SusE facilitates starch uptake independent of starch binding in B. thetaiotaomicron

Summary

The Bacteroides thetaiotaomicron starch utilization system (Sus) is a model system for nutrient acquisition by gut Bacteroidetes, a dominant phylum of gut bacteria. The Sus includes SusCDEFG, which assemble on the cell surface to capture, degrade and import starch. While SusD is an essential starch-binding protein, the precise role(s) of the partially homologous starch-binding proteins SusE and SusF has remained elusive. We previously reported that a non-binding version of SusD (SusD*) supports growth on starch when other members of the multi-protein complex are present. Here we demonstrate that SusE supports SusD* growth on maltooligosaccharides, and determine the domains of SusE essential for this function. Furthermore, we demonstrate that SusE does not need to bind starch to support growth in the presence of SusD*, suggesting that the assembly of SusCDE is most important for maltooligosaccharide uptake in this context. However, starch binding by proteins SusDEF directs the uptake of maltooligosaccharides of specific lengths, suggesting that these proteins equip the cell to scavenge a range of starch fragments. These data demonstrate that the assembly of core Sus proteins SusCDE is secondary to their glycan binding roles, but glycan binding by Sus proteins may fine tune the selection of glycans in the environment.

Graphical abstract

Introduction

The human gut microbiota performs critical tasks that promote host health and development (Round & Mazmanian, 2009, Flint et al., 2008, Wardwell et al., 2011, Stappenbeck et al., 2002) including the mutualistic break down of complex carbohydrates (fiber) from our diet, (Flint et al., 2008, Flint et al., 2017). Dietary fiber processing begins at the bacterial cell surface via the action of one or more glycoside hydrolases or polysaccharide lyases that liberate smaller oligosaccharides that are harvested by the same or neighboring bacterial species (Cockburn & Koropatkin, 2016). For example, organisms in the gut such as Ruminococcus bromii process resistant starch into small fragments that can then cross-feed other species (Ze et al., 2012). However, other bacteria such as Bacteroides thetaiotaomicron may employ a more “selfish” mechanism for the processing of specific glycans such as yeast cell wall α-mannan in which the bacterium breaks α-mannan into fragments that it can select for uptake, limiting cross-feeding (Cuskin et al., 2015). Understanding the molecular mechanisms employed by gut microbes to utilize dietary carbohydrates can foster the development of dietary strategies to manipulate the microbiota and improve health.

The Gram-negative Bacteroidetes are abundant members of the intestinal ecosystem in part due to their ability to efficiently acquire carbohydrates (Martens et al., 2009, Ding & Schloss, 2014). The Bacteroidetes employ complexes of proteins, termed Sus-like systems, that localize to the cell surface and act in concert to bind, degrade, and import glycans. Sus-like systems are encoded in polysaccharide utilization loci (PUL), gene clusters that are transcriptionally activated in response to a distinct carbohydrate substrate (Martens et al., 2009). Some Bacteroidetes dedicate ~ 20% of their genomes to encoding PUL (Martens et al., 2011). All PUL have at least one pair of proteins that share homology to SusC, a putative TonB-dependent transporter, and SusD, a starch-binding protein, from the Starch utilization system (Sus) of Bacteroides thetaiotaomicron (Martens et al., 2009, Terrapon et al., 2015). Detailed biochemical studies of the glycolytic machinery and glycan-binding proteins from PUL that target pectin, β-glucan, xyloglucan, chitin, and xylan, among others, have helped elucidate a general model of these systems by which carbohydrates are initially degraded at the cell surface and oligosaccharides are selected and imported into the cell via a SusCD-like complex (Ndeh et al., 2017, Tamura et al., 2017, Larsbrink et al., 2014, Larsbrink et al., 2016, Rogowski et al., 2015).

The B. thetaiotaomicron Sus is an eight-gene locus that encodes five outermembrane proteins, SusCDEFG, involved in starch uptake (Tancula et al., 1992) (Fig. 1). X-ray crystallographic structures of SusDEFG revealed eight distinct starch-binding sites among the four proteins (Koropatkin et al., 2008, Koropatkin & Smith, 2010, Cameron et al., 2012). SusG is the sole extracellular Sus enzyme and is required to process starch into maltooligosaccharides that are imported via the putative TonB-dependent transporter (TBDT) SusC (Reeves et al., 1996, Shipman et al., 1999). In the periplasm, maltooligosaccharides are depolymerized by SusA, a neopullulanase, and SusB, an α-glucosidase (Smith & Salyers, 1991, D'Elia & Salyers, 1996a, Gloster et al., 2008). The disaccharide maltose is recognized by the regulatory protein SusR, and activates sus transcription (D'Elia & Salyers, 1996b).

Figure 1. Overview of the Starch Utilization System (Sus) in B. thetaiotaomicron.

Starch is bound by the starch-binding lipoproteins SusDEF and the α-amylase SusG cleaves the polysaccharide to generate maltooligosaccharides that can be internalized by the TonB-dependent transporter SusC. The neopullulanase and α-glucosidase SusA and SusB, respectively, process the maltooligosaccharides to glucose that is introduced to the cytoplasm through an unknown transporter. The inner membrane-spanning protein SusR senses the disaccharide maltose and subsequently drives transcription of the sus locus.

One important role of SusD during glycan transport is the capture of maltooligosaccharides for efficient import via SusC, which leads to increased levels of these sugars in the periplasm and triggers transcriptional activation of sus via SusR (Cameron et al., 2014). We have demonstrated that cells with a starch- binding-deficient allele of SusD, termed SusD*, are much less sensitive to the presence of maltooligosaccharides in the environment, requiring 100–1000× more sugar to achieve maximum transcription of the sus operon. SusD* expressing cells also lag longer than SusD-expressing cells when cultured on starch or maltooligosaccharides but not on glucose. However, the addition of small amounts of maltose, which upregulates sus and but does not require Sus for uptake, partially relieves the lag seen with SusD* cells. These data suggested to us that the single starch-binding site of SusD is not required during steady-state growth of B. thetaiotaomicron on starch and maltooligosaccharides (i.e. when the sus operon is already upregulated) implying that SusE and SusF may provide substrate binding during transport (Cameron et al., 2014). Seminal work performed by Salyers and colleagues demonstrated that SusC and SusD interact and that SusEF may modulate the SusCD interaction (Cho & Salyers, 2001). The recent structures of two SusCD-like complexes from B. thetaiotaomicron revealed that the SusD-like protein sits on top of the SusC-like transporter, with the putative substrate-binding site facing the channel (Glenwright et al., 2017). In one of these complexes, multidomain PUL-encoded proteins resembling SusE and SusF co-purified with the transporter, suggesting that other PUL-encoded proteins may interact with the SusCD-like complex. Co-immunoprecipitation of Sus-like proteins that target host sugars such as heparin, heparin sulfate, and other N-linked glycoproteins also demonstrate that the cell-surface glycan-binding proteins physically interact (Cartmell et al., 2017, Renzi et al., 2011).

In this study, we investigated how the SusDEF lipoproteins and their starch-binding sites contribute to starch and maltooligosaccharide utilization by B. thetaiotaomicron. Using targeted mutations in SusDEF, we demonstrate that only SusE can compensate for the loss of starch binding by SusD (SusD*) during growth on maltoheptaose. Most strikingly, we observed that the starch-binding function of SusE is not required to support growth of the SusD* mutant on maltoheptaose. Finally we reveal how the SusDEF starch-binding lipoproteins drive the preference of the Sus complex for maltooligosaccharides of different size ranges. Together, these results show that SusD and SusE do not require starch-binding sites to direct maltooligosaccharides through SusC, yet SusDEF have a profound impact on the selection of maltooligosaccharides. These insights enhance our understanding of polysaccharide substrate capture by B. thetaiotaomicron and, more broadly, carbohydrate utilization in the Bacteroidetes.

Results

SusE compensates for the loss of the SusD starch-binding site during growth on maltoheptaose

Enzymatic cleavage by the SusG enzyme liberates starch-derived maltooligosaccharides that are the primary substrates of the remaining four Sus outermembrane proteins. To address the roles of SusE and SusF during growth on maltoheptaose, a model maltooligosaccharide, we created a series of genetic deletions and point mutants in the genes encoding these proteins and recombined these back into the sus operon via allelic exchange in a Δtdk strain of B. thetaiotaomicron, referred to throughout as the wild-type strain. Alleles of susDEFG that have been mutated to abolish starch binding are labeled as Sus* mutants, and all have been previously characterized (Table 1). All strains used in this study or created within our previous studies are listed in Table S1. To determine how the SusEF proteins in our strains affect growth on starch or maltooligosaccharides, cell growth was monitored anaerobically in a 96-well plate reader. For each growth experiment, all strains were back-diluted 1:200 from minimal media with glucose or maltose, as noted, into both the test substrate (starch or maltooligosaccharide) as well as a glucose control. All of the 34 mutants reported in this study grew as the parent strain on glucose (Fig. S1), and achieve exponential phase at nearly the same time, verifying that all strains in each set of growth experiments were started at the same O.D. Therefore despite the limitations of the plate reader in resolving growth at low O.D., we feel confident that the dramatic differences in lag that we see with some of our phenotypes reflects biological differences among the strains and is not an artifact of our experimental set-up. When necessary, we performed a second growth experiment in culture tubes in order to better resolve strain difference.

Table 1.

Mutant Sus Alleles Used In This Study

Mutant allele	Mutations (by protein residue number)	Effect	Reference
SusD*	W98A/N101A/W320A	No starch binding	(Cameron et al., 2014)
SusEb*	W192/K221A/Y229A/N252A	No starch binding at the Eb domain	(Cameron et al., 2012)
SusEc*	R326A/W336A/R350A	No starch binding at the Ec domain	(Cameron et al., 2012)
SusE*	W192/K221A/Y229A/N252A/R326A/W336A/R250A	No starch binding at the Eb or Ec domains	(Cameron et al., 2012)
SusF*	W177A/K208A/W222A/D231A/W287A/K323A/N356A/W396A/W442A/R456A	No starch binding at the Fa, Fb or Fc domains	(Cameron et al., 2012)
SusGD498N	D498N	Catalytically inactive (nucleophile mutant) SusG	(Koropatkin & Smith, 2010)
SusEC21A	C21A	Periplasmically localized SusE	(Cameron et al., 2012)
SusFC20A	C20A	Periplasmically localized SusF	(Cameron et al., 2012)

As previously observed, B. thetaiotaomicron cells expressing SusD* grow on 2.5 mg/ml (2.17 mM) maltoheptaose despite an extended lag compared to the wild-type parent strain (Fig. 2a). This lag is associated with reduced transcriptional activation of the sus operon, presumably due to inefficient maltooligosaccharide uptake and reduced glycan levels in the periplasm (Cameron et al., 2014). Growth on maltoheptaose is not dependent on extracellular processing by the surface enzyme SusG as SusD*ΔG cells grow the same as the SusD* strain. Additionally, cells lacking SusDEFG cannot grow on maltoheptaose, demonstrating that SusC alone is not sufficient to support growth on maltoheptaose. In order to better resolve the apparent lag between the wild type and SusD* strain, a second growth experiment with biological triplicate cultures on 2.5mg/ml maltoheptaose was performed in culture tubes and the O.D.₆₀₀ was assessed in a spectrophotometer to resolve lower O.D.₆₀₀ readings (Fig. 2b). The growth of these strains revealed a biphasic curve with an early exponential expansion containing similar specific growth rates at O.D.₆₀₀ = 0.01 between wild-type (0.26 ± 0.13 hr⁻¹) and SusD* (0.27 ± 0.05 hr⁻¹). A larger, second exponential phase revealed an increased growth rate at O.D.₆₀₀ = 0.2 for wild-type (0.68 ± 0.07 hr⁻¹) compared to SusD* (0.40 ± 0.09 hr⁻¹). SusD* displays an initial lag before growth as well as a slower second exponential growth rate relative to wild-type, and both contribute to the apparent lag observed in the plate reader. The precise difference between growth rates in the culture tubes vs the plate reader is unclear, but not entirely unexpected when growth conditions change. We suspect that evaporation in the plate reader set-up may contribute to these effects. The growth defects seen in SusD* are likely representative of the apparent lag phenotypes observed in other strains in this study.

Figure 2. SusE Compensates for the Loss of the SusD Starch Binding Site During Growth on Maltoheptaose.

(A,C) Average growth curves of B. thetaiotaomicron strains expressing SusD* in different sus mutant backgrounds cultured on 2.5 mg/ml maltoheptaose. Identical growth experiments in glucose are shown in Fig S1ab. (B) Average growth curves of wild-type (WT) and SusD* B. thetaiotaomicron cultured on 2.5 mg/ml maltoheptaose with absorbance measured manually in a spectrophotometer. Biological triplicate cultures were cultured in glucose then back-diluted 1:800 into minimal media with maltoheptaose. (D) Western blot using anti-SusF serum against whole cell lysates from WT, SusD*, SusD*ΔEG, and SusD*ΔFG cultures. Strains were cultured in minimal media containing 5 mg/ml maltose and were arrested in logarithmic phase then normalized by O.D.₆₀₀ for loading in SDS-PAGE.

The ability to grow on maltoheptaose is abolished when the SusD* allele is combined with loss of SusEFG, hinting at a compensatory role for SusE and/or SusF. Upon further analysis, individual deletions of susEG and susFG revealed that the loss of SusE, but not SusF, prevented the growth of cells expressing SusD* (Fig. 2c). This phenotype is not the result of a polar mutation from the susE deletion as immunoblots of whole cells with anti-SusF antibodies demonstrate the expression of susF (Fig. 2d). These data suggest a unique role for SusE during growth on maltooligosaccharides.

SusE and SusF share multiple structurally homologous but functionally divergent domains

That SusE and SusF display divergent functions within Sus is noteworthy as SusE and SusF contain conserved structural characteristics suggesting redundant roles during starch catabolism. Both proteins have multimodular structures comprised of β-sheet rich starch-binding domains in tandem with an N-terminal Ig-like fold domain (Cameron et al., 2012). Proline residues are present between sequential domains of both proteins that presumably limit conformational flexibility, with the exception that SusE contains a putative flexible linker between the N-terminal domain and its first starch-binding domain, the Eb domain (see Fig. 3b for schematic). SusF has three starch-binding domains, Fa, Fb, and Fc and SusE has two, Eb and Ec, named for their homology to the domains of SusF (Fig. 3b). Among all five domains, Ec has the highest binding affinity for maltoheptaose and starch (Cameron et al., 2012). The Ig-like fold of SusE (E_N) was predicted to be similar to that of SusF (F_N).

Figure 3. SusE and SusF are composed of Functionally Unique, But Structurally Homologous Starch-binding Domains.

(A) Average growth curves of B. thetaiotaomicron strains expressing single starch-binding site mutants of SusE on 2.5 mg/ml maltoheptaose. Identical growth experiments in glucose are shown in Fig S1c. (B) Schematic of chimeric SusE and SusF domain structures. The K_d (µM) of each chimera for maltoheptaose as measured by isothermal titration calorimetry (average of 3 replicates) is reported. The K_d (µM) of the recombinant wild-type proteins is displayed for reference, as previously reported (Cameron et al., 2012). A box is placed around constructs that support growth in B. thetaiotaomicron expressing SusD*. NA denotes chimeric constructs that did not express in E. coli. NT denotes chimeric constructs not tested for recombinant expression. (C) Average growth curves of B. thetaiotaomicron strains expressing chimeric SusE and SusF on 2.5 mg/ml maltoheptaose. The mutants that displayed growth were SusD*ΔFG (light purple), SusD*E-FbΔFG (green), and SusD*E-FcΔFG (orange). Identical growth experiments in glucose are shown in Fig S1d.

To identify the domains of SusE that promote growth on maltoheptaose in the SusD* strain, we utilized mutant alleles of susE that we have previously reported that contain only one viable starch-binding site (Cameron et al., 2012). The SusEb* and SusEc* expressing strains, in which the Eb and Ec domain respectively cannot bind starch, were exchanged into the B. thetaiotaomicron chromosome in place of susE. Expression of these single-binding mutants with the SusD* allele revealed that both SusD*Eb* or SusD*Ec* strains grow on maltoheptaose with similar kinetics, demonstrating that both of the SusE starch-binding sites can compensate for the SusD* mutation (Fig. 3a). This was surprising as the K_d of Ec for maltoheptaose is ~ 20 µM and 50-fold higher than both SusEb and SusD (Cameron et al., 2012). Despite a K_d of 1 mM for maltoheptaose, native SusD efficiently supports growth on maltoheptaose emphasizing that high affinity binding to maltoheptaose is not required for transport through SusC (Fig. 2a).

Given that either starch-binding site of SusE is adequate for growth on maltoheptaose, we sought to determine which domains of SusE must be present for growth. Exploiting the structural homology of SusE and SusF, we created a set of chimeric proteins in which Eb and Ec or Fb and Fc were swapped individually or in combination between SusE and SusF (Fig. 3b). Immunofluorescent microscopy of fixed whole cells verified that these chimeric proteins were expressed by B. thetaiotaomicron and trafficked to the outer membrane (Fig. S2); we have demonstrated that cells expressing site-directed mutants of SusE and SusF (SusE^C21A and SusF^C20A) that retain the proteins in the periplasm cannot be labeled (Cameron et al., 2012). Recombinant expression and purification of these chimeras was performed to verify starch binding, however three of the seven mutants, SusE-Fb, SusE-Fbc, and SusF-Eb, were not soluble when expressed in E. coli. Since all of these proteins expressed and correctly trafficked in B. thetaiotaomicron we speculate that there may be some aspects of protein folding that are unique to B. thetaiotaomicron. Proteins that could be expressed in E. coli were assayed for maltoheptaose binding by isothermal titration calorimetry, and displayed K_d values similar to the wild-type proteins (Fig. 3b, Fig. S3) (Cameron et al., 2012).

When grown on maltoheptaose, chimeric SusE containing only one starch-binding SusF domain (SusD*E-FbΔFG and SusD*E-FcΔFG) was able to support growth as was the wild-type SusE allele in B. thetaiotaomicron expressing SusD* (Fig. 3c). If both domains Eb and Ec were replaced with Fbc (SusD*E-FbcΔFG), then B. thetaiotaomicron could not grow, indicating that at least one of the SusE starch-binding domains is required. While the SusE-Fbc mutant could not be isolated from E. coli, the SusE-Fb allele that supports growth also could not be obtained recombinantly, supporting that lack of soluble expression in E. coli does not necessarily mean a lack of functional or correctly folded protein in B. thetaiotaomicron. None of the SusF chimeras possessing Eb or Ec domains were able to rescue growth in the SusD* expressing mutant. These observations suggest that the SusE starch-binding domains are necessary, but also that its N-terminal domain may be important. Indeed, our SusE mutant allele expressing the SusF N-terminal Ig-like domain (SusD*E-FIgΔFG), did not support growth on maltoheptaose although this chimera is expressed on the surface of the cell (Fig. S2). Taken together, the ability of SusE to compensate for SusD* is dependent on its distinctive domains at both the N and C-terminus. However, we cannot rule out that the putative flexibility of SusE is also a required facet of its function. Note that the chimeric SusE proteins were created to retain the putative linker between the N-terminal domain and the following Eb or Fb domain.

Maltoheptaose binding by SusE is not required to promote growth

We have recently reported that immunoprecipitation of SusD results in co-isolation of both SusE and SusC, so we hypothesized that SusE provides a starch-binding function within the proximity of SusC to guide glycans into the transporter (Tuson et al., 2018). Previous work provided evidence for an interaction between SusE and SusF, so we reasoned that SusF might contribute to glycan import via an interaction with SusE proximal to the SusCD complex (Cho & Salyers, 2001). However, the SusD*E*ΔG (expressing wild-type SusF), SusD*E*ΔFG, and SusD*E*F*ΔG strains grow on maltoheptaose with an extended lag phase compared to the SusD*ΔFG that expresses a wild-type SusE (Fig. 4a). We tested whether the additional lag from the SusD*E* strain could be rescued by a SusF chimera that has one or more domains of SusE. Interestingly, the only chimeric SusF that had any effect on growth of the SusD*E* strain was SusF-Ec, which abolished growth (SusD*E*F-EcΔG, Fig. S4). Yet, SusF-Ec does not prevent growth when co-expressed with wild-type SusE (SusD*F-EcΔG, Fig. S4, S1j). We speculate that the Ec domain on SusF-Ec is interfering with the Ec domain on SusE*, preventing its function in maltoheptaose transport. We also tested if the starch-binding sites on SusG enhance growth on maltoheptaose in the presence of SusD*E*F* as our previous work suggests that SusG dynamically interacts with the Sus complex (Karunatilaka et al., 2014). However, a catalytically dead allele of SusG (SusG^D498N) does not improve growth on maltoheptaose (Fig. 4a).

Figure 4. SusE Does Not Require Starch-binding Sites for its Unique Functionality.

(A) Average growth curves of B. thetaiotaomicron strains expressing starch-binding deficient SusD*E* in various sus mutant backgrounds on 2.5 mg/ml maltoheptaose or (B) 2.5 mg/ml maltoheptaose supplemented with 0.5 mg/ml maltose. Identical growth experiments in glucose are shown in Fig S1e,f, and on 0.5 mg/maltose alone in Fig S5.

The major observable difference in growth between our mutant strains is in the apparent lag, which occurs when these strains are cultured on maltoheptaose, not glucose (Fig S1). We attribute this lag to a defect in starch “sensing”, or rather how quickly sus transcription is activated to support growth on starch; we have demonstrated previously that SusD* expressing cells require 100–1000× more sugar to maximally induce sus compared to wild-type cells (Cameron et al., 2014). However the addition of 0.5 mg/ml maltose, which does not require Sus for import and alone supports minimal growth of the strains (Fig S5), can upregulate sus and alleviate this lag (Cameron et al., 2014). To examine how sensing might be affecting our mutants, we cultured several SusD* mutant strains in minimal media with 5 mg/ml maltose, then subcultured into 2.5 mg/ml maltoheptaose with 0.5 mg/ml maltose. By inducing the sus operon with maltose, all strains grew comparably to wild-type with the exception of SusD*ΔEG, which now displayed growth on maltoheptaose but had a kinetic defect and lower maximum culture density (Fig 4b). Thus, it is likely that in the absence of maltose, a SusD*ΔEG strain cannot efficiently import and accumulate sugar to sufficient levels required to activate transcription and increase Sus protein levels on the cell surface; this type of kinetic coordination between the glycan levels for transcriptional activation and the activity of the periplasmic enzymes has been reported for the chondroitin sulfate and heparin targeting Sus-like systems in B. thetaiotaomicron (Raghavan et al., 2014). However, the growth defect of the SusD*ΔEG strain on maltoheptaose plus maltose suggests that even when sus is activated, the presence of SusE with SusD* is required for effective maltoheptaose import. Taken together, these data support that SusE, but not SusF or SusG, displays unique functionality during growth on maltooligosaccharides, and that its ability to support glycan uptake is not entirely dependent on its ability to bind sugar.

Substrate binding by SusE provides a competitive advantage during growth on starch

Although maltoheptaose is a useful substrate to study the mechanism of maltooligosaccharide uptake, we wanted to know if SusE is similarly important for utilization of a starch polysaccharide like amylopectin. SusG is required for growth on starch, but in-frame deletions of susEF affect SusG transcription (Reeves et al., 1997, Cameron et al., 2014). Therefore we used the periplasmically localized mutants SusE^C21A and SusF^C20A (Cameron et al., 2014) to test growth on starch in the absence of cell surface SusE and SusF. Cells were grown on maltose, then back-diluted into 5 mg/ml maize amylopectin with 0.5 mg/ml maltose to assess growth phenotypes. As before, 0.5 mg/ml maltose alone supports a low level of growth for all strains (Fig S6). B. thetaiotaomicron expressing SusD grows on starch without SusE or SusF on the cell surface (Cameron et al., 2014) (Fig. 5a, SusE^C21A and SusF^C20A). Cells expressing SusD* alone or in combination with SusE^C21A, SusF^C20A, or SusE* grow on starch but display a biphasic growth pattern with a more severe defect in the first phase. To assess the differences in growth rates, we quantified specific growth rates for each strain on glucose and starch during early (O.D.₆₀₀ = 0.35) and late (O.D.₆₀₀ = 0.75) exponential growth (Fig. S7). Across the strains, growth on glucose was within standard error at both time points. On starch, cells expressing SusD were slightly faster in the first phase though not significantly so from the SusD* expressing strains (Fig S7). However, during late exponential growth on starch there was a significant decrease in growth rates in strains that expressed SusD* compared to those that expressed wild-type SusD, though there was no statistically significant difference among the SusD* expressing strains (Fig. 5b).

Figure 5. SusE Provides a Competitive Advantage During Growth on Starch.

(A) Average growth curves of B. thetaiotaomicron strains expressing periplasmically localized SusE (C21A) and SusF (C20A) in combination with SusD on 5 mg/ml maize amylopectin supplemented with 0.5 mg/ml maltose. Identical growth experiments in glucose are shown in Fig S1g and for 0.5 mg/maltose alone in Fig S6. The horizontal pink line indicates an O.D.₆₀₀ of 0.75 at which specific growth rates, shown in panel B, were calculated for all strains. (B) Specific growth rates for all strains (n=3) grown in panel (A). Bars denoted with the letter “a” are not statistically significantly different from each other or WT. Bars denoted with the letter “b” are not significantly different from each other but are different from WT with a P value < 0.05. Statistically significant differences were determined using the two-tailed unpaired Student’s t test. (C,D) In vitro competitions of barcoded SusD* mutants in 5 mg/ml glucose or 5 mg/ml maize amylopectin with 0.5 mg/ml maltose. Relative abundance was calculated as the percent composition of a strain’s DNA relative to the total DNA in the sample.

In order to verify the apparent similarities in growth for the SusD* mutants, we cultured the SusD*E^C21A, SusD*F^C20A, and SusD*E* strains in culture tubes on 5 mg/ml amylopectin with 0.5 mg/ml maltose, as described earlier (Fig. S8a). To capture growth at lower initial concentrations of cells, we back-diluted biological triplicate cultures 1:800 into starch instead of 1:200 as performed for the plate reader assays. The overall phenotypes among these three strains appeared similar in both modalities and there are no significant differences in the apparent growth rates in culture tubes. Most notably SusD* E^C21A displayed a reduced maximum culture density with respect to SusD*F^C20A and SusD*E* suggesting that, although its growth rate is comparable to that of SusD*-expressing strains, SusE is needed to access the same level of starch in the media. Together, these observations support the idea that starch-binding by SusD is important for starch import, especially as starch may become limited in later time points. That the presence of SusE was not a requirement for growth was surprising, but SusG may be compensating for this, either via its additional starch-binding sites or by generating small oligosaccharides such as maltose and glucose (Koropatkin & Smith, 2010).

Although the growth rates displayed among the SusD* variants were similar, we wanted to know if the competitive fitness of B. thetaiotaomicron expressing SusD* was influenced by the presence of SusE, SusF, or SusE*. We performed in vitro competition experiments by co-culturing the SusD*E^C21A strain with either SusD*F^C20A or SusD*E*. These mutants were genetically tagged and passaged each day into minimal media containing glucose or starch and the relative abundance of each strain was quantified by qPCR (Fig. 5cd). The competitions in glucose may have displayed some stochastic changes in the abundance over time since it is unlikely the SusD*E^C21A mutant is better suited for growth on glucose than SusD*F^20A; this seems likely given the larger experimental error of the specific growth rates on glucose (Fig S7). Nonetheless, growth on starch resulted in a drastic decrease in the abundance of SusD*E^C21A, as it was outcompeted by both SusD*F^C20A and SusD*E* within the same two week time frame. These results not only underscore that SusE does not need to bind starch to support growth, but also raise the possibility that SusE provides a fitness advantage by facilitating glycan uptake from the environment.

Surface Starch-Binding Proteins Coordinate Oligosaccharide Uptake Based on Their Length

In the Bacteroidetes, oligosaccharides are generated from carbohydrate-active enzymes anchored to the cell surface or secreted via outermembrane vesicles into the environment (Rakoff-Nahoum et al., 2014, Elhenawy et al., 2014, Grondin et al., 2017). Our data suggest that surface glycan-binding proteins like SusD, SusE, and SusF are important for optimal glycan uptake through SusC, and we hypothesized that these starch-binding proteins could drive the preferential uptake of certain lengths of maltooligosaccharides. To test this we grew B. thetaiotaomicron on a mixture of maltooligosaccharides spanning 1 – 40 glucose units (a commercial preparation marked as DP10 – 40, but HPAEC-PAD analysis of the starting mixture revealed the presence of smaller sugars ranging from DP1 – 10. DP = degree of polymerization of a chain of α1,4-linked glucose residues) (Fig. S9a, S10). All HPAEC-PAD chromatograms reported in this study are compiled within Fig. S10. Glucose, maltose, and maltotriose all support the growth of B. thetaiotaomicron independent of Sus (Tancula et al., 1992), but the majority of glycans in the mixture are > DP7 (Fig. S9a). Longer sugars in the range of DP32 – 40 fell below our detection limit here, but were detected in subsequent experiments. We considered that some of the longer oligosaccharides (> DP7) might not be transported efficiently, or at all, in the absence of SusG. Consistent with this notion, we observed that ΔSusG grows to a noticeably lower maximum O.D. compared to wild-type (Fig. 6a). Furthermore, we found this trend to be consistent across all other strains that lack SusG (Fig. 6b). Growth on DP10 – 40 by SusD*, SusD*E^C21A, and SusD*F^C20A all displayed similar kinetics, likely due to the enzymatic activity of SusG. Conversely, SusD*ΔG, SusD*ΔFG, and SusD*E*ΔG grew similarly on the DP10 – 40 mixture but the combination of starch-binding deficient mutations and loss of SusG resulted in these strains exhibiting an even lower maximum O.D. Consistent with a unique and important role in oligosaccharide uptake, the growth of SusD*ΔEG had a notably longer lag and the lowest maximum O.D. among these strains, underscoring the importance of SusE during starch uptake. The growth of SusD*ΔEG on DP10 – 40 is likely supported by the presence of maltose which induces sus.

Figure 6. Starch-Binding Proteins Facilitate the Uptake of Maltooligosaccharides in a Size Dependent Manner.

(A) Average growth curves of B. thetaiotaomicron WT and ΔSusG strains on 5 mg/ml DP10 – 40 maltooligosaccharide mix. Identical growth experiments in glucose are shown in Fig S1h. (B) Average growth curves of B. thetaiotaomicron SusD* expressing mutants on 5mg/ml DP10 – 40. Identical growth experiments in glucose are shown in Fig S1i (C) Relative abundance of DP3 – DP40 maltooligosaccharides in cell-free culture supernatants at stationary phase (n=2). Maltooligosaccharides were purified from media components resulting in the loss of glucose and maltose. (D) Enlargement of representative small (DP3), medium (DP15) and large (DP30) maltooligosaccharides in cell-free culture supernatant of stationary phase cultures. Error bars indicate standard deviations across two biological replicates.

To examine which maltooligosaccharides are depleted over time during growth, we grew ΔSusG on DP10 – 40 and measured glycan content in cell free supernatents from lag, exponential, and stationary phase (Fig. S9b). Note that the preparation of media samples for HPAEC-PAD analysis resulted in the loss of glucose and maltose, and so these glycans are omitted from the analysis. Lag and exponential phase ΔSusG cultures had a similar distribution of oligosaccharides with a significant increase in the relative abundance of DP3 and DP4 in stationary phase (Fig. S9c). An increase in a sugar’s relative abundance within the sample indicates that it is either being generated from the break down of a larger oligosaccharide, or it is being imported less frequently compared to other sugars, or both. Because we can detect glycolytic activity in ΔSusG culture supernatants due to the presence of intracellular or secreted enzymes within culture supernatants, we attribute the increase of DP3 and DP4 abundance at least partially to enzymatic activity (Fig. S11). Interestingly, there was a dramatic decrease in DP5 – DP16 but an increase in DP17 content and steady levels of larger DP sugars from exponential to stationary phase (Fig S9c,d). If unidentified glycolytic activity in the supernatant were wholly responsible for the loss of mid-range glycans we would expect a concomitant decrease in larger glycans as well, rather than a discrete increase in glycans of a particular DP length. Therefore we conclude from these data that the SusCDEF complex can import, and may select for, maltoolisaccharides of DP5 – 16.

We investigated the contributions of SusE and SusF to maltooligosaccharide preference by characterizing the portfolio of oligosaccharides in the media from parallel cultures of SusD*ΔFG, SusD*ΔEG and ΔSusG cells at stationary phase from growth on DP10 – 40. Once again, these strains displayed different growth kinetics and maximum culture densities, suggesting inefficient or selective uptake of certain glycans in the mixture (Fig. 6b, Fig 6c inset). Like the first ΔSusG growth experiment, there is an enrichment of DP3 – 4 compared to other sugars, which may be because these are less frequently taken up, and there is likely some glycolytic activity in the media. Interestingly, sugars > DP23 were not found in ΔSusG cultures but were detected in SusD*ΔFG and SusD*ΔEG cultures, which may in part explain why these strains grow to a lower maximum O.D. SusD*ΔEG, which displays the most severe growth defect on DP10 – 40, is less efficient at taking up sugars DP7 – 18 and removes more of the small sugars 3 – 5 compared to ΔSusG and SusD*ΔFG cells. These data suggest that SusE is needed to efficiently access mid-range and longer sugars, and in the absence of SusE the cells may need to scavenge smaller sugars to support growth (Fig 6d). Although the energetics of maltooligosaccharide import in the Bacteroides is largely unknown, the uptake of mid-length glycans through the Sus likely minimizes the number of transport events needed to support growth and may be a cost-effective strategy for glycan capture.

Discussion

Members of the gut microbiota compete for carbon and energy sources to survive in the densely populated colonic environment and the Bacteroidetes that dominate this niche rely on sets of cell surface proteins to recognize, degrade, and import dietary polysaccharides. Glycan transport across the outermembrane is a critical feature of these systems that allows for the complete depolymerization of polysaccharides in the periplasm, which prevents the release of monosaccharides to neighboring species (Cuskin et al., 2015). Hence, understanding how these bacteria import carbohydrates can provide knowledge for the strategic manipulation of select species in the gut. Here, we investigate glycan uptake via the prototypical Sus of B. thetaiotaomicron. Our findings have uncovered that the transporter SusC imports maltooligosaccharides in a manner that requires SusD and SusE, but is not contingent on their capacity for starch-binding, suggesting assembly of the complex is most important. Furthermore, these data suggest that the protein-protein interactions that dictate Sus complex assembly can tune how the cell acquires starch. These unexpected observations raise further questions about the underlying mechanisms of glycan transport by homologous Sus-like systems, particularly the relationship between SusC-like TBDTs and their cognate glycan-binding proteins. Two recent crystal structures of B. thetaiotaomicron SusCD-like complexes reveal that the SusD-like protein sits atop the SusC TonB-dependent transporter (TBDT), with the ligand-binding site directed into the opening of the barrel (Glenwright et al., 2017). Molecular dynamics simulations of the complex for peptide import reveal that binding of the ligand by SusD protein and the internal plug of the TBDT stabilizes the closed complex, triggering the import event. In the absence of ligand, SusD can open to repeat this cycle. A difference between the two SusCD-like crystal structures was the presence of additional PUL-encoded lipoproteins that co-purified with the complex. The putative peptide-targeting SusCD-like complex included proteins BT2261 – 2264, with two lipoproteins, BT2262 and BT2261. BT2261 and BT2262 are comprised of Ig-like folds, akin to SusE and SusF, and wrap around the back of the CD-like complex; these proteins seems to affect the movement of SusD in molecular dynamics simulations, suggesting they contribute to the open/close mechanism of the transporter. However, the precise role of BT2261 and BT2262 in selection of ligand or bacterial growth has not been elucidated.

Glycan-binding lipoproteins akin to SusE and SusF, sometimes referred to as “SusE-positioned genes” or surface glycan binding protein-B (SGBP-B), functionally differ from SusD and its homologs. The presence of SusD-like proteins is required in some Sus-like systems like that for the uptake of xyloglucan in B. ovatus (Tauzin et al., 2016), but is less important for the uptake of levan and not at all for the uptake of fructo-oligosaccharides in B. thetaiotaomicron (Sonnenburg et al., 2010). SusEF-like SGBPs are encoded in PULs throughout the gut Bacteroidetes suggesting that they play a critical role in glycan capture (Martens et al., 2011, Martens et al., 2009). However, the importance and nature of their role may vary depending on the characteristics of the substrate. In the Sus, SusE appears to have evolved an intriguing role in maltooligosaccharide uptake (Fig. 7a). Our data support that SusC and SusD interact to form a high affinity transporter that facilitates efficient growth on maltoheptaose. SusD*, in the absence of other starch-binding proteins, cannot support growth, possibly by stably interacting with SusC and preventing the introduction of sugar into the transporter. This sort of closed complex absent of substrate is seen in the SusCD-like crystal structure of BT1763/1762 (Glenwright et al., 2017). SusE can uniquely resuscitate growth independent of its starch-binding sites, which we speculate is accomplished by SusE interacting with the core SusCD complex and allowing SusD to open for import. Curiously, the starch-binding sites on SusF and SusG do not enhance growth kinetics in a SusD*E* strain. This observation highlights the idea that starch-binding on the cell surface is not sufficient to assist in transport, and that the interaction of SusE with the SusCD transport complex may give it a privileged proximity near the opening of the pore.

Figure 7. A Model for Maltooligosaccharide Uptake Facilitated by the SusEF Starch-Binding Proteins.

Asterisks indicate a site-directed mutant that no longer binds starch. The Sus proteins are colored: SusC, purple; SusD, green; SusE, blue; SusF, magenta, SusG, pink. (A) The relative growth of the cells on maltoheptaose is displayed with more + indicating less lag time and more efficient growth. SusEF are not necessary for in vitro growth on maltoheptaose, but the presence of SusE is required in a SusD* background. (B) The relative growth of the cells on maltooligosaccharides of various DP is displayed with more + indicating a greater maximum O.D. and better utilization of the glycans within the mixture. The thickness of the arrow indicates the ability of the cells to take maltooligosaccharides of DP3, DP15 and DP30. Cells expressing SusCDEF (ΔSusG) can utilize all sizes of maltooligosaccharides, while cells expressing only SusE or SusF in a SusD* background are impaired for the uptake of large and mid- to-large glycans respectively.

That bacteria would evolve glycan uptake systems that import larger oligosaccharides is logical from an energetics viewpoint – it saves the energy of active transport when larger oligosaccharides can be imported at the same relative energetic “cost” as smaller ones. Although a different type of import system from the Sus, ATP-binding cassette (ABC) transporters for maltodextrins in Gram-positive bacteria such as Streptococcus pneumoniae and Eubacterium rectale have been observed to specifically target the uptake of oligosaccharides larger than maltose (Abbott et al., 2010, Cockburn et al., 2015). In these cases, the surface digestion of polysaccharide is coordinated with the specificity of the ABC transport solute-binding protein. For example, in Streptococcus pneuomoniae the surface enzyme SpuA digests glycogen into a ladder of fragments of up to at least 30 residues in length, and the MalX solute-binding protein aids in the transport of maltooligosaccharides of up to 12 residues, based upon the size of the glycans preferentially depleted from culture supernatants of cells grown on glycogen (Abbott et al., 2010). This is in contrast to E. coli, in which the maltose-binding protein MalE can apparently transport glycans up to eight or ten glucose residues, but growth defects are observed on maltooligosaccharides larger than maltoheptaose, due in part to the inefficiency of the uptake of these longer sugars (Ferenci, 1980, Wandersman et al., 1979). Despite differences in mechanism, these maltooligosaccharide ABC transporters and the Sus are similar in that they provide a strategy for the specific uptake of maltooligosaccharides, however the Sus may be unique in its ability to target even longer glycans.

Our data suggest that the collection of Sus proteins at the surface may dictate the length of maltoligosaccharides captured by B. thetaiotaomicron. Surprisingly, the ΔSusG strain is capable of importing even the longest range of oligosaccharides (DP25 – 40), as suggested from the lack of detection of these sugars in the culture supernatent, although they are detected in SusD*ΔEG and SusD*ΔFG cultures. Because these sugars are much larger than the substrates targeted by classically studied TonB-dependent transporters (Noinaj et al., 2010), it seems probable that SusC has been adapted to target larger substrates derived from polysaccharides. Indeed, that SusC-like transporters for other complex glycans such as α-mannan, rhamnogalacturonan-II, chondroitin sulfate, and heparin or heparan sulfate can import very large oligosaccharides has been suggested or observed (Cuskin et al., 2015, Ndeh et al., 2017, Raghavan et al., 2014, Cartmell et al., 2017). SusG is an endoamylase that can conceivably generate large oligosaacharides when hydrolyzing starch, yet in vitro digests yield glucose and maltose (Koropatkin & Smith, 2010). Perhaps at the cell surface, the product profile of SusG is altered to include longer sugars, or the Sus can also sequester long oligosaccharides generated from neighboring species. Our work has also shown that maltooligosaccharides of varying length can be imported with different efficiencies, and that starch-binding proteins can in part modulate these differences (Fig. 7b). We show here that SusE and SusF can greatly effect the uptake efficiency of maltooligosaccharides at the cell surface. SusD*ΔEG displays a much lower capacity to efficiently transport mid- and long-range maltooligosaccharides of DP5 – 16. This forces the SusD*ΔEG strain to grow on smaller sugars that provide a poor return on the energetic investment to actively import those substrates. SusD*ΔFG has a similar but less critical deficiency during uptake as well. That these surface glycan-binding proteins can adapt the cell to access different lengths of the same type of glycan is currently unexplored, but it may provide a mechanism explaining how bacteria can partition into nutrient niches or how they can share nutrients while foraging on the same glycans. The Bacteroidetes acquire large, complex, and heterogenous substrates for carbohydrate catabolism in order to remain competitive in the gut ecosystem. A mechanistic understanding of nutrient acquisition in the gut microbiome is a key prerequisite to intelligently manipulating this ecosystem to our benefit (Gordon, 2012, Hutkins et al., 2015).

Experimental Procedures

Bacterial strains and culture conditions

For these experiments and to generate all of the mutant sus strains used in these experiments, the B. thetaiotaomicron VPI-5482 Δtdk strain was employed to facilitate allelic exchange, as previously described (Koropatkin et al., 2008, Cameron et al., 2012), and is the parent strain for all of the mutations within this this work. For clarity we refer to the Δtdk strain as wild-type, as this parent strain retains a wild-type sus locus. Mutations were generated using the counterselectable allelic exchange vector pExchange-tdk as previously described (Koropatkin et al., 2008). The primers used in this study were synthesized by IDT DNA Technologies and are described in Table S3 of the Supplemental material. The sus alleles for all Sus* mutants are included in Table 1 and were validated in our previous studies, as referenced. A table of strains used in this study is provided in Table S1. SusE-F chimeric constructs were designed based upon the known structures of these proteins.

B. thetaiotaomicron was cultured in a 37°C Coy anaerobic chamber (5% H₂/10% CO₂/85% N₂) from freezer stocks into tryptone-yeast extract-glucose (TYG) medium (Holdeman et al., 1977) and grown for 24 hours, to an O.D. ~1.0. The following day cells were back-diluted 1:100 into Bacteroides minimal media (MM) including 5 mg ml⁻¹ glucose or maltose (Sigma) as noted and grown overnight (16 hours). For kinetic growth experiments in a plate reader, MM-glucose or MM-maltose grown cells were back-diluted 1:200 into MM with the experimental carbohydrate, and in parallel to MM with glucose. Thus both glucose controls and experimental starch and oligosaccharide grown cultures were started at the same initial O.D.₆₀₀ of 0.1 in the plate reader (or 0.02 in a 1cm path length spectrophotometer at O.D.₆₀₀). All glucose controls for each set of growth experiments are displayed in Supplemental Figure 1. The substrates used for comparison to parallel glucose-grown cultures included: 2.5 mg ml⁻¹ (2.17 mM) maltoheptaose (Carbosynth), 5 mg ml⁻¹ maize amylopectin (Sigma), or 5 mg ml⁻¹ DP10 – 40 (Elicityl). Kinetic growth experiments were performed at 37°C in 96 well plates and O.D.₆₀₀ were recorded every 10–30 min. All plate reader growth experiments were performed in 3 – 10 replicates and the averages are reported in each figure here. However, all biological experiments were repeated at least twice to verify consistent growth phenotypes from day to day. Manual growth curves were performed in 16mm glass culture tubes and O.D.₆₀₀ measurements were taken from the same tubes by a Genesys20 spectrophotometer. The spectrophotometer was blanked with media plus maltoheptaose including those curves for the starch media, which has a background absorbance of 0.2. B. thetaiotaomicron was cultured in culture tubes as previously specified except cultures were started from 1:800 back-diluted overnight grown cells in MM plus glucose, or maltose for the starch-grown cultures. Specific growth rates were calculated as the ΔO.D.₆₀₀ Δtime⁻¹ O.D.₆₀₀⁻¹. The change in O.D. was calculated over a 40 minute – 1.5 hour duration.

Cloning and recombinant protein expression

Chimeric SusE and SusF alleles were PCR amplified from genomic DNA for ligation-independent cloning into pETite N-His vector (Lucigen Madison, WI) according to the manufacturer’s instructions. The N-terminal primers introduced a TEV-cleavable site between the mature protein that lacks a signal sequence and the His tag as well as a mutation C21A and C20A of SusE and SusF respectively to produce soluble TEV-cleavable His-tagged proteins. Chimeric SusEF-containing pETite plasmids were transformed into Rosetta (DE3) pLysS cells (EMD Biosciences) and plated on LB agar containing 50 µg ml⁻¹ kanamycin (Kan) and 20 µg ml⁻¹ chloramphenicol (Cm). After 16 hours of growth at 37°C, colonies from plates were used to inoculate 1 l of terrific broth plus Kan and Cm for growth at 37°C. Cultures were grown to an O.D.₆₀₀ of ~0.6 before 0.5 mM IPTG was added and cells were grown at room temperature (~20°C) for an additional 20h. Cells were harvested by centrifugation and cell pellets were flash frozen in liquid N₂ until purification.

Recombinant protein purification

Cell pellets were resuspended in 50 mls of His Buffer (25 mM NaH₂PO₄, 500 mM NaCl, 20 mM imidazole, pH 7.4) and were lysed by sonication. Lysates were centrifuged at 20,000 × g to remove intact cells and the soluble lysate containing. His-tagged chimeric SusEF proteins were purified using a 5 mL Hi-Trap metal affinity cartridge (GE Healthcare) charged with NiSO₄ according to the manufacturer’s instructions. Cell lysates were applied to the column in His Buffer and proteins were eluted with an imidazole (20–300 mM) gradient. The His-tag was removed with a 2 hour incubation with recombinant TEV (1:100 molar ratio of TEV to protein) at room temperature, followed by an overnight incubation at 4°C while dialyzing into His buffer with 20 mM imidazole. Affinity purification using a Ni-charged Hi-Trap affinity cartridge was repeated to remove the His-tag, uncut protein and His-tagged TEV, while the cleaved protein was collected as the flow through. The protein was then dialyzed against a storage buffer (20 mM HEPES, 100 mM NaCl, pH 7.0) and concentrated using a Vivaspin 15 (10,000 MWCO) centrifugal concentrator (Vivaproducts, Inc.).

Isothermal titration calorimetry

ITC measurements were performed on a low volume (300 µl sample cell) TA instruments NanoITC. Maltoheptaose (Sigma) solutions were prepared using the same dialysis buffer as the proteins, and thus protein and titrant were in the same buffer for all experiments. For each experiment 97 – 120 µM protein was placed in the sample cell. The reference cell was filled with deionized water. After the cell temperature was equilibrated to 25°C, an initial injection of 0.25 µl of maltoheptaose was performed followed by 25 subsequent injections of 2 µl of 7 mM maltoheptaose (Sigma). The sample cell was stirred at 250 rpm and the resulting heat of reaction was measured. Data were analyzed using the NanoAnalyze software package (TA instruments) by fitting to an independent binding model and fixing N to the number of known binding sites in the protein. Each titration was performed in triplicate.

Growth competition experiments and Quantitative PCR (qPCR)

SusE/F strains used in this competition experiment were tagged with either pNBU2-tag 11 or 14 (Koropatkin et al., 2008). Three biological replicates of SusE/SusF strains were passaged each day for two weeks using a daily 1:100 back-dilution into MM containing 5 mg ml⁻¹ glucose or 5 mg ml⁻¹ maize amylopectin supplemented with 0.5 mg ml⁻¹ maltose. Genomic DNA was harvested from cultures on days 0, 1, 2, 3, 4, 6, 8, 10, and 14 using DNeasy Blood & Tissue kit (Qiagen). Genomic DNA quantification was performed with a Mastercycler® ep realplex (Eppendorf), using KAPA SYBR® FAST qPCR Master Mix and 100 nM SusE and SusF primers, for 40 cycles of 95 °C for 3 s, 55 °C for 8 s, 72 °C for 20 s, followed by a melting step to determine amplicon purity. Samples were normalized to a DNA standard curve of genomic DNA from each respective strain. Relative abundance was calculated as the percent composition of a strain’s DNA relative to the total DNA in the sample. Primers used for qPCR are listed in Supplemental Table S3.

Western blotting

B. thetaiotaomicron strains were grown in MM containing 5 mg ml⁻¹ maltose and were harvested at O.D.₆₀₀=0.6. Cells were lysed in SDS sample buffer and 15 µg of total protein was loaded into an SDS-PAGE. SusF was detected in B. thetaiotaomicron whole cell lysates by western blot using custom rabbit polyclonal primary antibodies and horseradish peroxidase conjugated goat anti-Rabbit IgG secondary antibody (Sigma) (Cameron et al., 2012).

Immunofluorescence

B. thetaiotaomicron strains were grown in 5 ml minimal Bacteroides medium supplemented with 5 mg ml⁻¹ maltose to an O.D.₆₀₀ of 0.6 and then harvested via centrifugation (7,000 × g for 3 min) and washed twice with phosphate- buffered saline (PBS). Cells were resuspended in 0.25 ml PBS, and 0.75 ml of 6% formalin in PBS was added. Cells were incubated with rocking at 20°C for 1.5 h and then washed twice with PBS. Cells were resuspended in 0.5 to 1 ml blocking solution (2% goat serum, 0.02% NaN₃ in PBS) and incubated for 16 h at 4°C. Cells were centrifuged and resuspended in 0.5 ml of a 1:100 dilution of custom rabbit anti-SusE or anti-SusF antibody sera in blocking solution and incubated by rocking for 2 h at 20°C. Cells were washed with PBS and then resuspended in 0.4 ml of a 1:500 dilution of Alexa Fluor 488 goat anti-rabbit IgG (Life Technologies) in blocking solution and incubated with rocking for 1 h at 20°C. Cells were washed three times with an excess of PBS and then resuspended in 20 µl of PBS plus 1 µl of ProLong Gold antifade (Life Technologies). Cells were spotted on 0.8% agarose pads and imaged using an Olympus IX70 inverted confocal microscope. Images were processed with Metamorph Software.

Thin layer chromatography

Cell-free culture supernatant was collected from stationary phase ΔSusG cultures grown in minimal media containing 5 mg ml⁻¹ maltose. Supernatant was added to minimal media containing 5 mg ml⁻¹ DP10 – 40 to make up 5% of the reaction volume. Reactions were incubated at 37°C for 1 hour and then flash frozen. These reactions were spotted onto TLC Silica gel 60 F254 20×20 cm glass plates (Millipore) and separated with the solvent acetonitrile:ethyl acetate:isopropanol:water (40:10:25:525 until the solvent front was within 1 cm of the top of the plate. The sugars were then stained with 0.3% (w/v) N-(1- napthyl)ethylenediamine, 5% (v/v) sulfuric acid in methanol and heated until spots developed.

HPAEC-PAD Maltooligosaccharide Analysis

Samples were processed by the GlycoAnalytics Core at the University of California San Diego. Crude supernatant samples were passed over a PGC cartridge (poly-graphitized charcoal) HyperSep™ Hypercarb™ SPE Cartridges (Thermo Scientific), washed with 5 ml water and bound oligosaccharides were eluted with 30% Acetonitrile solution containing 0.1% TFA. This purification results in the loss of glucose and maltose. Eluted oligosaccharides were dried, resuspended in water, and injected on HPAEC-PAD. Oligosaccharide profiling was performed using BioLC CarboPac PA100 column (4×250mm) with PA100 (4×50 mm) guard column at a flow rate 1 ml min⁻¹. Pulsed amperometric detection with a gold electrode and standard quad waveform was used for carbohydrate analysis. The elution gradient was as follows: 0.0 – 20.0 min isocratic flow with 19 mM sodium hydroxide containing 7 mM sodium acetate, 20 – 70 min linear gradient of 0 mM – 400 mM sodium acetate. Maltrin-100 and Maltrin-200 were used as standards to compare the elution time for each oligosaccharide to verify the degree of polymerization. The area under each peak was calculated using Chromeleon™ 6.8 Chromatography Data System software and the DP values were assigned based on the retention time (min). Relative abundance is calculated as the percent composition of an oligosaccharide’s peak area relative to the total area of all peaks in the sample.

Supplementary Material

Supp FigS1. Figure S1. Compiled Growths on Glucose.

(A–J) Average growth curves of B. thetaiotaomicron strains on glucose. Panels are organized chronologically through the main text, and labeled according to the main figure that they correspond with. Note that the concentration of glucose used matches the concentration of the substrate used in the corresponding main text figure.

Supp FigS8. Figure S8. Growth and Specific Growth Rates of SusD* Strains on 5 mg/ml Starch with 0.5 mg/ml Maltose.

Average growth curves from biological triplicate cultures of SusD*E^C21A, SusD*F^C20A, and SusD*E* B. cultured on 5 mg/ml starch and 0.5 mg/ml maltose with absorbances measured manually using a spectrophotometer. The background absorbance of starch is 0.2 from a media blank lacking starch.

Supp FigS9. Figure S9. Maltooligosaccharides Profile Cell-free Culture Supernatants During ΔSusG Growth on DP10 – 40 Maltooligosaccharide Mix.

(A) Relative abundance of maltooligosaccharides in the DP10 – 40 maltooligosaccharide mixture as measured by HPAEC-PAD (n=1). (B) Growth curves (n=3) of ΔSusG on 5mg/ml DP10 – 40. L, E, and S denote when the cell-free supernatant samples for Lag, Exponential, and Stationary phases were taken. (C) Relative abundance of maltooligosaccharides from ΔSusG cell-free culture supernatants at lag, exponential, and stationary phase growth (n=3). Mid-range and longer oligosaccharides fell below our limit of detection in the stationary phase samples. Maltooligosaccharides were purified from media components resulting in the loss of glucose and maltose. (D) Enlargement of small (DP3), medium (DP15) and large (DP30) maltooligosaccharides in cell-free culture supernatant of ΔSusG cultures. Error bars indicate standard deviations across three biological replicates.

Supp TableS1

Table S1. Mutants Used in this Study

Supp TableS2

Table S2. Relative Abundance of Maltooligosaccharides in Culture Supernatant During Growth on DP10 – 40.

Supp TableS3

Table S3. Oligonucleotides Used in this Study

Supp FigS10. Figure S10. Compiled HPAEC-PAD Chromatogram Traces.

HPAEC-PAD chromatograms for each experimental replicate in this study. Representative Maltrin 100 and 200 standards that were used to determine oligosaccharide retention times are shown here with annotated peaks. Peak intensity as measured by pulsed amperometric detection (PAD) is displayed as nano Coulombs (nC). Quantification of the sugars in each experiment is presented in Table S2.

Supp FigS11. Figure S11. Stationary Phase ΔSusG Supernatants Display Glycolytic Activity.

Thin layer chromatography indicating the presence of glycolytic enzymes from stationary phase ΔSusG cultures grown in minimal media containing maltose. Ladder contains 5 mmols of glucose, maltose, and maltotriose. Supernatant (2 µl) was collected from cell-free stationary phase ΔSusG cultures. DP10 – 40 lane contains 100 µg of DP10 – 40 from MM containing 5 mg/ml DP10 – 40. Supernatant+DP10 – 40 contains both the ΔSusG culture supernatant and MM-DP10 – 40. All samples were incubated at 37°C for 1 hour before being loaded on the TLC plate. Degradation products, primarily maltotriose, are detected in the final lane indicating the presence of glycolytic enzymes in the culture supernatant.

Supp FigS2. Figure S2. Surface Staining of SusE and SusF Chimeras.

(A,B) B. thetaiotaomicron cells expressing a chimeric SusE or SusF in place of the wild-type alleles were visualized by immunofluorescence. Strains were grown in 5 mg/ml maltose and until mid-logarithmic phase. Non-permeabilized, formalin-fixed cells were probed with custom anti-SusE (A) or anti-SusF (B) serum then stained with anti-rabbit IgG conjugated to Alexa Fluor 488. Top and bottom panels show phase and fluorescence images, respectively. Scale bar = 5 µm.

Supp FigS3. Figure S3. Isothermal Titration Calorimetry (ITC) of SusE and SusF Chimeras With Maltoheptaose.

(A–C) All titrations were performed in triplicate with buffer-matched protein and ligand samples (in 20 mM Hepes, 100 mM NaCl pH 7.0) and at 25 °C. In each case, the upper graph shows the raw heat signal for 25 × 2.0 µl injections of maltoheptaose into protein; the bottom graph shows the integrated heats with fits to an independent binding model, with N fixed to the number of known binding sites in the protein. Representative isotherms are shown for one titration and the values shown are averages ± standard error for the three titrations.

Supp FigS4. Figure S4. Chimeric SusF-Ebc Inhibits Growth in B. thetaiotaomicron Expressing SusD*E*.

Average growth curves of B. thetaiotaomicron strains expressing starch-binding deficient SusD*E* combined with chimeric SusF on 2.5 mg/ml maltoheptaose.

Supp FigS5. Figure S5. Growth of B. thetaiotaomicron on 0.5mg/ml Maltose.

Average growth curves of WT and B. thetaiotaomicron strains expressing starch-binding deficient alleles on 0.5 mg/ml maltose.

Supp FigS6. Figure S6. Growth of Periplasmically Localized SusE/F Strains on 0.5 mg/ml Maltose.

Average growth curves of B. thetaiotaomicron strains expressing starch-binding deficient SusD allele and periplasmically localized SusEF mutants on 0.5 mg/ml maltose.

Supp FigS7. Figure S7. Specific Growth Rates for Periplasmically Localized SusE/F Strains on 5 mg/ml Starch with 0.5 mg/ml Maltose.

Specific growth rates are calculated at O.D.₆₀₀ 0.35 and 0.75 by calculating the change in O.D.₆₀₀/40 minutes/O.D.₆₀₀ of the sample. Bars denoted with the letter “a” are not statistically significantly different from each other or WT. Bars denoted with the letter “b” are not significantly different from each other but are different from WT with a P value < 0.05. Statistically significant differences were determined using the two-tailed unpaired Student’s t test.