The project aimed to develop Trichoderma reesei as a cost-effective platform for producing therapeutic proteins like antibodies. Major barriers were protease degradation and incorrect glycosylation. Through protease gene deletions and fermentation optimization, production levels were increased from 50 mg/L to over 7 g/L for antibodies and interferon. Glycoengineering was used to modify T. reesei's glycosylation pathway to produce mammalian-like Man5 and G0 glycoforms important for antibody functionality. Deletions of genes like alg3, pmt1, and additions like Stt3, GlsII, MnsI, GnTI/II altered glycosylation for therapeutic protein production.

1. TEKNOLOGIAN TUTKIMUSKESKUS VTT OY
Next Generation Biotherapeutics Production System: The
Filamentous Fungus Trichoderma reesei
Chris Landowski
VTT Industrial Biotechnology
PEGS 2016

2. 2
Project Background
The aim of the project was to create a therapeutic production system to
make a more cost effective production platform using Trichoderma reesei
A multi-year collaboration between:
Novartis
Glykos Finland Oy
VTT Technical Research Centre of Finland
Protein production
 mainly antibodies, but also other non-antibodies such as cytokines
and growth factors
Glycoengineering
 antibodies need to have correct glycan forms
 Man5, G0, FG0, and higher forms
Major barriers to production: protease activity and intracellular degradation

3. 3
 Originally a tropical soil fungus, a soft rot ascomycete
 A well known industrial producer of cellulolytic and
hemicellulolytic enzymes and heterologous proteins. The
native enzymes are used in food, feed, detergent, textile,
pulp & paper, and biorefinery applications
 Has a GRAS status for manufacture of food products
 Has a favorable glycosylation pattern, high proportion of
GlcNac2Man5 structure (Stals et al., 2004, Glycobiology 14,
725-737)
 An efficient protein secretor; the highest reported protein
production levels > 100 g/l (Cherry and Fidantsef, 2003,
Curr. Opin. Biotechnol 14, 438-443)
 Good process performance in large bioreactors. Amenable
to continuous culture mode.
Trichoderma reesei

4. 4
Examples of recombinant protein expression in T. reesei at VTT
Protein Yield g/l Reference
Bovine chymosin 0.15 Harkki et al., 1989, Bio/Technol 7, 596-603
Antibody Fab fragments 0.15 Nyyssönen et al., 1993, Bio/Technol. 11, 591-
595.
Phlebia radiata laccase 0.14 Saloheimo and Niku-Paavola, 1991,
Bio/Technol. 9, 987-990.
Trichoderma harzianum chitinase 8 Margolles-Clark et al., 1996, Appl. Environ.
Microbiol 62(6):2145-51.
Melanocarpus albomyces laccase 0.9 Kiiskinen et al., 2004, Microbiology, 150, 3065-
3074.
T. reesei tyrosinase 3 Selinheimo et al., 2006, FEBS J. 273, 4322-
4335.
Coprinus cinereus cutinase
Dipodascus lipase
Yeast-derived lipase
3
2
5
Unpublished
Aspergillus niger glucose oxidase 14 Unpublished
Fungal xylanase 1 13 Unpublished
Fungal xylanase 2 28 Unpublished

5. 5
Target protein expression strategies – full-length IgG
antibodies and interferon α2b
cbh1p carrier cbh1t HygR cbh1 3’ flank
IFNαCBHI
cbh1p carrier cbh1t cbh1p carrier cbh1t amdS cbh1 3’ flank
HC LCCBHICBHI
cbh1p carrier cbh1t cbh2t carrier cbh1p HygR cbh1 3’ flank
LCCBHICBHI HC
Both chains expressed from a tandem or inverted double construct in the cbh1 locus
Secretion carrier strategy:
-The target protein is fused with the most abundant native secreted protein, CBHI
-The target protein is separated from the carrier during secretion in golgi by the
host’s KEX2 peptidase
Interferon α2b
IgG antibodies
Initial production level ~50 mg/l
Initial production level ~50 mg/l

6. 08/04/2016 6
Barrier to production: proteases
There are 40 or more secreted
proteases in the culture
supernatant
Protease inhibitors were used to
profile which classes were most
problematic
Using the inhibitors these
proteases could be purified and
identified
Inhibiting serine proteases with
PMSF stabilized immunoglobulin
and aspartic proteases with
pepstatin A to a lesser degree
Immunoglobulin incubated in T. reesei culture
supernatant treated with and without protease
inhibitors

7. 7

8. 8
Proteases destroy peptide bonds
Major classes:
Serine
Cysteine
Aspartic
Threonine
Glutamic
Metalloproteases
T. reesei needs secreted proteases to break down protein substrates to provide nutrients and for cell wall remodeling
functions; 50 putative secreted proteases can be predicted

9. 9
Protease characterization → protease gene deletion series
 Criteria used for selecting the critical proteases for deletion from the production strain
 Spiking of supernatant with target proteins – inhibitor studies
 Isolation of specific protease groups with inhibitor affinity chromatography –
activity tests with target proteins, identification by mass spectrometry
 Deletion of individual protease genes in target protein producing strains, analysis
of the effect
 Abundance of the protease in the extracellular proteome or at mRNA level
 Expression of selected proteases in Pichia pastoris – activity analysis
 A series of protease gene deletions was made in a production strain
 The successive transformations were made by recycling the pyr4 marker gene
(FOA back-selection)

10. 10
Spiking of T. reesei supernatant with the target proteins
 Both the heavy chain and interferon α2b were degraded
 Chymostatin and SBTI (soybean trypsin inhibitor) enhanced IgG stability → the major problem is serine proteases
 Pepstatin enhances interferon α2b stability → the major problem is aspartic proteases
untreated chymostatin SBTI chymostatin/SBTI
118 kD
97 kD
54 kD
37 kD
29 kD
17 kD
1h 19h 1h 19h 1h 19h 1h 19h 20 hours0 time
25 kD
20 kD
15 kD
10 kD
50 kD
37 kD
Antibody heavy chain in the WT strain supernatant Interferon α2b in 6-protease
deletion strain supernatant

11. 11
Protease isolation with pepstatin affinity chromatography
 Affinity chromatography of T. reesei supernatant with pepstatin resin yielded one major protein and several
minor bands
 The purified protease fraction was active against IgG antiibody
 Similar affinity purification from a pep1 deletion strain showed that the major purification product is PEP1
104 kD
94 kD
51 kD
36 kD
28 kD
F1 F2 F3
150 kD
104 kD
94 kD
104 kD
94 kD
51 kD
36 kD
28 kD
19 kD
Fractions in SDS gel Fraction F3 activity against IgG Isolation from a pep1 deletion strain

12. 12
Protease characterization with the help of zymography
 One major and one minor protein band showing activity against IgG were isolated with
aminobenzamidine affinity chromatography
 Analysis of the protease deletion strains showed that the major band was TSP1 (trypsin) and the
minor one was SLP1 (major subtilisin)
104 kD
94 kD
52 kD
36 kD
28 kD
20 kD
Aminobenzamidine affinity chromatography
50 kD
38 kD
28 kD
100 kD
slp1
tsp1
Analysis of protease deletion strain supernatants
SDS gels cast with IgG, renatured and
stained after running

13. 13
Proteases found in the supernatant
 pep1, (42 kD), aspartic*
 pep2, aspartic*
 pep3, aspartic*
 pep4, (42 kD), aspartic*
 pep5, aspartic*
 pep8, aspartic
 pep9, aspartic*
 pep11, aspartic
 pep12, aspartic
 tsp1, (26 kD), trypsin-like*
 slp1, (93 kD), subtilisin*
 slp2, (58 kD), subtilisin
 slp3, subtilisin
 slp7, (60 kD), subtilisin
 slp8, (41 kD), subtilisin*
 gap1, (26 kD), glutamic*
 gap2, glutamic*
 tpp1, (65 kD), sedolisin
 sep1, (59 kD), serine endoprotease*
 amp1, (55 kD), aminopeptidase
 amp2, aminopeptidase*
Total of 34 proteases found – at variable abundances
The ones in red have been deleted from a strain (21)
*Are in the multiple deletion strain
 Four metalloproteases
 Three carboxypeptidases
 A cysteine endoprotease
 A cysteine protease
 A dipeptidyl peptidase
 Two aminopeptidases
 kex2, serine protease

14. 14
Sequential protease deletions –seven first steps
Total protease activity against fluorescent casein substrate was measured from small-scale cultures
pep1 tsp1 slp1 gap1 gap2 pep4 pep3

15. 15
Seeking improvements in interferon α2b production by
individual protease deletions
 Five different protease
genes were deleted in the
interferon producing strain
M577 (carrying 8 deletions)
 Immunoblot of bioreactor
samples
 The parental strain
produced
 0.84 g/l of interferon
 Four of the new protease
gene deletions enhanced
interferon production
 amp2 to 2.42 g/l
 slp7 to 2.11.g/l
 pep9 to 1.03 g/l
 sep1 to 1.38 g/l

16. 16
Protease deletion strains grow faster than
parental strain
faster
Aber Futura biomass signal (capacitance) and CO2 transfer rates
13 ∆(pep1 tsp1 slp1 gap1 gap2 pep4 pep3 pep5 pep2 sep1 slp8 amp2 pep9)
Total protein secreted by Δ6
strain increased 22%
Protease deletions may have
triggered a starvation response
related to nitrogen
The series was continued to a
strain with 13 protease genes
deleted, growth also improved
Only the deletion of the
subtilisin-like protease slp2
caused problems with
sporulation/growth.
Problems addressed by
lowering the slp2 level with
RNAi and promoter switch
approaches

17. 17
Antibody production – lots of unused potential
 Protease deletion strains together with fermentation optimisation work improved the IgG
antibody production levels up to 7.1 g/l
 The secretion carrier CBHI is produced in the fermentations at levels up to 38 g/l. This
equates theoretically to antibody levels of approximately 29 g/l.
1096532 8741 std
Heavy chain
Light chain
1096532 8741 std

18. 18
Interferon α2b production fermentation
 The production strain has 9
protease gene deletions and
an slp2 RNAi construct
 Almost all interferon cleaved
from the secretion carrier
 Interferon level increases
until the fourth culture day
and starts declining
 With a protease inhibitor
cocktail a level of 10.7 g/l
has been reached
standards
Immunoblot analysis

19. 19
Expression of an IGF1 fusion protein
The strain was able to producing 7.9 g/L of native IGF1 when bound to the carrier CBHI. The strain had
13 protease deletions. When inhibitors were added to the culture the level could be raised to 20 g/L
immunoblot using IGF1 antibody- carrier fusion
75 kD
50 kD
37 kD
25 kD
20 kD
15 kD
10 kD
100 kD
150 kD
250 kD
anti-CBHI
0.05 µl
supernatant
0.025 µl
supernatant
IGF1 std
75 kD
50 kD
37 kD
25 kD
20 kD
100 kD
150 kD
15 kD
10 kD
250 kD
immunoblot using IGF1 antibody- carrier fusion with
inhibitor treatment

20. 20
Expression of the protease sensitive FGF21
An earlier strain with 5 proteases deleted produced only 130 mg/L of the same 10 kD product. This is a 18x
improvement.
FGF21 was stabilized with inhibitors. The major stabilized form was 17 kD and was present at 3.5 g/L. The full length
product was observed at a level of 200 mg/L.
Supernatant samples were diluted so that 0.1 µl could be loaded per well. The strain had 13 proteases deleted
Δ(pep1 tsp1 slp1 gap1 gap2 pep4 pep3 pep5 pep2 sep1 slp8 amp2 slp7).
no inhibitors with inhibitors
d2 d3 d4 d5 d2 d3 d4 d5
75 kD
50 kD
37 kD
25 kD
20 kD
15 kD
10 kD
100 kD
150 kD
250 kD
standards

21. 08/04/2016 21
Production yields with different target proteins
Antibodies produced in Δ7 strain, IFN in Δ9 strain, and IGF1 in Δ13 deletion strain as
fusion with CBHI carrier. However, this is far from the maximal theoretical output.
For Mab01 we should be able to reach 29 g/L, based upon carrier expression level.

22. 22
Protein production conclusions
 Proteases degradation was discovered to be a major bottle neck for production
of therapeutic proteins - IgG antibodies, interferon α2b, IGF1, FGF21 - in
Trichoderma reesei
 The secreted proteases and their activities towards the target proteins were
analysed
 A production strain series of up to 13 protease gene deletions was made.
Several protease deletions enhanced the growth rate of the strain. Only the
deletion of slp2 gene caused problems in growth/sporulation.
 With the help of protease elimination and fermentation optimisation the
production levels of selected therapeutic proteins, IgG antibodies and
interferon α2b, could be increased from about 50 mg/l to over 7 g/l .

23. 23238.4.2016
Glycoengineering to form mammalian-like
glycan structures
 Trichoderma reesei naturally has around 80% GlcNac2Man5 structure
 Traditional pathway
 Alg3 pathway

24. 24
Best approaches – clean Man5%N-glycanonantibody
Modifications:
• ∆pmt1 [protein O-mannosyltransferase]
• Stt3 [oligosaccharyltransferase]
• α-1,2-mannosidase I
Man5 glycoform – Traditional pathway

25. 25
Alg3 pathway
alg3 is an α-1,3-mannosyltransferase

26. 26
Best approaches – Clean G0
Modifications:
• ∆alg3 [α-1,3-mannosyltransferase]
• ∆pmt1 [protein O-mannosyltransferase]
• Stt3 [oligosaccharyltransferase]
• GlsII [α-glucosidase II]
• MnsI [α-1,2-mannosidase I]
• GnTI & GnTII [GlcNAc transferases]
%N-glycanonantibody
G0 glycoform - alg3 pathway

27. 27
Glycoform Genotype Glycoform
Functionality
Aglyco any No ADCC, deficient CDC
Man3 STT3, ΔEndoT, ΔALG3, (GlsII or Endo-
mannosidase)
enhanced ADCC, normal CDC
Man5 STT3, MnsI, ΔEndoT enhanced ADCC; normal CDC
G0 Man3; GNTI, GNTII enhanced ADCC, normal CDC
FG0 G0; fucose synthesis pathway,
fucosyltransferase
normal ADCC, normal CDC
G2 G0; GalT (galactosyltransferase) enhanced ADCC, enhanced CDC (C1q)
G2F G0F; GalT (galactosyltransferase) normal ADCC, enhanced CDC (C1q)
EndoT= endo-β-N-acetylglucosaminidase; STT3= oligosaccharyltransferase; alg3 =α-1,3-mannosyltransferase;
MnsI = α-1,2-mannosidase I; GNT= GlcNAc transferase; glsII=α-glucosidase II
Glycoengineering summary

28. 28288.4.2016
T. Reesei produced fungal-type O-glycosylation
The innermost hexose unit was resolved to be mannose
A Pmt gene was deleted
As a result, O-glycosylation was reduced in mAbs
No negative impact on growth or antibody production was observed
22000 23000 24000 25000 26000 m/z
LC
22000 23000 24000 25000 m/z
LC
O-glycosylated
WT
no G-eng
Δpmt
O-glycosylated
Fungal O-glycosylation was removed

29. 29
Trichoderma reesei biotherapeutics manufacturing process
Cell line development Fermentation Downstream processing Product quality
Mature system for expression
of industrial enzymes (50-100
g/L routine)
Short development time for
MCB (2 months should be
achievable)
Single step transformation with
minimum strain selection
(targeted integration)
Can be used for expression of
wide variety of therapeutic
proteins
High titers, e.g. mAbs (7.6 g/L)
and interferon 2b ~8 g/l
High expression levels,
allowing for smaller reactors
Standard microbial reactors
used, can be adapted to single
use reactors
Defined media possible, no
components of animal origin
Low cost of media
Short fermenter times (4-7
days)
Short process development
times (6 months from gene to
GMP manufactured DS for
PoC)
Well suitable for continuous
manufacturing
Target protein secreted into
media at high titer
Simple primary harvesting
procedure
No virus inactivation, removal
or validation necessary
No inclusion bodies
Very high expression levels,
robust cells, therefore low host
cell protein contamination
achievable
Naturally afucosylated, ADCC
enhanced
Glycoprotein profile Man5,
Man3, G0, FG0, G2
Fucosylation
Low/no O-glycosylation
Only single O-mannoses, if
any
GRAS designation
Homogeneous glycoprofile

30. 30
The consortium
VTT Technical Research Centre of Finland Ltd.
 Christopher Landowski, Anne Huuskonen, Ann
Westerholm-Parvinen, Eero Mustalahti, Georg
Schmidt, Dhinakaran Sivasiddarthan, Maija Pollari,
Merja Penttilä, Markku Saloheimo
Novartis
 Ramon Wahl, Benjamin Sommer, Christian
Ostermeier, Bernhard Helk
Glykos Finland Ltd.
 Jari Natunen, Anne Kanerva, Anne Leppänen, Hanna
Salo, Anna-Liisa Hänninen, Noora Salovuori, Heidi
Salminen, Annamari Heiskanen, Maria Blomqvist, Titta
Kotiranta, Anne Olonen, Virve Pitkänen, Henna
Pynnönen, Jari Helin, Annika Kotovuori, Olli Autio,
Päivi Pihkala, Risto Kajanne, Juhani Saarinen