Weak links: Advancing target‐based drug discovery by identifying the most vulnerable targets

Mycobacterium tuberculosis remains the most common infectious killer worldwide despite decades of antitubercular drug development. Effectively controlling the tuberculosis (TB) pandemic will require innovation in drug discovery. In this review, we provide a brief overview of the two main approaches to discovering new TB drugs—phenotypic screens and target‐based drug discovery—and outline some of the limitations of each method. We then explore recent advances in genetic tools that aim to overcome some of these limitations. In particular, we highlight a novel metric to prioritize essential targets, termed vulnerability. Stratifying targets based on their vulnerability presents new opportunities for future target‐based drug discovery campaigns.


INTRODUCTION
Despite the availability of antibiotics, Mycobacterium tuberculosis (Mtb)   is the world's leading bacterial killer, claiming an estimated 1.3 million lives each year. 1 Treatment of drug-susceptible tuberculosis (TB) consists of a standard 4-to 6-month course of four antimicrobial drugs.This lengthy treatment regimen can have side effects that compromise patient adherence, thereby contributing to the rise of drug resistance. 2,3Indeed, TB is now thought to account for one-third of all deaths associated with antimicrobial resistance. 1 Controlling the TB pandemic is going to require multiple interventions, one of which is the development of new drugs.To this end, a fundamental goal of the TB field is to develop a regimen of new drugs with novel targets and/or mechanisms of action capable of treating both drug-sensitive and drug-resistant TB in weeks rather than months.
How are we going to do this?This review will briefly discuss the two main strategies to discover new TB drugs and address some of the limitations of each approach.We will then discuss the recent development This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.© 2024 The Authors.Annals of the New York Academy of Sciences published by Wiley Periodicals LLC on behalf of The New York Academy of Sciences.
of new genetic tools that help overcome some of these limitations, with an emphasis on a novel approach to define a target quality known as vulnerability.Vulnerability allows the stratification of essential genes by the amount of inhibition necessary to produce a fitness cost in bacteria.While the focus of this review is on Mtb, the approaches are generalizable to other bacterial pathogens and the concepts are applicable to target ranking in other diseases, such as malaria and cancer.

FROM PHENOTYPIC SCREENS TO TARGET-BASED DRUG DISCOVERY AND BACK
Current antitubercular discovery efforts primarily employ one of two strategies: phenotypic whole-cell screens or target-based approaches. 4Alternative (and more recent) approaches such as target-based whole-cell screening, virtual screening, and others are exciting new modalities being brought to bear on TB, and the reader is referred elsewhere for a discussion of these approaches. 5,6n phenotypic screens, compound libraries are screened to identify whole-cell active inhibitors of Mtb growth.Phenotypic screens have been very successful.All current drugs used to treat TB were discovered from whole-cell screens, from the first antitubercular, streptomycin, in 1944 to the most recent antitubercular, pretomanid, in 2019.However, phenotypic screens have limitations.These include: (1) the frequent rediscovery of inhibitors of so-called "promiscuous" Mtb targets such as MmpL3, DprE1, and QcrB, whose activity can be inhibited by structurally diverse compounds 7 ; (2) as typically applied, phenotypic screens are blind to the target and mechanism of action (MOA) of a hit compound, and defining the target and MOA can be difficult; (3) false positive hits such as pan-assay interference compounds (PAINs) are not always easily avoided 8 ; and (4) growth conditions of in vitro screens can differ significantly from in vivo infection, thereby running the risk of finding potent inhibitors of in vitro growth that are irrelevant during infection. 9er 90 years of whole-cell screening have made treating TB a reality.And given the success of this approach, it is important to continue to perform whole-cell screens to identify new drug candidates not subject to pre-existing drug resistance.That said, it seems unlikely that whole-cell screening alone will yield compounds that will be sufficient to dramatically reduce TB treatment time.
To achieve reductions in TB treatment time, a biology-driven approach could be useful.The objective of this strategy is to determine which biological mechanisms, when inhibited or poisoned alone or in combination, could lead to a rapid clearance of viable bacteria and/or reduction of drug-tolerant bacterial subpopulations that drive the need for months of combination chemotherapy.Members of such critical biological pathways could then become targets for so-called target-based drug discovery (TBDD).
In contrast to phenotypic screens, TBDD begins with the selection of a compelling biological target.The target, typically a protein, is purified and used to identify compounds that modulate in vitro target function, typically assaying for inhibition.Hit compounds are then tested for whole-cell activity.Target-driven approaches have a significant advantage in that targets can be rationally prioritized to select those that (1) are not subject to pre-existing drug resistance or that can even reverse resistance 10 ; (2) have a structurally exploitable active site; (3) are localized on an extracellular surface to potentially facilitate antibiotic access; and (4) are based on a biologic rationale that inhibiting the target could contribute to treatment shortening.
Given the throughput of biochemical screens and dramatically higher throughput of virtual screens, orders of magnitude more compounds and an expanded chemical space can be screened in TBDD compared to phenotypic screens.
So, with all these potential advantages, why is TBDD not the primary mechanism of discovering new antituberculars?The short answer is that it has not worked yet, at least not for antibacterials.

TARGET-BASED ANTIBACTERIAL DISCOVERY: A PROMISING BUT CHALLENGING APPROACH
In 1998, the whole genome sequence of Mtb became available, 11 and 3924 open reading frames were identified.Five years later, 614 of these genes were predicted to be essential for Mtb growth in in vitro culture. 12With this information in hand, the path forward seemed clear: pick essential Mtb genes whose protein products have a ligandable active site, perform TBDD, and new antibiotics would be found.
Unfortunately, while TBDD has been broadly applied to Mtb, 13 not a single campaign has resulted in a drug that is being used to treat TB patients.In the TB drug clinical pipeline, 14 only two (VXc486 and BVL-GSK098) out of 32 antibacterials were discovered through TBDD. 10,15periences with TBDD for other bacterial pathogens have not fared much better. 16As a result of these failures, many have abandoned TBDD and returned to phenotypic screens to discover new antibiotics.
However, we know that TBDD can work.TBDD has achieved notable successes in treating certain cancers, 17 psychiatric disorders, 18 and viral infections. 19Why then is this drug discovery modality failing to discover new antibacterials?

CONSIDERING TARGET QUALITIES WHEN SELECTING A GOOD DRUG TARGET
Say an essential gene whose protein product has a theoretically druggable activity is identified, the protein is purified, a suitable assay is developed, and then used to identify hit compounds from a highthroughput small molecule screen.Hit compounds are then tested on live Mtb, and disappointingly no growth inhibition is observed, and the campaign is ultimately declared a failure.
When TBDD fails, the failure is often attributed to compound liabilities. 16,20These include failure modes such as: (1) low compound permeation through the cell envelope; (2) the compound is metabolized, modified, or degraded by bacterial enzymes; and/or (3) the compound is effluxed out of the cell and away from its target (Figure 1).These are major failure modes for hit compounds.Indeed, once such liabilities are known, one can try to engineer around them using medicinal chemistry to achieve more potent compounds. 21,22For example, spectinomycin is a potent translation inhibitor but has poor antitubercular activity due to drug efflux.Structure-based design was used to generate spectinomycin analogs that are much more potent because they avoid drug efflux. 23,24But beyond the chemical properties of the compounds used in the screen, it is equally important to consider the target, because not all targets are equally attractive (Figure 1).While not an exhaustive list, an ideal target for TBDD would be druggable, essential for growth in vitro (to facilitate drug discovery) and in vivo, and vulnerable.

Druggability
Druggability refers to the likelihood that a target can be bound and has its activity modulated by a ligand.Druggability is often determined by the complementary chemical properties of the target (e.g., an enzyme active site) and a potential ligand.Targets with high druggability are theoretically more likely to be successfully targeted by small molecules. 25[28] For an antibacterial, it is likely the target must additionally be selectively druggable, that is, the potential ligand binding site must be sufficiently different from a potential human homolog to avoid toxicity.

Essentiality
Essential genes encode proteins or RNAs that are indispensable for the growth of the organism in a given condition.Almost all antibacterials target the products of genes essential for in vitro growth. 29us, essential genes are, in principle, reasonable candidates for drug discovery as their inhibition by a small molecule would impair bacterial growth or viability.Of the compounds in the TB clinical pipeline with known MOA, most inhibit an in vitro essential target. 14Growing evidence suggests that gene essentiality is largely but incompletely conserved between Mtb isolates, 30,31 and thus, it is important to ensure that a target chosen for TBDD is essential beyond the commonly used reference Mtb strains.Lastly, in vitro growth conditions can never mimic the full complexity of the infection environment in vivo.
Thus, it is important to ensure that a gene prioritized for TBDD encodes a product that is essential for Mtb growth in in vivo environments that more closely mimic those found in humans.
In Mtb, there are 743 genes essential for in vivo growth, 32,33 of which ∼500 can be considered druggable. 34How does one pick the most promising of the 500 targets for TBDD?Beyond druggability and other considerations, we and others have proposed that vulnerability is an important prioritization metric to consider. 13,31,35

Vulnerability
Antibacterial compounds rarely achieve full target inhibition.Thus, the all-or-none definition of essentiality afforded by traditional genetic approaches such as transposon insertion sequencing (TnSeq) and gene deletion fails to discern some of the most attractive bacterial targetsthose whose incomplete inhibition results in major fitness costs.[37][38][39][40] We quantitatively defined this expression-fitness relationship as gene vulnerability (Figure 2). 31Vulnerability relates the magnitude of gene expression inhibition with the resulting decrease in organismal fitness, thus describing gene essentiality as a continuous trait.In this framework, essential genes exist along a gradient of vulnerability that reflects the fitness costs of graded gene silencing.Vulnerable genes encode products where minimal inhibition is sufficient to reduce fitness, while invulnerable genes encode products that can endure higher levels of inhibition without reducing bacterial fitness.
It is easy to imagine that, all else being equal, vulnerable targets may more often result in successful TBDD campaigns than invulnerable targets.Vulnerable targets require less inhibition in order to reduce bacterial fitness, and thus compounds derived from a TBDD campaign may not need to be highly potent in order to achieve sufficient target inhibition to manifest whole-cell activity.Even compounds with incomplete envelope permeability or subject to xenobiotic metabolism and/or efflux may still achieve sufficient levels of target engagement for a vulnerable target to reduce bacterial fitness.Medicinal chemistry can then be performed to engineer around compound liabilities and improve on-target activity.Depending on the bacterial target, the maximum fitness cost of inhibiting a vulnerable target may also be much greater than inhibiting an invulnerable target. 31

Quantifying target vulnerability
Quantification of gene vulnerability requires the systematic titration of gene product levels and the determination of resulting bacterial fitness.This rules out all-or-nothing genetic approaches like TnSeq (as currently implemented in Mtb) and gene deletion approaches.
Prior methods to determine the vulnerability of Mtb targets relied primarily on regulated promoters or regulated proteolysis.For example, Mizrahi and colleagues studied vulnerability of coenzyme A (CoA) biosynthesis in Mtb by replacing endogenous gene promoters with tetracycline-regulated promoters. 36The authors could then use different concentrations of the tetracycline analog, anhydrotetracycline, to vary the levels of target gene expression.The authors found that biosynthetic enzymes in the CoA pathway required different levels of expression to maintain bacterial fitness.Strikingly, and consistent with prior work, CoaA/PanK could be expressed at levels below the detection limit by western blot without reducing bacterial fitness, potentially explaining why TBDD efforts to drug CoaA/PanK have failed. 41 a complementary approach, the Rubin and Schnappinger labs developed regulated proteolysis systems in Mtb. 35,42,43Here, a socalled degron tag is appended to the C-terminus of a target protein.In the most widely used system, this degron is recognized by a conditionally expressed adaptor protein SspB, which delivers the degron-tagged substrate to the ClpXP1P2 protease for degradation.Target-protein levels can be tuned by varying the expression levels of SspB.Using such a degron approach, Rubin and colleagues showed in M. smegmatis that depletion of essential antibiotic targets had widely varying effects on bacterial growth. 354][45][46] But these methods are labor-intensive, suffer from intrinsic limitations (e.g., nonnative expression levels with tetracycline-regulated promoters and intolerance of some proteins to C-terminal degron tags), and do not easily scale to study all potential Mtb drug targets.Moreover, because they are laborious, these methods do not easily translate to the study of Mtb clinical strains to determine vulnerability conservation.

CRISPRi OVERCOMES PRIOR METHODOLOGICAL LIMITATION IN STUDYING TARGET VULNERABILITY
To overcome some of the limitations described above, we recently developed a genome-scale CRISPR interference (CRISPRi) method to quantify gene vulnerability in Mtb (Figure 3).This approach allows the targeted and predictable tuning of gene expression and quantification of resulting bacterial fitness.In parallel, we developed a mathematical model that can correlate the predicted degree of target inhibition with the resulting fitness cost to quantify target vulnerability.We provide a general description of the approach and its findings below and refer the reader to Bosch et al. for details. 31 Our CRISPRi system employs a nuclease-dead Cas9 protein derived from S. thermophilus (Sth1 dCas9). 47dCas9 is guided to its target site in the genome by a single guide RNA (sgRNA). 48Both dCas9 and the sgRNA are under tightly regulated tetracycline-inducible promoters, thus enabling the conditional control of Mtb gene expression based on the presence or absence of tetracycline analogs like anhydrotetracycline or doxycycline.Sth1 dCas9 is different from the more commonly employed S. pyogenes dCas9 49,50 in that Sth1 Cas9 recognizes a more complex protospacer adjacent motif (PAM). 47The PAM is a sequence within the DNA target recognized by Cas9 that licenses R-loop formation as the sgRNA binds to the complementary DNA target.We found that Sth1 dCas9 can recognize noncanonical PAMs at the cost of the magnitude of target knockdown.Target knockdown could be predictably tuned across an approximately 200-fold range simply by targeting different PAM sequences. 47[51][52] Using this Sth1 dCas9 system, we constructed a pooled, genomewide library consisting of 96,700 sgRNAs targeting 98.2% of all annotated Mtb genes (Figure 3A).We biased the library toward sgRNAs targeting predicted in vitro essential genes 53 because knockdown of these genes was expected to reduce bacterial fitness and enable vulnerability quantification in standard laboratory culture medium.Each essential gene was targeted with an average of ∼100 sgRNAs that span the available range of predicted knockdown efficiencies by targeting different PAMs and shortening the length of the sgRNA targeting sequence.
With a tunable knockdown approach in hand, we next used this pooled Mtb CRISPRi library to perform a competitive growth experiment.We sampled sgRNA abundance by deep sequencing every 2.5-5 generations (Figure 3B).These data allowed us to empirically determine sgRNA strength, an approximation for the magnitude of target knockdown.sgRNA strength was normalized from 0 to 1, with 1 representing the strongest sgRNAs leading to high predicted levels of target knockdown.Using the sgRNA depletion data, we then sought to measure three properties for every gene that quantitatively reflect target vulnerability (Figure 3C): (1) the maximum fitness cost ("beta_max") imposed by strong target knockdown; (2) the fitness cost imposed by partial target knockdown ("M"); and (3) the phenotypic lag ("gamma") between the induction of CRISPRi and the time to observe an impact on bacterial fitness.The integration of these three factors can be used to define target vulnerability (Figure 3D).

ESSENTIAL Mtb GENES EXIST ALONG A GRADIENT OF VULNERABILITY
Consistent with prior observations, we found that essential Mtb genes vary widely in their vulnerability (Figure 4).Genes encoding the targets of the two most potent first-line TB drugs, isoniazid (inhA) and rifampicin (rpoB), were among the upper quartile of vulnerable genes, lending credence to the validity of vulnerability estimates to nominate valuable therapeutic targets.However, we also found 116 genes  to be even more vulnerable than inhA under these growth conditions, highlighting the potential for future TBDD efforts.
What are the most vulnerable genes in Mtb?Not surprisingly, many vulnerable genes encode products involved in central dogma processes: transcription, translation, and DNA replication.These findings are consistent with biological expectations and the considerable number of whole-cell active inhibitors targeting these processes.Yet, the analysis also brought granularity to these pathway-level findings.For example, within the process of DNA replication, large differences in vulnerability exist.The gyrase subunits gyrA and gyrB and the replicative polymerase dnaE1 are highly vulnerable, whereas genes involved in Okazaki fragment maturation (polA and ligA) are highly invulnerable.Within the process of protein translation, ribosomal proteins and tRNA synthetases are almost universally vulnerable, while amino acid biosynthesis is largely invulnerable, highlighting tRNA aminoacylation as a chokepoint in translation.Def, an enzyme that removes the formyl group from the initiating Met residue during protein translation, is essential but highly invulnerable in Mtb, providing a potential explanation for failed drug discovery efforts toward this target. 13Also unsurprisingly, many genes involved in cell envelope biosynthesis were highly vulnerable.At the pathway level, mycolic acid biosynthesis was more vulnerable than either of the two other major cell envelope polymers, peptidoglycan and arabinogalactan.But within the biosynthetic pathways of these latter two polymers, highly vulnerable enzymes such as murB and murX (peptidoglycan) and dprE1 (arabinogalactan) exist.
Beyond those genes and pathways expected to be critical for Mtb growth, there existed many unexpected vulnerabilities as well.For example, the components of the Clp protease complex, currently attracting much attention in TB drug discovery field, 54   work showing that some metabolic enzymes are maintained at much higher levels than required to maintain metabolic flux. 56Coenzyme A biosynthesis is also invulnerable, potentially providing an explanation for the failure to develop drugs that inhibit Mtb CoaA as mentioned above. 13Consistent with prior results, we identified coaBC as a choke point within coenzyme A biosynthesis, 36 but even this gene is not as vulnerable as targets in more vulnerable pathways.

WHY ARE ESSENTIAL GENES DIFFERENTIALLY VULNERABLE?
Why does Mtb express vulnerable genes at levels so close to those required for optimal fitness?That is, why not just make more of them?
Drawing parallels to haploinsufficient genes in eukaryotes, vulnerable genes may be dosage stabilized such that both under and overexpression reduce cellular fitness. 57The biochemical context of a target could also contribute to its vulnerability.For example, genes whose products are part of macromolecular complexes or constitute a chokepoint in a pathway 58 may be vulnerable.Of note, successful antimicrobials have been shown to be enriched for compounds targeting chokepoints. 58 the other hand, there are numerous potential reasons why a gene may be invulnerable.Higher-than-required levels of gene product could (1) impart robustness to stochastic changes in gene expression 59 ; (

GENETICS DOES NOT EQUAL PHARMACOLOGY
All genetic approaches have their limitations that must be taken into account.First, biochemical and genetic inhibition of targets are not the same. 63While small molecules can have diverse mechanisms of enzyme inhibition or agonism, CRISPRi only mimics the effect imposed by a noncompetitive inhibitor. 64For example, GSK-286 is a preclinical compound that blocks cholesterol catabolism in Mtb by functioning as an agonist of the adenylyl cyclase rv1625c. 65Second, the depletion of a target is not the same as the inhibition of its functional activity by a small molecule. 66For example, although a small molecule may selectively inhibit the enzymatic but not scaffolding function of an enzyme, CRISPRi will necessarily inhibit both.Such is the case for the Mtb ClpP2 protease subunit, whose chaperone binding activity is essential, but its proteolytic activity is not. 67Third, invulnerability should not be equated with undruggability.For example, BioA is an enzyme involved in the synthesis of the essential cofactor biotin.BioA is essential but highly invulnerable in media lacking biotin. 68And yet, the antibiotic amiclenomycin can covalently bind to and inhibit BioA to block Mtb growth. 69This fact highlights that the lifetime of the drug-target complex must be considered.Long-term target engagement, as in the case of covalent inhibitors, could increase small molecule efficiency against a less vulnerable target. 70Fourth, CRISPRi may cause a polar effect (i.e., any operonic gene downstream of the dCas9 binding site may be silenced in addition to the targeted gene 50 ).The vulnerability of a gene can thus be overestimated as it may reflect the compound effect of its own inhibition and any other gene located downstream in the same operon.Importantly, however, we found that the polar effect was not a main driver of target vulnerability in Mtb. 31

WHAT DOES THE FUTURE OF VULNERABILITY HOLD?
Vulnerability is influenced both by genetic background and growth conditions. 30,71We showed that while vulnerability was highly correlated between the reference lineage 4 strain H37Rv and the lineage 2 strain HN878, vulnerability was not universally conserved. 31This incomplete conservation of vulnerability could predict differential antibacterial susceptibility.Thus, it will be important to determine the conservation of vulnerability in different Mtb clinical isolates.Just as important, vulnerability to date has been quantified under replicating conditions in standard Mtb culture media.Future efforts should determine vulnerability under conditions that more closely mimic those found during human infection.

CONCLUSIONS
To curb the TB pandemic, we must discover new TB drugs that target biological processes whose inhibition or corruption can shorten treatment time.Recent technological advances hold great promise toward this goal.Here, we outline one such advance that focuses on improving our ability to identify promising TB drug targets.Vulnerability allows the prioritization of those drug targets that require minimal inhibition to block growth, and deprioritization of those expected to require very high levels of inhibition.By combining such target-quality assessments and biology-driven insight into key treatment-shortening processes, it is hoped that new drugs that outperform existing therapies may be discovered.The approach we outline here is readily generalizable to other bacterial pathogens, and the concepts are applicable to other diseases, such as malaria and cancer.

1
Target-based drug discovery (TBDD) can fail because of compound liabilities and choosing the wrong target.Compound liabilities such as low permeation, metabolization, and efflux (left) could be addressed by medicinal chemistry.To further increase the chances of TBDD success, target qualities such as target essentiality and vulnerability in vitro and in vivo (right) should be considered too.

2
Schematic depicting the concept of vulnerability as a continuous trait.Minimal knockdown of a vulnerable gene (dark purple) results in a severe fitness cost, while minimal knockdown of an invulnerable gene does not impact cell fitness.Until recently, both a method to systematically titrate gene expression and a model for the resulting expression-fitness relationship were unavailable for Mtb (see alsoRef.31).Abbreviations: E, essential; NE, nonessential.

3
Overview of the CRISPRi functional genomics platform used to quantify target vulnerability in Mtb.Upon library construction (panel A), triplicate cultures were passaged for approximately 30 generations in the presence (CRISPRi on) or absence of ATc.At the indicated time points, genomic DNA was harvested and sgRNA targeting sequences amplified for deep sequencing (panel B).The relative fitness of individual strains (sgRNA level) was quantified by the sgRNA log2 fold change (L2FC) over time (+ATc/-ATc) (panel C).Gene vulnerability (panel D) was then quantified by fitting a logistic regression curve through the data points of all sgRNAs targeting a given gene (i) and imputing the behavior of all possible sgRNA strengths not measured in our CRISPRi library (ii).The total fitness cost associated with all theoretical sgRNAs was then combined into a single value using an integral (iii), which we refer to as gene vulnerability (see alsoRef.31).Abbreviations: ATc, anhydrotetracycline; CRISPRi, CRISPR interference; sgRNA, single guide RNA.

F I G U R E 4
Circos plot depicting all targeted Mtb H37Rv genes (dots) with their respective vulnerability.The inner purple lines represent more negative vulnerability values, with the most vulnerable genes located closest to the center of the circle.Red dots indicate genes with a vulnerability in the upper quartile of vulnerability (filled red dot = confident vulnerability call; unfilled red dot = low-confidence call).Genes encoding the targets of first-line TB therapy (rpoB, inhA, and embAB) are highlighted in blue.The outer ring represents the gene-level L2FC value at 28.8 generations.The phthiocerol dimycocerosates (PDIM)/phenolic glycolipid (PGL) locus (gray) contains, as expected, no vulnerable genes (see also Ref. 31).
) enable a rapid cellular response to environmental changes to alter pathway flux faster than upregulating gene expression; (3) represent partial functional redundancy; (4) reflect moonlighting, where a single protein performs multiple functions and elevated protein levels are required to perform all functions60 ; and (5) reflect target levels required under growth conditions not modeled in standard laboratory culture.Invulnerability could also arise if the gene in question is part of a negative feedback loop.61CAN VULNERABILITY INCREASE THE SUCCESS OF TBDD FOR TB?The ability to systematically quantify target vulnerability presents novel opportunities for TBDD.All else being equal, vulnerability can differentiate between the large pool of essential and druggable targets, enabling the prioritization of highly vulnerable targets and deprioritization of highly invulnerable targets.While some vulnerable targets such as DprE1, TopA, EfpA, and tRNA synthetases are currently being prosecuted within the framework of the TB Drug Accelerator, 62 many additional targets remain unexplored.These include many noncanonical but potentially compelling targets like those involved in protein secretion (SecYEG) and folding (GroES and GroEL2), metabolism (NadD, Dxs1, AroF, and PurB), chromosome replication (DnaE1 and DnaA), and cell division (FtsZ).Time will tell whether TBDD campaigns directed against highly vulnerable targets have more favorable outcomes than TBDD campaigns prioritized based on gene essentiality alone.