Reduced genome bacteria are powerful models for dissecting biological function and streamlined platforms for industrial production. These bacteria have less of the genetic redundancy that often obscures the functions of genes and pathways, simplifying modeling of cellular processes (1
). Bacteria with experimentally reduced genome sizes have shown improved properties for industrial applications such as increased genomic stability (3
), faster growth rates (5
), improved transformation efficiency (6
), and ease of genetic manipulation (7
), all of which facilitate introduction and maintenance of engineered pathways. However, experimental genome reduction can also result in substantial growth defects (8
) and loss of robustness to environmental conditions (10
) if the physiology of the bacterium is not well understood and taken into consideration. An appealing alternative approach would be to utilize bacteria with naturally reduced genomes compared with related species. Such bacteria would effectively be “pre-evolved” with the benefits of a reduced genome but with the robustness of environmental strains.
is an emerging model for bacteria with naturally reduced genomes and has excellent properties as an industrial platform. Z. mobilis
is a member of the highly studied class Alphaproteobacteria but has at least 1,000 fewer genes than closely related species (a total of 1,915 protein-coding genes [11
]). This natural genome reduction has created a streamlined metabolism that efficiently ferments sugars to ethanol using the Entner-Doudoroff pathway (13
), aiding metabolic modeling efforts (15
) and highlighting the promise of Z. mobilis
as a biofuel producer. Despite its reduced genome, Z. mobilis
is free living (i.e.
, not reliant on another organism [16
]) and grows quickly to high densities in standard rich medium (e.g., buffered yeast extract and glucose [17
]). In contrast, other reduced genome models are often endosymbionts with fastidious growth requirements and dependence on a host (18
). Z. mobilis
growth is robust to diverse environmental conditions including the presence or absence of atmospheric oxygen (aerotolerant anaerobe), high concentrations of ethanol (up to 16% [vol/vol]) (19
), and some but not all inhibitors found in plant-derived biofuel fermentation substrates (20
). Its safety profile (generally regarded as safe [21
]), ease of manipulation in aerobic settings (22
), and excellent anaerobic fermentation properties (14
) indicate that Z. mobilis
is both an outstanding model for basic biology of Alphaproteobacteria and a rising industrial workhorse.
Genes required for robust growth of Z. mobilis
across conditions are understudied, hindering both its use as a reduced genome model and rational engineering efforts to optimize biofuel production. In these contexts, generally and conditionally essential genes are of particular interest because they must be maintained by reduced genomes and are linked to core cellular processes such as carbon metabolism that impact biofuel production (15
). Moreover, reduced genomes are thought to harbor a larger fraction of essential genes than larger genomes (18
), underscoring the importance of such genes in bacteria with less genetic redundancy. Although transposon (Tn) mutagenesis is often used at the genome scale to identify essential genes (23
), previous attempts to apply this approach to Z. mobilis
were unsuccessful (25
), possibly due to the polyploid nature of the Z. mobilis
chromosome (26 – 28
). Regardless, Tn or other gene disruption approaches alone cannot be used to phenotype genes in the condition where they are essential, since this results in cell death.
With the goal of defining and characterizing essential genes, we previously developed a CRISPRi (clustered regularly interspaced short palindromic repeats interference) gene knockdown system for Z. mobilis
). CRISPRi targets genes for knockdown using a single guide (sg)RNA which directs a catalytically dead (d)Cas9 nuclease to a complementary gene target where the sgRNA-dCas9 complex binds and blocks transcription (30
). Our Z. mobilis
CRISPRi system has several important advantages: it is isopropyl β-D-1-thiogalactopyranoside (IPTG) inducible, titratable with subsaturating inducer or using mismatched sgRNAs (31
), and stably integrated into the chromosome without the need for selection and knocks down multi-copy genes simultaneously—features which enable interrogation of both non-essential and essential genes in this plausibly polyploid bacterium.
Here, we combine comparative and functional genomics approaches to establish Z. mobilis as a genetic model for Alphaproteobacteria. We use genome-scale CRISPRi to identify genes that are generally or conditionally essential depending on the presence of environmental oxygen. We find that generally essential Z. mobilis genes represent a core set of genes that are highly conserved across Alphaproteobacteria. Our sets of Z. mobilis aerobic and anaerobic essential genes contain several surprising players, revealing an oxygen-dependent requirement for the RecJ DNA repair protein and critical roles for the ATP synthase and Rhodobacter nitrogen fixation (Rnf) complex in the maintenance and utilization of the ion-motive force (IMF) during anaerobic growth. Our studies provide a genetic window into how a naturally reduced genome can be both streamlined and robust.
Alphaproteobacteria are fascinating models for fundamental biological processes as well as microbial powerhouses for industrial production of green energy. Despite this importance, core genes in Alphaproteobacteria are understudied due to the lack of both a simplified model genome and genetic tools capable of phenotyping all genes. This work advances our understanding of alphaproteobacterial genes by establishing Z. mobilis as a streamlined, model microbe with a naturally reduced genome and employing a genome-scale, CRISPRi strategy for comprehensive phenotyping. We identified generally and conditionally essential genes that underpin the aerotolerant lifestyle of Z. mobilis and found broad conservation of these genes in other Alphaproteobacteria. Our analysis of Rnf and ATP synthase respiratory complexes highlights critical functions in maintaining/consuming IMF and points to a key role of Rnf in isoprenoid synthesis. Both our CRISPRi strategy and resulting insights into core gene functions are readily applicable to Alphaproteobacteria and beyond.
CRISPRi inducibility and titratability are invaluable for identifying conditional essentiality phenotypes. Although Tn-seq remains broadly useful for identifying non-essential gene phenotypes without the need to design guide libraries, CRISPRi has some clear advantages. Because genes are instantly inactivated during Tn library construction, genes that are essential under the condition the library was constructed in are lost from the pool before downstream phenotyping experiments can begin. The only way to mitigate this issue would be to construct a new Tn library for every condition assayed—a considerable and impractical burden for investigators that are interested in probing many conditions to evaluate gene networks. CRISPRi inducibility mitigates this issue by separating library construction from fitness assays. As a result, we were able to define genes essential for aerobic growth despite constructing our CRISPRi library aerobically. An important example from our study is recJ
, which has been defined as “essential” in previous alphaproteobacterial Tn screens (64
), but what we show is conditionally essential during aerobic growth and dispensable anaerobically. Other genes that are canonically associated with oxygen stress (e.g., sod
) may also appear to be generally essential if Tn libraries are constructed aerobically. Titrating CRISPRi knockdowns allowed us to investigate the phenotypes of conditionally essential genes in the condition for which they are essential. This enabled us to define the mutator phenotype of recJ
and IMF altering phenotypes of rnf
genes. Given the portability of CRISPRi (e.g., Mobile-CRISPRi [76
]), we anticipate its broad utility in characterizing conditional essentiality.
We report that the broadly conserved respiratory complexes, Rnf and ATP synthase, are conditionally essential during Z. mobilis
anaerobic growth and provide possible causes for their essentiality. Rnf complexes are widespread across prokaryotes and couple ion-motive force (Na+
) with reversible Fd/Fld:NAD+
oxidoreductase activity (that is, oxidation of Fdred
and reduction of NAD+
and vice versa) to accomplish diverse biological roles (68
). In some anaerobes, Rnf is required for energy conservation: oxidation of Fdred
is coupled with ion transport to establish ion-motive force necessary for ATP generation. Recently, the Rnf-ATP synthase supercomplex from Thermotoga maritima
was purified and conclusively demonstrated to operate in this manner in vitro
). In other organisms, ion transport occurs in the reverse direction, with ion-motive force powering the transfer of electrons from NADH to Fd/Fld. Canonically, Fdred
then acts as a biologically powerful reductant and donates electrons to nitrogenase, a required step in N2
). Z. mobilis
is known to fix N2
anaerobically in minimal medium lacking biologically available nitrogen (ammonium) (80
). Under these conditions, rnf
and other nif
cluster genes associated with N2
fixation are upregulated accordingly (81
), and their functions may be further regulated by protein phosphorylation status (82
The present study extends the role of Rnf by demonstrating that Rnf is required for anaerobic growth and MEP pathway flux in Z. mobilis
, even when ammonium is abundant. Here, using a multi-modal approach (genetic sensitivity to electrochemical gradient disruption, ThT membrane potential assay, and metabolomics), we demonstrate that Rnf is a consumer of the ion gradient (Fig. 4
), suggesting that it uses IMF to power reduction of Fd/Fld. Furthermore, we show that rnfE
knockdown results in accumulation of MEP isoprenoid biosynthesis pathway intermediates (Fig. 4C
). We suggest that MEP pathway flux is disrupted due to a deficiency in Fdred
, which donate electrons to the iron-sulfur cluster enzymes IspG and IspH in the MEP pathway (22
). Thus, this downstream disfunction of the essential MEP pathway may explain the essentiality of Rnf for anaerobic growth, in part or in full. Isoprenoids produced through the MEP pathway are biologically invaluable. In bacteria, isoprenoids act as electron carriers (quinones), pigments (carotenoids), membrane components (hopanoids), and signaling molecules (84
). Given that isoprenoid biosynthesis is essential under all conditions for Z. mobilis
and references) (29
), the link between Rnf and MEP pathway flux during anaerobic growth suggests a separate mechanism may exist for generating Fdred
to maintain flux during aerobic growth. Indeed, ferredoxin (fdxA
; ZMO0220) and Fd/Fld reductase (fpr
; ZMO1753) were both conditionally essential for aerobic growth in our screen. Given that fpr
is also one of the most highly upregulated genes following oxygen exposure (22
), we speculate that these genes may replenish Fdred
aerobically, while Rnf performs this function anaerobically.
Ongoing engineering efforts aim to increase isoprenoid production by Z. mobilis
for industrial use as therapeutics, food additives, fragrances, and sustainable biofuels (87
). For example, Martien et al. observed that upregulation of fpr
, and the iron-sulfur cluster assembly operon (suf
) upon oxygen exposure coincided with improved flux through the MEP pathway, suggesting that manipulation of these genes could yield isoprenoid production strains with high MEP pathway flux (22
). This proposal is supported by a work in E. coli
, where overexpression of Fpr, FldA, and MEP pathway enzymes increased flux through the MEP pathway (88
). Furthermore, Khana et al. demonstrate that the iron-sulfur cluster enzymes IspG and IspH can modulate bottlenecks in the MEP pathway, and the authors point to electron-resupplying accessory proteins as possible engineering targets to further increase MEP pathway flux (83
). The present work adds Rnf to this toolkit of potential engineering targets for enhancing isoprenoid production by Z. mobilis
, especially for anaerobic industrial fermentations.
We also find that the F1
ATP synthase is conditionally essential for anaerobic growth. We further demonstrate that the F1
ATP synthase performs the essential function of hydrolyzing ATP to pump ions across the inner membrane and establish IMF (i.e., running in reverse). This role is consistent with previous conclusions surrounding Z. mobilis
ATP synthase function (71
), including work showing that addition of CCCP affects membrane potential but not intracellular concentration of ATP (75
). Our work both confirms this literature and additionally demonstrates that this function is essential for Z. mobilis
Our screen and CCCP sensitivity data also revealed an interesting contrast between the effects of rnf
knockdown on aerobic fitness. Specifically, while rnf
knockdown appears to have no detectable impact on aerobic growth, knockdown of atp
results in a mild fitness defect in the presence of oxygen (Fig. 4A
; Table S6
). This discrepancy could be explained by the possibly redundant function of atp
with the Z. mobilis
aerobic respiratory chain, which has an elusive function but may contribute to ion gradient generation and/or detoxification of oxygen (44
Given the clear connection between Rnf and F1
ATP synthase in Thermotoga maritima
), it is tempting to speculate an equal-but-opposite model for Z. mobilis
wherein the F1
ATP synthase establishes an ion gradient that is then directly used by Rnf. It is especially tempting given the long-anticipated discovery of a IMF-dissipating counterpart to F1
ATP synthase function (72
). Our data, however, point to additional biological complexity. If the main function of the F1
ATP synthase was to supply IMF for Rnf function, atp
knockdown should yield a similar accumulation of MEP metabolites as occurs for rnf
knockdown. However, we did not observe a consistent trend in MEP pathway metabolites for atp
knockdown strains (Fig. S7
). Multiple possible explanations exist. For example, disruption of the ion gradient by atp
knockdown likely has widespread effects on cellular physiology (e.g.,
in motility and transport) which may convolute biological outcomes, there may be additional ion translocators contributing to the ion gradient, and there may be other MEP pathway regulatory mechanisms at play (22
Herein, we establish Z. mobilis as a valuable genetic model for Alphaproteobacteria and deepen our understanding of how this naturally reduced genome bacterium adapts to the presence or absence of environmental oxygen. We anticipate that its use as a streamlined model with decreased genetic redundancy will simplify functional analysis of the basic biology of Alphaproteobacteria. We further expect that the novel biology gleaned from this study will aid in the development of prolific Z. mobilis strains for biofuel production.
MATERIALS AND METHODS
Strains and growth conditions
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% H2,
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.
General molecular biology techniques
Plasmids and oligonucleotides are listed in Table S1
-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).
Alphaproteobacterial genome size analysis
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.
Comparative analysis of conserved and essential genes
) 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
). See Table S3
for additional information on digital resources and links to custom scripts.
Z. mobilis Mobile-CRISPRi individual gene and gene library construction
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
). 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 H2
O 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.
Transfer of the Mobile-CRISPRi system to the E. coli or Z. mobilis chromosome
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
). 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.
Analysis of Mobile-CRISPRi reverse polarity in Z. mobilis or E. coli using a fluorescent reporter operon
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.
Generation of predicted transcription units from RNA-seq data
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
). 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
Library growth experiment
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.
Sequencing library samples
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).
Counting sgRNA sequences
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
Comparisons between conditions
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
Classifying gene essentiality and fitness defects
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).
Analysis of individual CRISPRi knockdown strains for essentiality using spot dilution assays
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.
Analysis of recJ CRISPRi knockdown strain mutation rate in the presence of rifampicin
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 measurement of Z. mobilis atp and rnf partial knockdowns with ion-motive force disruption by CCCP
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.
Detection of membrane potential using Thioflavin T fluorescence
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.
Z. mobilis growth curves in RMG or ZRDM
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
Liquid chromatography-mass spectrometry metabolomics for atp and rnf knockdown strains
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 H2
O: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).