The genetics of aerotolerant growth in an alphaproteobacterium with a naturally reduced genome

A defining feature of Z. mobilis is its naturally small genome of only 1,915 protein-coding genes, which are encoded across one ~2.06-Mb chromosome and four endogenous plasmids (0.03–0.04 Mb each) (12). In contrast, the vast majority of free-living Alphaproteobacteria carry many additional genes (Fig. 1A), including some of the closest Z. mobilis relatives, such as Sphingomonas wittichii and Rhodobacter sphaeroides (~5,300 and ~4,200 genes, respectively) (Fig. 1B) (32, 33). Endosymbiont Alphaproteobacteria (e.g., Rickettsia and Wolbachia) (Fig. 1A) have even fewer genes than Z. mobilis but rely on their host to complement various missing gene functions (34, 35).

To evaluate Z. mobilis as a model alphaproteobacterium, we compared the Z. mobilis strain ZM4 genome against seven genetically tractable Alphaproteobacteria (“Tractable Alphas”) chosen based on the availability of both a genome sequence and an essential gene list curated from Tn screens (36 – 41). Given the low genetic diversity within the genus Zymomonas (17), ZM4 serves as a strong representation of the genus overall. We found that of the 890 protein-coding genes that are conserved across all seven Tractable Alphas, Z. mobilis has orthologs for 701 (~79%) (Fig. S1; Table S2). As expected, this group of genes was enriched for core cellular processes such as translation (false discovery rate [FDR] = 2.9e⁻⁷) and central metabolic processes (FDR = 9.1e⁻⁶). Z. mobilis also carries relatively few unique genes that do not share orthologs in any of the Tractable Alphas (346 unique genes, fewer than any of the Tractable Alphas) (Fig. 1B); these genes largely encode horizontally transferred elements (e.g., CRISPR-associated immune systems) and hypothetical proteins (Table S2). Genes carried by other Alphaproteobacteria but not by Z. mobilis are largely related to oxidative phosphorylation (e.g., type I NADH dehydrogenase and cytochrome c oxidase) and some tricarboxylic acid (TCA) cycle components that are known to be absent in Z. mobilis (Table S2) (42 – 46).

To interrogate the genetic requirements of this small-genome model species, we constructed a whole-genome knockdown library using our previously developed Z. mobilis CRISPRi system (29, 47), a task made further feasible by its minimalistic genome. For nearly all Z. mobilis genes (~98%, 1,895/1,933), we successfully designed and cloned four corresponding sgRNAs intended to provide strong knockdown (i.e., the sgRNA sequence is fully complementary to the gene of interest). For the remaining 38 genes, the library contains 1–3 sgRNAs per gene. When assessing gene-level fitness, we considered the behavior of the median sgRNA, a strategy which aims to neutralize any biases incurred from ineffective, off-target, or “bad seed” sgRNAs (48). The library also contains 1,000 non-targeting sgRNAs that act as internal controls.

CRISPRi is inherently polar, where knockdown of an upstream gene within a transcription unit (TU) causes concurrent knockdown of the downstream gene(s) (30, 49). Conversely, “reverse polarity” is a poorly characterized CRISPRi phenomenon in which knockdown of a downstream gene causes decreased expression of the upstream gene(s) (50, 51). Previous studies have observed reverse polarity in Bacillus subtilis (49) and to a lesser extent in Escherichia coli (48). In contrast, reverse polarity was largely undetected in Mycobacterium tuberculosis (52). To investigate if reverse polarity occurs in Z. mobilis, we designed sgRNAs targeted along a reporter TU encoding two fluorescent proteins. Our findings revealed that reverse polarity is indeed present in Z. mobilis, as knockdown of the downstream sfGFP-encoding gene resulted in decreased expression of the upstream mScarlet reporter, with the strongest reverse-polar effects observed when targeting nearest to the 5′ end of the downstream gene (5.4-fold mScarlet knockdown when targeting the 5′ end of sfGFP and 1.8-fold when targeting 3′ end) (Fig. S2). Similar results were observed in E. coli (3.1-fold mScarlet knockdown when targeting the 5′ end of sfGFP and no detectable mScarlet knockdown when targeting 3′ end of sfGFP) (Fig. S2). Together, these results underscore the known importance of interpreting CRISPRi results at the level of TUs. To account for this, we note predicted TUs (Table S5) when reporting gene phenotypes.

To understand how Z. mobilis employs its small genome to flourish in oxygen diverse environments, we utilized the whole-genome CRISPRi library to identify genes that are conditionally essential for aerobic or anaerobic growth and genes that are generally essential regardless of condition. Library cultures were grown in flasks with saturating concentrations of IPTG (1 mM) to induce CRISPRi and were maintained in the exponential phase for ~10 doublings at which point samples were taken to measure sgRNA spacer abundance by Illumina sequencing (Fig. 2A).

Likely due to its small genome size, Z. mobilis harbors a larger fraction of essential genes than other model organisms. We identified 320 genes that were essential across both conditions, conservatively defined by log₂ fold change ≤ −3 and significance ≤ 0.05 (Fig. 2B; Table S6). A similar approach (i.e., LFC < −2) recovered 79% of known E. coli essential genes from a genome-scale CRISPRi screen (54), suggesting that our cutoffs largely guard against false positives. Of the 320 generally essential Z. mobilis genes, 289 are protein coding; remarkably, these genes represent ~15% of the Z. mobilis genome. In comparison, the E. coli and B. subtilis genomes are made up of only ~7% (307/4,131) and ~6% (257/4,245) protein-coding essential genes, respectively (55 – 57). Thus, Z. mobilis models the unique features of small-genome, free-living species in a way that other model bacteria cannot.

To verify these phenotypes outside of the pooled library context, we chose 31 generally essential genes to assay individually using a new validation sgRNA that was not included in the library. Each knockdown strain was serially diluted and spotted onto plates containing IPTG inducer alongside a non-targeting control and then analyzed following growth for plating and colony size defects (see Materials and Methods). Initially, we successfully recapitulated phenotypes for 27 of the 31 essential gene knockdowns tested (Fig. 2C; Fig. S3). Non-recapitulated phenotypes may result from false positives, specificity to the pooled context, differences in essentiality phenotype during growth on a plate versus in liquid culture, or ineffective validation sgRNAs. For example, we were at first unable to validate the generally essential gene ispG (ZMO0180) using the new validation sgRNA. However, when we instead used an sgRNA from our library with an LFC close to the gene median, we successfully validated the ispG phenotype outside of the pooled context (Fig. 2C; Fig. S3), bringing the total recapitulated phenotypes to 28 of 31 (~90%).

Comparison of the gene essentiality profile of Z. mobilis to the seven Tractable Alphas further solidified its value as a model, small-genome alphaproteobacterium. We found that ~88% of the 289 generally essential Z. mobilis genes have homologs that are also essential in at least one other alphaproteobacterium, with ~66% sharing essential homologs in most (≥4) Tractable Alphas (Fig. 2D; Table S2), thus demonstrating substantial agreement between core essential genes of Z. mobilis and other Alphaproteobacteria. As expected, essential genes conserved across most Tractable Alphas included fundamental cellular processes such as transcription, translation, and protein export. Genes uniquely essential in Z. mobilis were functionally enriched for carbon metabolism (e.g., glycolysis and the pentose phosphate pathway), reflecting a low redundancy, “catabolic highway” (58) ideal for industrial engineering (Fig. 2D; Table S7). Other unique essentials included plasmid-borne, predicted DNA-binding proteins. Given that the native Z. mobilis plasmids can be cured without loss of viability (59), we speculate that these proteins are negative regulators of toxic genes that contribute to plasmid maintenance.

Tn libraries are often constructed aerobically, obscuring the distinction between genes that are conditionally essential for aerobic growth versus those that are generally essential. In contrast, CRISPRi libraries can be generated aerobically without perturbing genes during the construction process, enabling discovery of genes that are conditionally essential for aerobic growth. Following the same quantitative approach as above, we next identified 37 conditionally essential genes for aerobic growth (LFC ≤ −3 and significance ≤ 0.05 for aerobic growth but not for anaerobic growth) (Fig. 2B and 3A; Table S6). In multiple cases, Z. mobilis has redundant enzymes performing similar functions, except one is oxygen tolerant and the other is not. For example, nrdA (TU: ZMO1039) and nrdF (TU: ZMO0443), which encode class I ribonucleotide reductases (RNRs) that synthesize deoxyribonucleotides for DNA replication and repair, are conditionally essential for aerobic growth, whereas class III RNRs are oxygen sensitive (60). Similarly, genes encoding the pyruvate dehydrogenase complex, which converts pyruvate to acetyl-CoA aerobically or anaerobically (61), including pdhAB (TU: ZMO1605-1606; pdhA, pdhB), pdhC, and lpd (TU: ZMO0510-ZMO0513; pdhC, ZMO0511, lpd, and ZMO0513), are conditionally essential for aerobic growth because the other enzyme responsible for this function (pyruvate formate lyase) is oxygen sensitive (62). The presence of redundant aerobic essentials in an otherwise streamlined genome points to a key role for aerotolerant growth in the natural environment of Z. mobilis.

Consistent with previous work that identified Z. mobilis genes with increased expression following a shift from anaerobic to oxygen-replete conditions (22), knockdown of known oxidative stress genes such as sod (superoxide dismutase; TU: ZMO1060) and iron-sulfur cluster assembly (suf) genes including sufE (TU: ZMO1067-1068; sufE, ZMO1068), ZMO0422 (homolog of transcriptional regulator IscR), and ZMO0428 (TU: ZMO0422-0429; ZMO0422, sufB, sufC, sufD, sufS, ZMO0428, and ZMO0429) also showed oxygen-specific fitness defects (Fig. 3A and B; Table S6).

Interestingly, the DNA repair gene recJ (TU: ZMO1231) was also found to be conditionally essential for aerobic growth, and this phenotype was validated outside the pooled context (Fig. 3A and B). Although non-essential in E. coli (63), previous work has shown that recJ is essential across the Alphaproteobacteria (64) and in Deinococcus radiodurans (65, 66) for aerobic growth. However, these studies did not (or could not, in the case of strict aerobes) compare the recJ phenotype during aerobic versus anaerobic growth. recJ knockdowns in Z. mobilis showed no apparent growth phenotype under anaerobic conditions. To understand the role of RecJ in alphaproteobacterial physiology, we examined the mutation frequency in recJ knockdown strains. We found that RecJ suppresses the mutation frequency in both aerobic and anaerobic conditions. Because full knockdown of recJ is lethal aerobically, we used a subsaturating concentration of inducer to create partial knockdowns. To calculate mutation frequency, we measured the number of rifampicin-resistant colony forming units (CFU) in non-targeting and recJ partial knockdown strains. Knockdown of recJ increased the mutation frequency in both aerobic (3.6-fold) and anaerobic conditions (6.8-fold; Fig. 3D); thus, we establish a role for RecJ in mutation suppression in Z. mobilis that may be applicable to other Alphaproteobacteria. However, the mechanistic rationale for recJ essentiality during aerobic growth remains elusive and requires further studies. We were additionally able to associate an oxygen-specific phenotype with unannotated or poorly annotated genes including the TU encoding ZMO1961-ZMO1962 (uncharacterized protein and putative histidine kinase, respectively) and ZMO1790 (TU: ZMO1790; predicted heme-related transporter [67]) (Fig. 3A and B; Table S6). We report these and other uncharacterized genes with phenotypes from our screen in Table S7. Finally, we identified ferredoxin/flavodoxin (Fd/Fld)-based redox systems specifically required for aerobic growth including fdxA (TU: ZMO1753) and fpr (Fd/Fld reductase; TU: ZMO0220). As we will highlight in the next section, such redox functions play a similarly indispensable role in anaerobic growth, though the relevant genetic players may differ by condition.

We next examined the 68 genes that were conditionally essential for growth in the absence of oxygen (LFC ≤ −3 and significance ≤ 0.05 for anaerobic growth but not for aerobic growth) (Table S6). Strikingly, TUs encoding either the Rnf complex (TU: ZMO1808-1814; rnfH, rnfE, rnfG, rnfD, rnfC, rnfB, and rnfA) or the F₁F_O ATP synthase (TU-1: ZMO0238-0241; atpH, atpA, atpG1, and atpD). TU-2: ZMO0242; atpC. TU-3: ZMO0669-0671; atpG2, atpF) were conditionally essential for anaerobic growth. Of note, while rnf knockdown was inconsequential for growth in the presence of oxygen (median aerobic LFC for rnf genes = −0.2), atp knockdown resulted in a mild aerobic fitness defect (median aerobic LFC for atp genes = −1.2) (Table S6).

The F₁F_O ATP synthase reversibly couples ion transport with the synthesis or hydrolysis of ATP. Rnf, originally named for its role in Rhodobacter nitrogen fixation, is similarly reversible and couples ion transport with electron transfer between ferredoxin and pyridine nucleotides (68). A previous Tn study by Deutschbauer et al. also observed that rnf genes are required for anaerobic growth of Z. mobilis; however, the mechanism underlying this phenotype was not explored further (67). The same group proposed ZMO1842 to be rnfF (TU: ZMO1842-1842; possibly rnfF, rseC) based on cofitness with other rnf genes. This gene was also conditionally essential for anaerobic growth in our screen (Table S6), but it presently is not annotated as rnfF and thus was not included in our subsequent rnf analyses.

We sought to understand the potential mechanism driving the rnf and atp anaerobic growth phenotypes. First, we tested whether the complexes depend on the electrochemical gradient using carbonyl cyanide 3-chlorophenylhydrazone (CCCP), an ionophore that dissipates the proton gradient. We reasoned that partial knockdown of rnf and atp genes could act as a sensitized genetic background against which to test CCCP challenge. Given that the atp genes are encoded across three operons in Z. mobilis, we targeted one gene per atp operon for knockdown (atpA, atpC, and atpF). For rnf knockdown, rnfE was targeted. Partial knockdown was induced with a subsaturating concentration of IPTG inducer with the goal of achieving an intermediate level of fitness, rather than a specific expression level (Fig. S4). We found that, under anaerobic conditions, rnf and atp knockdowns had increased sensitivity to a normally sub-lethal concentration of CCCP, demonstrating their interdependence on IMF (Fig. 4A). Consistent with our screen results, when the assay was repeated aerobically, the rnf knockdown mutant behaved like the non-targeting control while atp knockdowns retained some sensitivity to CCCP, though to a lesser extent than was seen in the absence of oxygen (Fig. 4A).

Given that the reactions carried out by both complexes are reversible, multiple models could explain the synergy of the ion gradient with F₁F_O ATP synthase and Rnf. In each case, interdependence with the ion gradient could indicate either that the complex functions to establish the gradient (a producer) or that it depends on the gradient to carry out its essential function (a consumer). Thus, we next investigated the directionality of both complexes during anaerobic growth.

Two lines of evidence support the hypothesis that Rnf operates as a membrane potential consumer during anaerobic growth. First, a ThT membrane potential assay, wherein increased ThT fluorescence indicates stronger membrane potential (69, 70), revealed that partial knockdown of rnfE yielded stronger membrane potential than an uninduced control (Fig. 4B; Fig. S5). Second, mass spectrometry metabolomics analysis of the rnfE partial knockdown uncovered an accumulation of intermediate metabolites in the methylerythritol 4-phosphate isoprenoid biosynthesis pathway. Iron-sulfur cluster enzymes involved in the MEP pathway require reduced ferredoxin/flavodoxin (Fd_red/Fld_red) (Fig. S6) (22). We observed ≥ four-fold accumulation of two intermediates, MEP and MEcDP (Fig. 4C). Thus, we infer that rnfE knockdown hinders Fd/Fld reduction, leading to a bottleneck in MEP pathway flux. As such, our data are consistent with the conclusion that, in the absence of oxygen, Rnf uses the ion gradient to power the otherwise unfavorable transfer of electrons from NADH to Fd/Fld, which in turn drives essential processes such as isoprenoid biosynthesis (Fig. 4D).

Conversely, analysis of the F₁F_O ATP synthase agrees with previous evidence suggesting the ATP synthase operates as a membrane potential generator (71 – 75). Partial knockdown of atpC resulted in lower membrane potential than an uninduced control (Fig. 4B). Furthermore, we reasoned that quantifying the ATP/ADP ratio for the atpC partial knockdown versus an uninduced control using metabolomics may illuminate the reaction direction. For example, if partial knockdown of atp genes results in an accumulation of ATP, this would suggest that, in the absence of knockdown, the F₁F_O ATP synthase is functioning to hydrolyze ATP and pump ions outside the cytoplasm. Indeed, the ATP/ADP ratio increased relative to the non-targeting control for the two atp knockdowns with the clearest signal (Fig. 4E). This accumulation of ATP in the atp knockdown mutants indicates that during anaerobic growth in rich medium, the F₁F_O ATP synthase hydrolyzes ATP to generate IMF (Fig. 4F).

Strains and media recipes are listed in Table S1. Escherichia coli was grown in LB broth, Lennox (BD240230) at 37°C aerobically in a flask with shaking at 250 rpm, in a culture tube on a roller drum, or in a 96-well deep well plate with shaking at 900 rpm. Zymomonas mobilis was grown at 30°C aerobically or anaerobically (anaerobic chamber with 5% CO₂, 5% H_2, and balance N₂) in rich medium glucose (RMG), Zymomonas rich defined medium (ZRDM), or Zymomonas minimal medium (ZMM) either statically in a culture tube or deep well plate for polarity experiments, in flasks with stirring at 150 rpm for library experiments, or in a Tecan Sunrise microplate reader statically with 60 s shaking at high intensity prior to optical density at 600 nm (OD₆₀₀) reads every 15 min for growth curves. Media were solidified with 1.5%–2% agar for growth on plates. Antibiotics were added when necessary: E. coli (100 µg/mL ampicillin (amp), 100 µg/mL carbenicillin (carb), 20 µg/mL chloramphenicol (cm), and Z. mobilis (100 µg/mL cm). Diaminopimelic acid (DAP) was added at 300 µM to support growth of dap⁻ E. coli strains. Isopropyl β-D-1-thiogalactopyranoside (0.1–1 mM) was added where indicated. Strains were preserved in 15% glycerol at −80°C.

Plasmids and oligonucleotides are listed in Table S1. pir-dependent plasmids were propagated in E. coli strain BW25141 (sJMP146). Plasmids were purified using the GeneJet Plasmid Miniprep Kit (Thermo K0503), the QIAprep Spin Miniprep Kit (Qiagen 27106), or the Purelink HiPure Plasmid Midiprep Kit (Invitrogen K210005). Plasmids were digested with restriction enzymes from NEB and ligated using T4 DNA ligase (NEB M0202). DNA fragments were amplified using Q5 DNA polymerase (NEB 0491). PCR products were purified using the Monarch PCR & DNA Cleanup Kit (NEB T1030). Plasmids were transformed into electrocompetent E. coli cells using a Bio-Rad Gene Pulser Xcell using the Ec1 program (0.1-cm cuvette, 1.80 kV, and 1 pulse). Oligonucleotides were synthesized by Integrated DNA Technologies (Coralville, IA) or Agilent (Santa Clara, CA). Sequencing was performed by Functional Biosciences (Madison, WI) or the University of Wisconsin Biotechnology Center Next Generation Sequencing Core (Madison, WI).

Genome size analysis was performed using all complete National Center for Biotechnology Information (NCBI) genome entries for Alphaproteobacteria with a specific genus (i.e., not “Candidatus,” “alpha,” or “uncultured”) as of 7 July 2021. The median number of gene coding sequences (CDS) was calculated for each species and each genus.

OrthoFinder (89, 90) was used to identify orthologous groups among organisms (Table S2). Annotated genomes of Alphaproteobacteria associated with published Tn-seq experiments (“Tractable Alphas”) obtained from the NCBI and corresponding published lists of essential genes for these organisms (36 – 41) were analyzed. The model organism E. coli served as a comparator (55, 56). See Table S3 for additional information on digital resources and links to custom scripts.

sgRNAs were designed to knock down all genes in Z. mobilis ZM4 using a custom Python script and GenBank CP023715.1 as detailed in reference (29). sgRNA-encoding sequences were cloned between the BsaI sites of Mobile-CRISPRi (MCi) plasmid pJMP2480. Methodology for cloning individual sgRNAs was described previously in detail (29, 47). Briefly, two 24-nucleotide (nt) oligonucleotides encoding an sgRNA were designed to overlap such that when annealed, their ends would be complementary to the BsaI-cut ends on the vector.

The pooled CRISPRi library was constructed by amplification of sgRNA-encoding spacer sequences (Table S4) from a custom-pooled oligonucleotide library (SurePrint G7221A, Agilent) followed by ligation into the BsaI-digested MCi plasmid. Specifically, three pools of sgRNA-encoding inserts were generated by PCR amplification with primers oJMP197 and oJMP198 (Z1-genes), oJMP463 and oJMP464 (Z2-controls), and oJMP465 and oJMP466 (Z3-mismatches) from a 78-nt custom-pooled oligonucleotide library with the following conditions per 300 µL reaction: 60 µL Q5 buffer, 9 µL GC enhancer, 6 µL 10 mM each dNTPs, 15 µL each 10 µM forward and reverse primers, 6 µL 10 nM oligonucleotide library, 3 µL Q5 DNA polymerase, and 186 µL H₂O with the following thermocycling parameters: 98°C, 30 s; 15 cycles of the following: 98°C, 15 s; 56°C, 15 s; 72°C, 15 s; 72°C, 10 min; and 10°C, hold. Spin-purified PCR products were digested with BsaI-HF-v2 (R3733; NEB), and the size and integrity of full-length and digested PCR products were confirmed on a 4% agarose e-gel (Thermo). The BsaI-digested PCR product (without further purification) was ligated into a BsaI-digested MCi plasmid as detailed in reference 47. The ligation was purified by spot dialysis on a nitrocellulose filter (Millipore VSWP02500) against 0.1 mM Tris, pH 8 buffer for 20 min prior to transformation by electroporation into E. coli strain BW25141 (sJMP146). Cells were plated at a density of ~50,000 cells/plate on 150 mm LB-2% agar plates supplemented with carbenicillin. After incubation for 18 h at 37°C, colonies (~1,300,000 [Z1-genes], 700,000 [Z2-controls], and 1,950,000 [Z3-mismatches] for >30 coverage/oligonucleotide) were scraped from the agar plates into LB and pooled and the plasmid DNA was extracted from ~1 × 10¹¹ cells (10 mL at OD₆₀₀ = 33 [Z1], 25 mL at OD₆₀₀ = 11 [Z2], and 10 mL at OD₆₀₀ = 36 [Z3]) using a Midiprep Kit. This pooled Mobile-CRISPRi library was transformed by electroporation into E. coli mating strain sJMP3049 (20 ng plasmid DNA plus 90 µL electrocompetent cells, plated at a density of ~30,000 cells/plate on 150 mm LB-2% agar plates supplemented with carbenicillin and DAP). After incubation for 18 h at 37°C, colonies (~935,000 [Z1-genes], 124,000 [Z2-controls], and 247,000 [Z3-mismatches]) were scraped from the agar plates and pooled and resuspended in LB with DAP and 15% glycerol at OD₆₀₀ (40 [Z1], 17 [Z2], and 34 [Z3]) and aliquots of the pooled CRISPRi libraries were stored as strains sJMP2618, 2619, and 2620 (Z1-genes, Z2-controls, and Z3-mismatches, respectively) at −80°C. Combined results from Z1-genes and Z2-controls are reported in this paper.

The MCi system was transferred to the Tn7att site on the chromosome of Z. mobilis by tri-parental conjugation of two donor strains—one with a mobilizable plasmid (pTn7C1) encoding Tn7 transposase and a second with a mobilizable plasmid containing a Tn7 transposon encoding the CRISPRi system—and the recipient strain Z. mobilis ZM4. All matings used the E. coli WM6026 donor strain, which is pir⁺ to support pir-dependent plasmid replication, dap ⁻, making it dependent on diaminopimelic acid for growth, and encodes the RP4 transfer machinery required for conjugation. A detailed mating protocol for strains with individual sgRNAs was described previously (29, 47). Briefly, E. coli strains were grown ~16 h and Z. mobilis strains were grown ~24 h from single colonies. Cultures were spun at 4,000 × g for 5 min, and the cell pellets were washed twice with and equal volume of media (no antibiotic or DAP). For E. coli recipients, 100 µL of the washed donor and recipient strains was added to 700 µL LB and pelleted at ~4,000 × g and the cells were placed on a nitrocellulose filter (Millipore HAWP02500) on an LB plate supplemented with DAP and incubated at 37°C, ~2 h. For Z. mobilis recipients, 100 µL of the washed culture of donor strains and 500 µL of the recipient strain were added to 300 µL RMG and pelleted at ~6,000 × g and the cells were placed on a nitrocellulose filter (Millipore HAWP02500) on an RMG plate supplemented with DAP and incubated at 30°C, ~24 h. Cells were removed from the filter by vortexing in 200 µL media, serially diluted, and grown with selection on LB-cm plates at 37°C (E. coli) or selection on RMG-cm plates at 30°C (Z. mobilis).

For pooled library construction, aliquots of the Tn7 transposase donor strains (sJMP2618, 2619, and 2620) and pooled library strains were thawed and diluted to OD₆₀₀ = 10 in RMG. Overnight cultures of the E. coli Tn7 transposase donor (sJMP2591) and Z. mobilis recipient strain (sJMP412) were spun down for 10 min at 4,000 × g, and the OD₆₀₀ was normalized to 10. For each library, 2 mL of each strain was mixed and centrifuged for 10 min at 6,000 × g. Pelleted cells were spotted on two RMG agar plates and incubated for 39 h at 30°C prior to resuspension in RMG, serial dilution, and plating ~40,000 CFU/150 mm RMG-cm plates solidified with 2% agar followed by incubation for 72 h at 30°C. Cells were scraped from plates and resuspended in RMG + 15% glycerol, the density was normalized to OD₆₀₀ = 9, and aliquots were stored at −80°C as strains sJMP2621, 2622, and 2623. Efficiency of trans-conjugation (colony forming units on RMG-cm vs. RMG) was ~1 in 10⁴.

An operon encoding mScarlet and sfGFP reporter genes was cloned into the Mobile-CRISPRi vector (pJMP2367) at the PmeI site. Nine sgRNAs per reporter gene were then designed and individually cloned into this plasmid vector, and the Mobile-CRISPRi constructs were transferred to Z. mobilis as described above.

For Z. mobilis, cultures from single colonies were grown in triplicate in 1 mL RMG-cm in a 96-well deep well plate covered with AeraSeal at 30°C without shaking for ~24 h. Z. mobilis was subcultured 1:1,000 inoculum:medium into two 96-well deep well plates containing either RMG or RMG + 1 mM IPTG and incubated as described above. The plates were centrifuged at 4,000 × g for 10 min, the supernatant was removed, and the resulting cell pellets were resuspended in 1 mL PBS. Next, 200 µL of the cell suspension was transferred to a clear-bottom black microplate and measured in a Tecan Infinite microplate reader. Cells were shaken for 30 s (linear amplitude: 2.5 mm) prior to measurement of cell density (OD₆₀₀), GFP fluorescence (482/515 nm excitation/emission), and mScarlet fluorescence (560/605 nm). Experiments were repeated three times with 2–4 biological replicates each.

E. coli experiments were performed similarly, with the following modifications: LB was used in place of RMG, overnight growth was in 300 µL with shaking at 37°C for ~18 h, E. coli was subcultured 1:10,000 inoculum:medium and grown 6–7 h, and E. coli cells were further diluted 1:2 in PBS (to OD₆₀₀ ~ 0.3–0.6) prior to measurement to avoid cell shadowing in dense culture.

OD-corrected fluorescence is reported relative to a no-sgRNA fluorescent control. One sgRNA (“S2,” which targets mScarlet) proved toxic to Z. mobilis and thus was excluded from analysis for this organism.

We used previously published RNA sequencing (RNA-seq) data (GSE139939) gathered from Z. mobilis ZM4 cultures grown to exponential and stationary phase in RMG under aerobic and anaerobic conditions to generate putative transcription unit assignments for Z. mobilis (11). Briefly, after adapter trimming with Cutadapt version 2.10 (91) and quality trimming with Trimmomatic version 0.39 (92), we aligned replicates from each condition and growth phase separately to the ZM4 reference chromosome (GenBank CP023715.1) and large plasmids (GenBanks CP023716.1, CP023717.1, CP023718.1, and CP023719.1) using Rockhopper version 2.03 (93 – 95) to generate a set of predicted TUs for each condition and growth phase. Using custom Python scripts and Scipy version 1.5.2 (96), predicted transcript isoforms that covered the same genes for each growth phase were then parsed from Rockhopper output files and merged through single linkage hierarchical clustering using a pseudo-distance metric between two transcripts of “max size of compared transcripts in basepairs - number of basepairs that overlap” and a cophenetic cutoff of 100 bp. For merged transcripts, the most extreme boundaries were used. TU assignments are included in Table S5.

Z. mobilis CRISPRi libraries were incubated either in air or in an anaerobic chamber. The Z. mobilis CRISPRi libraries (sJMP2621, 2622, and 2623) were revived by the addition of 100 µL total frozen library stocks (OD₆₀₀ = 9, 8:1:6 ratio Z1:Z2:Z3) into 100 mL RMG (starting OD₆₀₀ = ~0.01) in a 500-mL flask and incubated with stirring at 150 rpm at 30°C until OD₆₀₀ = ~0.2–0.3 (~8 h) (initial timepoint = T _i) in duplicate. These cultures were diluted back to OD₆₀₀ = 0.01 in RMG + 1 mM IPTG in duplicate and incubated until OD₆₀₀ = ~0.2–0.3 (~8 h) (final timepoint = T _f), at which point they were again diluted back to OD₆₀₀ = 0.01 in RMG + 1 mM IPTG in duplicate and incubated until OD₆₀₀ = ~0.2–0.3 (~8 h). Cells were pelleted at 6,000 × g from 30 mL (T _i) or 10 mL (T _f), washed with 1× PBS and stored at −20°C for DNA extraction. Aerobic and anaerobic experiments were each done twice on separate days.

DNA was extracted from cell pellets with the DNeasy gDNA Extraction Kit (Qiagen) according to the manufacturer’s protocol, resuspending in a final volume of 100 µL with an average yield of ~50 ng/µL. The sgRNA-encoding region was amplified using Q5 DNA polymerase (NEB) in a 100-µL reaction with 2 µL gDNA (~100 ng) and primers oJMP697 and oJMP698 (nested primers with partial adapters for index PCR with Illumina TruSeq adapter) according to the manufacturer’s protocol using a Bio-Rad C1000 thermal cycler with the following program: 98°C, 30 s, and then 16 cycles of the following: 98°C, 15 s; 65°C, 15 s; and 72°C, 15 s. PCR products were spin purified and eluted in a final volume of 20 µL for a final concentration of ~20 ng/µL. Samples were sequenced by the UW-Madison Biotech Center Next Generation Sequencing Core facility. Briefly, PCR products were amplified with nested primers containing i5 and i7 indexes and Illumina TruSeq adapters followed by bead cleanup, quantification, pooling, and running on a NovaSeq 6000 (150 bp paired-end reads).

sgRNA-encoding spacer sequences were counted using the seal.sh script from the BBTools package (release: 28 March 2018). Briefly, paired FASTQ files from amplicon sequencing were aligned in parallel to a reference file that contained the spacer sequences cloned into the library. Alignment was performed using k-mers of 20 nucleotides in length—equal to the length of the spacer sequence. For more information on digital resources and links to custom scripts, see Table S3.

Log₂ fold change and confidence intervals were computed using edgeR (97). Briefly, trended dispersion of sgRNA-encoding spacers was estimated and imputed into a quasi-likelihood negative binomial log-linear model. Changes in abundance and the corresponding false discovery rates were computed for each spacer in each condition. Finally, gene-level fitness scores were obtained by calculating the median LFC of the spacers targeting each gene; gene-level significance was calculated by computing the Stouffer’s P-value (poolr R package) using the FDRs of the spacers targeting each gene (Table S8). For more information on digital resources and links to custom scripts, see Table S3.

Genes with significance ≤ 0.05 and median LFC ≤ −3 under aerobic and anaerobic conditions were considered generally essential. Genes with significance ≤ 0.05 and median LFC ≤ −3 under only one condition were classified as conditionally essential. Genes with significance ≤ 0.05 and median LFC ≤ −1 in one or both conditions were classified as having a fitness defect in the respective condition(s).

Phenotypes of gene knockdowns identified in the pooled screen were assayed by constructing individual Z. mobilis strains with sgRNAs targeting 31 putative generally essential genes, 20 putative aerobic essential or aerobic-specific fitness defect genes, 24 putative anaerobic essential or anaerobic-specific fitness defect genes, and 6 non-targeting control sgRNA strains. Guides used for follow-up experiments were distinct from the ones found in our pooled screen but were designed using the same parameters. Individual colonies from each strain were picked in duplicate (two biological replicates) into 1 mL RMG-cm into 96-well deep well plates, covered with AeraSeal, and incubated at 30°C for 24 h to saturation without shaking. Cultures were serially (1:10) diluted in RMG in a sterile V-bottom 96-well microplate and spotted onto RMG or RMG + 1 mM IPTG agar plates (OmniTray, Thermo 242811, 33 mL 2% agar medium) using a 96-well manual pinning tool (V&P Scientific; VP407A pin tool, VP381N microplate aligner, and VP380 OmniTray agar plate aligner) in triplicate (three technical replicates) and incubated either aerobically or anaerobically at 30°C for 72 h. Growth was measured using a fitness score and colony size score. Fitness scores represent the number of serial dilutions resulting in a visible spot, relative to a non-targeting control (i.e., a knockdown mutant with a fitness defect that produced spots for five dilutions, in comparison to a non-targeting control strain that produced spots for six dilutions, would have a fitness score equal to −1). Colony sizes were scored as small, large, or control-like. Growth was scored by three individuals; the median fitness score and the most prevalent colony size score are reported.

Acquisition of rifampicin resistance by a Z. mobilis recJ knockdown strain (sJMP10367) compared with a non-targeting control strain (sJMP10366) and parent strain (sJMP10365) was assayed as follows: ~2.4 × 10⁶ cells resuspended off plates into RMG (200 µL of a 10⁻⁵ dilution of cells normalized to OD₆₀₀ = 9) were plated in triplicate on RMG + 20 µM IPTG and incubated at 30°C for 90 h to induce partial knockdown. ~2.6 × 10⁸ cells from each replicate resuspended off plates into RMG (200 µL of cells normalized to OD₆₀₀ = 12) were plated in quadruplicate on RMG or RMG + 4 µM rifampicin (rif) and incubated aerobically and on RMG or RMG + 8 µM rif and incubated anaerobically, at 30°C in the dark for 96 h. Colony forming units were compared for growth ± rif.

Fitness challenge experiments were performed in clear 96-well microplates inoculated 1:50 (inoculum:medium) in duplicate from a saturated 5-mL overnight culture grown in RMG under either aerobic or anaerobic conditions. For anaerobic experiments, the 96-well microplate was placed in the anaerobic chamber overnight prior to inoculation. atp knockdown (sJMP6104, sJMP6105, and sJMP6109), rnf knockdown (sJMP6103), and non-targeting CRISPRi (sJMP6101) strains were grown in RMG with sublethal concentrations of inducer (50 µM IPTG for Fig. 4; Fig. S5; 25–50 μM IPTG for Fig. S4), carbonyl cyanide 3-chlorophenylhydrazone (8 µg/mL, from stock dissolved in DMSO), and/or DMSO (0.5%) as a control. Growth data were collected in a Tecan Sunrise plate reader for ~48 h. Experiments were repeated three times with two biological replicates each.

Actual relative fitness was measured by calculating the empirical area under the curve (auc_e) for each growth curve using the Growthcurver R package (version 0.3.1) (98) relative to the DMSO-only control for each strain. Following the multiplicative theory for synergy (49), predicted relative fitness for a combination of two conditions was calculated by multiplying the actual relative fitness values obtained from each condition individually (i.e., induction and CCCP). Actual and predicted fitness values are reported relative to those of the non-targeting control strain.

Overnight cultures of rnfE knockdown (MT271), atpC knockdown (MT272), and a non-targeting CRISPRi strain (MT270) were inoculated from single colonies in ZRDM + 70 µg/mL cm in duplicate and grown at 30°C in an anaerobic chamber. Cultures were diluted in fresh ZRDM-cm + 10 µM ThT to OD₆₀₀ = 0.04. Fifty micromolar-IPTG inducer was added as indicated. Two hundred microliters from each biological replicate was transferred to a 96-well microplate (flat, black, clear bottom, Greiner) in triplicate. Cells were grown at 30°C with shaking in a BioTek Synergy HTX Multimode Reader in an anaerobic chamber. OD₆₀₀ and ThT fluorescence (optical filters 460/40 nm and 528/20 nm for excitation and emission, respectively) were measured every 15 min for 30 h. Average ThT fluorescence and OD₆₀₀ values were corrected by subtracting background values from abiotic controls along the time course.

Aerobic cultures grown overnight in 5 mL RMG were washed twice in ZMM and then resuspended in ZMM to original volume. The washed cultures were diluted 1:50 (2 µL into 98 µL) into growth medium (RMG or ZRDM) in a 96-well microplate and grown ~24 h in a Tecan Sunrise microplate reader in the anaerobic chamber. Growth curves represent four to six biological replicates across two to three experiments (Fig. S8).

Z. mobilis was grown at 30°C in an anaerobic chamber. Z. mobilis cultures of atp knockdowns (sJMP6104, sJMP6105, and sJMP6109), rnfE knockdown (sJMP6103), and non-targeting CRISPRi (sJMP6101) strains were grown overnight, stationary, in 10 mL RMG in test tubes from single colonies, diluted ~1:1,000 into 60 mL RMG in 125 mL flasks, and grown with stirring (120 rpm) until OD₆₀₀ = ~0.7–0.9 to use as inoculum for experimental flasks. Experimental flasks were inoculated to OD₆₀₀ = 0.05 in 100 mL ZRDM + 50 µM IPTG in 250-mL flasks with stirring. From this flask, 30 mL was transferred into each of three separate 125-mL flasks with stirring. When these cultures reached OD = 0.5, 10 mL from each experimental flask was harvested by vacuum filtration onto a 0.45-µm nylon membrane filter (47 mm in diameter). Cells were resuspended off the filter into 1.5-mL ice cold solvent (40:40:20 acetonitrile:methanol:water), and the solvent was stored in a microcentrifuge tube at −80°C until ready for same-day processing.

As an internal standard, wild-type Z. mobilis ZM4 was grown in ZMM + labeled glucose (Table S1). Wild-type Z. mobilis was prepared as described above, with the following modifications: overnight RMG cultures were diluted into 10 mL ZMM + labeled glucose, a single experimental flask containing 30 mL ZMM + labeled glucose was inoculated, and 10 mL was collected in duplicate from the experimental flask.

Collected samples were spun in a microcentrifuge at max speed for 10 min at 4°C. Each sample’s supernatant was mixed 1:1 with supernatant from the labeled glucose internal standard and dried under a nitrogen gas stream.

Metabolomics LC-MS analyses (Table S9) were conducted utilizing a Vanquish ultra-high-performance liquid chromatography (UHPLC) system (Thermo Scientific), coupled with a Q Exactive hybrid quadrupole-Orbitrap mass spectrometer (Thermo Scientific) using electrospray ionization in negative-ion mode. The chromatography was performed at 25°C with a reverse-phase C18 column of 2.1 × 100 mm with 1.7 µm particle size (Water from Acquity UHPLC BEH). Solvent A (97:3 H₂O:methanol with 10 mM tributylamine adjusted to pH 8.2 using 10 mM acetic acid) and Solvent B (100% methanol) were used in a gradient manner: 0–2.5 min with 5% B, 2.5–17 min with a linear gradient from 5% B to 95% B, 17–19.5 min with 95% B, 19.5–20 min with a linear gradient from 95% B to 5% B, and 20–25 min with 5% B. The flow rate was constant at 0.2 mL/min. For targeted metabolomics, the eluent was injected into the MS for analysis until 18 min, after which the flow was redirected to waste. The MS parameters included full MS-SIM (single ion monitoring) scanning between 70 and 1,000 m/z and 160 and 815 m/z for the targeted metabolomics and MEP metabolite-specific methods, respectively. The automatic control gain (ACG) target was 1e6, with a maximum injection time (IT) of 40 ms and a resolution of 70,000 full width at half maximum (FWHM).