Metabolic modelling of a bacterialanimal symbiosis

1
Metabolic modelling of a bacterial/animal
symbiosis
Sandy Macdonald and Gavin Thomas Department of
Biology University of York Angela
Douglas Department of Entomology Cornell
University USA
2
Outline

• The Buchnera / aphid symbiosis
• The Buchnera genome
• Construction and analysis of the Buchnera
  metabolic model

3
Many insects have obligate bacterial symbionts

• Aphid symbionts
• Buchnera sp. APS Acyrthosiphon pisum
• Buchnera sp. SG - Schizaphis graminum
• Buchnera sp. BP - Baizongia pistacea

• Ant symbionts
• Blochmannia floridanus
• Blochmannia pennsylvanicus

• Tsetse fly symbionts
• Wigglesworthia glossinidia brevipalpis

4
Aphids phloem sap feeding pests
The pea aphid, Acyrthosiphon pisum.

• About 5000 different species
• Major crop pests.
• Restricted diet of phloem sap.
• All contain a primary symbiont and often a
  secondary symbiont.

5
Buchnera aphidicola sp. APS is the primary
symbiont of the pea aphid

• The bacteria are located in specialised insect
  cells called bacteriocytes in their body cavity.
• They are surrounded by an aphid-derived
  bacteriocycte membrane.

• TEM of bacteriocyte cytoplasm, showing coccoid
  Buchnera.
• The Buchnera are unculturable
• Vertically transmitted to aphid offspring via
  the ovary.

• The function of the symbiosis is nutritional.
• Phloem sap poor in essential amino acids
  (histidine, isoleucine, leucine, lysine,
  methionine, phenylalanine, threonine, tryptophan
  and valine).
• These are known to be provided by the symbiont.

6

• The Buchnera APS genome
• Small - 0.64 Mb
• 607 genes (569 protein coding genes)
• AT rich - 73 AT
• Almost 90 of the genes have known functions in
  E. coli K-12.
• Virtually no transcription regulation.
• No IS or phage elements
• Genome reannotated in BuchneraBASE.

7
The genome sequences of primary insect symbionts
reveal clear descent from an ancestral
?-proteobacterium
Ancestor 3.5 Mb 200 My ago
8
The Buchnera genome has undergone reductive
evolution by a series of deletion and
inactivations of existing genes

• Initial large deletions removed large regions of
  the genome.
• Followed by smaller deletions and gene
  inactivation.
• Genomes are still relatively co-linear
  (syntenic).

9
The genome supports the proposed symbiotic
function
10
Understanding the symbiosis

• Exploiting E. coli to rapidly produce a high
  quality metabolic reconstruction of Buchnera APS

11
The iJR904 metabolic model of E. coli K-12
Metabolic model which contains the reactions
catalysed by 904 gene products from E. coli.

• Built from biochemical/genetics literature for
  E. coli
• High quality reconstruction
• Contains 931 unique biochemical reactions
• Mass and charge balanced
• e.g.
• ch2o sl26da --gt 26dap-LL succ
• c26dap-LL ltgt 26dap-M
• c26dap-M h --gt co2 lys-L

Central metabolism
12
Construction of the metabolic network

• Method
• Map all orthologues between APS and K-12.
• Manually assign reaction codes for APS gene
  products.
• Store all relationships in BuchneraBASE.

www.buchnera.org
13
Construction of the metabolic network (2)
Removing isolated reactions During the reductive
evolution, some pathways are in the process of
being lost and still have remnants left. APS has
a number of isolated enzymes, e.g. SerC, which
were included in the original mapping.
14
Construction of the metabolic network (3)
Missing reactions Some pathways for EAAs are
not complete from the in silico reconstructions,
but are known to function in vivo. Infer
promiscuous enzymes to fill these gaps.
Full gene complement for synthesis of 4 EAAs
(histidine, tryptophan, threonine and lysine), as
well as the non-essentials arginine, cysteine and
glycine. Also very short of transporters. Many
have been inferred.
15
Carbon-skeleton based manual visualisation of
iGT196
196 gene products 240 compounds (39 of
iJR904) 263 reactions (27 of iJR904) 35 of
reactions for EAA biosynthesis.
Key Red hexagon high flux precursor Blue square
- EAA Red circle low flux precursor Blue
circle biomass component Grey triangle
inferred reaction
Thomas et al., (2009) BMC Systems Biology 324.
16
Understanding the symbiosis

• Analysis of the reconstruction using
  constraint-based modelling (flux balance analysis)

17
Flux balance analysis
The steady-state assumption states that for each
metabolite the sum of the fluxes producing that
metabolite is equal to the sum of the fluxes
consuming that metabolite.
18
Running FBA with the Buchnera model
FBA is essentially an optimisation problem solved
using linear programming, i.e. it finds the
optimal route of fluxes through the network to
get the desired output (the objective
function). Used reduced version of the biomass
reaction of E. coli as the objective function.
(0000050) ACP (0000050) fmnh2 (0000050)
pnto-R (0000050) gthrd (0000050) thmpp
(0000050) sheme (0000050) btn (0000050) hemeO
(0203000) gtp (0126000) ctp (0136000) utp
(0027600) murein5p5pp (0025400) dctp
(0176000) phe-L (0241000) thr-L (0054000)
trp-L (0326000) lys-L (0146000) met-L
(0428000) leu-L (0090000) his-L (0276000)
ile-L (0281000) arg-L (0087000) cys-L
(0582000) gly (0402000) val-L (0007000) spmd
(0000010) fad (0025400) dgtp (0024700) dttp
(0024700) datp (45731800) atp (0001000)
amp (0002150) nad (0000050) nadh (0000130)
nadp (0000400) nadph (0000006) coa
(0000050) accoa (0000003) succoa (0730200)
pi (0730200) ppi To make this a symbiotic
model the biomass reaction modified to include
the exported EAA component. An estimate of EAA
export was obtained empirically using the pea
aphid-Buchnera symbiosis reared on chemically
defined diets. It varied among the amino acids,
from 22 (histidine and tryptophan) to 50
(threonine) of the amount synthesised. Model
required significant tweaking to get it to
function had to build iteratively.
19
The Buchnera metabolic network is highly
constrained
The primary constraint that is used to assess the
output of the network are the uptake fluxes - 5
main precursors used by the network.
Optimal growth flux (5.21) is unusually only
reached by essentially a single solution, i.e.
a single distribution of internal fluxes, as
judged by flux variability analysis.

• A number of dead end metabolites are produced
  in significant quantities in the 5.21 model by
  Buchnera
• Adenine by product of spermidine biosynthesis
• 5-methylribose by product of spermidine
  biosynthesis
• S-ribosylhomocysteine by product of cysteine
  biosynthesis
• Succinate by product of lysine biosynthesis

20
The Buchnera metabolic network is fragile

• Dogma is that biological networks have evolved to
  be robust, i.e. tolerant of some disruption.
• Deleted individual genes in silico and measured
  resulting growth flux.
• E. coli iJR904 serves as a control.
• Deletion of 84 of the genes in iGT196 led to a
  gt99 decrease in growth flux (19 for iJR904).
• Similar results using linear minimisation of
  metabolic adjustment (linearMOMA).

21
Can the aphid control the metabolic function of
its symbiont?
Designed a qualitative experiment to demonstrate
the principle that the aphid could manipulate the
metabolic output of the bacterium just by
changing the inputs.
If this network is always on then this is a
plausible way to control the symbiosis.
Possibly at the level of transporter activity
in the bacteriocyte membrane.