THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
The ideal bioreactor
General features of animal cells Ideal bioreactors Metabolic network analysis Systems biology
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Microbial versus animal cell culture
ammonia glucose amino acids
glucose
biomass biomass product acetate product lactate
amino acids ammonia
Fed batch to avoid overflow metabolism Model based with on-line measurements Very high cell density (>100 g-DW/L)
Bolus feeding Heuristic approaches Low cell density: 2-10 g-DW/L
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Key differences (1)
Complex medium with two partially substitutable major substrates
• No simple one-substrate feed strategy
Cell cycle: use of cell numbers rather than mass
• Size (and hence yields) varies with growth rate • Numbers are delayed indicators of behaviour • Many triggers of (delayed) cell death
Apoptosis
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Medium
Quantity
Glucose, lactate Essential amino acids
Quality
Water: RO/MQ Balanced salts (note P) Buffer: bicarbonate, HEPES Fe chelator Detoxification
• •
Arg, His, Ile, Leu, Lys, Met, Phe, Thr, Trp, Tyr, Val, Cys GLN, ALA, NH3, Asn, Asp, Glu, Gly, Pro, Ser
• • • • •
Antioxidants: pyruvate, lipoic acid (GSH, ascorbate) Decoys: Pluronic F68 (albumin) Osmolarity: 280-360m Osm/kg pH 7.0-7.4 30-40 °C (37°C)
Non-essential amino acids
Physical parameters
Oxygen, CO2 Trace elements
B-vitamins, choline, myoinositol, ethanolamine, metals Lipids, nucleotide bases
•
Stimuli: growth factors and attachment factors
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Key differences (2)
Complex medium with two partially substitutable major substrates Use of cell numbers rather than mass
• No simple one-substrate feed strategy and lots of side-issues • Size (and hence yields) varies with growth rate • Numbers are delayed indicators of behaviour • Many triggers of (delayed) cell death
Apoptosis
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Numbers versus mass
0.10
1700
SS4 SS7
0.08 qm, mmol/109 cells/h
1500 Cell volume, µm3 SS1 SS6 SS2
0.06
1300 SS3 SS5 SS8
0.04
1100
0.02
0.00
900 0.00 0.02 µ, h-1 0.04 0.06
0.00
0.02 µ, h-1
0.04
0.06
Arg, His, Ile, Lys, Met, Phe, Thr, Tyr, Val
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Shift-up experiment
Cells were taken from a CSTR operated at 0.41 d-1 Inoculated in fresh medium Biomass rate: 1.54 (0.07) d-1 Number accumulation 0-8h: 0.31 (0.10) d-1 12-24h: 1.66 (0.17) d-1
Total cell volume, µm ml
3
-1
2x108 108 8x107 6x107 4x107 2000 1800 1600 1400 1200 1000 0 6 12 18 Time, hours 24 105 8x104 6x104 4x104 2x105 Cell number, cells ml
-1
Cell volume, µm
3
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Key differences (3)
Complex medium with two partially substitutable major substrates Cell cycle: use of cell numbers rather than mass
• No simple one-substrate feed strategy and lots of side-issues • Size (and hence yields) varies with growth rate • Numbers are delayed indicators of behaviour • Many triggers of (delayed) cell death
Apoptosis
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Cell cycle and cell death
In bioreactor
Apoptosis (“suicide”)
•
Internal (“BCL-2”)
Stress signal Release of Apaf-1 from BCL-2 and Cyt C from mitochondria Activation of caspase-9 Proteolytic cascade
• • • • • • • •
• •
External (Fas/TNF)
Stimulation of Fas or TNF receptors
Lack of growth factor Starvation (glutamine, glucose) Byproducts (lactate, ammonia) High or low oxygen Sub-optimal temperature, pH shock, osmolarity shock Injury (mechanical/chemical) Loss of plasma potential Swell & burst
Necrosis (“murder”)
Process can take many hours
cells remain metabolically active
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Monitoring biomass
Viable & dead cell counts represent a natural structure of cell culture models, but Numbers do not correspond to mass
• Average mass increases with growth rate • Need sizing or protein content to go with counts to link
metabolism to growth
Number rates represent historical data
• Viable cell number: mass growth rate 8-12 hours earlier • Dead cell number: apoptosis 6-48 hours earlier • Can use flow cytometry to monitor apoptosis
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Ideal bioreactors
Batch processes
• Batch • Bolus fed batch • Fed batch • Continuous culture • Homogenous perfusion • Heterogeneous perfusion
Continuous processes
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Batch processes
Secondary metabolite
• Production growth independent • Yield ~ time integral of viable, productive cell mass
5 4 3 2 1 5 4 3 2 1 50 100 150 200 0 0 50 100 150 200
5 4 3 2 1 0 0 50 100 150 200
0 0
rapid death unproductive
rapid death productive
slow death productive
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Specific productivity
0.5
rMAb, mg (g-wet)-1h-1
0.4 0.3 0.2 0.1 0.0 0.00 0.02 0.04 0.06
Growth rate, h-1
Murine hybridoma
Lloyd (2000) Cytotechnology 34: 59–70.
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Productivity
Assume exponential growth and decline Titre ~ CI = CI growth + CI decline
CI growth
xf 1 = ∫ xdt = (x f − x0 ) CI decline = ∫ xdt = kD µ
Titre effectively proportional to peak cell density, xf For a given maximum cell yield, titre from growth is inversely proportional to growth rate Production in the decline phase depends on why growth terminated (e.g., EAA depletion= no product) and is inversely proportional to the death rate
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
What limits xf?
Static culture
• • • • • • • •
pH or oxygen limited Glutamine depletion Keep salt buffer constant, except adjust NaCl for osmolarity Enriching all essential nutrients in proportion 2-3X essential nutrients possible Substrate inhibition (glucose)? Waste products from degradation? Salt, nutrient imbalance?
Standard medium Enriched medium
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
What limits xf?
3 60 50
Bolus feeding
Cells, 109 cells/L
• • •
Restore essential nutrients as they run low using nutrient concentrates Not much better than enriched batch Product inhibition
2
40 30
1
20 10
0 6 4 2 0
0 20 40 60 80 100 120 140 160
0 350 300 250 200
180
Ammonia, mM
Time, h
mOsmol/kg
MAb, mg/L
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Fed batch: growth control
Nutrient limitation
• • •
Feeding can typically increase xf and thus titre 2-3 fold Accumulation of ammonia and lactate cause severe inhibition and death Rate of accumulation (µxYxp) rather than actual level most important
Product inhibition
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
4 Cells, 109 cells L-1 3 2 1
Stress versus absolute level
40
1.2 [cells], 10 cells L
-1
[Glucose], mM qglc, pmol cell day
-1
NH3 step change
9
1.6
-1
12.0 9.0 6.0 3.0 1.2 0.8 0.4 0.0 0 50 100 150 Time, h 200 250
0.8 0.4 0.3 0.2 0.1 0.0 0 50
30 T2, hours
mOsmol kg-1
0 400 350 300 60 40 20 0 6
-1
20
100 150 Time, h
200
250
-1 -1
vOUR, mM/h qoxy, pmol cell day
4.00 3.00 2.00 0.12 0.11 0.10 0.09 0 2.5 50 100 150 Time, h 200 250
[Glutamine], mMqgln, pmol cell day
5.00
3.0 2.5 2.0 1.5 0.8 0.6 0.4 0.2 0.0 0 50 100 150 Time, h 200 250
MAb, mg L-1
10
0
1
2
3
4
5
[Initial ammonia], mM
NH3 feed
24 [Ammonia], mM 18 12 6 0 1.8 [Cells], 109cells/l 1.2 0.6 0.0
-1
15.00 [NH3], mM 12.00 9.00 6.00 3.00 1.00 YNH3/GLN 0.75 0.50 0.25 0.00 0 50
[Alanine], mM YALA/GLN
mM
4 2 0 0 48 96 Time, h 144 192
2.0 1.5 1.0 0.5 0.0 0 50 100 150 Time, h 200 250
0
30
60
Time, h
90
120
100 150 Time, h
200
250
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Fed batch: growth control
Nutrient limitation
• • • • •
Feeding can typically increase xf and thus titre 2-3 fold Accumulation of ammonia and lactate cause severe inhibition and death Rate of accumulation (µxYxp) rather than actual level most important If µxfYxp=Kstress defines xf then xf(0.5µ) = 2 xf(µ) => CIgrowth(0.5µ) = 4 CIgrowth(µ) Often Yxp also lower at lower growth rate!
Product inhibition
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Effect of slowing culture down
20 18 16 14 12 10 8 6 4 2 0 0 1 2 3 4 5 6 7 8
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Fed batch: growth control
Nutrient limitation
• • • • • • • •
Feeding can typically increase xf and thus titre 2-3 fold Accumulation of ammonia and lactate cause severe inhibition and death Rate of accumulation (µxYxp) rather than actual level most important If µxfYxp=Kstress defines xf then xf(0.5µ) = 2 xf(µ) => CIgrowth(0.5µ) = 4 CIgrowth(µ) Often Yxp also lower at lower growth rate! Can be difficult due to “memory” and without causing apoptosis Gambhir et al. (1999) J Biosci Bioeng 87: 805-810 (double fed batch) Total growth and hence inhibitor accumulation rate limited by supply at constant feed rate
Product inhibition
Yxp further reduced by limiting glc & gln simultaneously Limiting fed batch: DSinYsp=constant
-M2 + M3 +2~P +H2 = 0
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Stoichiometry
AS BP GXmet
γX
− 1 0 0 0 0 1 0 0 0 X + s + 0 0 0 0 1 0 0 0 0
0 1 0 0 −1 0 M 1 0 − 1 0 0 − 3 0 M 2 p − 1 1 0 0 0 0 1 M3 = 0 + 2 1 0 p2 0 − 1 1 0 0 −1 1 ~P 0 0 H2 0 0 − 1 0 − 2 1
E.g. R4: 0X + 0p1 + 0p2 + 0s +0M1- 1M2 + 1M3 +2~P +H2 = 0
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Rates
Rates are defined for each reaction (v1, v2, …, v6) and have unit mmol/g-DW/h.
rp1 v5 rx = v2 = γ v rs = −v1 = A v = = BT v rp v 2 6
T T
rmet
rM 1 v1 − v2 − v3 v3 − v4 rM 2 = GTv v4 − v5 − v6 = rM 3 = r~P − v1 − 3v2 + 2v4 + v5 r v3 − 2v5 H2
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
ACC = IN − OUT + GENERATED
∆(Cmet X ) = 0 − 0 + rmet X∆t
d (Cmet X ) = rmet X dt dCmet dX X + Cmet = rmet X dt dt dCmet 1 dX = rmet − Cmet dt X dt dCmet = rmet − µCmet = 0 ( PSS ) dt
+ rmet − µCmet ≅ rmet = 0 (rmet >> µCmet )
Mass balance
rmet = 1.1 mmol/g-DW/h
v3
M2
v4
µCmet = 0.1 µmol/g-DW/h
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Rates and balances
T T
rp1 v5 rx = v2 = γ v rs = v1 = − A v = = BT v rp v 2 6 rM1 v1 − v2 − v3 v3 − v4 rM 2 = GTv = 0 rmet = rM 3 = v4 − v5 − v6 r~P − v1 − 3v2 + 2v4 + v5 r v3 − 2v5 H2
5 equations, 6 unknown
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Example
(known) v3 = v4 = 2v5 v6 = v5 = rp2 v3 = 2v5
X 3~P v1 ~P v2 M1 v3 M2 2~P,“H2” v4 M3 v5 ~P Product 1
rp1 = v5 v2 = 3 v5 4
v1 = 2 3 v5 = −rs 4
substrate
2“H2”
v6 Product 2
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Creating the model
1156 reactions 277 transport 872 metabolites Compositional data tough for animal cells. Need to make several assumptions.
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Using the model
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Metabolite balancing
Simplified model
• Lumped biosynthesis (not essential) • Removal of reactions known to have low activity in tumor • •
cells (transcript analysis would be better) 59 reactions = 28 measured + 31 unknown 32 metabolite balances, but two redundant
1 degree of freedom 2 balances for measurement check
• Inequality constraints greatly reduces solution space
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
GLC CAR
3 1
GLN
38 19 21 66
GLUc
39 40
NH3 ASPc GLY ALAc ASN
42
60 34 35 36 37 61
GLUc ASN ASPc
G6P
2
R5P
20
NUC
65
16 15
F6P F1,6dP
4 64
NADH NADPH
63
SER
O2
O2 62 LIP CITc
28 29
LIP
11 14
9
GLN
41 17
GLUc
43 44 ASP 45
c
G3P
5 10
PROT
18
SER
13
3PG NUC PEP
6 7
AcCoA SER
OOAc
30 31
GLY
46
ALAc
50
12
GLY LAC
ALAc 52 PYRc PRO ASPc 51 OOAc
49
PYRc
8
MALc
32
α-KGc
GLUc 47 GLN
53 48
33
PYRm MI TOCHONDR IA
26
MALm
25 24 23
NH3
54
NH3 GLUm OOAm PYRm
OOAm
22
α-KGm
55 ASPm 56 ALAm
CITm
59 27
57
O2 ATP
GLUc
58
CYTOSOL
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Some conclusions
Cytosolic transamidase activity accounts for 2040% of GLN amido-N. Rest is released via GLNase GDH activity low. Energy from GLN only from “burning” of excess carbon from ALAAT and ASPAT reactions in TCA cycle No net transport of cytosolic glutamate to mitochondria
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
GLC CAR
3 1
GLN
38 19 21 66
GLUc
39 40
NH3 ASPc GLY ALAc ASN
42
60 34 35 36 37 61
GLUc ASN ASPc
Glutamine main contributor to NEAA in proteins Nitrogen in DNA/RNA bases Energy from glutamine derived from NADH generated as excess carbon is “burned”, e.g., Gln (5C) to Asp (4C) Ammonia from deamidation in mitochondria. Small amount from GDH Alanine: result of high Pyr and high Glu in cytosol? Reduced Gln metabolism with simultaneously low glucose and glutamine.
G6P
2
R5P
20
NUC
65
16 15
F6P F1,6dP
4 64
NADH NADPH
63
SER
O2
O2 62 LIP CITc
28 29
LIP
11 14
9
GLN
41 17
GLUc
43 44 ASP 45
c
G3P
5 10
PROT
18
SER
13
3PG NUC PEP
6 7
AcCoA SER
OOAc
30 31
GLY
46
ALAc
50
12
GLY LAC
ALAc 52 PYRc PRO ASPc 51 OOAc
49
PYRc
8
MALc
32
α-KGc
GLUc 47 GLN
53 48
33
PYRm MI TOCHONDR IA
26
MALm
25 24 23
NH3
54
NH3 GLUm OOAm PYRm
OOAm
22
α-KGm
55 ASPm 56 ALAm
CITm
59 27
57
O2 ATP
GLUc
58
CYTOSOL
THE UNIVERSITY OF QUEENSLAND
AUSTRALIA
Cell cycle
General pattern
• • • • • • •
Cyclin X accumulates Activates CDK Starts new phase Cyclin ubiquitinated & destryed Start again Includes autophosphorylation Translates linear signals (gradual accumulation) to steep sigmoidal signals
Kinase/phosphatase cascade
Four interlinked signaling cascades