6g gives 1mg In AEX 3.6g pellet yielded 50µg (expect 500µg). This suggests 10% recovery from a 2µm DEAE CIM column.

Experiments currently being done 1)CL4B followed by large S1000 at pH6- requires reformulation of existing column. Care will be taken to ensure that protein is not released from the column during equilibration to the lower pH. 2)PEG optimisation- to provide a specific VLP enrichment step. Material from a recent S1000 run will be precipitated and recovered. If this is not very efficient then this method will be abandoned. 3)Addition of glycerol (10%) to SEC/AEX buffers- prevention of aggregation and precipitation 4)HPLC calibration curve- use of rHBc and BioRad markers 5)VLP1 lysis pH- use of HPLC to monitor solubility of VLP optimal pH Lysis pH6 Spin CL4B pH6 S1000 pH6 TFF (conc) AEXTFF (desalt) TFFFilter 8M urea (1M final) 10% glycerol Urea removed Summary Quality of both VLP1 and VLP2 increased by harvesting at 24hrs Material from 3P fermentation of VLP2 gives acceptable product by SEC SEC purification has been successfully scaled up from 25g to 100g cell pellet SEC purification of VLP2 can be performed with hold between CL4B and S1000 columns S1000 column requires conditioning before use Cim-DEAE at pH 6.0 is possible polishing step