What mechanisms underlie the transitions responsible for the diverse shapes observed in the living world? Although bacteria exhibit a myriad of morphologies1, the mechanisms responsible for the evolution of bacterial cell shape are not understood. We investigated morphological diversity in a group of bacteria that synthesize an appendage-like extension of the cell envelope called the stalk2,3. The location and number of stalks varies among species, as exemplified by three distinct subcellular positions of stalks within a rod-shaped cell body: polar in the genus Caulobacter and subpolar or bilateral in the genus Asticcacaulis4. Here we show that a developmental regulator of Caulobacter crescentus, SpmX5, is co-opted in the genus Asticcacaulis to specify stalk synthesis either at the subpolar or bilateral positions. We also show that stepwise evolution of a specific region of SpmX led to the gain of a new function and localization of this protein, which drove the sequential transition in stalk positioning. Our results indicate that changes in protein function, co-option and modularity are key elements in the evolution of bacterial morphology. Therefore, similar evolutionary principles of morphological transitions apply to both single-celled prokaryotes and multicellular eukaryotes.
Subscribe to Journal
Get full journal access for 1 year
only $3.90 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Tax calculation will be finalised during checkout.
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
Genomic data of strains sequenced in this study are deposited in GenBank/EMBL/DDBJ under accession numbers AWGD00000000 (Asticcacaulis sp. AC460), AWGE00000000 (Asticcacaulis sp. AC466), AWGC00000000 (Asticcacaulis sp. AC402), AWGF00000000 (Asticcacaulis sp. YBE204) and AWGB00000000 (Asticcacaulis benevestitus DSM16100).
Young, K. D. The selective value of bacterial shape. Microbiol. Mol. Biol. Rev. 70, 660–703 (2006)
Wagner, J. K. & Brun, Y. V. Out on a limb: how the Caulobacter stalk can boost the study of bacterial cell shape. Mol. Microbiol. 64, 28–33 (2007)
Stove-Poindexter, J. L. & Cohen-Bazire, G. The fine structure of stalked bacteria belonging to the family Caulobacteraceae. J. Cell Biol. 23, 587–607 (1964)
Pate, J. L. & Ordal, E. J. The fine structure of two unusual stalked bacteria. J. Cell Biol. 27, 133–150 (1965)
Radhakrishnan, S. K., Thanbichler, M. & Viollier, P. H. The dynamic interplay between a cell fate determinant and a lysozyme homolog drives the asymmetric division cycle of Caulobacter crescentus. Genes Dev. 22, 212–225 (2008)
Poindexter, J. S. Biological properties and classification of the Caulobacter group. Bacteriol. Rev. 28, 231–295 (1964)
Gonin, M., Quardokus, E. M., O’Donnol, D., Maddock, J. & Brun, Y. V. Regulation of stalk elongation by phosphate in Caulobacter crescentus. J. Bacteriol. 182, 337–347 (2000)
Wagner, J. K., Setayeshgar, S., Sharon, L. A., Reilly, J. P. & Brun, Y. V. A nutrient uptake role for bacterial cell envelope extensions. Proc. Natl Acad. Sci. USA 103, 11772–11777 (2006)
Schlimpert, S. et al. General protein diffusion barriers create compartments within bacterial cells. Cell 151, 1270–1282 (2012)
Hughes, H. V. et al. Protein localization and dynamics within a bacterial organelle. Proc. Natl Acad. Sci. USA 107, 5599–5604 (2010)
Brown, P. J., Hardy, G. G., Trimble, M. J. & Brun, Y. V. Complex regulatory pathways coordinate cell-cycle progression and development in Caulobacter crescentus. Adv. Microb. Physiol. 54, 1–101 (2009)
Schmidt, J. M. & Stanier, R. Y. The development of cellular stalks in bacteria. J. Cell Biol. 28, 423–436 (1966)
Kuru, E. et al. In situ probing of newly synthesized peptidoglycan in live bacteria with fluorescent d-amino acids. Angew. Chem. Int. Edn Engl. 51, 12519–12523 (2012)
Aaron, M. et al. The tubulin homologue FtsZ contributes to cell elongation by guiding cell wall precursor synthesis in Caulobacter crescentus. Mol. Microbiol. 64, 938–952 (2007)
de Pedro, M. A., Grunfelder, C. G. & Schwarz, H. Restricted mobility of cell surface proteins in the polar regions of Escherichia coli. J. Bacteriol. 186, 2594–2602 (2004)
Brown, P. J. et al. Polar growth in the alphaproteobacterial order Rhizobiales. Proc. Natl Acad. Sci. USA 109, 1697–1701 (2012)
Curtis, P. D. & Brun, Y. V. Getting in the loop: regulation of development in Caulobacter crescentus. Microbiol. Mol. Biol. Rev. 74, 13–41 (2010)
Ohta, N., Lane, T., Ninfa, E. G., Sommer, J. M. & Newton, A. A histidine protein kinase homologue required for regulation of bacterial cell division and differentiation. Proc. Natl Acad. Sci. USA 89, 10297–10301 (1992)
Mallarino, R. et al. Two developmental modules establish 3D beak-shape variation in Darwin’s finches. Proc. Natl Acad. Sci. USA 108, 4057–4062 (2011)
Loehlin, D. W. & Werren, J. H. Evolution of shape by multiple regulatory changes to a growth gene. Science 335, 943–947 (2012)
Werner, T., Koshikawa, S., Williams, T. M. & Carroll, S. B. Generation of a novel wing colour pattern by the Wingless morphogen. Nature 464, 1143–1148 (2010)
Chan, Y. F. et al. Adaptive evolution of pelvic reduction in sticklebacks by recurrent deletion of a Pitx1 enhancer. Science 327, 302–305 (2010)
Ronshaugen, M., McGinnis, N. & McGinnis, W. Hox protein mutation and macroevolution of the insect body plan. Nature 415, 914–917 (2002)
Darwin, C. On the Origin of Species by Means of Natural Selection, ix + 502 pp. (John Murray, 1859)
Collins, T. J. ImageJ for microscopy. Biotechniques 43, 25–30 (2007)
Ochman, H., Elwyn, S. & Moran, N. A. Calibrating bacterial evolution. Proc. Natl Acad. Sci. USA 96, 12638–12643 (1999)
Poindexter, J. S. Selection for nonbuoyant morphological mutants of Caulobacter crescentus. J. Bacteriol. 135, 1141–1145 (1978)
Wan, Z., Brown, P. J., Elliott, E. N. & Brun, Y. V. The adhesive and cohesive properties of a bacterial polysaccharide adhesin are modulated by a deacetylase. Mol. Microbiol. 88, 486–500 (2013)
Merker, R. I. & Smit, J. Characterization of the adhesive holdfast of marine and freshwater caulobacters. Appl. Environ. Microbiol. 54, 2078–2085 (1988)
Roy, A., Kucukural, A. & Zhang, Y. I-TASSER: a unified platform for automated protein structure and function prediction. Nature Protocols 5, 725–738 (2010)
Huang, X. Q. & Miller, W. A. Time-efficient, linear-space local similarity algorithm. Adv. Appl. Math. 12, 337–357 (1991)
Katoh, K. & Standley, D. M. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol. Biol. Evol. 30, 772–780 (2013)
Waterhouse, A. M., Procter, J. B., Martin, D. M., Clamp, M. & Barton, G. J. Jalview version 2–a multiple sequence alignment editor and analysis workbench. Bioinformatics 25, 1189–1191 (2009)
Tamura, K. et al. MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Mol. Biol. Evol. 28, 2731–2739 (2011)
Williams, K. P., Sobral, B. W. & Dickerman, A. W. A robust species tree for the alphaproteobacteria. J. Bacteriol. 189, 4578–4586 (2007)
JW, T. Bias and confidence in not quite large samples. Ann Math Stats 614 (1958)
Thanbichler, M., Iniesta, A. A. & Shapiro, L. A comprehensive set of plasmids for vanillate- and xylose-inducible gene expression in Caulobacter crescentus. Nucleic Acids Res. 35, e137 (2007)
Horton, R. M. PCR-mediated recombination and mutagenesis. SOEing together tailor-made genes. Mol. Biotechnol. 3, 93–99 (1995)
Vasilyeva, L. V. et al. Asticcacaulis benevestitus sp. nov., a psychrotolerant, dimorphic, prosthecate bacterium from tundra wetland soil. Int. J. Syst. Evol. Microbiol. 56, 2083–2088 (2006)
Syvanen, M. Evolutionary implications of horizontal gene transfer. Annu. Rev. Genet. 46, 341–358 (2012)
We thank members of the Brun laboratory and C. Fuqua for comments on the manuscript. We thank D. Kysela, V. Hughes and V. Silvanose for help in environmental sampling and phylogenetic analysis, L. Zhuo and C. Huang for help in statistical analysis, S. Shaw for advice on quantitative image analysis, M. Hahn, M. Lynch and R. Raff for discussions on evolution, and the Center for Genomics and Bioinformatics at Indiana University for help in sequencing. We thank the Indiana University Light Microscopy Imaging Center for their help with OMX super-resolution microscopy, supported by National Institutes of Health grant S10RR028697-01, and the Indiana Molecular biology Institute electron microscopy facility at Indiana University for their help with electron microscopy. We thank M. Thanbichler, J. Poindexter, P. Caccamo and P. Viollier for providing us with Caulobacter strain and plasmids, J. Poindexter, J. Peterson, J. Lindquist and A. Quinones for help in locating the strain collection of the late Jack Pate from which we obtained some of the A. excentricus and A. biprosthecum strains used in this study, and M. Wortinger, S. Green, E. Quardokus and J. (Wagner) Herman for early work with Asticcacaulis that helped set the stage for this study. This work was supported by National Institutes of Health grant GM051986, National Science Foundation grant MCB0731950 and by a grant from the Indiana University Metabolomics and Cytomics Initiative (METACyt) program, which was financed, in part, by a major endowment from the Lilly Foundation. P.J.B.B. was supported by National Institutes of Health National Research Service Award AI072992.
The authors declare no competing financial interests.
Extended data figures and tables
Extended Data Figure 1 SpmX localization precedes, and is required for, stalk synthesis in Asticcacaulis.
a, Stalk synthesis occurs at the junction between the cell body and stalk in A. biprosthecum and A. excentricus. Cell surface proteins were pulse-labelled with TRSE, excess TRSE was removed and cells were allowed to grow for seven to eight doublings. The cell body and cell body proximal stalk label were diluted by cell elongation, whereas the cell body distal part of the stalk remained stained, indicating that new material is incorporated at the base of the stalk both in A. biprosthecum (left) and in A. excentricus (right). b, Fluorescent d-amino acids track stalk peptidoglycan synthesis to its base. Cells were stained with the fluorescent d-amino acid 7-hydroxycoumarin-amino-d-alanine (HADA) (blue), which labels regions of peptidoglycan synthesis, in A. biprosthecum (left) and A. excentricus (right). c, d, SpmX localization precedes stalk synthesis in Asticcacaulis. Time-lapse microscopy tracks the dynamics of native SpmX–eGFP localization in A. biprosthecum (c) and A. excentricus (d), images taken every 50 (c) and 30 (d) minutes, respectively. Arrows indicate the early localization of SpmX–eGFP in the swarmer compartment of pre-divisional cells. Schematics of dynamic localization patterns are shown on top. e, Structured illumination micrographs of the localization of SpmX–eGFP at different stages of the life cycle in A. biprosthecum (top) and in A. excentricus (bottom). Cells were stained with the red outer membrane protein stain TRSE and SpmX–eGFP was expressed from its native chromosomal locus. SpmX–eGFP localizes at the future position of stalk synthesis in the daughter swarmer-cell compartment of the pre-divisional cell. f, g, SpmX is required for stalk elongation in A. biprosthecum (f) and A. excentricus (g). Cells expressing a xylose-inducible spmX allele were depleted of SpmX and the cell body and stalks were stained with the red outer membrane protein stain TRSE. Cells were then grown in the presence or absence of the xylose inducer. Schematics of the pulse-chase results are shown below each group. h, i, SpmX is required for the initiation of stalk peptidoglycan synthesis. Cells were stained for 5 min with HADA. The blue HADA staining correlates with SpmX–eGFP at the base of the stalks in A. biprosthecum (h) and A. excentricus (i). Dashed arrows indicate where peptidoglycan synthesis co-localizes with the SpmX–eGFP foci at the base of the stalks. Arrows indicate the localization of SpmX–eGFP in the swarmer cell or the swarmer-cell compartment. j, k, HADA staining of the spmX− mutant in A. biprosthecum (j) and A. excentricus (k). HADA staining no longer produces the foci seen at the base of stalks in wild-type strains. All data are representative of at least two biological repetitions analysing at least 100 cells each. Scale bars, 1 µm.
Extended Data Figure 2 DivJ localizes to the base of the stalk, and the effects of SpmX overexpression and phosphate starvation on stalk synthesis.
a, DivJ localizes to the base of the stalk (left) and the divJ− mutant still synthesizes stalks (right) in A. biprosthecum. b, Localization dynamics of DivJ–eGFP tracked by time-lapse microscopy, images taken every 80 min. Note that DivJ–eGFP only localizes in the daughter cell after cytokinesis (arrow). c, SpmX is required to localize DivJ (top) but DivJ is not required to localize SpmX (bottom) in A. biprosthecum. d, Micrographs of wild-type A. excentricus cells overexpressing SpmX–eGFP. e, Structured illumination micrographs of wild-type (left) and spmX− (right) A. excentricus cells growing under low phosphate conditions. Cells (red) are stained with TRSE and the polar holdfast (green) is stained with a fluorescent wheat-germ agglutinin lectin. Phosphate starvation induces polar stalk synthesis in the A. excentricus spmX− mutant, pushing the holdfast to its tip as shown in the right schematic. f, g, Micrographs of wild-type (left, with SpmX–eGFP) and spmX− (right) A. biprosthecum (f) or A. excentricus (g) grown under phosphate starvation. h, Transmission electron micrographs of A. excentricus grown under phosphate starvation. The diffuse structure around the cell body and the stalk is a sheath whose synthesis is induced in response to environmental stress39. All data are representative of at least two biological repetitions analysing at least 100 cells each. Scale bars, 1 µm.
a, b, Expression of SpmX–eGFP in strains used in this study. All fusion proteins were expressed from a xylose-inducible promoter on a replicating plasmid in the presence of 0.05% (w/v) xylose. SpmX–eGFP expression was measured by quantifying the fluorescence intensity of the fusion proteins. Corrected fluorescence is calculated as (integrated fluorescence − integrated background)/area. Measurements for each strain were done in duplicate and at least 500 cells were quantified in each case. Error bars, standard error of the mean. c, Western blot of SpmX–eGFP fusion proteins expressed from either the chromosomal locus or a replicating plasmid in different species probed with anti-GFP antibody. Wild-type strains were used as controls. Note that only SpmXAB–eGFP seems to have a clipping/degradation pattern. Data are representative of three biological repetitions. d, Western blot of various chimaeric/heterologous SpmX–eGFP fusions when expressed in the A. excentricus spmX− mutant, probed by anti-GFP antibody. Note that the relative amounts of expression correlate with the results from quantitative fluorescence analysis (a). The size of SpmX–eGFP fusions ranges from 76 to 120 kDa, and free eGFP is expected to be around 31 kDa. Data are representative of three biological repetitions. e, f, SpmX–mCherry and SpmX–eGFP share the same localization pattern in A. biprosthecum (left, Pearson r = 0.79 ± 0.1) and A. excentricus (right, Pearson r = 0.81 ± 0.08) cells. Both strains are expressing SpmX–eGFP fusions from the chromosomal locus and the SpmX–mCherry fusions from a replicating plasmid. Scale bars, 1 µm. Data are representative of two biological repetitions. g, The elevated cytoplasmic fluorescence in A. biprosthecum may correlate with the clipping/degradation pattern of SpmXAB–eGFP. Data are representative of two biological repetitions analysing at least 100 cells each. Error bars, standard deviation.
Full length or chimaeric SpmX–eGFPs were expressed in wild-type A. biprosthecum and A. excentricus as indicated on the left. Representative pictures are shown. All data are representative of three biological repetitions analysing at least 100 cells each. Scale bars, 1 µm.
Extended Data Figure 5 Overview of the subcellular localization quantification method and characteristics of newly sequenced genomes.
a, Output images provided by the customized ImageJ-based software package. Fluorescent (left) and corresponding phase contrast (right) images are shown. The cell boundary is represented in green or orange according to the side of the cell. The major and the minor axes of the cell are represented in cyan. In each cell, fluorescent foci of SpmX–eGFP (white spot, left panel) were detected and outlined by a yellow circle centred to their respective sub-pixel resolution positions. The dark blue line links the focus coordinates to its relative position on the major axes. b, Schematic representation of the polar coordinate system in which each point is defined by three coordinates: the distance from the pole or from the major axis (r1, r2), the angle relative to the major axis (θ1, θ2) and the distance from the mid-cell (d1, d2). A null angular coordinate means that the focus is localized at the tip of the cell pole, whereas 90° means that the focus is localized on cell sides. The symbol dP represents the distance between the pole (P or P′) and the cell boundary (red dot). F1 and F2 represent subpolar and lateral localized objects with their respective coordinates. c, Variation of the position of discrete points on the cell boundary relative to the pole (P or P′) as determined by the polar coordinates (n = 100). Each point represents the normalized position on the major (rb/dp × cosθb) and minor (rb/dp × sinθb) axis of discrete positions along the cell boundary in the pole region. The shaded region indicates the interquartile range as a measure of the dispersion between the upper and lower quartiles of values observed for discrete values of θb. d, Schematic showing how localization of SpmX–eGFP is quantified by measuring the angle (θ) and the radial distance (r) of each focus using the geometric centre of the pole as the origin of a polar coordinate system within a normalized cell body (see a–c). Red, purple and yellow coloured circles represent SpmX in C. crescentus (CC), A. excentricus (AE) and A. biprosthecum (AB), respectively. P, S and L in parentheses are shorthand to denote the native polar, subpolar and lateral stalk positioning of their respective SpmX proteins. e, Density plot of the absolute values of measured angles (|θ|) of heatmaps in Fig. 2b (same n). Measurements are binned for every 10°. Error bars in the angle profiles denote standard deviation of the sample evaluated by the Jackknifing method (see Methods). f, The general characteristics of five Asticcacaulis genomes. PD denotes the permanent draft status of the genome. g, Differential interference contrast micrographs of the sequenced strains in (f) except for A. benevestitus, which was previously published39. Scale bars, 1 µm. h, Phylogenetic tree inferred from SpmX sequences of different species. The alignment from Extended Data Fig. 8 was used to infer this tree based on the maximum likelihood method. Note that because E. coli and P. aeruginosa do not have spmX orthologues, species from the Hyphomonadaceae family serve as the outgroup (Maricaulis maris and Oceanicaulis alexandrii). We estimated the statistical support for each node by performing 500 bootstrap repetitions. The overall SpmX protein tree topology matches that of the species tree (Fig. 1d). The only exception is that Brevundimonas subvibriodes is placed differently, owing to the difficulty of resolving its phylogeny40. Scale, number of substitutions per position.
a, Microscopic images and heatmaps of A. excentricus (left) or A. biprosthecum (right) expressing SpmX–mCherry fusions from the chromosomal locus. Note that the localization as well as the function in stalk synthesis is identical to that of the respective spmX::spmX-egfp and wild-type strains (Figs 1 and 2). b, c, Microscopic images of the A. excentricus spmX− (left) and A. biprosthecum spmX− (right) strains expressing non-tagged SpmX. Note that SpmXAB(L) induces mostly polar stalks in the A. excentricus spmX− mutant (b) and SpmXAE(S) induces mostly subpolar stalks in the A. biprosthecum spmX− mutant (c). Both SpmXs were able to complement the stalkless phenotype. d, e, Microscopic images and heatmaps of expression of SpmX–mCherry fusions in the A. excentricus spmX− (left) and A. biprosthecum spmX− (right) mutants. The phenotypes are identical to that of the non-tagged SpmX (as well as SpmX–eGFP in Fig. 3). f, Expression of SpmXAE(S)–mCherry (left) and SpmXAE(S) (right) in wild-type A. excentricus both induce synthesis of many stalks, as is the case for SpmX–eGFP (Extended Data Fig. 2). All data are representative of three biological repetitions analysing at least 100 cells each. Scale bars, 1 µm.
Extended Data Figure 7 SpmX is required for stalk positioning in Asticcacaulis and the integrity of SpmX is critical for its function in A. biprosthecum.
a, Holdfast (red) localizes to the pole both in A. biprosthecum (left) and in A. excentricus (right) spmX− mutants. b, Dual-labelling images of SpmX–eGFP (green) and holdfast (red) in the A. biprosthecum spmX− mutant complemented by SpmXAE(S) (left) and the A. excentricus spmX− mutant complemented by SpmXAB(L) (right). c, SpmXAE(S) localizes and initiates stalk synthesis at the subpolar position in the A. biprosthecum spmX− mutant (left). SpmXAB(L) localizes to polar or subpolar locations before triggering either polar and subpolar stalk synthesis in the A. excentricus spmX− mutant (right). We noticed that SpmX–eGFP sometimes localizes in the stalk, probably because of stalk outgrowth from its base because it should not be able to diffuse to positions in already synthesized stalks, where diffusion is constrained by cross-bands9 (Fig. 1a). d, From top to bottom: 1, SpmX–mCherry (825 amino acids + mCherry, 121 kDa) localizes to the base of stalks. This fusion protein was expressed from its native chromosomal locus (spmX::spmX-mcherry); 2–4, different truncated alleles of SpmX fail to complement the stalkless phenotype of the A. biprosthecum spmX− mutant. Both C-ter− SpmX–mCherry (1–150 amino acids + mCherry, 45 kDa) and TM− SpmX–mCherry (1–750 amino acids + mCherry, 111 kDa) were expressed from the native spmX chromosomal locus replacing the natural allele. N-ter− SpmX–mCherry (150–825 amino acids + mCherry, 104 kDa) was expressed from a replicating plasmid (YB7129) in the A. biprosthecum spmX− mutant. e, Western blot of truncated SpmX–mCherry fusions detected with anti-mCherry polyclonal antibodies. Data are representative of two biological repetitions. Micrographs are representative of three biological repetitions analysing at least 100 cells each. Scale bars, 1 µm.
Extended Data Figure 8 Alignment of SpmX sequences from species/strains used in this study (as shown in Fig. 1d).
Sequences were aligned in Jalview (http://www.jalview.org/). Coloured residues denote conservation across species. Vertical dashed lines denote the boundaries of the N-terminal muramidase domain (black) and the two transmembrane domains (red), respectively. Chimaeras were constructed using the end of the muramidase domains as the boundary between N- and C-terminal moieties (Fig. 3). Red arrows indicate the three main species used in this study: A. biprosthecum, A. excentricus and C. crescentus, respectively. From top to bottom, the order of the sequences reflects the tree topology as deduced from Fig. 1d. The coloured vertical lines next to the names of the species/strains denote their respective stalk positioning as shown in Fig. 1d (yellow, bilateral; purple, subpolar; red, polar). The numbers next to the names of the species/strains indicate the lengths of their respective SpmX proteins.
Extended Data Figure 9 Expression of SpmXAB(L)–eGFP or SpmXAE(S)–eGFP in C. crescentus does not induce ectopic stalk synthesis.
a, b, The localization of SpmXCC(P)–eGFP/SpmXCC(P)–mCherry and the phenotypes of the strains are identical. c–e, Asticcacaulis SpmX–eGFP variants localize mostly to the pole in wild-type C. crescentus (left) and C. crescentus spmX− (right) strains. No ectopic stalks were observed, indicating that the alternative targets are lacking in C. crescentus. It should also be noted that the expression of variants of SpmXAE–eGFP can cause cell filamentation. f, SpmXCC–eGFP localizes to the pole in the A. biprosthecum spmX− strain and cannot rescue stalk synthesis. All data are representative of three biological repetitions analysing at least 100 cells each. Scale bars, 1µm.
About this article
Cite this article
Jiang, C., Brown, P., Ducret, A. et al. Sequential evolution of bacterial morphology by co-option of a developmental regulator. Nature 506, 489–493 (2014). https://doi.org/10.1038/nature12900
Nature Communications (2021)
CrvA and CrvB form a curvature-inducing module sufficient to induce cell-shape complexity in Gram-negative bacteria
Nature Microbiology (2021)
Nature Communications (2021)
Cryo-EM structure of the bacteria-killing type IV secretion system core complex from Xanthomonas citri
Nature Microbiology (2018)
Rapid, precise quantification of bacterial cellular dimensions across a genomic-scale knockout library
BMC Biology (2017)