Bacterial cyclic β‐(1,2)‐glucans sequester iron to protect against iron‐induced toxicity

Abstract

Cellular iron homeostasis is critical for survival and growth. Bacteria employ a variety of strategies to sequester iron from the environment and to store intracellular iron surplus that can be utilized in iron‐restricted conditions while also limiting the potential for the production of iron‐induced reactive oxygen species (ROS). Here, we report that membrane‐derived oligosaccharide (mdo) glucan, an intrinsic component of Gram‐negative bacteria, sequesters the ferrous form of iron. Iron‐binding, uptake, and localization experiments indicated that both secreted and periplasmic β‐(1,2)‐glucans bind iron specifically and promote growth under iron‐restricted conditions. Xanthomonas campestris and Escherichia coli mutants blocked in the production of β‐(1,2)‐glucan accumulate low amounts of intracellular iron under iron‐restricted conditions, whereas they exhibit elevated ROS production and sensitivity under iron‐replete conditions. Our results reveal a critical role of glucan in intracellular iron homeostasis conserved in Gram‐negative bacteria.

Introduction

Iron is essential to nearly all living organisms, due to its vital role in redox enzyme systems that are essential for several biological processes, such as respiration, photosynthesis, and DNA synthesis 1, 2. The ability of iron to exist in two oxidation states, ferrous (Fe²⁺) or ferric (Fe³⁺), poses problems of toxicity and insolubility. Bacteria have evolved diverse mechanisms to counter the problems associated with iron‐restricted availability and potential toxicity due to iron overload. In order to maintain proper iron homeostasis, bacteria utilize efficient iron uptake systems that enable them to scavenge various forms of iron from the environment under iron‐restricted conditions. In general, this involves secretion and uptake of high‐affinity ferric iron chelators called siderophores. Certain pathogenic bacteria are also able to utilize host–iron complexes as iron sources to fulfill their iron needs 3, 4. The ferrous form of iron, on the other hand, is soluble and is presumed to diffuse into the periplasm via porins present in the outer membrane and then transported into the cytoplasm via a ferrous iron transporter (Feo). However, little is known about how Fe²⁺ is stabilized in the periplasmic space, as ferrous iron is very unstable, and can readily be oxidized to ferric iron under aerobic conditions.

Free iron is toxic to the cell as it causes the production of reactive oxygen species (ROS) by the Fenton reaction. Bacteria produce iron storage proteins, such as ferritin, bacterioferritin, and DPs, that serve as intracellular iron reserves that can be utilized under iron‐restricted conditions and serve to limit the potential for iron‐mediated ROS formation 1, 5. The majority of the iron inside the cell is in such iron storage forms or in metalloproteins. However, a fraction of the intracellular iron, predominately in the ferrous form, is presumed to be complexed with a low‐molecular weight ligand or carrier to create a “free” (redox‐active or chelatable) accessible iron pool 6, 7. Although earlier studies have suggested the existence of an accessible free ferrous iron pool in the cell, the nature of the free iron pool and its role in iron trafficking and homeostasis have not been established 6, 8.

Here, we focus on iron uptake and metabolism by Xanthomonas campestris pv. campestris (Xcc), a Gram‐negative phytopathogen of cruciferous plant 9. Xanthomonads produce the siderophore xanthoferrin under iron‐restricted conditions, and iron metabolism plays a critical role in their virulence 10, 11, 12, 13. In order to gain insight into iron uptake and metabolism, we screened a transposon‐induced mutant library of Xcc to identify mutants overproducing siderophores. Interestingly, among such mutants we identified several extracellular polysaccharide (EPS)‐deficient mutants that exhibited growth deficiencies under iron‐restricted conditions. In this study, we addressed how exopolysaccharide production in Xcc is linked to iron homeostasis.

The xanthomonads produce characteristic extracellular polysaccharides (EPS) which consist largely of the exopolysaccharide xanthan and a secreted cyclic β‐(1,2)‐glucan, which plays an important role in virulence 14, 15. Xanthan, the major constituent of EPS, is a polymer of repeating pentasaccharide units with a trisaccharide side chain 16. The secreted exopolysaccharide is a cyclic β‐(1,2)‐glucan that contains 16 glucosyl residues with 15 β‐1,2 linkages and one α‐1‐6 linkage 14.

Cyclic β‐(1,2)‐glucan is a member of the family of osmoregulated periplasmic glucans (OPG) that are ubiquitous in Gram‐negative bacteria and constitute one of the major intrinsic components of the bacterial envelope, being as high as 5–20% of the total cellular dry weight 17, 18. Although glucan is primarily localized in the periplasm, several bacteria also produce secreted glucan 19. Earlier studies suggested diverse, albeit poorly understood biological roles of secreted and periplasmic glucans, particularly in host–bacteria interactions, osmoadaptation, and cell signaling 17, 19, 20. However, despite glucan being present in many Gram‐negative bacteria, their physiological role has not been established.

Here, we report the discovery of an important role of cyclic β‐(1,2)‐glucan in bacterial iron homeostasis. We show that cyclic β‐(1,2)‐glucan of X. campestris binds to the ferrous form of iron and promotes growth under iron‐limited conditions. Evidence indicates that periplasmic glucan sequesters iron in the periplasm, protecting the bacterial cells by limiting the potential for free iron‐induced toxicity under iron‐replete conditions, and promotes growth under iron‐restricted conditions. Xanthomonas campestris and Escherichia coli mutants blocked in the production of β‐(1,2)‐glucan exhibit growth deficiencies under iron‐restricted conditions and are also sensitive to iron‐induced toxicity.

Results

Cyclic β‐(1,2)‐glucan is required for growth under iron‐restricted conditions

Extracellular polysaccharide (EPS)‐deficient mutants of Xanthomonas campestris pv. campestris (Xcc) (B5 and B12; transposon‐induced mutants in the xanA and gumD, respectively, which were identified by a genetic screen) exhibited overproduction of siderophores, an indicator of iron starvation (Fig EV1A). Further, we determined whether EPS‐deficient mutants of Xcc exhibited growth deficiencies under iron‐restricted conditions. To achieve iron‐limiting conditions, we employed 2,2′‐dipyridyl (DP), a ferrous iron‐specific chelator. In a rich medium (peptone–sucrose; PS), the growth rates of wild‐type and EPS mutant strains (B5 and B12) were similar (doubling times of ~3 h; Appendix Table S1). The EPS‐deficient mutants B5 and B12 exhibited significant growth deficiencies in media containing 2,2′‐dipyridyl, compared to the wild‐type strain (Appendix Table S1). Furthermore, addition of either exogenous iron (50 μM FeSO₄) or EPS isolated from the wild‐type strain rescued the growth deficiency exhibited by either mutant strain (Appendix Table S1). We also performed growth assays with the extracellular ion chelator diethylenetriaminepentaacetic acid (DETAPAC) and impermeant chelator bathophenanthroline disulfonate (BPS) and in low‐iron medium, which further confirmed the iron specificity of the growth defect (Figs 1A–F and EV1B–D; Appendix Table S1; Appendix Fig S1).

Figure EV1. Role of cyclic β‐(1,2)‐glucan in Xanthomonas campestris growth under iron‐restricted conditions.

1. EPS‐ and glucan‐deficient ndvB mutants overproduce siderophore under iron‐restricted conditions. Production of siderophore was detected on chrome azurol sulfonate (CAS) plates. Different strains of Xcc were grown and spotted on CAS plates as indicated and grown for 24 h. Appearance of orange halo indicates production of secreted siderophore.
2. Growth assay of Xcc 8004, ndvB, and ndvB(pHM1J)⁺ in the presence of impermeable iron chelator bathophenanthroline disulfonate (150 μM BPS).
3. Growth assay in 150 μM BPS + 50 μM FeSO₄.
4. Growth assay in 150 μM BPS + 50 μM MnCl₂.
5. Growth assay in 150 μM intracellular ferrous iron chelator 2,2′‐dipyridyl (DP).
6. Growth assay in 150 μM DP + 50 μM FeSO₄.
7. Growth assay in 150 μM DP + 50 μM ZnSO₄.

Data information: (B–G) Data shown are mean ± SD (n = 3).Source data are available online for this figure.

Figure 1. Glucan promotes bacterial growth under iron‐restricted conditions.

• A–F
  Growth analysis of wild‐type Xcc8004, EPS‐deficient mutants (B5 and B12), ndvB, and ndvB (pHM1J)⁺ grown in PS medium (A) or PS medium containing 50 μm extracellular ion chelator diethylenetriaminepentaacetic acid (DETAPAC) (B) or supplemented with either (C) secreted glucan (0.5 mg/ml), (D) xanthan (0.5 mg/ml), (E) FeSO₄ (40 μM), or (F) MnCl₂ (40 μM). Data shown are mean ± SD (n = 3).

Source data are available online for this figure.

To identify the EPS‐associated component that promoted growth under iron‐restricted conditions, we carried out high‐performance liquid chromatography (HPLC) on EPS isolated from the wild‐type Xcc strain on a C‐18 reverse‐phase column. Two major peaks were observed with retention times of 1.4 and 2.1 min corresponding to wild‐type EPS (Fig EV1A). Interestingly, addition of the abundant 1.4‐min fraction, corresponding to xanthan, did not rescue the growth defect exhibited by the EPS‐deficient mutants (Appendix Table S1). In contrast, the 2.1‐min fraction could rescue their growth deficiency under iron‐restricted conditions (Appendix Table S1). Xanthan and glucan fractions isolated from EPS by both HPLC and differential precipitation were reanalyzed by HPLC to confirm single peak (Fig EV1B). Further analysis of the 2.1‐min eluate by matrix‐assisted laser desorption ionization–mass spectrometry (MALDI–MS) and comparison with the control matrix peaks (without glucan sample; Fig EV2A and B) yielded a quasimolecular ion at mass‐to‐charge ratio (m/z) 2632.8 (Fig 2A), which matched that for an [M + Na]⁺ ion of the unsubstituted 16‐member cyclic glucan of Xanthomonas campestris 21. We further performed acid hydrolysis of the 2.1‐min fraction with 1 M HCl, derivatized it with N,O‐bis(trimethylsilyl)trifluoroacetamide (BSTFA), and subjected it to gas chromatography–mass spectrometry (GC–MS). As shown in Fig 1H, a prominent GC peak having an m/z ratio of 204 was detected. A search of the National Institute of Standards and Technology (NIST) database of known compounds yielded a match to glucopyranose pentakis‐O‐trimethylsilyl, indicating that the constituent monomer of the cyclic glucan is glucose (Figs 2B and EV2C).

Figure EV2. Controls for MALDI‐MS and GC‐MS analysis.

• A, B
  MALDI–MS analysis of the control matrix without glucan sample. The m/z values are shown in varied scale. Shown are the m/z scales of 1,000–7,000 (A) and 500–4,500 (B).
• C
  GC–MS analysis of acid‐hydrolyzed and N, O‐bis(trimethylsilyl)trifluoroacetamide (BSTFA)‐derivatized isolated glucan. Peaks in the range of 3.2–12.7 min of retention time (RT) are derivative intermediate of trimethylsilyl (TMS), derivatization agent. The rest of the peaks yielded a match to glucose‐TMS derivatives, indicating that the constituent monomer of the cyclic glucan is glucose.

Source data are available online for this figure.

Figure 2. Identification of cyclic β‐(1,2)‐glucan from Xanthomonas campestris (Xcc) extracellular polysaccharide (EPS).

1. MALDI–MS analysis of the 2.1‐min RT eluted fraction from the wild‐type Xcc8004 EPS. Shown is one quasimolecular ion at mass‐to‐charge ratio (m/z) 2632.8, which matched the calculated mass for an [M + Na]⁺ ion based on an unsubstituted 16‐member cyclic glucan (inset) of Xanthomonas campestris. Peaks with m/z of below 1,100 were matrix‐derived peaks after comparing the MALDI–MS of control matrix only (see also Figs EV2A and B, and EV4D).
2. Top: GC–MS analysis of acid‐hydrolyzed and BSTFA‐derivatized, 2.1‐min RT eluted fraction from the EPS of wild‐type Xcc strain. Bottom: matching spectrum from NIST database, corresponding to glucopyranose pentakis‐O‐trimethylsilyl (see also Fig EV2C).
3. Quantification of total EPS and secreted (S) and periplasmic (P) glucans by phenol–sulfuric acid colorimetric assay from 2‐day‐old cultures. Secreted xanthan and glucan were isolated from the cell‐free culture supernatant by addition of KCl [1% (w/v)] and 2 volumes of ethanol, to precipitate xanthan. Supernatant containing cyclic β‐(1,2) was purified by BioGel P4 size‐exclusion chromatography. To isolate periplasmic glucan, cell pellet was treated with 1% trichloroacetic acid (TCA) and was purified by BioGel P4 size‐exclusion chromatography. Data shown are mean ± SD (n = 3).

Source data are available online for this figure.

In several genera including Xanthomonas and Rhizobiaceae, the glycosyltransferase NdvB (for nodule development B) has been shown to be required for the synthesis of cyclic β‐(1,2)‐glucans from UDP‐glucose 18, 22, 23. To evaluate the contribution of cyclic β‐(1,2)‐glucan to promotion of growth under iron‐restricted conditions, we generated an insertional null mutant (ndvB) in Xcc. Knockout of ndvB compromised the production of the extracellular cyclic β‐(1,2)‐glucan, but had no effect on the production of xanthan (Figs 2C and EV3A). The ndvB mutant exhibited a growth deficiency under iron‐restricted conditions and overproduced siderophore, phenotypes that could be rescued by either exogenous supplementation with iron or glucan (Figs 1A–F and EV1A, E–G; Appendix Table S1). Furthermore, the growth defect exhibited by glucan‐deficient ndvB mutant in medium containing DETAPAC, BPS, and DP could be rescued by addition of FeSO₄ but not by MnCl₂ or ZnSO₄, which further corroborated the iron specificity of the growth defect (Figs 1A–F and EV1E–G; Appendix Table S1). As glucan is normally synthesized under low‐osmotic conditions and accumulates in the periplasmic space of Gram‐negative bacteria 17, 19, we isolated glucan from the periplasmic fraction by size‐exclusion chromatography followed by analysis by HPLC and MALDI–MS. The ndvB mutant strain is compromised in its production of periplasmic glucan (Fig EV3C). Periplasmic glucan isolated from the wild‐type strain exhibited, at HPLC retention time of 2.1 min, a quasimolecular ion with an m/z ratio of 2632.8, identical to that of the secreted extracellular glucan (Fig EV3D).

Figure EV3. Identification of cyclic β‐(1,2)‐glucan from Xcc .

Cyclic β‐(1,2)‐glucan is the EPS‐associated moiety that promotes growth under iron‐restricted conditions. EPS was isolated from 2‐day‐old cultures of wild‐type Xcc8004; glucan‐deficient ndvB mutant; and glucan‐deficient mutant harboring the complementing plasmid [ndvB (pHM1J)⁺]. The isolated EPS was analyzed and separated by high‐performance liquid chromatography (HPLC) on the C‐18 column and eluted with gradient of 10–60% acetonitrile (ACN), pH 7, in 50 mM phosphate buffer.

1. Left: HPLC elution profile of wild‐type Xcc8004 EPS. Shown are the peaks corresponding to 1.4‐ and 2.1‐min RT eluate. Center: HPLC profile of EPS isolated from the ndvB mutant (glucan‐deficient and xanthan‐proficient) strain, as shown by the presence of 1.4‐min RT eluate (xanthan) peak and absence of 2.1‐min RT eluate peak. Right: HPLC elution profile of ndvB(pHM1J)⁺ EPS showing peaks corresponding to 1.4‐ and 2.1‐min RT eluate.
2. HPLC profile of collected 1.4 (xanthan) and 2.1 (cyclic β‐(1,2)‐glucan) RT elute from Xcc8004 EPS. Xanthan and glucan fractions isolated from EPS by either HPLC fractionation or differential precipitation were reanalyzed by HPLC to confirm single peak.
3. EPS‐deficient (B5 and B12) and ndvB mutants are deficient in periplasmic glucan production. Elution profile of periplasmic glucan isolated from different strains of Xcc on size‐exclusion P4 BioGel column. Periplasmic glucan was isolated by treating the cells with 1% TCA, and cell‐free supernatant was neutralized by the addition of ammonium hydroxide, concentrated in a rotary evaporator, and further purified by BioGel P4 size‐exclusion chromatography. Fractions (1.5 ml) were collected at a flow rate of 20 ml/h and assayed for carbohydrate using the phenol–sulfuric acid colorimetric assay. Fractions containing cyclic β‐(1,2)‐glucan were pooled and lyophilized (indicated by arrows).
4. MALDI–MS analysis of purified cyclic β‐(1,2)‐glucan obtained by BioGel P4 size‐exclusion chromatography. Pooled fractions containing the cyclic glucan from the wild‐type Xcc8004 and ndvB (pHM1J)⁺ and the corresponding fractions from the B5, B12, and ndvB mutants were analyzed by MALDI–MS. Shown is the presence of one quasimolecular ion at mass‐to‐charge ratio (m/z) 2632.8 (indicated by arrows), only in the wild‐type Xcc8004 and ndvB (pHM1J)⁺, which matched the calculated mass for an [M + Na]⁺ ion based on an unsubstituted 16‐member cyclic glucan.

Source data are available online for this figure.

Glucan binds ferrous iron

How does glucan promote growth of Xanthomonas under iron‐restricted conditions? A clue came from the observation that the characteristic reddish‐purple color complex formed in the presence of dipyridyl and ferrous iron 24 disappeared in the growth media after addition of either EPS or glucan but not that of xanthan (data shown for glucan; Appendix Fig S2). This observation suggested that a component of EPS may be binding to ferrous iron in the growth medium (Appendix Fig S3).

To check the iron‐binding constituent of EPS, we isolated extracellular glucan and xanthan from wild‐type Xcc and major and minor constituents of EPS, respectively. We used a colorimetric method to determine free iron using the Fe²⁺‐specific chelator ferrozine, which forms a magenta‐colored Fe²⁺–ferrozine complex 25. We performed a series of experiments to assess the ability of different polysaccharides such as lipopolysaccharide (LPS), mannan, and xanthan from the commercially available sources, along with xanthan and glucan isolated from Xcc to bind Fe²⁺. Isolated glucan from Xcc sequestered a significantly higher amount of iron (~20‐fold) compared to other bacterial polysaccharides (Fig 3A). To determine whether the iron‐binding ability is limited to only cyclic β‐(1,2)‐glucan from Xanthomonas, we tested a variety of commercially available glucans from yeast, barley [β‐(1,3)‐glucan], and Euglena gracilis. While all glucans exhibited some iron‐binding ability, it was twofold or more less than that of the periplasmic cyclic β‐(1,2)‐glucan from Xanthomonas (Fig 3B).

Figure 3. Glucans bind Fe²⁺ iron.

1. Iron (Fe²⁺)‐binding potential of different polysaccharides [used at 5.56 μmol (glucose equivalent)] using the Fe²⁺–ferrozine colorimetric assay. Lipopolysaccharide (LPS; E. coli 0111:B4), mannan (Saccharomyces cerevisiae), and isolated and commercial xanthan (purchased from Sigma‐Aldrich), and extracellular glucan from Xcc.
2. Colorimetric iron‐binding assay. Glucan from various sources binds Fe²⁺ iron. Extracellular (ECG) or periplasmic (PPG) glucan isolated from the wild‐type Xcc8004 or the ndvB (pHM1J)⁺; and commercial glucan from yeast, barley, and Euglena (purchased from Sigma‐Aldrich). Glucans from various sources were incubated with FeSO₄ at room temperature for 15 min, and the amount of free iron was determined using the standard concentration of ferrozine–Fe²⁺ complex.
3. Iron binding measured by inductively coupled plasma‐optical emission spectrometry (ICP‐OES). Xanthan and various glucans were mixed with 2 mM FeSO₄ and dialyzed against deionized water before the determination of Fe²⁺ content by ICP‐OES. Data shown are mean ± SD (n ≥ 3). ** indicates P‐value < 0.01 and *** indicates P‐value < 0.001 by paired Student's t‐test.
4. Specificity of glucan–Fe²⁺ interaction. Purified Xcc β‐(1,2)‐glucan was incubated with 100 μM FeSO₄ and increasing concentrations of Ca²⁺, Co²⁺, Ni²⁺, Mn²⁺, Mg²⁺, or Zn²⁺, and free or unbound iron was measured by Fe²⁺–ferrozine colorimetric assay.

Data information: (A, B, D) Data shown are mean ± SD (n = 3).Source data are available online for this figure.

In an independent approach to determine iron binding, glucans (383 μM; 0.38 μmol/ml; 1 mg/ml) were mixed with 2 mM FeSO₄ (2 μmol/ml in 200 μM ascorbic acid) and then dialyzed against 200 μM ascorbic acid (pH 4.2), before Fe²⁺ content was determined by inductively coupled plasma‐optical emission spectrometry (ICP‐OES). All glucans bound 40–70 μmol of Fe²⁺ (40–70 μmol of Fe²⁺/100 μmol of glucan), whereas xanthan exhibited very poor Fe²⁺ binding (Fig 3C). In the wild‐type Xcc strain grown under iron‐replete condition, the molar ratio of iron and periplasmic glucan was ~1, which correlates with in vitro glucan–iron binding assays (Fig 3; Appendix Fig S4). However, under iron‐restricted conditions the iron‐to‐periplasmic glucan ratio was approximately 10‐fold lesser than under iron‐replete condition (~0.1) (Appendix Fig S4).

To determine the specificity of the glucan–Fe²⁺ interaction, we assessed Fe²⁺ binding in the presence of a 20‐fold molar excess of different divalent cations including Mg²⁺, Mn²⁺, Ni²⁺ Co²⁺, and Zn²⁺ and also with the ferric (Fe³⁺) form of iron (Fig 3D; Appendix Fig S5). Figure 3D shows that Mg²⁺, Mn²⁺, Ni²⁺, and Co²⁺ did not exhibit a significant competition with Fe²⁺ for glucan binding, whereas Zn²⁺ in 20‐fold molar excess caused an approximately 60% reduction in the capacity of glucan to bind iron.

Glucan sequesters iron in the periplasm to maintain normal cytoplasmic iron levels

Our findings that glucan promotes growth under iron‐restricted conditions and both extracellular and periplasmic glucans bind Fe²⁺ iron suggest that glucan is involved in maintaining intracellular iron homeostasis. We thus performed a series of experiments to determine the role of glucan in intracellular iron homeostasis. First, we used inductively coupled plasma‐optical emission spectrometry (ICP‐OES) to measure the iron content in wild‐type cells of Xcc, the ndvB mutant, and the EPS‐deficient mutants grown under iron‐replete and iron‐restricted conditions. Growth under iron‐restricted conditions caused an overall reduction in cellular iron levels in the wild type, ndvB mutant, and the EPS‐deficient mutants. However, ndvB mutant strain contained ~2‐fold less iron compared to the wild‐type strain, when grown under iron‐restricted conditions (Fig 4A). Addition of glucan in the iron‐restricted medium could reverse the lower iron content exhibited by the ndvB and EPS‐deficient mutants (Fig 4A). Furthermore, ICP‐OES measurement indicated that the levels of zinc, magnesium, and manganese were unaffected by the lack of glucan biosynthesis (Fig EV4D). We then performed a series of experiments to determine the role of glucan in iron uptake or storage. For iron uptake assays, we measured incorporation of radiolabeled ⁵⁵Fe²⁺ into iron‐deprived cells. The wild type, ndvB mutant, ndvB complemented with pHM1J, and EPS mutants B5 and B12 incorporated the same amount of radiolabeled Fe²⁺ over the 60‐min time‐course of the experiment (Fig EV4C). Since the glucan‐deficient mutant was not deficient in Fe²⁺ uptake, we next wanted to know whether glucan plays a role in iron storage or localization.

Figure 4. Glucan is required to maintain normal cellular iron homeostasis in Xcc.

1. Analysis of total iron levels measured by inductively coupled plasma‐optical emission spectrometry (ICP‐OES). Cells were grown under either rich media (PS), iron‐restricted (PS + 150 μM DP) media, or glucan‐supplemented iron‐restricted media (PS + 150 μM DP + 0.5 mg/ml glucan) till the late exponential growth phase. Bacterial cells were harvested and washed, and then, iron content was determined using ICP‐OES. Data shown are mean ± SD (n = 3). ** indicates P‐value < 0.01 determined by paired Student's t‐test (two‐tailed with equal variance).
2. Co‐elution of periplasmic glucan and radiolabeled iron–glucan complex. Periplasmic glucan was isolated from the ⁵⁵Fe²⁺‐loaded cells and analyzed by BioGel P4 size‐exclusion chromatography. Fractions (1.5 ml) were collected at a flow rate of 20 ml/h and assayed for glucan using the phenol–sulfuric acid colorimetric assay, and radioactivity was measured to detect the presence of ⁵⁵Fe.
3. Iron localization assay using atomic absorption spectroscopy. Two sets of bacterial cells were grown in iron‐restricted media (PS + 150 μM DP) till late exponential phase. Cold iron (50 μM FeSO₄) was added to one set of bacterial cultures and incubated for 1 h at 28°C. Bacterial cells were harvested, washed from both the sets of bacterial cultures. Periplasmic and cytoplasmic fractions were collected and freeze‐dried, and then, iron content in each fraction was determined by inductively coupled plasma‐optical emission spectrometry (ICP‐OES). Iron content is represented on the Y‐axis as microgram of iron present per milligram of proteins in the respective fractions. Data shown are mean ± S.E. (n = 3). * indicates P‐value < 0.05 and ** indicates P‐value < 0.01 by paired Student's t‐test.

Source data are available online for this figure.

Figure EV4. Iron localization assay using radiolabeled iron.

1. Periplasmic and cytoplasmic fractions were isolated from the cells collected over the 60‐min time‐course of the ⁵⁵Fe²⁺ uptake, and radioactivity was measured to detect the presence of ⁵⁵Fe. Incorporated iron is represented on the Y‐axis as picomole of iron present per milligram of proteins in the respective fractions. * indicates P‐value < 0.05, ** indicates P‐value < 0.01, and *** indicates P‐value < 0.001, between the data obtained from mutants and the data obtained from wild‐type and complementing strains, respectively, by paired Student's t‐test.
2. Iron localization assay in the presence of glucan (1 mg/ml) in the uptake medium. Periplasmic and cytoplasmic fractions were isolated from the cells collected over the 60‐min time‐course of the ⁵⁵Fe²⁺ uptake, and radioactivity was measured to detect the presence of ⁵⁵Fe. Incorporated iron is represented on the Y‐axis as picomole of iron present per milligram of proteins in the respective fractions.
3. EPS and glucan biosynthesis mutants did not exhibit significant defect in ferrous iron uptake. Radiolabeled ⁵⁵FeCl₃ was incubated in 1 M L‐ascorbate to reduce it into ⁵⁵Fe²⁺. Ferrous iron transport was initiated after addition of 0.4 μM ⁵⁵Fe²⁺ to the bacterial cell cultures of Xcc 8004, B5, B12, ndvB, and ndvB(pHM1J)⁺ grown under iron‐restricted media (PS + 150 μM DP). The total amount of radiolabeled ⁵⁵Fe²⁺ incorporated into the cells was measured using scintillation counter. The data presented represent means from two independent experiments (each with three replicates).
4. Analysis of total zinc, magnesium, and manganese levels in different strains of Xcc measured by inductively coupled plasma‐optical emission spectrometry (ICP‐OES).

Data information: (A, B, D) Data shown are mean ± SD (n = 3).Source data are available online for this figure.

To understand the fate of radiolabeled iron in the cells, we fractioned the periplasmic and cytoplasmic contents from the iron‐loaded cells and measured the amount of radiolabeled iron. Cells were grown under iron‐restricted conditions in the presence or absence of exogenous glucan and then incubated with radiolabeled ⁵⁵Fe²⁺, and its accumulation in the periplasmic and cytoplasmic fractions was determined by liquid scintillation counting. A significantly higher amount of radiolabeled iron was detected in the periplasmic fraction in the wild type compared to the ndvB and EPS‐deficient mutants (Fig EV4A). By contrast, a higher amount of radiolabeled iron was detected in the cytosolic fraction of the glucan‐deficient ndvB mutant as well as in the EPS‐deficient mutants compared to the wild‐type strain. Furthermore, addition of glucan to the uptake medium rescued the altered radiolabeled iron accumulation exhibited by the ndvB and EPS‐deficient mutants (Fig EV4B). Since the majority of the glucan accumulates in the periplasm in Gram‐negative bacteria, we reasoned that some fraction of the Fe²⁺ that enters the periplasm is sequestered by glucan.

As a first test of this prediction, we isolated periplasmic glucan from the iron‐loaded cells after the incorporation of radiolabeled iron and analyzed it by size‐exclusion chromatography. Elution profile of the glucan and radiolabeled iron indicated that there was considerable overlap of the eluate retention time of radioactivity and periplasmic glucan (within the fractions 25–40; Fig 4B). These results supported the hypothesis that some fraction of the ⁵⁵Fe²⁺ iron entering in the periplasmic space is sequestered by glucan. To provide additional support for this model, iron‐deprived cells obtained after growth under iron‐restricted conditions were loaded with cold iron by preincubating them with media containing 50 μM FeSO₄, followed by washing with medium lacking iron. The iron content of periplasmic glucan isolated from the iron‐deprived and iron‐loaded cells, as well as that in the remaining cell pellet, was then measured by ICP‐OES. The iron content in periplasmic fraction of the glucan‐deficient ndvB mutant was nearly 2.5‐fold less than that of the wild‐type strain. In contrast, a significantly higher amount of iron was detected in the cytosolic fraction of the glucan‐deficient ndvB mutant compared to the wild‐type strain (Fig 4C).

β‐(1,2)‐Glucan protects cell from iron‐induced reactive oxygen species (ROS)

An excess of free iron is known to generate ROS by the Fenton reaction 5. Based on our finding that glucan can sequester ferrous iron in the periplasm and serve as an iron reservoir, we examined the role of glucan in iron‐induced ROS generation. In these experiments, we treated wild‐type and glucan‐deficient Xcc cells with an excess of FeSO₄ (0.25 mM) or FeSO₄ + dipyridyl. The accumulation of ROS was measured by exposing the cells to 2′,7′‐dichlorodihydrofluorescein diacetate (H₂DCFDA), which is converted to the highly fluorescent compound 2′,7′‐dichlorofluorescein (DCF) by ROS, and fluorescence was measured by flow cytometry (Figs 5A–E and 6B) and fluorometry (Fig 6A). There was a significantly higher ROS production as evidenced by increased DCF fluorescence in the ndvB mutant and EPS‐deficient mutants compared to the wild‐type Xcc strain when exposed to excess iron. By contrast, treatment with dipyridyl significantly reduced the ROS levels in these mutants (Figs 5A–E and 6A). Furthermore, addition of either glucan or EPS significantly reduced iron‐induced ROS production, whereas xanthan had little effect on their production (Appendix Table S2). Since ROS causes DNA damage leading to cell death, we examined the viability of Xcc cells exposed to 500 μM FeSO₄. In contrast to wild‐type Xcc cells, ndvB and EPS mutants exhibited increased susceptibility to iron‐induced ROS production (Fig 6C). In addition, we also performed survival assay of bacterial strains under 150 μM FeSO₄, 150 μM FeSO₄ + 1 mM H₂O₂, and 1.5 mM H₂O₂ (Appendix Fig S6A–C). EPS‐ and glucan‐deficient mutants exhibited no significant difference in survival under 150 μM FeSO₄ (Appendix Fig S6A). However, mutants displayed significantly less survival than wild type in 150 μM FeSO₄ + 1 mM H₂O₂ (Appendix Fig S6B), but significantly better survival in 1.5 mM H₂O₂ (Appendix Fig S6C). These results correlate with iron content quantified using ICP‐OES, that is, less intracellular iron content in EPS‐ and glucan‐deficient mutants under PS or iron‐depleted medium but more intracellular iron under iron‐replete media (Fig 4A; Appendix Fig S7).

Figure 5. Role of glucan in the iron‐induced ROS generation.

• A–E
  Xanthomonas campestris: (A) wild‐type Xcc 8004, (B) B5 (EPS‐deficient), (C) B12 EPS‐deficient), (D) ndvB (glucan‐deficient), and (E) ndvB(pHM1J)⁺ (complementing strain); cells were treated in the presence or absence of either FeSO₄, H₂O₂, or FeSO₄ + DP and were analyzed for ROS by monitoring DCF fluorescence with flow cytometry.

Source data are available online for this figure.

Figure 6. Glucan protects bacterial cells from iron‐induced oxidative stress.

1. Xanthomonas campestris cells treated with FeSO₄, H₂O₂, and FeSO₄ + DP were analyzed for ROS by monitoring DCF fluorescence with fluorometry.
2. The percent of ROS‐positive Xanthomonas campestris cells treated with FeSO₄, H₂O₂, and FeSO₄ + DP was determined by flow cytometry at 3 h after treatment. Shown are the mean percentages of ROS‐positive cells exceeding the fluorescence of 99% of untreated cells.
3. Iron sensitivity assay. Cells were grown overnight in PS medium, and cultures were diluted in fresh media at a density of 1 × 10⁶ cells/ml and treated in the absence or presence of 500 μM FeSO₄. The viability of cultures was measured at the indicated time point by dilution plating. Percent survival was determined as the number of viable cells at each time point divided by the number of viable cells before exposure to FeSO₄.

Data information: (A–C) Data shown are mean ± SE (n = 3). * indicates P‐value < 0.05, ** indicates P‐value < 0.01, and *** indicates P‐value < 0.001 compared with data obtained from iron‐treated wild type and mutant by paired Student's t‐test.Source data are available online for this figure.

To further determine the role of glucan in the protection from iron‐induced ROS, we extended our experiments to wild‐type E. coli (MG1655), and ΔmdoG and ΔmdoH mutants blocked in periplasmic glucan production. E. coli cells were grown in low‐osmotic medium, in the absence or presence of FeSO₄ (0.5 mM) or FeSO₄ + dipyridyl, and accumulation of ROS was measured by flow cytometry (Fig 7A and B) or fluorometry (Fig 7C). There was a significantly higher ROS production in the ΔmdoG and ΔmdoH mutants compared to the wild‐type E. coli strain when exposed to high iron.

Figure 7. Role of glucan in the iron‐induced ROS generation conserved in Escherichia coli .

1. Wild‐type E. coli (MG1655) and ΔopgG (ΔmdoG; JW1035) and ΔopgH (ΔmdoH; JW1037) mutants blocked in periplasmic glucan synthesis were grown in LOS medium, and cells were treated in the presence or absence of either FeSO₄, H₂O₂, or FeSO₄ + DP and were analyzed for ROS by monitoring DCF fluorescence with flow cytometry.
2. The percent of ROS‐positive E. coli cells treated with FeSO₄, H₂O₂, and FeSO₄ + DP was determined by flow cytometry at 3 h after treatment. Shown are the mean percentages of ROS‐positive cells exceeding the fluorescence of 99% of untreated cells.
3. Escherichia coli cells treated with FeSO₄, H₂O₂, and FeSO₄ + DP were analyzed for ROS by monitoring DCF fluorescence with fluorometry.

Data information: (B, C) Data shown are mean ± SE (n = 3). *** indicates P‐value < 0.001 compared with data obtained from iron‐treated wild type and mutant by paired Student's t‐test.Source data are available online for this figure.

Discussion

Iron homeostasis in bacteria is tightly regulated as cells must accumulate iron under iron‐replete conditions to support the essential cellular redox functions while also limiting its potential to induce oxidative damage. In this study, we have identified a novel role for bacterial β‐(1,2)‐glucan in sequestering ferrous iron in the periplasm, thus creating an intracellular iron reservoir while also protecting cells against free iron‐induced ROS production. Glucan has a large effect on iron homeostasis since Xcc mutants blocked in glucan production are deficient in growth under iron‐restricted conditions and harbor low levels of intracellular iron (Figs 1 and 4A). Both secreted and periplasmic glucans bind Fe²⁺ and the interaction was specific as other divalent metal ions and ferric form of iron exhibited only weak competition for glucan binding. However, the structural features of interaction between glucans and iron involved remain to be determined. While we find a strong effect of secreted and periplasmic cyclic β‐(1,2)‐glucans in Xcc, it could be possible that this may be a rather common role of this molecule, as glucan is ubiquitously present in many Gram‐negative bacteria.

Previous reports of molecular modeling and function studies have suggested that cyclic β‐(1,2)‐glucan can adopt a conformational structure with polar cavities and cage‐like structure allowing it to sequester “guest molecules” in a noncovalent fashion, including formation of inclusion complexes 18, 26, 27. It is pertinent to note that it has been reported that secreted cyclic glucan provides antibiotic resistance in Pseudomonas aeruginosa, grown on biofilms by sequestering aminoglycoside antibiotics in the periplasm 28.

Although the majority of the iron in bacterial cells is bound to proteins, a considerable amount of free iron exists in the cell in the form of ferrous ions bound to low‐molecular weight iron‐binding ligands. This is generally assumed to form a transit pool of iron in the cell that is trafficked between metalloproteins and other iron storage proteins 6, 29. Several putative low‐molecular weight iron‐binding ligands such as nucleotides, amino acid, and polypeptides have been proposed to be involved in the free iron pool 7, 8. However, the role of these putative iron‐binding ligands in iron trafficking and the cellular iron pool has not been demonstrated.

In this study, we have provided several lines of evidence demonstrating that glucan sequesters ferrous iron in the periplasm and that such iron–glucan complexes are important components of the intracellular free iron pool. First, in our Fe²⁺ uptake assays, a significantly higher amount of iron was detected in the periplasmic fraction of the glucan‐proficient wild‐type Xcc strain, in contrast to the glucan‐deficient mutants (Fig EV4). Second, co‐elution of periplasmic glucan with radiolabeled ⁵⁵Fe²⁺ by size‐exclusion chromatography analysis and measurement of iron content in the glucan isolated from the periplasm indicated that a substantial amount of the ferrous iron is sequestered by the periplasmic glucan (Figs 4 and EV4).

Although glucan is localized primarily in the periplasm, several bacteria also produce secreted glucan 17, 18. We have found that extracellular supplementation of cyclic β‐(1,2)‐glucan could rescue the growth deficiency exhibited by the glucan‐deficient Xcc mutants under iron‐restricted conditions. Our in vitro uptake assays with glucan–iron complexes provided evidence that glucan is imported into the periplasm of glucan‐deficient Xcc mutants, in contrast to the glucan‐proficient wild‐type Xcc strain (Appendix Fig S8). It is possible that similar to siderophore‐mediated uptake, glucan is transported by a yet unidentified outer membrane receptor. It has been reported that Sphingomonas sp., a Gram‐negative bacterium, can transport (import) alginate, a high‐molecular weight polysaccharide (25,000 Da), through complex “superchannel” consisting of flagellin‐like proteins and several TonB‐dependent receptors 30, 31. Previous studies have shown that NdvA/ChvA, an ATP‐binding cassette transporter protein, is involved in the secretion of glucan in the periplasm and to the extracellular medium in Agrobacterium tumefaciens and Rhizobium meliloti 32, 33. We speculate that similar to glucan export, bacteria may also utilize yet unidentified glucan import machinery. Uptake of glucan–Fe²⁺ complexes may not entirely account for the restoration of the growth exhibited by the glucan‐deficient Xcc mutants upon exogenous glucan supplementation. Alternatively, secreted glucan may be involved in sequestering ferrous iron, produced by the cell‐associated ferric reductase, and thereby concentrate ferrous form of iron in close proximity to bacterial cells to facilitate its uptake.

It has been shown that in many bacteria including E. coli, osmotic regulation of glucan synthesis occurs primarily at the level of modulation of glucan synthesis enzymatic activity rather than by regulation of gene expression 34, 35. The production of periplasmic glucan in some bacteria such as Brucella is not affected by osmolarity 35, 36. Interestingly, glucan production analysis indicated that in Xanthomonas, production of glucan is induced by approximately threefold under iron‐restricted conditions and is suppressed by twofold under high‐osmotic conditions (Fig EV5A). Furthermore, E. coli MG1655 strain also exhibited an approximately 25% increase in periplasmic glucan production under iron‐restricted conditions (Fig EV5B). Expression analysis of ndvB and gum genes indicated that there was no significant change in the expression under iron‐restricted conditions. However, under high‐osmotic conditions (PS + potassium glutamate), the expression of ndvB and gum genes was reduced approximately twofold (Fig EV5C). This suggests that in Xanthomonas, increased production of glucan under iron‐restricted conditions appears to occur at a post‐transcriptional level. Interestingly, measurement of periplasmic iron in Xcc indicated that there is an approximately twofold reduction in iron content under high‐osmotic conditions (PS medium containing 250 mM potassium glutamate) compared to low‐osmotic PS medium (Fig EV5D). It is also pertinent to note in this regard that Xcc, which colonizes the plant xylem vessels containing xylem sap, is very low in solutes and presumably low in iron content 11, 13, 37.

Figure EV5. Bacterial cyclic β‐(1,2)‐glucan production induced under iron‐restricted conditions.

1. Periplasmic glucan (PPG) production of Xcc 8004 was assessed under iron‐restricted (PS + 100 μM DP), iron‐replete (PS + 100 μM FeSO₄), and high‐osmotic (PS + 100 μM potassium glutamate) media.
2. Estimation of periplasmic glucan production in E. coli MG1655 strain grown under rich LB and iron‐restricted media.
3. Expression analysis of gumD and ndvB genes under rich PS, iron‐restricted, and high‐osmotic media using qRT–PCR.
4. Periplasmic iron estimation of the wild‐type Xcc 8004 grown under low‐osmotic PS media or PS media supplemented with 250 mM potassium glutamate to make high‐osmotic PS media.
5. Glucan and xanthan estimation by the acid hydrolysis method. Quantity of purified glucan and xanthan was determined before and after dialysis of polysaccharide–Fe²⁺ complex against 200 μM L‐ascorbate through a 2‐kDa dialysis membrane.
6. Alkaline phosphatase assay of isolated periplasmic and cytoplasmic fractions from different Xcc strains.
7. Glucose‐6‐phosphate dehydrogenase assay of isolated periplasmic and cytoplasmic fractions from different Xcc strains.

Data information: Data are mean ± SD (n = 3). *P < 0.05, **P < 0.01, and ***P < 0.001, calculated by paired Student's t‐test.Source data are available online for this figure.

We have shown that glucan‐deficient mutant of Xcc as well as E. coli contains higher levels of ROS when exposed to excess iron. In several bacteria, it has been shown that iron storage protein ferritin is involved in storage of iron and promotes growth under low‐iron conditions 1. Furthermore, it has been shown earlier that an excess of free iron leads to oxidative stress and other deleterious effects as it catalyzes OH‐radical formation, which can be suppressed by complexing iron with chelators such as dipyridyl or ferrozine or overexpressing the iron storage protein ferritin 5, 29, 38, 39. We thus propose that glucan promotes growth under iron ‐restricted conditions to a certain extent; however, its major contribution appears to buffer down the potential deleterious effect of iron, as suggested by the experiments on ROS sensitivity.

Materials and Methods

Bacterial strains and culture conditions

Xanthomonas campestris pv. campestris (Xcc) strains were grown on peptone–sucrose agar (PSA) or in PS broth at 28°C shaking at 200 rpm, as described previously 40. Escherichia coli strains were grown in Luria–Bertani (LB) medium at 37°C. For glucan production in E. coli and Pseudomonas syringae pv. syringae (Pss), strains were grown in low‐osmotic salt medium (LOS) 41 at 28°C for Pss and 37°C for E. coli. For Agrobacterium tumefaciens, the strain was grown in low‐osmotic TY medium [0.5% tryptone and 0.3% yeast extract] as described earlier 35. Bacterial strains and plasmids used in this study are listed in Appendix Table S3.

Screen for isolating siderophore‐overproducing mutants of Xcc

Transposon (Tn5) mutagenesis was done in the Xcc8004 wild‐type background by introducing the transposon mutagenesis suicide plasmid pRL27 from E. coli by conjugation 42. For siderophore production assay, Xcc strains were grown on peptone–sucrose agar (PSA)–chrome azurol sulfonate (CAS) siderophore indicator plates containing 100 μM 2,2′‐dipyridyl 43, 44. Plates were incubated at 28°C for 24 h. Appearance of orange halo indicative of secreted siderophore production was scored by measuring the halo diameter. A total of 10,000 colonies from fifty independent matings were screened on PSA‐CAS plates to identify siderophore‐overproducing mutants. We identified seven independent siderophore‐overproducing transposon‐induced mutants in xanA, xanB, and gumD genes involved in extracellular polysaccharide (EPS) biosynthesis. In Xanthomonas, xan and gum operons are involved in the synthesis of extracellular polysaccharide xanthan and are highly conserved among Xanthomonas spp. Two transposon‐induced mutants, B5 (xanA::mTn5) and B12 (gumD::mTn5), that were deficient in extracellular polysaccharide and overproduced siderophore were further characterized in this study. The transposon insertions in B5 and B12 were located at positions corresponding to codon 343 and codon 342 of the xanA (locus tag: XC_3608; 450 a.a.) and gumD (locus tag: XC_1660).

Siderophore production and growth assays

For siderophore production assay, Xcc strains were grown on peptone–sucrose agar (PSA)–chrome azurol sulfonate (CAS) siderophore indicator plates containing 100 μM 2,2′‐dipyridyl as described previously 43. Plates were incubated at 28°C for 24 h. Appearance of orange halo indicative of secreted siderophore production was scored by measuring the halo diameter. For growth assay, Xcc cells were grown in 5 ml PS liquid media to exponential phase (OD₆₀₀ = 0.6), and 0.2% culture was inoculated in 20 ml fresh PS medium with or without 100 μM 2,2′‐dipyridyl. For supplementation experiments, FeSO₄ (50 μM), EPS (0.5–1.8 mg/ml), glucan (0.3–1.8 mg/ml), and xanthan (0.5–3.6 mg/ml) were added to the cultures and shaken at 28°C, 200 rpm. OD₆₀₀ was measured using a UV/visible spectrophotometer (Ultrospec 2100 pro; Cambridge, UK). Detailed methods for screening for isolating siderophore‐overproducing mutants of Xcc and construction of ndvB mutant are described in the Appendix.

Construction of ndvB mutant of Xcc

All standard molecular biology and genetics were performed as described previously 45. PCR amplifications were performed with Taq polymerase (Thermo Scientific) and AccuTaq polymerase (Sigma‐Aldrich, USA) according to the manufacturer's instructions. Restriction digestion and ligation reactions were performed with enzymes obtained from New England Biolabs (USA) as per the manufacturer's instructions. To obtain the ndvB mutant of Xcc, a 500‐bp internal fragment of the ndvB gene was PCR‐amplified with primers NDV pk18KO F EcoRI: GCGAATTCGTTCGAGTTTTCCGCCGGAAAGCACGAT and NDV pk18KO R HindIII: GCAAGCTTTACTGCGCAT AGTCCTGCGC GGCA, and cloned in pK18mob plasmid 46. The resulting recombinant plasmid was electroporated into competent cells of Xcc8004 wild‐type strain to obtain Km^r single recombinant, which was further confirmed by PCR and sequencing. Integration of pK18mob results in nonpolar mutation. For complementation of glucan‐deficient ndvB mutant, the wild‐type ndvB gene was amplified from the Xcc 8004 wild‐type strain using primer pairs NDV Comp HindIII: GCAAGCTTAGGAGGACAGCTGTGTCTGCAGCTTCCCCCGCACC and NDV Comp R EcoRI: GCGAATTCTTAGCCCGGC AACCGTACCT GGAGCT and cloned in broad host range pHM1 vector.

Separation, identification, and quantification of cyclic β‐(1,2)‐glucan

Extracellular polysaccharide was isolated from the cell‐free culture supernatant of wild‐type Xcc8004 strain by acetone precipitation as described earlier 47. In brief, EPS was precipitated by adding two volumes of ice‐cold acetone and left overnight at 4°C, and centrifuged at 7,969 g for 20 min. The collected EPS precipitate was dried, and the dry weight was determined. The total carbohydrate content was estimated by the phenol–sulfuric acid method using D‐glucose as the standard. To identify the EPS‐associated component that promotes growth under iron‐restricted conditions, EPS was fractionated on a C‐18 reverse‐phase HPLC column (Agilent 1100 Series, USA) with a gradient of 50 mM phosphate buffer with 10% acetonitrile (ACN), pH 7, as (A) and 60% ACN with 50 mM phosphate buffer, pH 7, as (B) at a flow rate of 1.0 ml/min at 25°C for 30 min. The sample injected was 50 μl from 1 mg/ml of stock equally. The separated components were monitored by a UV detector at a range of 214 nm. Data recording and processing was done using ChemStation software (Agilent 1100). Two major peaks were observed at 1.4‐min RT eluate fraction corresponding to standard xanthan, and 2.1‐min RT eluate, which complemented the growth deficiency of EPS‐deficient mutants of Xcc. Xanthan and glucan fractions isolated from EPS by either HPLC fractionation or differential precipitation were reanalyzed by HPLC to confirm single peak (Fig EV3B).

Purification, separation, and quantification of secreted and periplasmic cyclic β‐(1,2)‐glucans

Purification and separation of secreted and periplasmic cyclic β‐(1,2)‐glucans from Xcc were performed as described earlier 27, 48, 49. Cyclic glucans present in the cell‐free culture supernatants were purified and analyzed after precipitation and removal of xanthan by centrifugation by the addition of 1% KCl (w/v) final concentration and 2 volumes of ethanol. The supernatant containing glucan was lyophilized and dissolved in 5% (v/v) acetic acid and purified by size‐exclusion chromatography on a BioGel P4 column. To isolate periplasmic glucan, cells were pelleted at 8,000 g for 30 min and washed with 30 mM Tris–HCl buffer, pH 8, and the pellets were extracted with 1% trichloroacetic acid (TCA) to release periplasmic glucan. The cells were then removed by centrifugation and the supernatant was neutralized by the addition of ammonium hydroxide, and the neutralized extract was concentrated by rotary evaporation. The dried supernatant was dissolved in 5% (v/v) acetic acid, and glucan was purified by separation on size‐exclusion chromatography. For BioGel P4 size‐exclusion chromatography, a column (1.5 × 42 cm) was equilibrated and eluted with 5% (v/v) acetic acid. Fractions (1.5 ml) were collected at a flow rate of 20 ml/h and assayed for carbohydrate content by the phenol–sulfuric acid method using d‐glucose as the standard using the phenol–sulfuric acid method [50; Fig EV3C]. Fractions containing cyclic glucans were pooled and lyophilized. Purified glucan was further analyzed by MALDI–MS and GC–MS as described below.

Matrix‐assisted laser desorption ionization–mass spectrometry (MALDI–MS)

Matrix‐assisted laser desorption ionization–mass spectrometry analysis was done as described previously 21. The lyophilized 2.1‐min RT eluate fraction or purified glucan was redissolved in double‐distilled water and mixed in a ratio of 1:5 with the matrix (2,5‐dihydroxybenzoic acid, 10 g/l dissolved in 10% aqueous ethanol solution). One microliter of the resulting solution was deposited onto a stainless steel target, and the solvent was evaporated under a gentle stream of warm air. Experiment was done on a time‐of‐flight mass spectrometer (Bruker Daltonics Ultraflex III, Germany) equipped with nitrogen laser (337 nm wavelength, 2.5 ns pulse width). Analysis of the spectrum was done based on quasimolecular ions obtained by the MALDI–MS method. The 2.1‐min RT eluate and purified glucan gave one quasimolecular ion at mass‐to‐charge ratio (m/z) 2632.8, which matched the calculated mass for an [M + Na]⁺ ion based on an unsubstituted 16‐member cyclic glucan of Xanthomonas campestris 29. As a control, we analyzed the control matrix without glucan sample. The m/z peaks in the range of 500–1,100 were derived from the matrix (Fig EV2A and B).

Gas chromatography–mass spectrometry

The 2.1‐min RT eluate fraction or purified glucan was hydrolyzed with 1 M HCl for 4 h at 100°C and derivatized with N,O‐bis(trimethylsilyl)trifluoroacetamide (BSTFA) by incubating the sample in BSTFA and acetonitrile at 60°C for 80 min and subjected to GC–MS on a Shimadzu GC‐2010 Plus system (Shimadzu Corporation, Japan). The MS data (total ion chromatogram (TIC)) were acquired in the full‐scan mode (m/z of 50–500) at a scan rate of 1,000 amu using the electron ionization (EI) with electron energy of 70 eV. The acquired spectrum was searched against standard NIST‐05 library/WILEY (Fig EV2C). The list of complete library search report of GC–MS peaks. Peaks in the range of 3.2–12.7 min of retention time (RT) are derivative intermediate of trimethylsilyl (TMS), derivatization agent. The rest of the peaks yielded a match to glucose‐TMS derivatives, indicating that the constituent monomer of the cyclic glucan is glucose (Fig EV2C).

Glucan–iron binding assays

For colorimetric iron‐binding assay, glucan isolated from Xcc and commercial glucan from yeast, barley, and Euglena were incubated with 100–1,000 μM of FeSO₄ (100 nmol to 1 μmol in 200 μM ascorbic acid, pH 4.2) for 15 min at room temperature. To determine unbound or free iron, ferrozine (Sigma‐Aldrich, St. Louis, USA) was added at a final concentration of 1 mM (1 μmol) as a chromogenic ferrous iron chelator as described previously 25. Fe²⁺ chelator ferrozine forms a magenta‐colored Fe²⁺–ferrozine complex. The amount of free iron was determined using the standard concentration of ferrozine–Fe²⁺ complex and was expressed in μM using a molar extinction coefficient of 27,900 M⁻¹ cm⁻¹ at 562 nm 51.

For competitive binding assays, glucan (1 mg/ml) was incubated with 100 μM FeSO₄ (in 200 μM ascorbic acid, pH 4.2) in the presence of increasing concentrations of Mg²⁺, Mn²⁺, Ni²⁺ Co²⁺, and Zn²⁺ for 15 min. In the order‐of‐addition experiment, glucan was first preincubated either with 100 μM FeSO₄ or with different divalent metal cations for 15 min, followed by addition of either metal cations or FeSO₄, respectively, for 15 min. The amount of iron bound was measured by either the ferrozine colorimetric or ICP‐OES method as described for the glucan–iron binding assay.

Membrane dialysis of Fe–glucan complex and binding assessment using ICP‐OES

We added excess of FeSO₄ (2 μmol) to glucan solution (1 mg/ml; 0.38 μmol) and performed dialysis with a 2‐kDa membrane (Spectrum Laboratories, Inc., Rancho Dominguez, CA, USA) against 200 μM ascorbic acid (pH 4.2). We also quantified glucan before and after the dialysis and did not observe substantial loss of glucan (Fig EV5E). However, all the unbound iron was removed, as we did not observe any leftover iron in negative control after the dialysis (Appendix Table S4). After dialysis, 10 μl was taken out separately for quantification by phenol–sulfuric acid carbohydrate assays 50, and the rest of the sample was dried in a SpeedVac concentrator. The dried samples were dissolved in 30% HNO₃ at 80°C overnight and diluted 10‐fold with MilliQ water. Iron content was determined by ICP‐OES (JY 2000 sequential ICP‐OES spectrometer; Jobin Yvon, Horiba, France). Quantification was carried out against aqueous standard of iron traceable to NIST (National Institute of Standards and Technology, India).

Iron uptake and localization assays

In vitro iron uptake assay was done as described previously 11, 12. In brief, cells were grown in iron‐restricted medium (PS + 150 μM DP) for 24 h, washed with 50 mM phosphate buffer, pH 7.4, and resuspended in Chelex‐100 (Sigma)‐treated PS to an OD₆₀₀ of 1.0 and kept at 28°C for 5 min. Iron (⁵⁵Fe²⁺) transport was initiated by addition of 50 μM of ascorbate‐reduced ⁵⁵FeCl₃ (specific activity 5 mCi/mg; BARC, Mumbai, India) by diluting the stock ten times with 1 M sodium ascorbate (to convert it as ⁵⁵Fe²⁺). Further, radiolabeled ferrous iron‐added cells (in Chelex‐100‐treated PS containing 500 μM ascorbic acid, pH 4.2) were incubated at room temperature. At different time points, 200 μl of the cell suspensions was layered onto di‐butyl phthalate–di‐octyl phthalate (1:1) mixture and centrifuged at 23,426 g for 1 min to stop the uptake during the 60‐min time‐course of the experiment. The aqueous layer was aspirated, and the pellet was resuspended in 1% (v/v) Triton X‐100, and accumulation of ⁵⁵Fe in the cells was determined by liquid scintillation counting (Tri‐Carb 2910 TR Liquid Scintillation Analyzer; Perkin Elmer, USA).

For iron localization, periplasmic and cytoplasmic extracts were isolated from the iron‐labeled cells using the PeriPreps™ Periplasting Kit (Epicentre, Madison, USA) as per the manufacturer's instructions, and ⁵⁵Fe in the extracts was determined by liquid scintillation counting. We performed alkaline phosphatase assay (a marker for periplasmic fraction) and glucose‐6‐phosphate dehydrogenase assay (a marker of cytoplasmic fraction) 52 of isolated periplasmic and cytoplasmic fractions from different Xcc strains to confirm that the periplasmic and cytoplasmic fractions have been separated faithfully and equivalently for the EPS/gumD mutants and the wild‐type strain (Fig EV5F–G). For the determination of iron bound to periplasmic glucan, periplasmic fraction isolated from the iron‐labeled cells was treated with 1% TCA and separated by BioGel P4 size‐exclusion chromatography, as described for the periplasmic glucan isolation. Glucan and ⁵⁵Fe in the size‐exclusion chromatography fractions were determined by the phenol–sulfuric acid method and liquid scintillation counting.

Estimation of intracellular iron content

For the determination of total iron content, cells were grown under iron‐replete (PS) or iron‐restricted medium (PS + 150 μM DP) to the late exponential phase and were washed twice with phosphate‐buffered saline (PBS), and iron content was determined by ICP‐OES.

For determining iron content in the periplasmic glucan, cells were deprived of iron by growing in iron‐restricted medium (PS + 150 μM DP) to the late exponential phase and then incubated in fresh PS media containing 200 μM ascorbic acid, supplemented with 50 μM FeSO₄. Periplasmic glucan was extracted by 1% TCA treatment and purified by BioGel P4 size‐exclusion chromatography, and iron content was determined by ICP‐OES.

Measurement of ROS

To assess ROS production in response to excess FeSO₄ treatment, we used the cell‐permeable nonfluorescent probe 2′,7′‐dichlorodihydrofluorescein diacetate (H₂DCFDA; Sigma‐Aldrich), which is converted to the fluorescent compound 2′,7′‐dichlorofluorescein (DCF) by ROS. Cultures (1 × 10⁹ cells/ml) were treated for 3 h with either FeSO₄ (250 μM for Xcc and 500 μM for E. coli), FeSO4 + DP (250 μM for Xcc and 500 μM for E. coli), or H₂O₂ (10 mM for Xcc and 2 mM for E. coli), and H₂DCFDA was added at a final concentration of 10 μM, and the cultures were incubated for 30 min in the dark at room temperature. Fluorescence was measured with a fluorescence reader (482 nm [excitation]/512 nm [emission]) (Synergy H1 Multi‐Mode Reader; BioTek, VT, USA) or by flow cytometry (Aria™ III flow cytometer; Becton Dickinson, CA, USA).

Detection of ROS production by flow cytometry

To assess the effect of exogenous EPS and glucan in ROS production in response to FeSO₄ treatment, Xcc cells were grown in the presence or absence of EPS and xanthan (0.9–1.8 mg/ml) to a cell density of 1 × 10⁹ cells/ml, as described in the growth experiments, and treated with FeSO₄ (250 μM for Xcc and 500 μM for E. coli), FeSO4 + DP (250 μM for Xcc and 500 μM for E. coli), or H₂O₂ (10 mM for Xcc and 2 mM for E. coli), for 3 h. The H2DCFDA was added at a final concentration of 10 μM and the cell cultures were incubated for 30 min in the dark at room temperature, and fluorescence was analyzed by flow cytometry (Aria™ III flow cytometer), equipped with a 488‐nm argon laser for excitation, 515 ± 15‐nm (with 495‐nm long‐pass mirror for FITC fluorescence) emission filter, and with a 100‐μm nozzle at 20 psi. The following PMT voltages were used: 401 (FSC), 500 (SSC), and 695 (FITC) for Xcc and 375 (FSC), 451 (SSC), and 660 (FITC) for E. coli. At least 10,000 cells were collected for each sample. The percent of positive cells was determined with FlowJo software and reflects the number of ROS‐positive cells exceeding the fluorescence of 99% of untreated cells. Further, the cells grown under different concentrations of EPS, xanthan, and dipyridyl as described in growth assay experimental procedure were subjected to excess FeSO₄ treatment and analyzed for ROS‐positive cells. Hydrogen peroxide, a well‐known ROS producer inside the cells, is used as positive control. Xanthomonas and E. coli cells were treated with 10 mM and 2 mM of H₂O₂, respectively, for 3 h, and we followed the same procedure of ROS detection as FeSO₄.

NaOCl and NiSO₄ treatment for measurement of ROS

Xcc cultures were treated with NaOCl in three stages. In brief, Xcc cultures were grown to a density of 1 × 10⁹ cells/ml and treated with freshly prepared NaOCl solution and incubated for 40 min in the dark at 30°C with gentle shaking. The concentrations of NaOCl used for exposure at the beginning of each stage were [in mmol (g cell⁻¹)] 0.66 (stage I), 4.65 (stage II), and 20 (stage III). After the third incubation, the free chlorine left in the flasks was quenched using sterile sodium thiosulfate. The cells were then treated with H₂DCFDA at a final concentration of 10 μM and incubated for 30 min in the dark at room temperature, and fluorescence was analyzed by flow cytometry as described above by treating with 2′,7′‐dichlorodihydrofluorescein diacetate (H₂DCFDA) as described above for the quantification of iron‐induced ROS‐positive cells. For NiSO₄ treatment, cells were treated with 100 μM NiSO₄ for 3 h. The H₂DCFDA was added at a final concentration of 10 μM and the cells were incubated for 30 min in the dark at room temperature, and fluorescence was analyzed by flow cytometry.

Survival assay against iron toxicity

Xcc cultures were grown to a cell density of 1 × 10⁹ cells/ml in 20 ml PS medium and treated with 500 μM FeSO₄. The number of surviving cells was determined by dilution plating on PSA plates from 0‐ to 12‐h cultures, at 2‐h time intervals. Percent survivability was calculated as the number of colonies obtained in the treated versus untreated cultures.

Alkaline phosphatase assay

Cell normalized periplasmic and cytoplasmic fractions were isolated from different strains of Xcc as mentioned above. 50 nmol of substrate (p‐nitrophenyl phosphate) was added to each fraction of protein in 100 μM glycine buffer (pH 10.4). Equal amounts of proteins were taken from both cytoplasmic and periplasmic fractions. Reaction samples were incubated in the dark at 37°C for 60 min. 50 μl of 3 M sodium hydroxide solution per 200 μl reaction volume was used to stop the reaction. Absorbance was monitored at 405 nm, and the enzyme activity was calculated from the standard prepared with known amount of pNPP and commercial alkaline phosphatase (NEB).

Statistical analysis

P‐values were determined by an unpaired two‐sample Student's t‐test, assuming unequal variances, by using the Microsoft Office Excel data analysis tool. P < 0.05 was considered statistically significant.

Author contributions

SJ and SSP conducted the experiments and analyzed the data. AM performed and analyzed ITC experiments to confirm glucan–iron interaction. BBP constructed the ndvB mutant and complementing clones. SC designed the experiments and wrote the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

Source Data for Expanded View

Review Process File

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

Source Data for Figure 6

Source Data for Figure 7

EMBO Reports (2018) 19: 172–186