Pseudomonas syringae Differentiates into Phenotypically Distinct Subpopulations During Colonization of a Plant Host

Pseudomonas syringae differentiates into phenotypically distinct subpopulations during colonization of a plant host Jos e S. Rufi an, 1 Mar ıa-Antonia S anchez-Romero, 2 Diego L opez-M arquez, 1 Alberto P. Macho, 1† John W. Mansfield, 3 Dawn L. Arnold, 4 Javier Ruiz-Albert, 1 Josep Casades us 2 and Carmen R. Beuz on 1 * 1 Depto. Biolog ıa Celular, Gen etica y Fisiolog ıa, Instituto de Hortofruticultura Subtropical y Mediterr anea, Universidad de M alaga-Consejo Superior de Investigaciones Cient ıficas (IHSM-UMA-CSIC), M alaga E-29071, Spain. 2 Departamento de Gen etica, Facultad de Biolog ıa, Universidad de Sevilla, Seville 1095, Spain. 3 Faculty of Natural Sciences, Imperial College, London SW7 2AZ, UK. 4 Centre for Research in Bioscience, Faculty of Health and Applied Sciences, University of the West of England, Bristol BS16 1QY, UK. Summary Bacterial microcolonies with heterogeneous sizes are formed during colonization of Phaseolus vulgaris by Pseudomonas syringae. Heterogeneous expression of structural and regulatory components of the P. syringae type III secretion system (T3SS), essential for colonization of the host apoplast and disease development, is likewise detected within the plant apoplast. T3SS expression is bistable in the homogeneous environment of nutrient-limited T3SS-inducing medium, suggesting that subpopulation formation is not a response to different environmental cues. T3SS bistability is reversible, indicating a non-genetic origin, and the T3SS HIGH and T3SS LOW subpopulations show differences in virulence. T3SS bistability requires the transcriptional activator HrpL, the double negative regulatory loop established by HrpV and HrpG, and may be enhanced through a positive feedback loop involving HrpA, the main component of the T3SS pilus. To our knowledge, this is the first example of phenotypic heterogeneity in the expression of virulence determinants during colonization of a non-mammalian host. SIGNIFICANCE The plant pathogen Pseudomonas syringae requires a type III secretion system (T3SS) to inject effector proteins into host cells and to cause disease. This study shows that expression of T3SS genes is activated in a heterogeneous fashion during colonization of plant tissues. Cell-to-cell differences in T3SS gene expression are likewise observed in the homogeneous environment of nutrient-limited culture medium, where an isogenic bacterial population bifurcates into lineages that express or not the T3SS. Differences in T3SS expression are non-heritable, are established through the action of a double-negative regulatory feedback loop and determine differences in plant disease severity. Phenotypic heterogeneity is therefore a factor that must be considered when portraying bacterial adaptation to plant niches. Introduction Bacterial infections involve spatial and temporal changes in gene expression that accompany the migration of pathogens from the site of invasion to target tissues. Pathogen progression inside the host is therefore accompanied by physiological adjustments to respond to different stimuli and microenvironments. However, phenotypic changes are not always deterministic, directly correlated with stimuli. Stochastic events such as an uneven distribution of regulators during cell division can produce cell-to-cell differences within a homogeneous microenvironment. This can lead to probabilistic determination of certain phenotypic traits, generally known as phenotypic heterogeneity or phenotypic variation (Davidson and Surette, 2008). Phenotypic heterogeneity has been known to take place in microbial clonal populations for decades (Bigger, 1944; Novick and Weiner, 1957). In certain cases, phenotypic Received 8 July, 2016; accepted 9 August, 2016. *For correspondence. E-mail [email protected]; Tel. (1134) 952 131959; Fax (11 34) 952 132001. † Present address: Shanghai Center for Plant Stress Biology, Shanghai Institutes of Biological Sciences, Chinese Academy of Sciences, Shanghai, 201602, China V C2016 Society for Applied Microbiology and John Wiley & Sons Ltd Environmental Microbiology (2016) 18(10), 3593–3605 doi:10.1111/1462-2920.13497 heterogeneity merely reflects the occurrence of cell-to-cell differences generated by molecular noise. In other cases, however, phenotypic heterogeneity reflects the occurrence of bistability, the bifurcation of a unimodal physiological state into two distinct states, generating two bacterial subpopulations or lineages. Bistability is usually the consequence of bimodal gene expression, which can be generated by a positive feedback loop as described in the E. coli lac operon (Novick and Weiner, 1957) or by a double negative feedback loop as in the lysis/lysogeny decision of bacteriophage lambda (Novick and Weiner, 1957; Herskowitz and Hagen, 1980). The literature on bacterial bistable switches has been enriched with interesting examples in the last decade (Dubnau and Losick, 2006; Davidson and Surette, 2008; van der Woude, 2011; S anchez-Romero and Casades us, 2014; van Vliet and Ackermann, 2015; Uphoff et al., 2016). In certain cases, the biological significance of bistability remains a mystery. In other examples, however, subpopulation formation may be viewed either as a division of labour within the population or as a bet-hedging strategy that may facilitate adaptation to environmental challenges (Veening et al., 2008). The importance of analysing phenotypic heterogeneity has been highlighted in the context of antibiotic exposure for animal and human pathogens, and in the colonization of animals hosts (Helaine and Holden, 2013; Arnoldini et al., 2014; Campbell-Valois et al., 2014; Claudi et al., 2014; S anchez-Romero and Casades us, 2014; Manina et al., 2015; Band et al., 2016; Conlon et al., 2016; Van den Bergh et al., 2016). In Salmonella enterica,for instance, phenotypic heterogeneity has been observed at several stages of host colonization including invasion of the intestinal epithelium, survival in macrophages and colonization of the gall bladder (Stecher et al., 2004; Saini et al., 2010; B€ aumler et al., 2011; Stewart and Cookson, 2012). Other examples that highlight the relevance of phenotypic heterogeneity in bacterial infections are bistable expression of the cholera toxin in Vibrio cholerae (Nielsen et al., 2010), and of the NO-detoxification system in Yersinia pseudotuberculosis (Davis et al., 2015). Despite increasing evidence supporting the notion that bacterial pathogens exploit non-genetic variation to adapt to mammalian hosts, little is known about the occurrence or potential impact of these processes in the adaptation of bacteria to non-animal hosts. In this work, we have addressed this issue in the archetypal plant pathogen Pseudomonas syringae (Mansfield et al., 2012). P. syringae is an academically relevant model pathogen of increasing economical impact in agriculture, with recent resurgence of old diseases and emergence of new ones (Shenge et al., 2007; Green et al., 2010). P. syringae enters the plant from the leaf surface through natural openings or wounds reaching the intercellular spaces of the leaf parenchyma, the apoplast, where it replicates. In the apoplast, P. syringae uses a type III secretion system (T3SS) to deliver effector proteins into the plant cell cytosol (Alfano and Collmer, 1997; Rohmer et al., 2004). Many of these effectors act to suppress plant defences to allow bacterial colonization (Macho and Zipfel, 2015). In this study, we show that phenotypic heterogeneity occurs during plant colonization by P. syringae. Structural and regulatory components of the P. syringae T3SS display heterogeneous expression within the plant apoplast, and bistable T3SS expression is detected in the homogeneous environment of nutrient-limited T3SS-inducing medium. T3SS bistability is reversible, supporting a nongenetic origin, and generates bacterial subpopulations with differences in virulence. To our knowledge, this is the first example of bacterial phenotypic heterogeneity in a nonmammalian host. Results Expression of the T3SS is heterogeneous within the plant apoplast Following the dynamics of fluorescently labelled P. syringae pv. phaseolicola populations during colonization of Phaseolus vulgaris, we observed that the size of bacterial microcolonies within the apoplast was heterogeneous (Fig. 1A and B). Since the apoplast is a complex and multifarious environment, heterogeneity might reflect adaptation of P. syringae to distinct microenvironments. An alternative possibility, however, is that heterogeneous colony size might result from random differences in the expression of virulence factors, as described in certain animal pathogens (Nielsen et al., 2010; Davis et al., 2015). To evaluate whether apoplast-growing bacteria could display heterogeneous gene expression, we applied single-cell methods to analyse transcriptional fusions to gfp of several P. syringae genes. Given the relevance of the T3SS in plant colonisation by P. syringae, we focused our study on T3SS genes. The choice was further supported by a report on the necrotrophic plant pathogen Dickeya dadantii showing that a plasmid-cloned type III promoter displayed phenotypic heterogeneity under laboratory conditions (Zeng et al., 2012). We generated transcriptional fusions to gfp downstream of chromosomelocated native copies of three genes encoding T3SS elements: hrpL, encoding an alternative sigma factor of the extracytoplasmic factor (ECF) family (Fouts et al., 2002), hrcU, the promoter-distal gene of the HrpL-controlled hrcQRSTU operon, encoding a structural component of the T3SS (Charkowski et al., 1997) and hopAB1, encoding a type III secreted effector involved in suppressing plant defences (Jackson et al., 1999). All three strains displayed wild-type virulence (Supporting Information Fig. S1). 3594 Rufi an et al. V C2016 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology,18, 3593–3605 Formation of bacterial subpopulations in a plant host 3595 V C2016 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology,18, 3593–3605 Bacterial distribution within a microcolony developing in the confines of the intercellular spaces of the leaf apoplast is heterogeneous (Supporting Information Fig. S2 and Supporting Information Videos S1 and S2) and can thus lead to apparent differences in fluorescence intensity as judged by microscopic examination. Therefore, to unequivocally associate potential differences in fluorescence to individual bacteria, we applied single-cell analyses to apoplastextracted bacteria. Microscopic analysis on apoplastextracted bacteria carrying the T3SS gene fusions to gfp revealed strong cell-to-cell differences in fluorescence, supporting that expression of the T3SS genes is phenotypically heterogeneous within the plant (Fig. 1C and D). Bacteria not expressing the genes were found for all three fusions both by microscopic examination (Fig. 1C), and by flow cytometry analyses (Fig. 1D), indicating that a subpopulation of bacteria that do not express the T3SS genes does appear during colonisation of the host plant tissue. P. syringae bifurcates into two subpopulations due to bistable expression of T3SS genes To ascertain whether the phenotypic heterogeneity observed for expression of the T3SS genes was a response to environmental cues or could have stochastic origin, we examined gene expression in the homogeneous environment of nutrient-limited Hrp-inducing medium (HIM) (Huynh et al., 1989) (Fig. 2). Growth in HIM triggers a signalling cascade that activates expression of HrpL, which in turn activates expression of all T3SS genes (Fig. 2A and B). A remarkable observation, however, was that all three expression patterns were heterogeneous in HIM, in contrast with those obtained in non-inducing medium (i.e., LB medium, Supporting Information Fig. S3). Heterogeneity was higher during exponential growth (24 h) than in stationary phase (48 h) (Fig. 2B). In all cultures, a fraction of bacterial cells carrying gfp fusions displayed fluorescence levels overlapping with those of non-GFP control bacteria (Fig. 2B, centre and right panels). This was particularly clear in exponentially growing bacteria (24 h), where expression of all three gfp fusions reached a bistable state (Fig. 2B, centre panels). Because the differences in T3SS expression between the two subpopulations were not all-or-none, we use the terms T3SS HIGH and T3SS LOW instead of T3SS ON and T3SS OFF . Bistability was no longer detected in stationary phase cultures (48 h), supporting a reversible and non-genetic origin for the differences observed between subpopulations (Fig. 2B, right panels). Bistability of the T3SS genes requires HrpL and is established through the HrpV/HrpG double-negative regulatory loop Because HrpL activates expression of hrcU and hopAB1 (Xiao and Hutcheson, 1994) (Fig. 2A), we considered the possibility that the bistable state might be passed down from HrpL to genes under its control. To test this hypothesis, we introduced into the strains carrying hrpL::gfp or hopAB1::gfp fusions, a plasmid carrying a copy of hrpL under the control of the lacZ promoter, which enables moderate, constitutive expression in P. syringae (Ortiz- Mart ın et al., 2010b), to evaluate its impact on gfp expression by flow cytometry (Fig. 3A). The bimodal distribution of chromosome-located hrpL::gfp expression (Fig. 3A in black) becomes unimodal in the presence of constitutively expressed HrpL (coloured). In the case of hopAB1::gfp, bistability is reduced, although not entirely abolished in the presence of plasmid-encoded HrpL, but the population displays a shift towards the T3SS HIGH state. These observations suggest that HrpL may play a central role in the establishment of bistability in the system. Bistability is often triggered by transforming a quantitative cell-to-cell difference into a qualitative difference through the action of one or more feedback loops (Veening et al., 2008). Two such feedback loops regulate the Fig. 1. Colony development and T3SS gene expression in P. syringae display phenotypic heterogeneity within the plant. A, B. Colony development is phenotypically heterogeneous within the plant apoplast. A: Representative confocal microscopy image of bean leaves inoculated with 5 310 6 cfu ml 21 of wild-type eYFP (yellow) 3 days post-inoculation (dpi). Boxes highlight small microcolonies among the typical and more abundant larger ones. Largest not-rounded microcolonies typically result from closely located colonies merging. Red corresponds to auto-fluorescence generated by chloroplasts. Scale bar corresponds to 100 lm. At least three independent experiments were carried out. B. Graph showing relative area of individual microcolonies. Colony area is expressed in percentage of total image area. C,D. Expression of the T3SS is phenotypically heterogeneous within the plant. C. Fluorescence microscopy images of apoplast-extracted bacteria from bean leaves 5 days post-inoculation with a 5 310 5 cfu ml 21 inoculum of each of the strains carrying chromosome-located transcriptional gfp fusions to the T3SS genes hrpL, hrcU or hopAB1. Left panels show GFP fluorescence, and right panels this image merged with that obtained from bacterial staining with the membrane dye FM4-64 (red). Inset shows a close up of the area of the corresponding image enclosed by a solid line square. Dotted line squares highlight other examples of bacteria displaying GFP fluorescence below the level of detection. Scale bar corresponds to 10 lm. D. Flow cytometry analysis of apoplast-extracted bacteria carrying hrpL::gfp, hrcU::gfp or hopAB1::gfp fusions, obtained from bean leaves 5 days after inoculation with a 5 310 5 cfu ml 21 inoculum. Non-GFP 1448A was also included as a reference to differenciate OFF and ON subpopulations. Data are represented as a dot plot (forward scatter [cellular size] versus GFP fluorescence intensity). All data were collected for 100,000 events per sample. At least three independent experiments with two replicates each were carried out. Figure shows typical results. 3596 Rufi an et al. V C2016 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology,18, 3593–3605 Formation of bacterial subpopulations in a plant host 3597 V C2016 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology,18, 3593–3605 expression of the T3SS genes in P. syringae:(i)apositive feedback loop controlled by HrpA, the main subunit of the T3SS pilus (Roine et al., 1997; Wei et al., 2000); (ii) a double negative feedback loop regulated by HrpV and HrpG (Wei et al., 2005) (Fig. 2A). We analysed the roles of these regulators in the establishment of T3SS bistability using mutants defective in these genes and/or plasmids carrying the individual genes under study. Although bistability in hopAB1::gfp expression was reduced in a DhrpA mutant, bimodal expression of the hrpL::gfp fusion was still observed in the absence of HrpA, thus making HrpA an unlikely candidate to be the molecular switch required to trigger the bistable state (Fig. 3B). Bistability of hopAB1::gfp was abolished in a DhrpG mutant (Fig. 3C). In turn, absence of HrpV increased the proportion of cells expressing higher levels of hopAB1::gfp (Fig. 3C). This happened regardless of the presence of HrpG, as indicated by the fact that a DhrpV mutation was epistatic over a DhrpG mutation. Constitutive expression of either regulator from a plasmid led to reciprocal results on hopAB1::gfp expression: a stronger bistable phenotype was detected upon overexpression of HrpG, and bistability was abolished in cultures that constitutively expressed HrpV (Fig. 3C). The effect that constitutive expression of these regulators have on hrpL::gfp closely matched the effect seen on hopAB1::gfp (Fig. 4B). A tentative interpretation of the above observations is that HrpG and HrpV may be key elements in T3SS bistability, perhaps due to the existence of cell-to-cell differences in the amount or activity of these regulators. Differences in gene expression correlate with differences in virulence Validation of our reductionist observations in HIM was pursued by analysis in planta.InP. syringae,theT3SSis necessary to suppress basal defences and T3SS absence limits bacterial proliferation within the plant and prevents the development of disease (Alfano and Collmer, 1997). Thus, if heterogeneous expression of the T3SS genes were associated to the heterogeneity observed in the size of apoplast-located microcolonies, T3SS HIGH and T3SS LOW subpopulations would be expected to differ in their ability to interact with the plant host. To test this possibility, we analysed the development of disease in leaves inoculated with T3SS HIGH and T3SS LOW bacterial subpopulations, sorted according to their level of expression of hopAB1 (Fig. 4A, upper panel). The sorted populations were both virulent, an observation consistent with the fact that bacteria expressing hopAB1 were detected in both subpopulations although in different numbers (Fig. 4A, centre panel and 4B). However, the development of disease symptoms was faster in leaf areas inoculated with the population expressing higher levels of hopAB1, and the symptoms were also stronger (Fig. 4B). The spread of the disease symptoms was also faster beyond the areas inoculated with the T3SS HIGH subpopulation, suggesting a more efficient colonization of distal tissues by this population. Thus, differences in T3SS gene expression appear to correlate with differences in virulence: namely, the T3SS HIGH subpopulation is more virulent than the T3SS LOW subpopulation. Discussion This study shows that colonisation of the plant apoplast by Pseudomonas syringae involves cell-to-cell differences in expression of its T3SS. Although microenvironments within the apoplast might provide different signals to control bacterial gene expression, heterogeneous T3SS expression is also detected in the homogeneous environment of nutrient-limited T3SS-inducing medium (HIM), thus making unlikely it results as a direct response to environmental cues. Flow cytometry analysis of cultures grown in HIM showed the occurrence of two bacterial lineages, one of which expressed the T3SS at high levels while the other did not show significant T3SS expression. This bistable pattern of gene expression appeared during exponential growth, and reverted to unimodal heterogeneity in stationary cultures, thus suggesting a non-genetic origin and making phase variation an unlikely one (van der Woude, 2011). A key factor in bistable expression of the P. syringae T3SS appears to be the HrpL sigma factor (Fouts et al., 2002), as indicated by the disappearance of the T3SS LOW Fig. 2. Expression of hrpL::gfp,hrcU::gfp and hopAB1::gfp is bistable during exponential growth in Hrp-inducing medium. A. Schematic representation of the two feedback regulatory loops operating on the expression of the T3SS genes: a positive feedback loop controlled by HrpA, and a double negative feedback loop regulated by HrpV, an anti-activator of the T3SS genes that binds to HrpS the enhancer-binding protein required for HrpL expression, and HrpG, which binds to HrpV acting as an anti-anti-activator. Expression of hrpA, hrpV,hrpG,hrcU and hopAB1, is HrpL-activated, although in the case of highly expressed hopAB1 some expression can still be detected in the absence of HrpL. B. Histograms show flow cytometry analysis of strains carrying chromosome-located transcriptional fusions to the T3SS genes hrpL (top row), hrcU (middle row) or hopAB1 (bottom row). Histograms show cell counts versus GFP fluorescence at 0 h (immediately after a 1:10 dilution into HIM of an overnight LB culture), 24 h after the dilution into HIM (exponential phase) or 48 h (stationary phase). Black histograms show non-GFP 1448A included as a reference. Coloured histograms show GFP fluorescence for the strains carrying the fusions as indicated in each case. Dotted lines show the results of a replicate experiment. All data were collected for 100,000 events per sample. At least three independent experiments with two replicates each were carried out. Figure shows typical results. 3598 Rufi an et al. V C2016 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology,18, 3593–3605 Formation of bacterial subpopulations in a plant host 3599 V C2016 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology,18, 3593–3605 subpopulation when expression of HrpL was uncoupled from its regulation (i.e., upon constitutive expression of HrpL from a plasmid). HrpL activates expression of more than 50 genes within the nutrient-limited plant leaf apoplast (Ferreira et al., 2006; Lam et al., 2014; Mucyn et al., 2014), including the hrp/hrc genes that encode the T3SS, and effector genes (Xiao et al., 1994; Fouts et al., 2002). Bistable expression was detected in hrcU, which encodes a structural component of the T3SS required for secretion and translocation (Charkowski et al., 1997), and in hopAB1, encoding a type III secreted effector involved in suppression of plant defences (Jackson et al., 1999). Hence, transmission of HrpL bistability appears to occur downstream the regulatory cascade, generating T3SS HIGH and T3SS LOW cells. Bistability is hindered either by deletion of hrpG or by constitutive expression of HrpV (Fig. 3C), suggesting that the HrpV/HrpG pair may constitute the bistable switch involved in turning quantitative differences in gene expression into qualitative differences. Although HrpA does not seem to be essential, it may contribute to bistability by increasing the number of bacteria with gene expression levels high enough to cross the threshold for activation. Heterogeneous expression of the P. syringae T3SS is also observed during growth within the apoplast, and T3SS HIGH and T3SS LOW cells are recovered from P. syringae apoplast colonies. This observation, together with the fact that T3SS HIGH and T3SS LOW subpopulations differ in virulence, suggests that heterogeneous T3SS expression may play a role in the adaptation of P. syringae to plant hosts. Unfortunately, a direct test is not feasible because genetic changes that abrogate bistability in laboratory medium (mutation of hrpG or plasmid-borne expression of HrpL or HrpV) alter T3SS expression mean levels in a way that impairs virulence (Ortiz-Mart ın et al., 2010a,b). While the ultimate significance of T3SS heterogeneity in P. syringae remains to be established, a tentative interpretation is that heterogeneous expression of virulence determinants may serve as a stealth strategy for defence evasion in the apoplast, an environment where unsuppressed plant defences are operative (Mitchell et al., 2015). This possibility is supported by theoretical studies indicating that phenotypic heterogeneity can have adaptive value in changing and/or hostile environments (Kussell et al., 2005; Kussell and Leibler, 2005), and by experimental evidence of host defence evasion by lineage formation in animal pathogens (Srikhanta et al., 2010; Lovell et al., 2011; Hern andez et al., 2012; Claudi et al., 2014; S anchez-Romero and Casades us, 2014; Manina et al., 2015). Moreover, the phenotypic heterogeneity observed in P. syringae T3SS genes could also play a role in adaptation to other stages of its life cycle. Among plant pathogens, high frequencies of T3SS polymorphism are detected in natural Arabidopsis-associat- ed populations of P. syringae, and less aggressive variants, increase their growth potential in mixed infections and have a fitness advantage in non-host environments (Barrett et al., 2011). Furthermore, natural isolates of P. syringae from agricultural and non-agricultural niches display differences in the phenotypic (but not the genotypic) structure of the populations (Morris et al., 2008). On these grounds, we tentatively propose that phenotypic heterogeneity in T3SS gene expression may protect P. syringae populations from plant defences. Furthermore, it seems conceivable that T3SS LOW subpopulations may persist undetected in plant leaves, in a fashion reminiscent of the animal pathogens that cause persistent and chronic infections. Whatever the case, our description of bistable expression of the P. syringae T3SS brings about the notion that subpopulation formation during infection is not restricted to animal pathogens. Experimental procedures Bacterial strains and growth conditions Bacterial strains used in this work are listed in Supporting Information Table S1, and plasmids are listed in Supporting Information Table S2. Escherichia coli and Pseudomonas syringae pv. phaseolicola (Pph) were grown at 378Cand288C, Fig. 3. Bistability of hrpL::gfp and hopAB1::gfp is abolished by constitutive expression of HrpL, deletion of hrpG or constitutive expression of HrpV. A. Flow cytometry analysis of HIM-growing bacterial strains carrying chromosome-located transcriptional fusions to the T3SS genes hrpL or hopAB1, carrying or not a plasmid expressing hrpL under the control of P lac , a constitutive promoter of moderate expression in P. syringae. Histograms show cell counts versus GFP fluorescence after 24 h of growth in HIM. Black histograms show fluorescence of the fusions in the absence of the plasmid. Coloured histograms show fluorescence of the fusions in the strain carrying the plasmid as indicated. Dotted lines show the results of a replicate experiment. B. Flow cytometry analysis of HIM-growing bacterial strains carrying chromosome-located transcriptional fusions to the T3SS genes hrpL or hopAB1, in different genetic backgrounds. Histograms show cell counts versus GFP fluorescence after 24 h of growth in HIM. Black histograms show fluorescence of the fusions in an otherwise wild-type background. Coloured histograms show fluorescence of the fusions in a strain carrying the DhrpA mutation. Dotted lines show the results of a replicate experiment. All data were collected for 100,000 events per sample. C. Flow cytometry analysis of HIM-growing bacterial strains carrying chromosome-located transcriptional fusions to the T3SS genes hopAB1 or hrpL, in different genetic backgrounds. Histograms show cell counts versus GFP fluorescence after 24 h of growth in HIM. Histograms show fluorescence of the fusions in each of the indicated genetic backgrounds. WT indicate the strain that only carries the indicated gene fusion. All data were collected for 100,000 events per sample. At least two independent experiments with two replicates each were carried out with similar results. 3600 Rufi an et al. V C2016 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology,18, 3593–3605 Fig. 4. Bacterial populations sorted according to hopAB1 expression display differences in virulence. A. Flow cytometry analysis of a HIM-growing culture of the strain carrying hopAB1::gfp. GFP fluorescence intensity is shown as a green histogram. Gates were drawn to separate hopAB1::gfp bacteria displaying fluorescence levels overlapping the 1448A non-GFP bacterial population (indicated with a line marked as low), used as a negative control (Grey histogram), from cells expressing high GFP levels (indicated with a line marked as high, and including the mode for the expressing population). After sorting, aliquots of sorted cells were run again through the cytometer to confirm the efficacy of the sorting process (below), and bacterial concentration adjusted to 1 310 6 cfu ml 21 . Some overlap caused by the dynamic and reversible nature of the process can be detected. B. Disease symptom progression in bean leaves inoculated with 1 310 6 cfu ml 21 of each of the sorted populations at 6 and 11 days post inoculation (dpi). Results from three replicate experiments are shown. Formation of bacterial subpopulations in a plant host 3601 V C2016 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology,18, 3593–3605