Integrated Human-Virus Metabolic Modelling Predicts Host-Based Antiviral Targets Against Chikungunya, Dengue and Zika Viruses

Current and reoccurring viral epidemic outbreaks such as those caused by Zika virus illustrate the need for rapid development of antivirals. Such development would be immensely facilitated by computational approaches that can provide experimentally testable predictions for possible antiviral strategies. A key factor that has not been considered fully to date in the study of antiviral targets is the high dependence of viruses to their host metabolism for reproduction. Here, we focus on this dependence and develop a stoichiometric, genome-scale metabolic model that integrates human macrophage cell metabolism with the biochemical demands arising from virus production. Focusing this approach to currently epidemic viruses Chikungunya, Dengue and Zika, we find that each virus causes specific alterations in the host metabolic flux towards fulfilling their individual biochemical demands as predicted by their genome and capsid structure. Subsequent analysis of this integrated model allows us to predict a set of host reactions, which when constrained can inhibit virus production. We show that this prediction recovers most of the known targets of existing antiviral drugs, while highlighting a set of hitherto unexplored reactions with either broad or virus specific antiviral potential. Thus, this computational approach allows rapid generation of experimentally testable hypotheses for novel antiviral targets within a host. SIGNIFICANCE STATEMENT A key challenge in combatting any new and emerging virus outbreaks is rapid drug development. In particular, generation of experimentally testable hypotheses through computational approaches is mostly lacking. Here, we address this gap by developing host-virus metabolic models for three viruses that cause current (or previously) epidemic viral outbreaks. We develop viral biomass functions using information from their genomes and physical structure, and incorporate these within a genome-scale metabolic model of human macrophage cells. The resulting integrated model allows us to predict host reactions, which when blocked, stop the system from attaining optimal viral production. These predictions recover currently known antiviral targets within human cells, and highlight a set of new reactions that are hitherto not explored for antiviral capacity.


INTRODUCTION
Rapid development of antiviral drugs for emerging and re-emerging viruses, such as the Zika virus, remains a significant challenge (1,2).Given that virus production within a host is intertwined with host immune response and metabolism (3), it is suggested that novel development of antivirals should take into account host processes (4,5).Indeed, viruses are entirely dependent on their hosts' cellular resources for their replication.This is highlighted by observed variations in virus production levels correlating with cell-to-cell variance in growth rate and phase (6), as well as virus infection leading to changes in host metabolism (7).In particular, virus infection leads to significant metabolic alterations in the host, in some cases resulting in up to 3-fold increase in glycolysis rates (7)(8)(9) and changes in ATP production rates (6).This observation can be seen as an emergent property of the combined host-virus metabolic system and could be related to changes in host cellular demands arising from viral production (10,11).More specifically, alterations in host metabolism upon infection can be understood as either viruses actively manipulating the host system to their advantage (12), or the additional draw of metabolic components for viral production simply resulting in a re-arrangement of host metabolic fluxes.
Regardless of its cause, the entanglement between host metabolism and viral production opens up the possibility to perturb the former, as a way of limiting the latter (9,12,13).To explore this possibility and towards understanding the potential interplay between host metabolism and the additional 'virus demand' on it, stoichiometric genome scale metabolic models and their optimisation through flux balance analysis (FBA) can provide ideal starting points as they are demonstrated to allow analysis of cellular physiology as an interconnected system (14,15).Integration of virus production in a host metabolic model has already been utilised to study the infection of bacteria with phage, indicating the presence of metabolic limitations on phage replication depending on host's metabolic environment (16).While this type of stoichiometric metabolic analysis can potentially be applied to any host-virus pair, it is particularly suited to Alpha-and Flavi-viruses.The rather simple physical and genomic structure of these viruses (17,18) allow straightforward construction of a pseudo biochemical reaction representing their production from constituting parts.This pseudo reaction can then subsequently be incorporated into a genome-scale metabolic model of any host.
Here, we develop and apply such an FBA approach to analyse host-virus metabolic entanglement.Using a stoichiometric metabolic model of a human macrophage cell, we establish an integrated virus-macrophage metabolic model for three viruses causing current (or previously) epidemic outbreaks: Chikungunya virus (CHIKV); Dengue virus (DENV); and Zika virus (ZIKV).These are representatives of the virus genera Alpha-(CHIKV) and Flavi-virus (DENV, ZIKV), which are positive-sense single-strand RNA-viruses with rather simple physical structures (17,18).Viruses of both families have been observed to infect monocyte derived macrophage cell lines (19)(20)(21) and are usually transmitted to humans via arthropod vectors, the most common being mosquitos of the Aedes genus (22,23).By analysing the integrated metabolic model, we find that viral production results in significant alterations in host metabolic fluxes.Subsequent analysis of this integrated model through linear optimisation allows us to predict a set of host reactions, which when constrained can inhibit virus production within the macrophage metabolic system.We show that this prediction recovers most of the known targets of existing antiviral drugs, while highlighting a set of hitherto unexplored reactions that either limit virus activity broadly or for a specific virus.Thus, this computational approach allows rapid generation of experimentally testable hypotheses for novel antiviral targets within a host cell.

RESULTS
Host metabolism displays alternative host-and viral-optimal states.We first used the integrated virus-macrophage stoichiometric metabolic model to interrogate potential changes in host metabolism upon virus infection.To do so, we considered two idealised scenarios; one in which the metabolic system is optimised for the functional requirements of the host cell as determined by a maintenance related biomass reaction (24) (host-optimal state), and another in which the metabolic system is optimised solely for a biomass reaction that represents the production of virus particles and that is derived from viral genomes (virusoptimal state) (see Methods and Figure S1 for details of virus biomass calculations).These two states provide the theoretical extremes of a continuum of metabolic states that can arise during virus infection.Whilst the first scenario aims to represent the normal physiological state of macrophage cells, the second state represents a thought experiment of the host metabolic fluxes being set for maximizing virus production.
To compare the host-and virus-optimal states of the model, we analyse the metabolic fluxes directly feeding into the biomass pseudo reaction (see Supplementary Files S1 for biomass reactions and S2 for flux values and ranges).As expected from linear optimisation, we find that these fluxes reflect the stoichiometric differences in the amino acid and nucleotide requirements of the host cell and the virus, thus achieving perfect fulfilment of host or virus biomass requirements.We conclude that stoichiometric differences in metabolic requirements for virus production vs. host maintenance, as summarised in Figure 1, result in different metabolic flux states of the host model.
The flux variability allowed in the system varies for viral-and host optima, indicating significant physiological changes in the host metabolism to meet viral demands.To understand how the flux changes at biomass level affect the metabolic system, we calculated the allowed flux variability for individual reactions in the model using either host-or virusbased optimisation (see Methods).Flux variability analysis (FVA) allows for a more robust analysis of different states of the model, compared to simply calculating optimal flux sets, which are shown to be subject to inaccuracies inherent in linear solvers used in flux balance analysis (25).We find that the median of the allowed optimal metabolic flux ranges, between host-and virus-optimal states, shows significant changes across subprocesses (Figure 2 and Supplementary File S2).In particular, the virus-optimal state displays significantly increased median flux for reactions associated with lipid metabolism and nucleotide biosynthesis, and significantly decreased flux for reactions associated with fatty acid biosynthesis and transport (including intracellular transport reactions).Besides these general overall trends across subprocesses, the virus-optimal state displays both increased or decreased median flux for specific reactions within each subprocess (see pie charts in Figure 2).These changes are in accordance with downstream requirements for fulfilling biomass requirements, and relate to interconnections among sub-processes.For example, the reactions from the lipid metabolism subprocess that show the most increase in their median fluxes (compared to host) involve ADP/ATP and phosphor, which are metabolites that link directly into the reactions of the nucleotide biosynthesis subprocess (and feeding into increased nucleotide requirement in the virus, see Figure 1).
The specific changes in the allowed flux ranges also highlight potential physiological changes.As an illustrative example, we show the extent of changes within the glycolysis pathway, where allowed flux ranges that can sustain virus-optima are wider compared to those that can sustain host-optima (Figure 2).The allowed ranges for glucose and oxygen uptake indicate that virus-optima can be sustained even under low uptake fluxes, indicative of the potential feasibility of anaerobic metabolism still sustaining virus production (26).Taken together, this comparison of host-and virus-optimal states show that the differences within the stoichiometric requirements of the different viruses and between the host cause largescale alterations in the host metabolic fluxes.
Enforcing host-optimal flux ranges on individual reactions in the model predicts antiviral targets that can suppress viral production.As the host-optimal and virus-optimal flux ranges within the integrated model differ, we hypothesize that the model can be constrained in a way to limit viral production (see Methods).To test and utilise this hypothesis, we use the integrated stoichiometric model to identify the host reactions, which, when constrained limit virus production the most.This analysis can be implemented in different ways, for example through constraining of flux values to zero (i.e.reaction 'knockouts').Applying such knock-outs, we find several reactions that limit virus-optima, but all of these also results in significant reduction in host-optima (Supplementary File S3).To identify if there any reactions that can perturb virus production, whilst maintaining the host viability, we constrained reaction fluxes to ranges that are derived from the FVA described above.In particular, we identified flux ranges that still allowed for the attainment of the host-optimal state, but were outside of the range allowed by the virus-optimal state (see Methods).This approach highlights a set of reactions that result in different levels of reductions in the virus optima of CHIKV, DENV or ZIKV, while not affecting the host-optima (as expected from the way we set the flux constraints, see Methods).We identify 29 reactions that can reduce the virus optima to below 80% of the original value for at least one virus (Supplementary File S4).Interestingly, many of these 29 reactions are interconnected, and are involved in the de novo synthesis of RNA nucleotides (both purine and pyrimidine pathways) and in amino acid interconversions (Figure 3).Particular examples include reactions directly involved in the synthesis of adenosine, guanosine and uridine/cytidine nucleotides, and upstream reactions such as those involving inosine monophosphate (IMP) and orotidine monophosphate (OMP).
These identified reactions are potential antiviral targets, in the sense that altering their fluxes can limit virus production within the host.Thus, we explored if these reactions match with known antiviral drug targets.Performing a literature analysis, we found that there are currently 10 antivirals, specific to RNA-viruses, and these target only 5 unique metabolic enzymes (see Supplementary Table 1).Of these 5 drug targets (and the associated drugs) one has been experimentally verified to be effective against CHIKV ( Inositol-5'-monophosphate dehydrogenase; IMPD) (27); and another against DENV (Dihydroorotate Dehydrogenase; DHORD9) (28).Whilst the other three targets have been verified to be effective against a number of RNA viruses (29), they are yet to be tested against CHIKV, DENV and ZIKV.
We found that out of these 5 known antiviral targets, all are implicated in our analysis.The three known antiviral target reactions involving the genes IMPD (27); DHORD9 (28); and Orotidine-5'-phosphate Decarboxylase (OMPDC) (29) are found to perturb virus optima for all viruses (Figure 3).The antiviral target S-adenosylhomocysteine hydrolase (AHC) ( 29) is predicted to effect only CHIKV optima, and only to a level higher than the 80% cut-off we used in the above analysis (we note that setting AHC reaction flux to zero abolishes virus growth for all three viruses (see Supplementary File S3)).Finally, CTP synthase, which has been indicated to exhibit an effect on several RNA viruses (29), is included in the model as two reactions which perform the same reaction utilising either ammonia (mediated by CTPS1) or glutamine (mediated by CTPS2) as a nitrogen source (30) and therefore not highlighted in our initial flux enforcement analysis focusing on single reactions.When we constrain both reactions associated with these two reactions simultaneously at host-derived flux ranges, a reduction in all virus optima is observed.
Identified flux enforcement effects are significant and arise from differences in hostand virus-stoichiometric requirements.The above results provide support for using integrated host-virus metabolic models for identifying host-based antiviral targets.However, any analysis based on stoichiometric metabolic flux optimisation, as done here, can be dependent on details of model implementation and assumptions (25).For example, while the host metabolic model and the biomass function that it incorporates are verified against experimental data (24), the model still assumes a specific media composition and uptake fluxes.To test if our predictions are robust against key assumptions in the model, we analysed the effects of variations in the media composition for the host model and the genomic sequence of the virus on the antiviral target prediction.In particular, we repeated the above analysis for 1000 alternative media uptake fluxes and 1000 point mutations for each of the three virus genomes, affecting the final virus biomass (see Methods).We found that the list of reactions with highest impact on virus biomass, while maintaining the host biomass, are qualitatively not altered with these changes in the model structure and we still recover known broad antiviral targets (Supplementary File S5 and Figures S2-4).
To further probe the significance of these reactions as effectors of virus production, we generated 6000 randomised virus biomass compositions using the three original virus biomass functions as a starting point (see Methods).Repeating the enforcement analysis for this randomized set thus allowed us to generate a distribution of effects of reactions on viruslike particle production in the host system, thus acting as a null hypothesis.Results from the enforcement of the original unaltered virus sequences, as well as the point mutated virus sequences (mentioned above), were compared against the population of 'randomised' viruses to assess the significance of the antiviral effect (see Methods).This allows derivation of a significance value for the effects resulting from each reaction, when the flux enforcement analysis is applied on it.When we rank reactions according to the significance of their effects, we find that the list of reactions shown in Supplementary File S4 are ranked among the top (Supplementary Files S5 and S6), i.e. these reactions cause biomass reduction for each virus that is statistically significant when compared to their effects on randomised virus-like biomass functions.There are a couple of exceptions only in the case of ZIKV, where the reactions mediated by cystathionine g-lyase (CYSTGL) and cystathionine beta-synthase (CYSTS) did not show any significance in their effect under flux enforcement.Additional statistical analysis showed that most of the reactions listed in Supplementary File S4 (27 out of 29) also showed significant differences in the magnitude of their effects among the three different viruses.In other words, whilst the reactions we highlight are not necessarily unique when comparing amongst CHIKV, DENV and ZIKV, their quantitative effects on virus production is significantly different for each individual species.This, combined with the fact that our randomisation process maintained the key features of stoichiometric differences among the host and virus-like biomass functions, highlight that the flux-perturbing effects of the identified reactions emerge from the core metabolic stoichiometric differences between host and the viruses.In particular, the fact that viruses use much higher levels of nucleic acids per biomass unit (Figure 1).

Many of the predicted additional host reactions effecting virus production can be
targeted by existing drugs.Considering the computationally predicted potential of the additional reactions identified as antiviral targets, we have searched for these reactions in a database of known inhibitor-like molecules (31).We found that 15 of these reactions already have known molecules, and in some cases existing drugs, targeting their catalysing enzymes (Figure 3, and full list in Supplementary File S4).These findings present experimentally testable predictions on host reactions, the disruption of which could limit virus production.It must be noted, however, that our computational analysis identifies flux enforcement based on differences in host-and virus-optimal states of the model, where 'enforcement' can mean either reduction or increase in a given flux.In contrast, most of the currently known molecules act as enzyme inhibitors (31) and would be expected to reduce metabolic fluxes.

DISCUSSION
We present a computational approach that combines application of FBA and FVA with development of integrated host-virus metabolic models.We show that this novel approach recovers the known metabolic antiviral targets within a human macrophage cell and predicts new potential targets.These predicted reactions fall primarily onto pathways involving nucleotides and amino acids that are differentially used by the host and virus.The results of this study are in line with an integrated perspective that views the virus as an additional metabolic burden on the host cells that could be met or avoided by tinkering of host metabolic fluxes.The observed overlap between predicted reactions and known antiviral drugs gives confidence to this integrated modelling approach and highlights its potential as a rapid prediction tool to guide experimental design.This can be especially useful in the case of new and emerging viruses for which limited clinical and experimental data may be available to inform drug target identification.
The integrated stoichiometric metabolic modelling approach focuses on metabolic changes as a driver of virus production, and does not consider factors associated with virus-host cell recognition, viral entry and release (32).Furthermore, the application of the linear optimisation on stoichiometric models (i.e.FBA and FVA) strictly assumes that host metabolism is at steady state, and thus prohibits analysis of the dynamics of cellular physiology.Such dynamics could be taken into account to a certain extent by imposing different flux constraints, which could be derived from proximal experimental data (16), through development of simplified metabolic temporal models (10,11), or by combining dynamics with linear optimisation on stoichiometric models (33,34).Additionally, the extent of the missing information in genome-scale stoichiometric models creates limitations on how much of the metabolic processes can be covered (35).
The future efforts into model curation and standardisation (36) would open up the possibility of extensive analysis of host-virus pairings from a metabolic stance.The presented findings already suggest that targeting host metabolic processes that are linked to host-virus compositional mismatches can be used to combat virus production without altering host functions.In particular, analysis of extended flux enforcement strategies such as flux limitations on double and triple reaction combinations might identify virus specific drug combinations.Combined with the future development of additional host-virus integrated models covering many cell and virus types can thus allow a fruitful route to computational guiding of experimental antiviral drug discovery.

Generation of virus biomass objective functions.
To implement the FBA approach to studying virus infections from a metabolic stance, we define a pseudo reaction accounting for the production of virus particles from its constituents.We call this reaction a virus biomass objective function (VBOF).To account for metabolic fluxes associated with the virus production, the VBOF needs to capture the stoichiometry of nucleotide, amino acid and associated energy metabolites relating to virus production, similar to biomass production function used for microbial metabolic models (37).We derive the metabolic stoichiometry of virus production from the viral genome sequence, the subsequently encoded proteins, the copy number of those proteins, and knowledge of the energetic requirements for peptide bonds and phosphodiester bonds.Details of this derivation is given below, while a schematic of VBOF generation is included as Figure S1.

Genome and protein information for the viruses. The genome sequences used in the
present study are obtained from the NCBI genomic database (38) using the following accession numbers and accessed in March 2016; Zika; NC_012532.1,Dengue; NC_001474.2,and CHIKV; NC_004162.2(original files are provided as Supplementary File S7).Viruses can be classified by their replication methods, known as the Baltimore Classification System (39), and depending on this classification, a viral particle may contain more than a single copy of the genome.This is factored into the calculation of the nucleotide counts.In the presented study, all studied viruses fall into Group IV classification: they replicate their positive single-stranded RNA (+ssRNA) genome via a negative ssRNA (-ssRNA) intermediate.Therefore, the counts of the nucleotides in the negative strand is equal to the counts of the complementary nucleotide in the positive strand, i.e. count of A on (+/-) strand = count of U on (-/+) strand, and similarly for G and C counts.The count for each RNA nucleotide (Adenosine (A), Cytidine (C), Guanine (G) and Uracil (U)) can be taken directly from the virus genome sequence: RNA utilises Uracil (U) in place of Thymine (T), therefore T must be replaced with U from the genome sequence readout.In this study, all the viruses have two categories of polyproteins that compose the proteome: structural and nonstructural.The amino acid sequence of these two polyproteins, and indeed any genome derived protein sequences, are obtained from gene annotations of the viral genomes as provided in the NCBI genome entries (see above for NCBI entries used).The different subcategories of the viral proteome may be differentially incorporated into a single virus particle.
For the viruses studied here, the structural and non-structural polyproteins are expressed in a ratio that is derived from the overall virus structure (i.e.proteins in the capsid, nucleocapsid, etc.) (18).The ratio is 1:240 for CHIKV (18), and 1:180 for DENV/ZIKV (17).More broadly, the ratio of different protein classes in a single virus particle can be derived from the overall virus structure or directly from literature / experimental evidence.

Calculating nucleotide investment per virus.
The total moles of each nucleotide in a mole of virus particle ( " #$# ) are obtained from their count in the virus genome ( " % ) and replication intermediates ( " & ), and multiplied by the genome copy number (C g ): where the indexation is over nucleotides.The moles of nucleotides are then converted into grams of nucleotide per mole of virus (g NTPS mol -1 virus ;  " , ), by multiplying  " #$# with the respective molar mass (g mol -1 ) of the nucleotides ( " , ): where the indexation is over nucleotides.Summing  " , over all nucleotides and combining this with the similar calculation for amino acids allows us to get the total molar weight of the virus in terms of nucleotides and amino acids (M v , see Equation 15below).Finally, the stoichiometric coefficients of each nucleotide in the VBOF are expressed as millimoles per gram of virus (mmol NTPS g -1 virus ;  " , ): where the indexation is over nucleotides.

Calculating amino acid investment per virus.
The total moles of each amino acid per mole of virus particle ( 3 #$# ) is obtained similarly using the sequence information of structural ( 3 45 ) and non-structural ( 3 ,5 ) proteins.Counts of each amino acid in these proteins is multiplied by the respective copy numbers of these proteins (C sp and C np ): where the indexation is over amino acids.C np is 1 for all viruses studied here, while C sp is 240 for CHIKV (18), and 180 for DENV/ZIKV (17).The moles of amino acids per mole of virus is then converted into grams of amino acid per mole of virus (g AA mol -1 virus ;  3 9 ), by multiplying  3 #$# with the respective molar mass (g mol -1 ) of each amino acid (M X ): where the indexation is over amino acids.Finally, the stoichiometries of each amino acid in the VBOF is expressed as millimoles per gram of virus (mmol AA g -1 virus ;  3 9 ): where the indexation is over amino acids.

Calculating ATP requirement for amino acid polymerisation (mmol g -1 virus
).The polymerisation of amino acid monomers requires approximately 4 ATP molecules per peptide bond (40), defined here as the constant k ATP ( = 4) The overall moles of ATP (A TOT ) required to form the structural (A SP ) and non-structural (A NP ) polyproteins are calculated from the respective amino acid counts: #$# =  67  45 +  87  ,5 where the indexation is over amino acids.From A TOT , we calculate the stoichiometry of ATP in the VBOF as millimoles per gram of virus (S ATP ): As ATP is hydrolysed in this process, the water requirement for polymerisation  ?@ $ is equal to that of ATP.The products from the hydrolysis of ATP (ADP, P i and H + ) are also accounted for in the VBOF (see Equation 16).

Calculating pyrophosphate (PP i ) liberation from nucleotide polymerisation (mmol g -1 virus
).The polymerisation of nucleotide monomers to form the RNA viral genome liberates a PP i molecule (40), defined here as the constant k PPi ( = 1).The overall moles of PP i (P TOT ) required to form the viral genome (P G ) and replication intermediates (P R ) are calculated from the respective nucleotide counts: To convert this into the PP i stoichiometry in the VBOF as millimoles per gram of virus (S PPi ), we again use the overall molar mass (g mol -1 ) of one mole of virus: Calculating total Viral Molar Mass.The total molar mass of the virus M v is calculated from the total mass of the genome and proteome components as: Final Construction of the VBOF.The left-and right-hand terms of VBOF are based on the above calculations of stoichiometric coefficients.The final stoichiometry for the VBOF (pseudo-reaction) is: " , + ⋯ +  3 9 + ⋯ +  =#5 +  ?@ $ →  =D5 +  5 E +  ?F +  55" (16) This pseudo-reaction accounts for the virus' biomass, and the energy requirements associated with its production, and can be incorporated into stoichiometric metabolic models of the host to represent the presence of a virus in that system.
Integration of iAB-AMØ-1410 and Chikungunya, Dengue and Zika Viruses.The VBOFs for the three viruses (CHIKV, DENV, and ZIKV) were integrated into three-separate instances of the 'host' macrophage model (iAB-AMØ-1410) (original files are provided as Supplementary File S7).In each individual case, the respective VBOF was appended into the existing macrophage model, with a lower flux bound of zero and an upper bound of infinity, reflecting no upper constraints on this flux (41).No other metabolites or reactions were added to any of the models.All of the individual flux bounds of the model reactions were used as previously set ( 24), but any bounds set to -1000 or 1000 are replaced with infinity, since the use of infinity, rather than arbitrarily large values, is shown to be a more robust approach to represent unbounded reactions in a linear programming model (41).We also confirmed that the use of arbitrary large bounds (such as -1000/1000) instead of infinity does not change the presented results qualitatively.A set of subprocesses, derived from known aggregatesubsystems (42), were appended as metadata to each individual host-virus model and linked with the pre-existing defined subsystems.A full description of the subsystems and mapping of reactions into these are supplied in Supplementary File S3.The used integrated model is provided in a computer readable (SBML) format with the publication.
Characterising the stoichiometric differences between host and virus.For both the host (iAB-AMØ-1410) and viruses (CHIKV, DENV and ZIKV) we have pseudo-reactions that capture the metabolic requirements for the maintenance/production of their respective biomass.By comparing these pseudo-reaction stoichiometries, we can quantify the differences in amino acid and nucleotide requirements to fulfil the host or virus objectives.
To do so, we calculate the fold change in nucleotide and amino acid usage by normalising their individual stoichiometric coefficients against the sum of stoichiometries of all metabolites present in the objective function (other than ATP): where indexation i is over nucleotides (or amino acids) and k is over all biomass precursors, and the subscript H and V indicate the use of the host and virus biomass functions respectively.A positive value indicates a higher usage for nucleotide (or amino acid) i by the virus than the host, whilst a negative value indicates a lower usage.

Comparison of host-and virus-optimised states.
For all analyses, the generated host-virus integrated models were optimised and reaction fluxes predicted using the linear optimisation approach known as flux balance analysis (FBA) (43).Linear optimisation is a mathematical technique that optimises a given function under a set of constraints defined by mathematical inequalities.In the context of metabolic models, the constraints correspond to limitations on reaction fluxes, while the function to be optimised can be defined as the flux in a specific reaction.While several biologically plausible objective functions can be defined (44), a common approach is to define a pseudo reaction that describes biomass production from its constituent parts, and then optimise the flux to this reaction, as we have done here.Since the set of constraints includes constraints on uptake reactions, this application of FBA results in prediction of optimal biomass production flux with respect to a specific uptake flux.In other words, FBA optimises for biomass yield from given substrates assumed to be present in the media.In this work, we apply FBA to optimise a combined host-virus metabolic system satisfy either the host or virus objective function (as described above) and study the resulting flux predictions.
To simulate a virus-optimal state, the models are optimised using the respective VBOFs of CHIKV, DENV and ZIKV viruses as the objective function, while to simulate a host-optimal state the models are optimised using the existing biomass maintenance reaction for the human macrophage as presented in (24).Besides running linear optimisation to find the optimal flux sets under each scenario, we have also performed a flux variability analysis (FVA) (45), which provides flux ranges for each reaction that still would allow attainment of a given host/virus optima.The FVA approach is shown to be more robust to instabilities associated with prediction and comparison of a single optimal flux sets (41).For each reaction in the model we compared the resulting flux ranges from FVA under host and virus optimisation, by evaluating the mean value of the allowed flux range for each individual reaction (A i ) and then collating the mean flux values for reactions associated with given subprocesses (aggregated subsystems) as a percentage of total flux through that process.
More formally, we define the differential distribution of reaction flux for each subprocess (i) between the host and virus optimised models in terms of a fold-change: where the indexation k is over all reactions of the model, while the indexation m is over Enforcement.Host-derived enforcement considers the effect of maintaining a metabolic system in a host-optimised state, whilst attempting to optimise the model for VBOF.For this approach, we systematically set individual lower and upper flux bounds of individual reactions to a specific flux range.For each reaction, this range (e r ) is derived from the corresponding minimum (F -) and maximum (F + ) flux values for that reaction obtained from VBOFs that were either derived from the original VBOFs or were completely randomly generated.To evaluate impact of small deviations from the original VBOFs, we generated variants of the original virus genomes through nucleotide substitution; for each virus we generated 1000 genome variants, where each variant contained 1, 2, 3, 4, 5 or 10 nucleotide substitutions.For the subsequent VBOF generation from these variant genomes, the genome and protein copy numbers were kept as in the original.To evaluate more variant VBOFs, we generated another 1000 genomes for each virus that were created from the original genome with a random number (between 0 and the total length of the genome) of nucleotide substitutions, and using randomly drawn structural and non-structural polyprotein copy numbers per virus particle.Finally, and in an attempt to generate a set of VBOFs that are far removed from the original ones, both in terms of genome sequence and the structural and non-structural protein numbers, we directly generated random VBOFs.We have implemented this by drawing 1000 sets of individual stoichiometries of biomass components from a uniform distribution on [a, b], where a and b are 1) ±99% of the original stoichiometric coefficients of a given virus, or 2) are ±99% of the average of all original stoichiometric coefficients of a given virus.These approaches to generating variant virus genomes give us set of sequences (and associated VBOFs) that are increasingly removed from the original virus VBOFs.For each randomised VBOF created, the host-derived flux enforcement analysis was repeated (with a recalculation of the bounds used for the individual enforcements) and the reactions that perturb virus optima the most when constrained were identified.This whole analysis resulted in 8000 randomised VBOFs and FBA simulations, the results of which are summarised as percentage impact of individual host reactions on virus optima for different sets of VBOFs (see Supplementary File S5).To compare results of flux enforcement analysis to that obtained from using randomised biomass functions, we used a one-way ANOVA and Tukey's honest significance tests for each individual virus against RNA nucleotides (the x-axes are labelled with the common short notations for these).All calculations and biomass formulations are as described in the Methods, and all biomass stoichiometric values are provided as Supplementary File S1.Abbreviations used for the compounds and reactions are as in Supplementary Table S2 and S3 respectively.Some of the identified reactions are interconnected, forming pathways.

Figures and
Pathways associated with subsystems are indicated by coloured reaction arrows: Orange, pyrimidine synthesis; Purple, pentose phosphate pathway; Blue, purine synthesis; Green, nucleotide biosynthesis.The starting metabolites into these pathways, Glutamine and D-Ribose 5-phosphate, are derived from glutamine biosynthesis and pentose phosphate pathways.Reactions targeted by a known antiviral or inhibitor are marked by white and red filled stars and circles, respectively.Complete list of antiviral compounds from which the matches were obtained are provided as Supplementary Table S1.Complete list of inhibitors and the associated reactions are provided as Supplementary File S4.A complete list of enforcement results for all reactions are provided as Supplementary Files S2 and S5. .
reactions that belong to subprocess i.The superscript indicates the use of flux values from host (H) and virus (V) optimised models respectively.A positive value indicates a higher mean flux in subprocess i in the virus-vs.host-optimised model, whilst a negative value indicates a lower mean flux.Reaction knockout and host-derived flux analyses.To find reactions which can preferentially alter virus optimised state of the model, we considered the effect of systematically constraining individual reactions.Knockout analysis.Knockout analysis considers the effect of systematically setting individual reaction fluxes to zero, and then attempting to maximise for VBOF.The knockout optima for the virus production reaction flux  LO is then compared to the original flux over this reaction;  PQ .Host-Derived

Figure 1 |
Figure 1 | Fold change difference in usage of amino acids and nucleotides between host

Figure 2 |
Figure 2 | Comparison of model fluxes between host optima and CHIKV, DENV and

Figure 3 |
Figure 3 | Reaction pathway schematic showing the top 29 reactions from host-derived

Figure S2 |
Figure S2 | Host-derived enforcement reactions analysed over varying model media

Figure S3 |
Figure S3 | Host-derived enforcement reactions analysed over varying model media

Figure S4 |
Figure S4 | Host-derived enforcement reactions analysed over varying model media