Optimized CRISPR Interference System for Investigating Pseudomonas alloputida Genes Involved in Rhizosphere Microbiome Assembly

Abstract

Pseudomonas alloputida KT2440 (formerly P. putida) has become both a well-known chassis organism for synthetic biology and a model organism for rhizosphere colonization. Here, we describe a CRISPR interference (CRISPRi) system in KT2440 for exploring microbe–microbe interactions in the rhizosphere and for use in industrial systems. Our CRISPRi system features three different promoter systems (XylS/P_m, LacI/P_lac, and AraC/P_BAD) and a dCas9 codon-optimized for Pseudomonads, all located on a mini-Tn7-based transposon that inserts into a neutral site in the genome. It also includes a suite of pSEVA-derived sgRNA expression vectors, where the expression is driven by synthetic promoters varying in strength. We compare the three promoter systems in terms of how well they can precisely modulate gene expression, and we discuss the impact of environmental factors, such as media choice, on the success of CRISPRi. We demonstrate that CRISPRi is functional in bacteria colonizing the rhizosphere, with repression of essential genes leading to a 10–100-fold reduction in P. alloputida cells per root. Finally, we show that CRISPRi can be used to modulate microbe–microbe interactions. When the gene pvdH is repressed and P. alloputida is unable to produce pyoverdine, it loses its ability to inhibit other microbes in vitro. Moreover, our design is amendable for future CRISPRi-seq studies and in multispecies microbial communities, with the different promoter systems providing a means to control the level of gene expression in many different environments.

Introduction

Pseudomonas alloputida KT2440 (formerly P. putida)^1,2 is an important model organism for many disciplines, including synthetic biology, plant–microbiome interactions, and bioremediation due to its versatile metabolism, tolerance to toxic compounds, and strong biofilm forming capabilities.³⁻⁵ KT2440 is a competitive maize rhizosphere colonist and exhibits chemotaxis toward maize root exudates.⁶⁻⁸ It is hypothesized that the secretion of chemicals helps plants shape their rhizosphere microbiome by recruiting beneficial bacteria, like KT2440, to developing roots.⁸ Other microbes can respond to and utilize these exudates as well, or they may find them toxic and inhibitory, so the ability of bacteria to successfully colonize the root is an effect of both the plant-secreted chemicals and the competition or cooperation between microbes in the community. Pseudomonads are known for their competitive abilities through their production of antibiotics,⁹ secretion of toxins via a type 6 secretion systems,¹⁰⁻¹² and production of siderophores such as pyoverdine to scavenge iron in iron-limited conditions.¹³⁻¹⁵ While we know that Pseudomonads and other microbes form stable communities on plant roots, we have a poor understanding of what traits are important for their rhizosphere colonization or their ability to compete with other microbes for limited resources. Studies with soil- and plant-associated Pseudomonads have used in vitro expression technology (IVET) to identify promoters that are activated in the maize rhizosphere,^16,17 Tn-Seq to identify genes required for root colonization or competitive behaviors,¹⁸⁻²² and transcriptomics to identify genes expressed in the rhizosphere.^7,23−26 While powerful, these approaches each have limitations, including finding only genes that are regulated or highly expressed in the rhizosphere, being unable to detect genes that are essential, or not permitting use of conditional mutant phenotypes. Further, these studies have largely been done in pure culture, preventing the detection of genes that are differently regulated or essential when in the presence of other organisms on the root.

Increasingly, studies are using CRISPR interference sequencing (CRISPRi-seq) in place of Tn-Seq,²⁷⁻²⁹ and several CRISPRi systems have been described for P. alloputida KT2440.³⁰⁻³⁵ These systems allow for modulation of gene expression by targeting a deactivated Cas9 (dCas9) to a specific location in the genome using a single guide RNA (sgRNA), thereby repressing the gene of interest as dCas9 binds (Figure 1A,B).³⁶ While each system has its merits, they are not suitable for use in CRISPRi-seq studies to study microbial interactions in the rhizosphere. One major concern is the use of non-native regulatory systems (such as RhaS/P_Rha, AraC/P_BAD, or LacI/P_lac, induced by rhamnose, arabinose, and IPTG, respectively), reducing their ability to be tightly controlled.^{10,30−35,37,38} With a lack of control can come toxicity within cells: it has been well characterized that high levels of dCas9 can nonspecifically bind to “NGG” PAM sites throughout the genome, thereby potentially blocking gene expression and reducing cell fitness.^39,40 Another concern is the use of plasmid-based systems (rather than integrated into the genome), which can lead to copy-number variation or loss of the system in the absence of selection.^31,35 Others have low-throughput means of cloning in sgRNAs.^31,34 Finally, for exploring plant–microbe and microbe–microbe interactions in the rhizosphere, the high inducer concentrations required for many of these systems could significantly alter the microbiome’s composition and metabolic properties, particularly if the plant or members of the rhizosphere microbial community produce or consume sugars such as arabinose or rhamnose.

Figure 1.

Schematic illustrations of CRISPRi, plasmids, and sgRNA cloning methods. (A) dCas9 is under the control of an inducible promoter on a mini-Tn7 integrated into the genome, while sgRNAs are constitutively expressed from a plasmid. (B) CRISPRi functions in one of two ways: dCas9 binds to a promoter, blocking RNA polymerase from initiating transcription, or it binds to a gene in the open reading frame, blocking RNA polymerase from elongating its mRNA transcript. (C) Plasmid map of pMRS-XylS-dCas9. This plasmid is identical to pMRS-LacI-dCas9 and pMRS-AraC-dCas9, aside from the promoter region between PacI and AvrII sites: LacI/P_lac and AraC/P_BAD, respectively. This plasmid is inserted into the chromosome of KT2440 between the Tn7-L and Tn7-R sites (yellow shading). Following insertion, the Gm^R-marker can be removed at flanking FRT sites using a flippase. (D) Plasmid map of pSpyB-GFP and sgRNA cloning scheme. Multiple promoters are available to modify sgRNA expression levels: P_EM7 (45% relative expression), P_BG35 (25%), P_BG42 (100%), and P_BG28 (3%). pSpyB-GFP can be digested with BsaI and separated via gel extraction. Twenty bp sgRNA oligos with 4 bp overhangs are annealed and ligated into pSpy (Golden Gate cloning). Following the transformation, screening is streamlined by the presence of GFP in the uncut vector. Plasmid map images created with BioRender.com.

As most of the current CRISPRi systems for KT2440 were designed for the bioproduction of chemicals,^30,32,34 they did not satisfy design requirements for our purposes. We aimed to satisfy the following conditions: dCas9 expression must be very low in the absence of an inducer, consistent between cells in the same population, and tunable when different concentrations of inducer are added. Importantly, the cloning of sgRNAs should be efficient and have a high throughput, and the CRISPRi system must have minimal effects on fitness and be maintained in the absence of selection. Some of the currently available CRISPRi systems have features that we chose to emulate in our design. The XylS/P_m and CymR/CuO promoter systems have been used to develop CRISPRi systems in Pseudomonas and Streptomyces, and each shows low levels of leakiness in the absence of an inducer.^31,41 These and other studies show that the use of aromatic compounds as inducers results in high levels of control over gene expression, and because inducers easily pass through cell membranes, there is no need to supply a transporter to ensure the inducer gets inside.^41,42 Moreover, lower inducer concentrations in the μM rather than mM range are sufficient.⁴³ The presence of a single copy of the dCas9 expression system on a mini-Tn7 transposon permits chromosomal integration at a neutral, intergenic site, thereby avoiding copy-number variation when plasmid-borne.^35,44 Peters et al. showed that a human codon-optimized Streptococcus pyogenes dCas9 functions better than wild-type dCas9 in P. aeruginosa, and Stolle et al. had success with a P. aeruginosa codon-optimized dCas9.^10,44 Finally, while using inducible sgRNA expression systems can lead to up to 300-fold repression of genes,⁴⁵ and similar levels of repression can be reached using an inducible dCas9 and constitutively expressed sgRNAs.³⁵ There are clear benefits to inducible dCas9 systems, given leakiness and potential growth defects observed when constitutively expressed.^40,46

Here, we describe a CRISPRi system designed for use in KT2440, using a Pseudomonas codon-optimized dCas9 whose expression is controlled by various regulatory systems that can be integrated into the chromosome with plasmid-based sgRNA expression systems varying in promoter strength. We compare the AraC/P_BAD and native XylS/P_m transcriptional activator systems to the commonly used LacI/P_lac classic repressor, showing that XylS/P_m often outperforms the LacI/P_lac and AraC/P_BAD systems. We show the benefits of each of these systems, including when they each exhibit low baseline leakiness and high levels of tunability, allowing for precise modulation of gene expression. We also show that maximal repression of gene expression can be attained with relatively low levels of sgRNA expression based on a comparison of promoters driving sgRNA levels that vary in strength. Importantly, we demonstrate for the first time both the use of CRISPRi to modulate microbe–microbe competitive interactions and in evaluating the role of a gene in rhizosphere colonization.

Results and Discussion

Construction of the CRISPRi System

To create a CRISPRi system designed for P. alloputida KT2440, we selected P. aeruginosa codon-optimized S. pyogenes dCas9,¹⁰ which has an “NGG” PAM site that is highly frequent in GC-rich Pseudomonads.⁴⁷ We chose the mini-Tn7 transposon pTn7-M⁴⁸ system as a chassis for our dCas9 expression system, containing a Km^R gene on the plasmid and a Gm^R gene on the transposon, which was modified as described briefly here. First, we added FRT sites flanking the Gm^R gene to facilitate creation of marker-less insertions.⁴⁹ Second, the translational enhancer BCD2, which can enhance expression and reduce transcriptional noise in Pseudomonas sp.,⁴⁸ was inserted into the MCS, followed by dCas9, creating pMRS-dCas9. Lastly, the desired promoter/regulatory systems were then inserted in PacI/AvrII sites to control dCas9 expression. This architecture provides flexibility in selecting which regulatory system will control dCas9 expression. As proof of concept, we selected AraC/P_BAD, LacI/P_lac, and XylS/P_m (Figure 1C and Table S1), which respond to arabinose, IPTG, and 3-methylbenzoate (3-MBZ), respectively.^43,44 Importantly, the pTn7-M transcriptional terminators were retained to limit read-through transcription. Simultaneously, vectors containing multimeric superfolder green fluorescent protein (msfgfp) instead of dCas9 were created with the same promoters, providing a means to quantify the tunability of each expression system with different inducer concentrations (Table S1).

Rather than integrating the sgRNA expression system into the mini-Tn7-based dCas9 system,^31,34 we selected pSEVA231 as a chassis, given the flexibility and stability of the Standard European Vector Architecture (SEVA) plasmids in P. alloputida.^50,51 pSEVA231 is a medium copy-number plasmid with the broad-host-range pBBR1 origin of replication, which is stable in many bacteria, and a Km^R-marker, as similar Gm^R variants of this plasmid had fitness costs to KT2440 in the absence of selection.⁵² The sgRNA is composed of the tracrRNA (transactivating CRISPR RNA), which binds to dCas9, and the crRNA (CRISPR RNA), which provides sequence specificity for directing dCas9 to specific genomic regions. We began by inserting a gBlock containing two BsaI cloning sites and the S. pyogenes (Spy) tracrRNA sequence, creating the plasmid pSpy.^36,52−54 We then introduced msfgfp between the BsaI sites, creating pSpy-GFP. Next, four synthetic constitutive promoters with different relative activities⁴⁸ were inserted at PacI/AvrII sites upstream of the sgRNA insertion site. Activities of these promoters are relative to the maximum reported for KT2440: promoters P_EM7 (pSpyA) at 45%, P_BG35 (pSpyB) at 25%, P_BG42 (pSpyC) at 100%, and P_BG28 (pSpyD) at 3% relative activity.⁴⁸ Finally, the BsaI sites facilitate Golden Gate cloning of sgRNAs with unique sticky ends, enabling directional cloning of sgRNAs.⁵² When the vector is digested with BsaI, the GFP gene is cut out, and successful clones will lack fluorescence, thereby streamlining the screening process (Figure 1D).

Impact of the dCas9 System on Growth

In this study, we used the Rf^R-derivative of P. alloputida KT2440.⁴⁴ With any of the dCas9 systems in the absence of an inducer, KT2440 exhibits slower growth (Supporting Figure S1). A common issue with CRISPRi is that dCas9 can be toxic to bacteria, particularly when highly expressed.^39,40 Presumably, this is due to dCas9 binding nonspecifically to “NGG” PAM sites, particularly in the absence of a sgRNA.³⁹ To resolve this issue, we designed a negative control where the sgRNA targets the Tn7-R region of mini-Tn7, named Tn7sg, theoretically tethering dCas9 to a neutral location on the chromosome and preventing interference with transcription in other regions of the genome. We performed growth curves to assess the fitness costs of the sgRNA vector. In uninduced conditions, all strains have similar growth patterns, and the presence of the sgRNA vectors does not cause additional growth defects. However, when dCas9 is expressed in each of the three promoter systems, strains harboring pSpyB-empty grow more slowly than either strains without a plasmid or with pSpyB-Tn7sg (Figure 2). We believe that this is due to the expression of the tracrRNA, which may be able to form a complex with dCas9. In the absence of crRNA, this untargeted dCas9 may be free to bind anywhere, potentially repressing genes contributing to growth. However, these fitness costs can be overcome with the use of crRNA controls that direct dCas9 to a neutral site, such as the mini-Tn7 used here. Thus, we chose to use Tn7sg as a negative control sgRNA in the rest of the experiments reported here.

Figure 2.

sgRNA control prevents dCas9 toxicity at high inducer concentrations. Growth curves of the XylS/P_m (A/D), LacI/P_lac (B/E), and AraC/P_BAD (C/F) expression systems in 0 and high levels of inducer. Blue = KT2440 + dCas9, red = KT2440 + dCas9 + pSpyB-empty, and green = KT2440 + dCas9 + pSpyB-Tn7sg. Data are the mean ± standard error of 3 replications.

Validation of the System in P. alloputida KT2440

To validate our system, we decided to assess the efficacy by targeting expression of an essential and a nonessential gene in KT2440. We selected the essential gene ftsZ, which forms the contractile ring during cell division, as its short-term repression results in an elongated cell phenotype, while long-term repression leads to cell death.³⁵ We chose pvdH as our nonessential gene target, since it is required for production of fluorescent pyoverdines, a readily detectable and measurable phenotype.⁵⁵ We designed 3 pvdH sgRNAs targeting the nontemplate strand, two in the promoter region and one in the open reading frame (ORF), and 5 ftsZ sgRNAs, all targeting the template strand of the ORF. Here, sgRNAs are named by the gene they are targeting and the number of base pairs the sgRNA is away from the beginning of the ORF: positive numbers target the ORF, and negative numbers target the promoter.

Using the XylS/P_m-dCas9 system, when comparing the phenotypes of the various sgRNAs targeting ftsZ or pvdH, we observed that the location of sgRNA binding influenced the extent of gene repression. Three ftsZ sgRNAs (+14, +30, and +33) were functional in repressing cell growth, while two (+92 and +162) were not. Many studies show that targeting the template strand of a gene only can repress gene expression when the promoter is targeted, while repression of the nontemplate strand functions well both in the promoter and open reading frame.^27,32,36,56 We believe the reason why our ftsZ sgRNAs are functional despite targeting the template strand is because they are close enough to the promoter to block transcription initiation. One pvdH sgRNA (−51) did not alter pyoverdine production, while two others (−77 and +6) were both able to repress pyoverdine production, though pvdH_–77 had higher levels of repression (Figure S2B). Consequently, we chose ftsZ₊₃₃ and pvdH_–77 sgRNAs for additional experiments exploring the limits of our CRISPRi system. For each regulatory/promoter system, when dCas9 is induced in cells with the ftsZ₊₃₃ sgRNA, there was significant cell death and cell elongation, and cells with the pvdH_–77 sgRNA had little to no visible pyoverdine production (Figure 3). In contrast, when CRISPRi was induced in strains harboring the negative control sgRNA Tn7sg, there were no changes in survival, pyoverdine production, or cell size (Figures 3 and S3).

Figure 3.

Validation of the CRISPRi system using ftsZ and pvdH sgRNAs. (A) Targeting ftsZ leads to a reduction in survival. The control sgRNA Tn7sg does not lead to cell death when induced. (B) Targeting pvdH leads to a loss of fluorescence on King’s B medium, indicating a lack of fluorescent pyoverdine production. (C) Targeting ftsZ for short periods of time in liquid culture leads to elongated cells. Scale bar: 20 μm.

Tunability of the CRISPRi System

Another common issue is leakiness in dCas9 expression in the absence of an inducer, which can produce confounding results^33,35 and, in worst-case scenarios, can lead to outcomes such as a 50% reduction in gene expression in uninduced conditions.⁴⁶ To assess whether our systems are leaky and to assess their response to increasing inducer concentrations, we examined the repression of pyoverdine production using pSpyB-pvdH_–77 for the XylS/P_m-, LacI/P_lac-, and AraC/P_BAD-dCas9 systems. ΔpvdH and pSpyB-Tn7sg control strains were included. Pyoverdine production was measured using fluorescence relative to culture biomass in a low iron-availability CAA medium. While all three systems showed a titratable response to increasing inducer concentrations (Figure 4), each exhibited varying extents of leakiness in the absence of an inducer. In the LacI/P_lac system, in the absence of an inducer, pyoverdine production by the pvdH_–77 strain was not significantly lower (p = 0.8761) than the amount produced by the Tn7sg strain. The AraC/P_BAD and XylS/P_m systems were both significantly (p < 0.0021) leaky based on the decreases in pyoverdine production by the pvdH_–77 compared to the Tn7sg strain of 32 and 16%, respectively (Figure 4C). Importantly, all systems exhibited tunability with increasing inducer concentration, albeit the LacI/P_lac and AraC/P_BAD-based systems requiring mM concentrations while the XylS/P_m-based system requiring μM concentrations to repress pyoverdine production to levels observed for the ΔpvdH mutant. A similar result, demonstrating the benefit of CRISPRi in tunably repressing even essential genes, was observed with ftsZ₊₃₃ sgRNA (Figure S2C).

Figure 4.

Evaluation of inducer concentration on the repression of pyoverdine production. Repression of pvdH with pSpyB-pvdH-77 in (A) XylS/P_m, (B) LacI/P_lac, and (C) AraC/P_BAD-based systems. Data is presented as normalized pyoverdine production, which is pyoverdine fluorescence (excitation 400 nm, emission 460 nm) divided by biomass (optical density at 600 nm) that is then normalized to pSpyB-Tn7sg with 0 μM inducer. Separate normalizations were performed for each regulatory system. Plots reflect results from one representative experiment out of 2–3 separate experiments. Pink = pSpyB-Tn7sg, purple = pSpyB-pvdH-77, and green = KT2440 ΔpvdH. Data are the mean ± standard error of 3 replications per experiment.

We also assessed how the extent of sgRNA expression influenced the sensitivity of our CRISPRi system. Our approach was to compare the repression of pyoverdine production by modulating the extent of sgRNA expression in a strain in which dCas9 expression is controlled by the XylS/P_m system. All plasmids contained the same sgRNA (pvdH_–77), but each differed in the strength of the constitutive promoters driving its expression, ranging in strength from 3% (pSpyD) to 100% (pSpyC) relative expression, based on the maximum reported previously for KT2440.⁴⁸ Interestingly, there was little difference in repressing pyoverdine production between the P_BG35 promoter (pSpyB) at 25% from the P_BG42 promoter (pSpyC) with the highest level of expression (100% relative activity) (Figure 5). Only promoter P_BG28 (pSpyD) at 3% relative activity provided an insufficient amount of sgRNA for effective repression of pyoverdine production. Our findings indicate that with the XylS/P_m-mediated dCas9 expression system, sgRNA abundance typically does not limit CRISPR interference when using moderate (pSpyB) to strong (pSpyC) promoters. This is consistent with other studies reporting that when dCas9 expression is not limiting, varying sgRNA expression levels can lead to different levels of CRISPRi.^45,46

Figure 5.

sgRNA expression level has little effect on CRISPRi efficacy. XylS/P_m-dCas9 with pSpyA, B, C, or D expressing pvdH_–77 sgRNA with increasing concentrations of 3-MBZ showed an increased level of repression of pyoverdine production. Data shown is in units of relative fluorescence: pyoverdine fluorescence (excitation 400 nm, emission 460 nm) divided by biomass (optical density at 600 nm), normalized to the 0 μM 3-MBZ measurement of each strain. Red = pSpyA-pvdH_–77, blue = pSpyB-pvdH_–77, yellow = pSpyC-pvdH_–77, and orange = pSpyD-pvdH_–77. Data are the mean ± standard error of 3 replications.

Consistency of dCas9 Expression in Our CRISPRi Systems

Given the results above indicate dCas9 expression may be limiting CRISPRi in our system, we wanted to compare how the different regulatory systems controlled gene expression. Our approach was to measure GFP expression of KT2440 in CTYE media containing the LacI/P_lac-, AraC/P_BAD-, and XylS/P_m regulatory systems, controlling GFP expression rather than dCas9 expression. We include constitutive P_BG35-GFP and promoterless controls to assess how varying inducer concentrations could affect GFP production. GFP expression increased to high levels in both the XylS/P_m- and AraC/P_BAD-based systems as inducer concentrations increased, while the LacI/P_lac-based system exhibited little increase in GFP production with increasing inducer concentration (Figure 6A–C), despite the use of higher concentrations of IPTG than above CRISPRi assays. The experiments with the LacI/P_lac-based system were repeated after recloning and retransforming KT2440 with independent constructs multiple times, and in each instance, the same result was obtained.

Figure 6.

Assessment of inducer concentrations on XylS/P_m, LacI/P_lac, and AraC/P_BAD regulatory systems, controlling GFP production. Cultures were exposed to inducer and incubated for 6 h before assessing GFP production. pMRS-BG35-GFP and pMRS-GFP were used as constitutive and promoterless-GFP controls, respectively. (A–C) GFP production was assessed by fluorimetry. Green = constitutive GFP, pink = inducible GFP, and purple = promoterless GFP. Data are the mean ± standard error of 3 replications. Data is expressed as normalized fluorescence, which is relative GFP fluorescence (excitation 485 nm, emission 525 nm) divided by biomass (optical density at 600 nm) that is then normalized to constitutive GFP with 0 μM inducer. (D–F) GFP production was assessed in individual cells by flow cytometry. The dashed line at 100 indicates the cutoff value for GFP expression.

Next, we wanted to investigate whether all cells in a population responded similarly to an inducer as the concentration increased or if the population exhibited an “all-or-none” response. This response is commonly observed in both the LacI/P_lac- and AraC/P_BAD-systems: at subsaturating inducer concentrations, cells are either very highly or minimally induced, and the proportion of which that respond increases with higher inducer concentrations.^57,58 Using the same GFP reporter systems, we examined the individual cell responses by flow cytometry. We set the cutoff value for GFP expression at 100, as 98% of promoterless-GFP cells lie below, and 95% of constitutive GFP cells lie above this value. Both the XylS/P_m- and AraC/P_BAD-GFP showed a tunable response to increasing inducer concentration, achieving levels that exceeded the constitutive control (P_BG35-GFP), while the LacI/P_lac-GFP system had a far lower level of expression per cell at the highest inducer concentration examined (Figure 6D–F). The XylS/P_m system exhibited a broader distribution of cell responses to all inducer concentrations than either the LacI/P_lac- or AraC/P_BAD-systems, suggesting that there could be greater heterogeneity in dCas9 expression in the XylS/P_m system. At the highest inducer concentrations, there was a small population (2.8 to 5.8%) of the XylS/P_m system that was uninduced, although this was not statistically different from the P_BG35-GFP control (4.9%, p = 1.0000). Neither the LacI/P_lac- or AraC/P_BAD-system showed the expected “all-or-none” response to the inducer, which could be partially explained by the lack of the AraE arabinose importer in the AraC/P_BAD-system.⁵⁹ While the LacI/P_lac-, AraC/P_BAD-, and promoterless-GFP systems all had very little GFP expression in the absence of an inducer (0.7–2%), the XylS/P_m system had 14% of cells expressing GFP when uninduced, a significant increase (p < 0.0001).

We also investigated how similarly individual cells responded to induction of CRISPRi with the three systems when cells were targeted with ftsZ₊₃₃ in a liquid culture in LB. CRISPRi was induced for 3 h; then, cell lengths were measured using microscopy. The average length of cells in uninduced conditions was 2.5 μm, so we set the length of “elongated” cells at 5 μm to account for cells that are in the process of dividing. In the absence of an inducer, there was evidence of leakiness, with 9% of AraC/P_BAD cells and 3% of XylS/P_m cells being elongated, compared to only 1% for LacI/P_lac (p ≤ 0.0015, Figure 7A). Following induction, only 38% of LacI/P_lac cells were elongated, compared to 64% of AraC/P_BAD and 67% of XylS/P_m, again demonstrating that the LacI/P_lac system is significantly less responsive than the other two (p < 0.0001, Figure 7B).

Figure 7.

Evaluation of the heterogeneity of the inhibition of cell division within a population of cells. The length of cells in the LacI/P_lac, AraC/P_BAD, and XylS/P_m-based systems with the ftsZ₊₃₃ sgRNA in the (A) absence and (B) presence of an inducer. Inducer concentrations were 1 mM IPTG, 66 mM arabinose, and 200 μM 3-MBZ, respectively. The dotted line at 5 μm represents two times the average length of cells under the uninduced conditions. Data are the compilation of 3 replications.

CRISPRi Functions in the Rhizosphere

We believe that CRISPRi and CRISPRi-seq technologies should soon be able to add to the findings that IVET, Tn-Seq, and transcriptome studies have made in identifying genes contributing to rhizosphere colonization and microbe–microbe interactions.^{7,16,18−26,60} Here, we illustrate that our CRISPRi system is functional in the rhizosphere of maize. For this proof of concept, we focus only on the AraC/P_BAD and XylS/P_m systems, given the different types of inducers (a sugar versus an aromatic compound). We selected repression of ftsZ (ftsZ₊₃₃ sgRNA) since we know that ftsZ repression leads to cell death (Figure 3). Moreover, we reasoned that the loss of viability would be reflected in decreased rhizosphere colonization (survival) by KT2440 regardless of when dCas9 expression is induced. Our approach included inoculating surface-sterilized, pregerminated seedlings with bacteria prior to growing plants in microcosms (Figure 8A), as described in the Materials and Methods section. The inducer was added at the time of planting, or for a subset of plants, 1 day after planting, and the extent of root colonization by KT2440 was ascertained by dilution plating over plant development. We also included strains with the Tn7sg sgRNA as a control for assessing rhizosphere colonization potential when CRISPRi targets a neutral, nonessential locus.

Figure 8.

CRISPRi in the rhizosphere. (A) Diagram of the plant growth system. (B–C) Survival of bacteria in the rhizosphere over a 4-day period in the presence or absence of (B) 33 mM arabinose or (C) 500 μM 3-MBZ. Bacteria harbored either AraC/P_BAD-dCas9 or XylS/P_m-dCas9 and pSpyB-ftsZ₊₃₃ or pSpyB-Tn7sg. Data are the mean ± standard error of 4–7 replications. Diagrams created with BioRender.com.

In the AraC/P_BAD-ftz₊₃₃ treatments, when exposed to arabinose at the time of planting, 73-fold fewer cells colonized the roots (p < 0.0001) 1 day post-planting compared to the no-arabinose control (Figure 8B). Additionally, when plants were watered with arabinose 1 day post-planting, we saw 11-fold fewer CFUs per root compared to plants without inducer (p = 0.0861). However, by day 4, rhizosphere colonization in arabinose-treated plants increased to levels comparable to those of plants without arabinose. Similarly, in the XylS/P_m system, when cells were exposed to 3-MBZ at the time of planting, 6-fold fewer cells colonized the roots 1 day after planting compared to the no 3-MBZ control (p = 0.0005, Figure 8C). Moreover, by 4 days after planting, rhizosphere colonization did not recover to the no-inducer control levels (15-fold fewer cells, p < 0.0001), which was not observed in the arabinose system (Figure 8B) since population sizes of the no-inducer and inducer treatments at day 4 were comparable. In contrast, unlike in the arabinose system, when plants were watered with 3-MBZ 1 day post-planting, we did not observe a decrease in rhizosphere competence at day 2 or later (p ≥ 0.3263, Figure 8C). Interestingly, rhizosphere colonization by AraC/P_BAD-Tn7sg was consistently greater than AraC/P_BAD-ftsZ₊₃₃ without arabinose (p = 0.0006, Figure 8B), indicating that the arabinose-inducible system is leaky in the absence of inducer, and/or the arabinose secreted by maize roots⁶¹ leads to low-level induction. This finding was not observed in the XylS/P_m system.

In the AraC/P_BAD-ftsZ₊₃₃ and XylS/P_m-ftsZ₊₃₃ treatments with an inducer throughout the experiment, there was an increase in CFU 4 days after planting after an initial population decline (relative to no-inducer). This suggested that either CRISPRi was transient or that we had selected for variants that had lost CRISPRi capabilities. To assess this possibility, we collected Km^R colonies (reflecting retention of the sgRNA plasmid) from roots at 4 days post-planting with and without inducer to determine whether the isolates retained CRISPRi functionality when re-exposed to inducer in a survival assay. In the AraC/P_BAD system, 92% of the isolates from plants without arabinose and only 8% of the isolates from plants treated with arabinose retained CRISPRi functionality, a significant change (p < 0.0001). In the XylS/P_m system, 75% of the isolates recovered from plants without 3-MBZ, and 100% of isolates from plants with 3-MBZ retained the CRISPRi functionality, which was not a significant difference (p = 0.2174). This regrowth and colonization of plant roots over time by isolates in the AraC/P_BAD system watered with arabinose reflects the strong selective advantage of cells that have mutations allowing them to escape ftsZ repression by CRISPRi, an outcome observed in other studies.⁶⁰ While providing a plausible explanation for our observations with the AraC/P_BAD system, it does not sufficiently explain the opposite outcome in the XylS/P_m system. It is conceivable that the concentration of the 3-MBZ inducer was lower than anticipated due to sorption onto surfaces or volatilization, leading to lower levels of dCas9 expression. As a consequence, the fitness costs of expressing dCas9 were lower in the XylS/P_m than the AraC/P_BAD system, thereby reducing selective advantages to cells that lost dCas9 functionality. Future studies should explore mechanisms contributing to the loss of CRISPRi functionality, how this outcome can be avoided, and why there are differences between the two regulatory systems.

CRISPRi can Modulate Microbe–Microbe Interactions

Pseudomonas species are known for their competitive behaviors, whether it is due to production of antibiotics,⁹ type 6 secretion systems,¹⁰⁻¹² or production of pyoverdine to scavenge iron under iron-limited conditions.¹³⁻¹⁵ In the interaction assay modeled after Lozano et al.,⁶² we spread maize rhizosphere and endosphere isolates onto solid King’s B medium to create lawns and then spotted aliquots of KT2440 strains on top of them. King’s B medium is a high-carbon medium with relatively low iron, causing Pseudomonads to produce large quantities of fluorescent siderophores (pyoverdines). We assessed the development of zones of inhibition over 4 days in the presence of exogenous iron and in media amended with 3-MBZ to induce the XylS/P_m-controlled CRISPRi system. Due to the ability of some isolates to utilize arabinose, we reasoned that 3-MBZ was a better inducer of CRISPRi since the concentrations were too low to support bacterial growth. On unamended media, zones of inhibition were observed around all pyoverdine-producing strains, and both this inhibition and the production of pyoverdine were relieved with the addition of 50 μM FeCl₃ (Figure 9). Repression of pyoverdine production by CRISPRi in the 3-MBZ amended plates relieved growth inhibition and pyoverdine production, as illustrated for maize endosphere isolate Lysobacter sp. E-23 (Figure 9). This was consistent with 8 isolates representing 3 phyla and 7 families, indicating that in these conditions, KT2440’s iron-dependent inhibition is caused by production of pyoverdine. These results suggest that CRISPRi is an effective tool for identifying genes contributing to KT2440–microbe interactions.

Figure 9.

CRISPRi can be used to modulate microbe–microbe interactions. CRISPRi of pvdH expression leads to loss of KT2440’s inhibitory activity. KT2440 cultures were spotted on the lawns of a maize isolate, Lysobacter sp. E-23, on King’s B medium containing 50 μM FeCl₃ or 1 mM 3-MBZ. Images were taken 48 h post-plating. Zones surrounding the KT2440 spots indicate inhibition of growth (white light), which aligns with the fluorescence of pyoverdine under UV light. Key indicates strain placement on plates: wild-type KT2440, ΔpvdH, XylS/P_m-dCas9 + pSpyB-Tn7sg, and XylS/P_m-dCas9 + pSpyB-pvdH_–77.

Conclusions

In this study, we designed a CRISPRi system optimized for use in P. alloputida KT2440, which we believe is suitable for future experiments using CRISPRi-seq. This system combines features from previously published CRISPRi systems in Pseudomonads, including a Pseudomonas codon-optimized dCas9 driven by the XylS/P_m, AraC/P_BAD, or LacI/P_lac promoters.^10,31,44 sgRNAs are expressed constitutively and can create a robust CRISPRi response, even with low levels of expression. The “plug-and-play” design of our pSpy vectors and the use of mini-Tn7-based dCas9 delivery system make this system flexible and easily adaptable for use in other organisms.³⁵ This system can precisely modulate gene expression, and its low background expression provides a platform for tightly regulating the expression of genes with minimal fitness costs.

Environmental nutrient composition is likely to strongly influence the concentration of the inducer required to elicit a response, as suggested by the varied responses in different growth media. This is most problematic with the AraC/P_BAD system, which is consistent with prior studies with Pseudomonads where there can be high levels of expression in the absence of an inducer that can be alleviated by the presence of glucose.³⁸ In our studies, glucose was a component of the medium for the flow cytometry experiments (assessing regulation at the single-cell level) and fluorimetry experiments (assessing promoter responsiveness at a population level), while it was absent in the media for assessing the effects of pvdH repression at a population level and ftsZ repression on cell length. Given that glucose activates catabolite repression and functions as a repressor of the P_BAD promoter,^58,63,64 it could explain our media-dependent differences in leakiness and variation in inducer concentrations necessary for eliciting a response. Our ability to control the leakiness of the AraC/P_BAD promoter may be due to the use of a wide range of growth media, highlighting how environmental context and promoter choice can influence CRISPRi performance. This is an important consideration when deploying CRISPRi in complex habitats such as in soil or in the rhizosphere or when selecting a regulatory system for controlling dCas9 expression.

Each promoter system assessed in these studies exhibited differences in inducer concentrations required for effective repression, leakiness, and uniformity of repression in a population of cells. While the LacI/P_lac system showed the lowest levels of leakiness, it also showed low levels of inducibility under most conditions. Performance of the AraC/P_BAD system varied considerably based on whether glucose was present, indicating its use in the rhizosphere or in multispecies microbial communities may be limited, given the potential for production and/or consumption of the arabinose inducer by the plant or other microbes in the vicinity, as well as the catabolite repressor glucose.⁶¹ Finally, the XylS/P_m system performed well in most circumstances tested here, with fairly low levels of leakiness in most conditions, the requirement of low inducer concentrations, and the broadest dynamic range of control. However, the XylS/P_m system’s utility may be limited by the volatility of this inducer molecule, and effective use may require gastight systems or reapplication of inducer to maintain CRISPRi over time. A benefit is that 3-MBZ is likely not as readily utilized by other microbes as sugars such as arabinose. Taken together, the XylS/P_m promoter system native to P. alloputida performed best overall. Importantly, we demonstrate for the first time that CRISPRi is functional in the rhizosphere and that CRISPRi can be used to modulate microbe–microbe interactions.

Materials and Methods

Plasmid Construction and Cloning

A full list of plasmids, gBlocks, primer sequences, and sgRNA oligonucleotides can be found in Tables S1 and S2. Q5 polymerase was used to amplify fragments for cloning into vectors. Colony PCR was performed to validate sequences of cloned cells using TaKaRa ExTaq polymerase following the protocol from Choi and Schweizer.⁶⁵ PCR cycle conditions were obtained from the user manuals for each polymerase.

The pSpy sgRNA expression system was constructed from pSEVA231 in compliance with the Standard European Vector Architecture (SEVA) guidelines.³ The constitutive promoters P_EM7, P_BG35, P_BG42, and P_BG28 were generated by annealing forward and reverse oligonucleotides, then digesting with PacI/AvrII, creating the respective vectors pSpyA, pSpyB, pSpyC, and pSpyD.⁴⁸ Immediately downstream of these promoters, two BsaI sites were introduced to be used for inserting sgRNAs, followed by a S. pyogenes (Spy) tracRNA sequence at Eco53kI/KpnI sites.^36,53,54 pSpy-GFP vectors were created by inserting gBlock msfGFP_pSpy directionally at BsaI sites. When the vector is digested, the msfgfp gene is cut out (visible on a gel), and the resulting clones will have lost their fluorescence, streamlining the verification process.

The Pseudomonas codon-optimized dCas9 expression system was based on pTn7-M, a Km^R vector with Gm^R on the mini-Tn7 transposon, with the following modifications.⁴⁸ First, we created a derivative in which the Gm^R gene could be excised post-insertion of the mini-Tn7 into the genome. We amplified the gentamicin resistance gene and flanking FRT sites from the plasmid pUC18 miniTn7T-Gm⁴⁹ using the primers Tn7T-F-L and Tn7T-R-AL, generating AatII sites on either end of the PCR product. This product was inserted into pTn7-M at AatII sites and then sequenced to confirm the directionality of the insertion. Next, the BCD2 translational enhancer gBlock was inserted at PacI/BamHI, which also added a BmtI site to the MCS of the vector. Pseudomonas codon-optimized dCas9 was amplified from plasmid pSW196 using primers Psd9-Fw-BmtI and Psd9-Rev-NotI. The product was inserted into the vector at the BmtI and NotI sites. Finally, promoters were inserted at PacI/AvrII sites upstream of BCD2. AraC/P_BAD was amplified from pJMP1237 using the primers AraBAD-F and AraBAD-R, LacI/P_lac was amplified from pJMP1161 using the primers pLac-F and pLac-R, and XylS/P_m was amplified from pSEVA238-DM using the primers PS-1 and PS-2.^43,44 A separate set of plasmids were generated with msfgfp in place of dCas9, allowing for measurements of GFP expression as a proxy for dCas9. BCD2-GFP was amplified from the plasmid pBG using the primers PS-1 and PS-2, digested with AvrII and SacI, and ligated into the mini-Tn7 vector.⁴⁸ Following insertion of GFP, the promoters AraC/P_BAD, LacI/P_lac, XylS/P_m, P_BG42, and P_BG35 were inserted at PacI/AvrII as described above.

Bacterial Growth Medium

Bacteria in this study were grown in the following media: Luria-Bertani (LB), Tryptic Soy Broth (TSB), King’s B medium (KB), Vogel-Bonner minimal medium⁶⁵ (VBMM), Super Optimal broth with catabolite repression⁵² (SOC), minimal 9 medium³¹ (M9) with 20 mM succinate (M9S), 21C with Tryptone and yeast extract⁶⁶ (CTYE), and Casamino acids medium⁵² (CAA). A 1 mM phosphate buffer (PB) was used to collect bacteria from the samples. Aqueous soil extracts (SE) were obtained by stirring soil in distilled water (100 g soil per liter) at 80 °C for 16 h and then centrifuging and filtering out solids, as described previously.⁶⁷ Murashige and Skoog (MS) plant growth medium with macro- and micronutrients (MSP01)⁶⁸ was supplemented with 1.65 g/L ammonium nitrate.

Bacterial Strains and Culture Conditions

Escherichia coli strain DH5α was used for cloning and maintenance of sgRNA plasmids, and strain DH5α λpir was used for mini-Tn7 plasmids, which had unstable origins of replication (R6K). E. coli strains were cultured in LB or SOC at 37 °C, and when appropriate, the medium was amended with 50 μg/mL kanamycin, 15 μg/mL gentamicin, and 10 μg/mL tetracycline. P. alloputida KT2440 and its derivatives used in this study can be found in Table S1. P. alloputida was cultured in LB, KB, VBMM, M9S, CAA, CTYE, and TSB and incubated at 30 °C, and when appropriate, the medium was amended with 50 μg/mL kanamycin, 30 μg/mL gentamicin, 25 μg/mL rifampicin, and 20 μg/mL tetracycline. For CRISPRi experiments, the inducers Isopropyl β-d-1-thiogalactopyranoside (IPTG), arabinose, and 3-methylbenzoate (3-MBZ) were added.

Electrocompetent E. coli cultures were prepared in advance and stored at −80 °C until use.⁵²E. coli was grown at 37 °C until the midexponential phase and then chilled and harvested using gentle centrifugation. Cells were washed with cold sterile water and resuspended in 10% glycerol. Electrocompetent P. alloputida cultures were prepared fresh on the day of the transformation using a sucrose competency protocol.^49,65 Plasmid DNA (50–100 ng) was then added, and cells were electroporated in 2 mm cuvettes at 2.5 kV. Cells were recovered using 1 mL of LB (P. alloputida) or SOC (E. coli) for 1–2 h and then plated onto selective media.

Conjugation of plasmids into P. alloputida occurred as outlined by Choi et al.^49,65 Aliquots of overnight cultures of an E. coli plasmid donor, P. alloputida recipient, and HB101 pRK2013 helper were mixed with 1 mL of LB. For the construction of strains with mini-Tn7 integrated into the chromosome, we also included SM10 λpir pTNS2 to provide the transposase. Mixtures were washed twice with LB, then spotted onto LB plates, and incubated for 5–24 h. Cells were collected, placed in 1 mL of 0.9% NaCl, and vortexed to resuspend before plating onto VBMM plus antibiotics and incubating at 30 or 37 °C.

Integration of Mini-Tn7 into P. alloputida Genome

dCas9 expression vectors were integrated into the genome of KT2440, and then the Gm^R-marker was removed following previously published methods.⁴⁹ Colony PCRs were used to confirm insertions with the primers Tn7-L/PPglmS-up and to validate the removal of the Gm^R-marker using the primers Gm-up/Gm-down.

sgRNA Design and Selection

Sequences and up/downstream regions of genes of interest were obtained from the Pseudomonas Genome Database.⁴⁷ SAPPHIRE and BPROM were used to predict promoters and −10/–35 sites of genes of interest.^69,70 CRISPOR was used to generate lists of available sgRNA target sites, which were manually parsed to select sgRNAs targeting either the promoter or first 200 bp of the ORF.⁷¹ As a negative control, a sgRNA targeting the Tn7-R region of the mini-Tn7 transposon was designed to account for sgRNA expression and the presence of the pSpy vectors. ftsZ sgRNAs were designed by Ian Blaby using gRNA-SeqRET.⁷² sgRNA oligonucleotides with BsaI overhangs were annealed and then ligated into BsaI-digested pSpy vectors. To verify insertion of these sgRNAs, colonies were screened for the presence of GFP expression (loss of GFP indicates successful transformations) and then confirmed via colony PCR using the primers PS-1 and PS-2 before submitting for sequencing to verify the construct.

Growth Curves

We performed growth curves in LB to measure the dCas9 toxicity at increasing inducer concentrations. Plates were incubated at 30 °C and shaken continuously in a BioTek Synergy H1 microtiter plate reader. OD600 was measured every 30 min for 24 h. Three biological replicates per condition were performed, and the entire experiment was repeated at least twice. Diauxic growth curves observed in these experiments are likely due to P. alloputida switching carbon sources as they run out over time, which is common in LB.⁷³

Creation of the pvdH Deletion Mutant

We deleted the pvdH gene using allelic exchange, as outlined in the protocol by Hmelo et al.⁷⁴ The upstream region of pvdH was amplified with the primers pvdH-Up-F-HindIII and pvdH-Up-R-overhang while the downstream region was amplified with pvdH-Down-F and pvdH-Down-R-BamHI. Splice overlap extension PCR was used to fuse the two fragments together, which were then digested with HindIII/BamHI and ligated into pEX18-Tc.⁷⁵ Colony PCR was performed using the M13-F and M13-R primers to amplify this cloning site, and amplicons were submitted for Sanger sequencing. The plasmid was conjugated into KT2440 using the protocols listed above and recovered on LA plates containing rifampicin and tetracycline. Counterselection to induce the removal of the plasmid from the chromosome again occurred using media containing 5% (w/v) sucrose. Colonies were screened by streaking onto King’s B plates to look for a lack of fluorescent pyoverdine production before confirming the deletion using colony PCR and the primers pvdH-seq-F and pvdH-seq-R.

In Vitro ftsZ and pvdH CRISPRi Assays

ftsZ CRISPRi strains were grown in LB overnight and then serially diluted and spotted onto LB plates containing the appropriate inducer. Plates were incubated overnight, and then growth was assessed visually. For pvdH, the same assays were performed, with the exception that King’s B medium was used. Both pyoverdine production and growth were assessed in these assays visually using a UV transilluminator to visualize the fluorescence of pyoverdine.

Microscopy

ftsZ₊₃₃ CRISPRi strains were grown in LB overnight and then diluted 1:20 in fresh LB. Cultures were incubated for 3 h, spiked with appropriate inducers, and incubated for 3 more hours before wet-mounting onto microscope slides. DIC imaging was performed on a Leica DMI8 microscope using the 63× objective under oil immersion. Experiments were performed in triplicate. Images were processed using Adobe Photoshop. Cell lengths were measured using ImageJ.⁷⁶ A minimum of 300 cells were counted per replicate before pooling for analysis. For determining the value of how long a cell must be to be “elongated,” we took the average cell length of cells without an inducer (2.5 μm) and doubled it (5 μm) to account for cells that may be dividing at the time of measurement.

Measurement of Pyoverdine Production and GFP Expression in Liquid Culture

Strains were grown in CAA overnight and then diluted to an OD600 of 0.1 in fresh CAA. Inducers were added to CAA, and then 200 μL of the medium containing inducers was added to the wells of a 96-well plate. 20 μL amount of culture was then added to each well. Plates were incubated at 30 °C and shaken continuously in a BioTek Synergy H1 microtiter plate reader. OD600 and pyoverdine fluorescence (excitation 400 nm, emission 460 nm) were measured hourly for 24 h. Three biological replicates per condition were performed, and the entire experiment was repeated at least twice. Fluorescence was normalized to cell density, and all measurements were normalized to the pSpyB-Tn7sg strain with no inducer (Figure 4) or the 0 μM 3-MBZ condition (Figure 5). For GFP experiments, the same protocol was used with some modifications: strains were grown in CTYE instead of CAA, and GFP fluorescence was measured instead of pyoverdine fluorescence (excitation 485, emission 525). Data were normalized to the constitutive GFP strain with no inducer (Figure 6).

Flow Cytometry

Strains were grown overnight in CTYE at 30 °C, diluted to an OD₆₀₀ of 0.1, then mixed 1:10 with fresh CTYE containing appropriate inducers, and grown at 30 °C with shaking for 6 h. Following incubation, cultures were fixed in a solution of 0.4% paraformaldehyde, and GFP expression was measured using a FACSCanto flow cytometer at the Iowa State University Flow Cytometry Facility. The value of 100 was chosen as a cutoff for GFP fluorescence, as 98% of promoterless-GFP cells were below this value, and 95% of constitutive GFP cells were above this value.

CRISPRi Rhizosphere Experiments

Phz51 maize seeds were generously provided by Thomas Lübberstedt.⁷⁷ Phz51 seeds were surface-sterilized by soaking in 70% ethanol for 10 min and then in 20% bleach for 10 min, washing 5 times with sterile water between each soak. Seeds were placed on germination paper moistened with a plant growth nutrient solution (1× MS + Captan fungicide (2 g/L)) and then rolled and placed vertically in a beaker with a small amount of the nutrient solution in the bottom.⁶⁸ Seedlings were incubated for 3–5 days until most seedlings had a 1 cm radicle emerging. Seeds were sorted so that each treatment had equal numbers of seeds with short/medium/long radicles and then placed in Petri dishes and covered with the bacterial inoculum (overnight culture of CRISPRi ftsZ₊₃₃ or Tn7sg strain, diluted to an OD600 of 0.1 in PB). Seeds were incubated for 15 min before planting. A subset of seeds were used at this point to calculate the number of bacterial cells per seedling present in the starting inoculum, described below. Syringe barrels (60 mL) were filled with a 2:1:1 mixture of Turface MVP/Turface Quick Dry/vermiculite. 4 h before planting, syringes were watered with a nutrient solution that was 0.5× aqueous soil extract and 0.5× MS. Some syringes received inducer (500 μM 3-MBZ or 33 mM arabinose). Seeds were placed in the syringes, filled to the top with more Turface/vermiculite, and watered with 5 mL more of the nutrient solution. To prevent loss of 3-MBZ via volatilization, syringes were covered with Whirl-Pak bags and ends were plugged with luer-lock caps. Collection of plants for enumerating bacteria on the roots occurred at 1, 2, and 4 days post-planting. After collection of the day 1 samples, all syringes were plugged and watered with additional plant nutrient solution for 15 min, with or without the inducer present, before opening the luer-lock to drain excess water to avoid creating saturated conditions. After being thoroughly drained, luer-lock caps were replaced.

Enumeration of KT2440 on the roots was performed by aseptically removing seedlings from the syringes, cutting off the shoot, and then placing the roots in 10 mL of PB. Samples were placed in a sonicating water bath for 5 min, vortexed for 30 s, and serially diluted before plating onto LB plates containing 25 μg/mL each of Km and Rf. Plates were incubated at 30 °C for 24–48 h and then counted. There were at least 4 replicates per condition per time point. Following enumeration, roots and shoots were each dried in a 60 °C oven for 48 h and then weighed.

Cells recovered from the roots were subjected to an assay to confirm that CRISPRi was still functional following their time in the rhizosphere. Twelve Km^R colonies per condition (± inducer) were selected from plates used to enumerate CFUs on roots 4 days after planting. As many different plants as possible were used to select colonies that were grown overnight before being subjected to the ftsZ survival assay described above. Colonies were scored as either maintaining CRISPRi (cells did not survive when placed on media with inducer) or losing CRISPRi (surviving on media with inducer).

We assessed whether either of the inducers at the concentrations used was harmful to plant growth. Root dry weights were unaffected by the 3-MBZ (p = 0.2305) and arabinose (p = 0.1563) at the concentrations used here (Figure S4A). However, maize seedlings exhibited water soaking damage on the leaves in all treatments, presumably due to the high humidity conditions created by covering the plants with Whirl-Pak bags (Figure S4B). These findings indicate that our experimental design has no apparent influence on plant growth that might influence experimental outcomes.

Microbial Interactions Assays

To measure microbial interactions, we used methods adapted from Lozano et al.⁶² Environmental microbes used in these interaction assays were isolated from the bulk soil, rhizosphere, or endosphere of maize plants grown in low-nitrogen soils in Iowa fields. Roots were washed to obtain the rhizosphere and ground using a ball mill to obtain the endosphere before plating.⁷⁸ Maize isolates tested here included bacteria from the genera Chryseobacterium sp. S-02, Arthrobacter sp. R-11, Dyella sp. R-26, Rhodococcus sp. R-03, Lysobacter sp. E-23, Luteibacter sp. E-22, Sphingobium sp. R-21, and Pseudomonas sp. E-05. Strains with “S” are from the soil, “R” from the rhizosphere, and “E” from the endosphere. Strains were grown in King’s B broth overnight, and then all environmental isolates were diluted 1:100 in fresh media and spread onto the surface of King’s B plates to create lawns. A subset of plates contained 50 μM FeCl₃ or 1 mM 3-MBZ. Once lawns were dry, each was spotted with wild-type KT2440, KT2440 ΔpvdH, KT2440-XylS/P_m-dCas9 + pSpyB-pvdH-77, and KT2440-XylS/P_m-dCas9 + pSpyB-Tn7sg. Plates containing 3-MBZ were placed in airtight plastic tubs to reduce the volatilization of the inducer. Plates were incubated at 30 °C for 96 h and imaged daily under bright light and UV. The assay was repeated three times.

Statistical Analysis

JMP Pro version 17 was used for statistical analysis. Pyoverdine production, log-transformed flow cytometry data, and plant dry weights were each analyzed by a one-way analysis of variance (ANOVA), followed by a Tukey–Kramer HSD multiple comparison posthoc test. For the microscopy data on cell length, the data was log₁₀-transformed for a two-way ANOVA (promoter/regulator and presence/absence of inducer were main effects), and treatments were compared by Tukey–Kramer HSD multiple comparison test. In the plant experiment, CFU data was log₁₀-transformed for two-way ANOVAs (sgRNA and presence/absence of inducer main were main effects), which were performed for each day of the experiment. Treatment means were compared using the Tukey–Kramer HSD multiple comparison test. To compare CFU count data over time, we performed another two-way ANOVA (day and sgRNA were the main effects), and a Student’s t test was used to compare colonization by day. For each regulatory system, a two-tailed Fisher’s exact test was used to compare cells subjected to the survival assay after recovery from the rhizosphere. The number of cells that retained CRISPRi functionality was compared to determine if they were statistically different when obtained from the inducer and no-inducer control treatments.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acssynbio.4c00312.

• Lists of primers, oligonucleotides, gBlocks, strains, and plasmids used in this work, and impact of dCas9 on KT2440 growth, different sgRNA target sites, and plant dry weight graphics when exposed to 3-MBZ and arabinose (PDF)

Author Present Address

^§ Corteva Agriscience, Johnston, Iowa 50131, United States

Author Contributions

L.J.H. conceived the study. M.N.R.S. and D.X.V. performed the experiments: M.N.R.S. designed the dCas9 expression system, and M.N.R.S. and D.X.V. designed the pSpy sgRNA expression system. M.N.R.S. curated the figures and performed data analysis. L.J.H. supervised the project and provided training and mentorship. M.N.R.S. wrote the first draft of the manuscript. All authors provided comments and approved the final manuscript.

The authors declare no competing financial interest.