Engineering Living Systems as Programmable Machines
Synthetic biology is the discipline of deliberately redesigning biological systems — not just editing individual genes, but rewiring the metabolic logic of an organism to do something new. Where classical genetic engineering asks "can I insert this gene?", synthetic biology asks "can I redesign this organism's entire metabolic routing so that carbon flows toward the compound I want instead of the ones I don't?" The distinction matters because it determines both the ambition of what's possible and the complexity of what's actually required to make it work.
For microalgae, synthetic biology is particularly compelling in theory. Algae are solar-powered biochemical factories: they take CO₂, water, and sunlight as inputs and can, in principle, produce virtually any carbon-containing molecule if you redirect their internal biochemistry correctly. The photosynthetic machinery — which converts light energy into chemical energy — is already there. The question is whether you can redirect that chemical energy away from biomass growth and toward your target compound.
In practice, this requires understanding how metabolic pathways are regulated — what genes are turned on when, what signals trigger accumulation of a compound, and critically, where carbon gets lost to competing pathways that you don't want. The last point is where most synthetic biology projects discover that the map is not the territory: a pathway that looks straightforward on a whiteboard turns out to be entangled with a dozen other cellular processes, and "simply" redirecting it has consequences that weren't in the model.
This module covers: the conceptual framework (Design-Build-Test-Learn), the main engineering strategies (promoter engineering, pathway overexpression, competitive pathway knockouts, heterologous expression), the key quantitative metrics you need to evaluate claims, the gap between published results and industrial reality — and, explicitly, how to read a synthetic biology paper critically. The last point is the most practical skill this module teaches.
How Synthetic Biology Projects Actually Proceed
The Design-Build-Test-Learn (DBTL) cycle is the operating framework of synthetic biology, borrowed from engineering. It replaces the traditional research model of hypothesis-experiment-publication with an iterative engineering loop that is explicitly designed to improve strains cycle by cycle. Understanding the cycle is essential for evaluating claims in the literature — and for understanding why progress is slower than press releases suggest.
What Synthetic Biologists Actually Do to a Cell
Four strategies dominate metabolic engineering of microalgae: promoter engineering to control gene expression levels, overexpression of biosynthetic pathway genes to increase flux toward the target compound, knockout of competing pathways to redirect carbon, and heterologous expression of genes from other organisms to enable entirely new biosynthetic routes. Each has different technical requirements and different types of published evidence supporting its use.
A promoter is the DNA sequence upstream of a gene that determines when the gene is expressed and at what level. Engineering a promoter means swapping a gene's natural promoter for a stronger one (constitutive overexpression), an inducible one (expression triggered by a specific signal like nitrogen starvation or light intensity), or a synthetic hybrid. In algae, promoter options are limited compared to bacteria — the library of well-characterised algal promoters is small. In Chlamydomonas, the HSP70A–RBCS2 promoter tandem is the workhorse for nuclear overexpression. In Nannochloropsis, the violaxanthin-chlorophyll a-binding protein promoter (VCP) has been used. In the chloroplast of Chlamydomonas, constitutive promoters (Patpβ, Prbcl) drive very high expression. Promoter engineering is usually the first lever pulled and the safest — it does not delete anything.
If astaxanthin is made via β-carotene via lycopene via geranylgeranyl diphosphate (GGPP) from the MEP pathway, overexpressing the rate-limiting enzyme in that chain increases flux toward astaxanthin. The most successful published example in algae: Couso et al. (2012) overexpressed phytoene synthase (PSY) in Chlamydomonas under a nitrogen-starvation-inducible promoter, achieving a 3× increase in total carotenoid content. The challenge is identifying which enzyme is actually rate-limiting — this requires either flux analysis or empirical testing of multiple candidates. Overexpressing the wrong enzyme does nothing. Overexpressing a regulatory enzyme (one that controls not just its own product but other branches) can have unintended consequences throughout the pathway.
If starch synthesis and lipid synthesis compete for the same Acetyl-CoA pool, knocking out the key starch biosynthesis enzyme (ADP-glucose pyrophosphorylase, AGPase) forces the cell to store carbon as lipid instead. This is the logic behind the Ajjawi et al. (2017) result in Nannochloropsis: they disrupted a transcription factor (ZnCys) that represses lipid accumulation, achieving 2–3× higher TAG content. The risk: knocking out a competing pathway often reduces growth rate because the cell has lost a metabolic buffer that normally helps manage carbon overflow. A strain that accumulates 3× more lipid but grows 50% slower may have identical or worse volumetric productivity. Growth rate must always be measured alongside titre in any knockout study.
Some compounds of commercial interest are not naturally produced by the most tractable algal chassis. The solution: clone the biosynthetic genes from an organism that does produce the compound and express them in the chosen chassis. Chlamydomonas has been used as a platform for expressing recombinant proteins from mammals (vaccine antigens, growth factors) and for producing terpene compounds whose biosynthetic genes come from plants or bacteria. The Triton Algae Innovations platform (San Diego) produces animal-free heme protein using Chlamydomonas as a chassis. The challenge is codon optimisation (the DNA sequence must be adjusted to match the host organism's codon usage preferences), protein folding (a protein that folds correctly in a bacterium may not fold correctly in a chloroplast), and cofactor availability (some enzymes require cofactors that the host organism may not produce in sufficient quantity).
The Skill That Distinguishes Depth from Survey
You will encounter synthetic biology papers claiming dramatic yield improvements in algae — 5× more astaxanthin, 3× more DHA, 10× higher protein content. Most of these results are real in the narrow sense that the experiment produced those numbers. Most of them are also not what they appear to be when you read past the abstract. This section gives you the specific questions to ask.
A paper reports "10× improvement in astaxanthin content" in Chlamydomonas reinhardtii. What the abstract doesn't mention: Chlamydomonas does not naturally produce astaxanthin at commercial levels; the comparison is to Chlamydomonas wild-type, not to Haematococcus commercial strains; the experiment was run in 50 mL flasks; the engineered strain grew 60% slower than wild-type; and no scale-up data was presented. The result is real and scientifically interesting. It is not a competitive threat to Cyanotech's Haematococcus production facilities. Reading the full paper instead of the abstract is the difference between understanding this and misunderstanding it.
| Metric | Definition | Why It Matters | What to Check |
|---|---|---|---|
| Titre | mg/L of target compound in culture broth at harvest | Determines downstream processing cost — higher titre means less volume to process per unit product | Was this measured at peak titre or end of batch? Was culture density normalised? |
| Yield | g product / g biomass (or g substrate consumed) | Measures metabolic efficiency — what fraction of carbon input ends up in the target product | Compare to theoretical maximum yield from the pathway stoichiometry; check if growth rate was affected |
| Volumetric productivity | mg/L/day of target compound | Determines fermenter/PBR size and capital cost; the most commercially relevant single metric | Always calculate this even if the paper doesn't report it: titre ÷ batch time |
| Specific productivity | mg product / g biomass / day | Normalises productivity to biomass level — useful for comparing strains at different cell densities | Check if volumetric productivity is high only because cell density is high, not because productivity per cell improved |
| Growth rate (μ) | Doublings per day; or specific growth rate h⁻¹ | Must be reported alongside titre — a high-titre slow-growing strain may be commercially worse than a lower-titre fast-growing one | Any result that shows high titre without reporting growth rate is incomplete; ask why it was omitted |
| Genetic stability | Trait retention across generations without selection pressure | Commercial strains must maintain their phenotype over hundreds of generations; laboratory strains often revert | Was the strain tested beyond 30 generations? Were transformants maintained on antibiotic selection (which masks instability)? |
Who to Follow and What the Literature Is Actually Telling You
Most synthetic biology research in algae happens in academic laboratories. Knowing which groups are producing work most likely to translate to commercial relevance is the skill that distinguishes informed monitoring of the field from reading paper titles. Below are the groups and companies whose work is most worth tracking in the context of commercial algae production for food, nutraceutical, and omega-3 applications.
Ajjawi et al. (2017, Nature Biotechnology) — the landmark Nannochloropsis CRISPR paper demonstrating 2–3× lipid increase via ZnCys transcription factor disruption. This is the most-cited result in algae metabolic engineering and remains the clearest demonstration that CRISPR can produce commercially meaningful yield improvement in a production strain. The gap: lipid titre and growth rate tradeoff was noted but not resolved. The Melis Lab at Berkeley has also produced important work on photosynthetic efficiency improvement (truncated light-harvesting antenna for higher biomass productivity), with pilot-scale data.
Stephen Mayfield's group is the leading academic-to-commercial translation engine for Chlamydomonas chloroplast engineering. Triton Algae Innovations (spinout) produces recombinant heme protein and animal-free food ingredients using engineered Chlamydomonas. The chloroplast transformation platform achieves very high expression of heterologous proteins — demonstrating commercial feasibility of algae as a recombinant protein platform. Most relevant for SustaBloom's watch list if recombinant production platforms become relevant in Phase 3 or 4 strategy.
The most directly India-relevant research group for SustaBloom's context. CSMCRI (now part of CSIR) has decades of work on Spirulina, Dunaliella, and Gracilaria cultivation under Indian coastal conditions. Balasubramanian's group has published on phycocyanin extraction optimisation, Spirulina productivity under Indian light conditions, and basic characterisation of marine microalgae from the Gujarat coast. Not primarily synthetic biology — more classical cultivation science — but the most relevant source of India-specific production benchmarks. Publications in Bioresource Technology and Algal Research.
Metabolic flux analysis and carbon partitioning in photosynthetic organisms. Not algae-specific, but producing the underlying metabolic modelling tools — flux balance analysis, isotope labelling experiments — that are being applied to algae by other groups. Their work on carbon allocation between starch, lipid, and protein pools in C3 plants directly informs how the same tradeoffs are being approached in microalgae. Worth reading one review paper per year from this group to stay current on the flux analysis methodology being applied to algae metabolic engineering.
Algenol uses genetically engineered Spirulina-like cyanobacteria (Synechococcus) for ethanol production — one of the few examples of a transgenic photosynthetic organism operated at pilot scale outdoors. LanzaTech uses engineered acetogenic bacteria (not algae) for gas fermentation but applies many of the same metabolic engineering principles. These cases are worth studying not for their algae genetics specifically, but for what they reveal about the operational reality of running engineered photosynthetic organisms outdoors: contamination management, genetic stability over operational time, and the delta between lab and outdoor productivity.
Five Structural Limitations That the Literature Understates
The metabolic burden problem
Every gene you overexpress in a cell consumes cellular resources — ribosomes, energy, precursors. When a cell is forced to maintain high expression of a foreign protein or a heterologous pathway, it diverts resources from growth. This manifests as reduced growth rate and often reduced genetic stability over time. In algae, which are already metabolically stressed by light-dark cycles and nutrient fluctuations, additional metabolic burden can reduce outdoor productivity significantly even when indoor lab results look positive. Most published synthetic biology papers are run under optimal, stable conditions that minimise this burden; outdoor production exposes it.
Position effects in nuclear transformation
When a transgene integrates into the nuclear genome, its expression level depends on where it lands — near transcriptionally active chromatin (high expression) or silenced heterochromatin (low or no expression). This position effect means that two transformants carrying identical constructs can show wildly different expression levels. Most papers screen multiple transformants and report the best performer. What they don't always disclose: most transformants showed poor expression, and the "best" transformant may be a statistical outlier. This is why stable, reproducible overexpression in algal nuclei remains technically challenging compared to chloroplast transformation (which avoids position effects) or bacterial engineering.
Silencing of introduced transgenes over time
Even when transformation succeeds and initial expression is high, many algal transgenes undergo transcriptional silencing after 20–50 generations in the absence of selection pressure. This is particularly common in Chlamydomonas nuclear transformants and appears to be related to repeat-induced silencing mechanisms analogous to RNAi. Commercial production runs a strain for hundreds to thousands of generations; a strain that silences its engineered trait at generation 50 is commercially worthless even if it produces spectacular results in a 2-week lab experiment. Long-term stability testing is the most important missing data point in the vast majority of published algae synthetic biology papers.
Pleiotropy — unintended consequences
Metabolic pathways are not discrete channels. They are embedded in regulatory networks where a change in one enzyme activity propagates through multiple other pathways. Knocking out the starch synthesis pathway to redirect carbon to lipids may secondarily affect the cell's ability to buffer excess reducing power during high-light stress — leading to oxidative damage under outdoor conditions that was invisible in the controlled-light lab experiment. These pleiotropic effects are hard to predict from pathway maps and often only manifest at scale or under production conditions. Papers that claim a clean phenotype in a knockout or overexpression strain without measuring broad metabolite profiles (metabolomics) are underreporting the complexity of their system.
The feedback regulation problem
Many biosynthetic pathways have feedback inhibition built in — the end product inhibits an early enzyme, preventing runaway accumulation. This is useful for cellular homeostasis but frustrating for metabolic engineers who want to accumulate large amounts of the end product. In the astaxanthin pathway, for example, accumulation of downstream products provides feedback onto upstream enzymes. Overexpressing the downstream enzyme may not increase flux if the upstream enzyme is already being inhibited by product accumulation. Solving this requires engineering not just the pathway genes but also the regulatory logic of the pathway — which is significantly more complex than simple overexpression and is why "straightforward" pathway engineering often produces less improvement than predicted.
The research is genuine and the progress is real. Nannochloropsis CRISPR knockouts produce significantly more lipid than wild-type under controlled conditions. Chlamydomonas chloroplast transformants produce recombinant proteins at commercially meaningful levels. Phaeodactylum CRISPR editing is established. The issue is not that the science is fake — it is that the gap between "works in a flask" and "works in a 10,000 L outdoor PBR under Indian monsoon conditions with a 5-year regulatory timeline" is not measured in months. It is measured in years to decades. A company building on this science today should position those results as the thesis for a Phase 3 or Phase 4 product, not as a near-term commercial differentiator.
Question 1 — What is the baseline? "7× higher titre" compared to what? If the comparison is to Nannochloropsis wild-type under standard conditions, the relevant question is how this compares to the commercially optimised wild-type strains used by Nannochloropsis EPA producers. Wild-type lab strains often produce significantly less EPA than optimised commercial lines (which have had years of classical selection applied to them). A 7× improvement over an unoptimised lab strain might be a 2–3× improvement over a commercially optimised line — still significant, but less alarming.
Question 2 — Which metric is "titre"? "Higher titre" means higher concentration in the culture flask. Ask: what was the volumetric productivity (mg/L/day)? If the engineered strain grows more slowly — which is common when beta-oxidation is knocked out, since beta-oxidation helps the cell manage fatty acid turnover — the higher titre may reflect longer cultivation time rather than faster production. Calculate or request the volumetric productivity figure.
Question 3 — How many transformants were screened? If the paper reports a single "representative transformant" out of an unspecified screening campaign, the 7× result may represent the tail of a distribution, not a reliably reproducible phenotype. Reproducibility across independently generated transformants is essential for commercial confidence.
Question 4 — "Photobioreactor conditions" — what scale? This phrase in an academic paper almost certainly means a 1–2 L laboratory PBR under controlled light and temperature. This is not outdoor production conditions. Ask: was the experiment run under simulated outdoor light cycles with temperature variation, or under stable controlled light at laboratory temperature? The beta-oxidation knockout may have consequences for cold stress tolerance that are invisible in a 25°C lab PBR but relevant at night temperatures in an outdoor system.
Question 5 — Genetic stability across generations? Was the strain tested for stability beyond 30 generations without antibiotic selection? If the paper was published in a single batch experiment without stability data, the commercial relevance is limited even if all other metrics are strong.
Your investor brief: "A paper reports 7× EPA titre improvement in engineered Nannochloropsis. We have assessed it against five criteria: the improvement is against an unoptimised baseline, so the real-world advantage over commercial lines is unclear; volumetric productivity data is needed to assess commercial relevance; scale is laboratory only; and no stability data is presented. We will monitor this group's next publication, which should include scale-up and stability data. This is not a near-term competitive threat to wild-type production — it is a 3–5 year research pipeline if the results hold at scale. We recommend no change to current strategy."
Design step: Define the target phenotype precisely before touching any cultures. What does "improved phycocyanin content" mean in commercial terms? Food-grade phycocyanin requires ≥0.4 A620/A280 absorbance ratio in the extract, and price is determined by purity and colour intensity. The relevant target is not just phycocyanin % DW in the biomass — it is also the quality and stability of phycocyanin under production conditions (phycocyanin degrades under high-temperature stress, so heat tolerance is a co-selected trait). Design step deliverable: a written phenotype specification with measurable numbers — "target: 15% DW phycocyanin in dried biomass under standard Indian outdoor conditions in June (high temperature month); phycocyanin absorbance ratio ≥0.5 after standard aqueous extraction protocol."
Build step: Apply UV mutagenesis (254 nm, 30–60 second exposure to a dilute culture, targeting ~1–5% survival rate) to the starting strain. Plate survivors on solid medium under growth conditions. This generates 500–2,000 colonies per mutagenesis run, each carrying a unique combination of point mutations. This is the "build" — you have created a population of variant organisms, each with a different genetic composition.
Test step: Screen colonies for phycocyanin content. The fastest method is Nile Red fluorescence for lipids (not directly applicable here) — for phycocyanin, use spectrophotometric screening of crude cell extracts from individual colonies suspended in buffer. This is labour-intensive but feasible: extract 200 colonies, measure A620 of each extract, rank by phycocyanin content. Select the top 10–15% for further characterisation. Subculture selected colonies for 30 generations without selection pressure and re-measure — this is the stability test, built into the Test step. Measure growth rate alongside phycocyanin content, because a high-phycocyanin slow-growing strain may have worse volumetric productivity than the wild-type. Success metrics: volumetric phycocyanin productivity (mg phycocyanin/L/day) ≥25% above wild-type; stable across 30 generations; no significant reduction in growth rate (less than 15% reduction acceptable).
Learn step: Analyse the distribution of improvements. Did most survivors show modest improvement (suggesting the wild-type is already near its optimum under these conditions), or was there a wide distribution (suggesting significant unexploited genetic variation)? What conditions produced the best improvement — did high-temperature mutagenesis survivors perform better under Indian summer conditions? This informs whether a second mutagenesis cycle is worth running and what conditions to use. Document everything — each DBTL cycle is an experiment whose findings inform the next cycle. After 3–5 cycles, a cumulative improvement of 30–50% in volumetric phycocyanin productivity is a realistic target based on published classical selection outcomes for carotenoid-producing organisms.
What is biologically credible: Chlamydomonas chloroplast transformation is the most established recombinant protein production platform in any microalga. Expression levels of up to 10–20% of total soluble protein are achievable for well-optimised transgenes in the chloroplast — this is comparable to E. coli expression levels for many proteins. The Mayfield Lab at UCSD and Triton Algae Innovations have demonstrated this commercially for food proteins (heme) and are advancing it for vaccines and industrial enzymes. The photosynthetic input substituting for glucose feedstock is real — the carbon and energy source is CO₂ and sunlight, which are free at the point of use.
The economic gap to test: The cost of industrial enzyme production is dominated not by feedstock cost but by capital cost, processing cost, and titre. Industrial enzyme producers (Novozymes, DSM) achieve titres of 50–100 g/L protein in optimised E. coli or fungal fermenters. Chlamydomonas photobioreactors achieve cell densities of 1–5 g/L dry biomass; even at 15% of total soluble protein, that is 0.15–0.75 g/L recombinant protein titre. The volumetric productivity gap between bacterial fermentation (highly optimised, decades of development) and Chlamydomonas PBR production is probably 50–200× in terms of titre and 10–50× in terms of volumetric productivity. The cost saving from avoiding glucose feedstock is not close to offsetting a 50–200× volumetric productivity disadvantage when PBR capital costs are factored in.
When does the thesis work: The Chlamydomonas recombinant platform makes commercial sense for high-value proteins where: (a) the protein cannot be produced in bacterial systems (because it requires eukaryotic post-translational modifications, or folds incorrectly in E. coli), (b) the market price for the protein is high enough (>$1,000/kg) to justify the lower titre, and (c) the photosynthetic origin provides a regulatory or marketing advantage (animal-free, sustainable production). Industrial enzymes sold at $5–50/kg do not meet criterion (b). Vaccine antigens, specialty food proteins, and pharmaceutical-grade proteins sold at $100–10,000/kg potentially do.
Specific assumptions to test in the pitch: What is the target protein, its market price, and the required purity specification? What titre has been demonstrated in the lab and at what scale? What is the cost model per gram of purified protein at 1,000 L PBR scale, including capital depreciation, energy, and downstream processing? Compare to the market price. If the cost model closes at a protein price above $200–500/kg and the founder has a specific customer with that requirement, the thesis is credible. If they are relying on commodity enzyme pricing, the model does not work and they know it or they don't.
The fundamental reason the tradeoff exists: Carbon and energy are finite resources inside a cell. Every gram of carbon directed toward a target product (lipid, pigment, recombinant protein) is a gram of carbon not directed toward new cell biomass. Every unit of ATP consumed to run the biosynthetic pathway to the target compound is a unit of ATP not consumed for DNA replication, protein synthesis for growth, and cell division. This is an inescapable zero-sum tradeoff at the level of cellular resource allocation. Cells that accumulate large amounts of a target compound grow more slowly because they are investing resources in product accumulation rather than biomass doubling. This is why astaxanthin accumulation in Haematococcus requires nitrogen starvation: the cell is induced to stop growing and redirect resources into stress-protection pigment synthesis.
Why it is worse in photoautotrophic algae compared to heterotrophic bacteria — reason 1: fixed energy input rate. A bacterium growing on glucose can be fed more carbon by increasing the glucose feed rate when you want more product — you can partially decouple growth rate from product formation by adjusting feed. An algae cell growing on light and CO₂ has a fixed maximum photosynthetic rate determined by light intensity and the capacity of the photosynthetic machinery. You cannot simply "feed more light" (beyond saturation) to compensate for the growth rate reduction. This means the total resource pool available for both growth and product formation is capped by photosynthetic rate.
Reason 2: Light-dark cycles impose temporal constraints. Outdoor algae undergo daily light-dark cycles. During the light period, carbon is fixed and energy is generated. During the dark period, some of the stored carbon is consumed for respiration. An engineered strain that accumulates more product during the light period may also consume more of it during the dark period if the product is located in metabolically active pools. Bacteria in bioreactors have no equivalent temporal constraint.
Reason 3: Metabolic burden compounds with light stress tolerance. In bacteria, metabolic burden from overexpressed transgenes primarily manifests as slower growth in a stable environment. In outdoor algae, the cell is also managing reactive oxygen species, UV radiation, and temperature fluctuations. A cell that is already metabolically stressed by these environmental factors and then additionally burdened by a high-expression transgene may show much greater growth rate penalties under outdoor conditions than under the stable lab conditions where the initial result was measured. This is why outdoor performance for engineered algae strains is almost always worse than indoor performance — the metabolic burden is magnified by environmental stress.
Overexpression: pushing flux into the pathway. You increase the expression level of an enzyme in the target pathway, typically the rate-limiting step, to push more substrate through toward the product. The assumption is that the bottleneck is enzyme concentration — that the substrate is available but the pathway can't process it fast enough.
Specific example from algae literature: Couso et al. (2012, Plant Physiology) overexpressed phytoene synthase (PSY), the first committed enzyme in carotenoid biosynthesis, in Chlamydomonas under a nitrogen-starvation-inducible promoter. Under nitrogen starvation conditions, the engineered strain accumulated 3× more total carotenoids than wild-type. The result was real and reproducible across transformants. The limitation: PSY overexpression was effective because PSY was genuinely rate-limiting — there was substrate (GGPP) available that PSY was not processing fast enough. If the upstream GGPP supply had been limiting instead, the PSY overexpression would have had little effect. Identifying the actual rate-limiting step requires either metabolite measurement or testing multiple enzymes.
Competitive pathway knockout: redirecting flux from an alternative route. You eliminate a competing pathway that consumes the same precursor, forcing the cell to route more carbon through the target pathway by default. The assumption is that the precursor pool is shared between the target pathway and a competing pathway, and removing the competing pathway increases precursor availability for the target.
Specific example: Ajjawi et al. (2017, Nature Biotechnology) disrupted ZnCys, a transcriptional regulator in Nannochloropsis that activates the starch synthesis and other competing pathways. The knockout removed the repression of lipid accumulation and achieved 2–3× higher TAG content. This worked because ZnCys was a regulatory bottleneck rather than just a single enzyme — knocking it out had a multiplying effect across multiple genes it regulated, effectively removing competition from several pathways at once.
Which is more likely to produce commercially viable results? Competitive pathway knockouts, when they work, tend to produce larger improvements because they address the system-level competition for resources rather than optimising a single enzymatic step. However, they also carry higher risk: removing a competing pathway can have pleiotropic consequences — the starch synthesis pathway, for example, is not just a carbon sink; it also plays a role in buffering excess reduced carbon during high-light stress. Removing it can impair stress tolerance in ways that only appear under outdoor conditions. Pathway overexpression is safer (no deletion risk) but typically produces smaller improvements (10–50% rather than 2–3×) because it only optimises one step of a multi-step system. The most powerful results in the literature combine both strategies — overexpress the target pathway while knocking out the primary competitor — which is why the most cited papers (including Ajjawi 2017) use this combined approach.