Sugar import suppresses Klebsiella pneumoniae Mucoidy in a cAMP-CRP-dependent manner Authors: Saroj Khadka, Gabriella M Gates, Drew J Stark, Katlyn Todd, Bennett A Allison, Jay Vornhagen, Laura A Mike Published: 2026 ABSTRACT AbstractHypervirulent Klebsiella pneumoniae (hvKp) causes severe invasive infections in community and hospital settings, often originating from gut colonization. Capsular polysaccharide (CPS) and its associated features are a key hvKp virulence factor. Altering CPS properties, such as mucoidy, in response to environmental cues enhances K. pneumoniae fitness. While several physical and nutrient cues influence mucoidy, the molecular mechanisms by which host-relevant signals, such as sugars, regulate mucoidy remain undefined. Here, we show that sugar import, not catabolism, broadly suppresses hvKp mucoidy through cAMP-CRP signaling and downregulation of the mucoidy regulator, rmpD. This sugar-dependent regulation is conserved across multiple hvKp strains. Although non-mucoid hvKp associate with gut mucin and epithelial cells at a greater frequency, both hypermucoid and non-mucoid states show reduced fitness for gut colonization. Our findings support that sugars are a key mucoidy regulatory cue and that the dynamic regulation of mucoidy contributes to efficient gut colonization by hvKp.Subject terms: Bacteriology, Pathogens Introduction Klebsiella pneumoniae is a Gram-negative bacterial pathogen responsible for a variety of nosocomial and community-acquired infections. It ranks as the fourth leading cause of bacterial infection-related deaths globally and accounts for an estimated 790,000 deaths annually1. K. pneumoniae typically transmits via the fecal-oral route and infects the lungs, bloodstream and urinary tract. K. pneumoniae strains are broadly classified into two pathotypes: classical (cKp) and hypervirulent (hvKp). HVKp particularly poses a public health threat due to its ability to cause severe invasive infections, such as pyogenic liver abscess, meningitis and endophthalmitis in both healthy and immunocompromised individuals. This is further exacerbated by the recent emergence of convergent isolates that are hypervirulent and antimicrobial resistant2,3. Hypermucoviscosity (hereafter ‘mucoviscosity’ is referred to as ‘mucoidy’) is a defining virulence feature of hvKp. Hypermucoid colonies exhibit a sticky phenotype linked to uniform capsular polysaccharide (CPS) chain length4,5. Although its precise role in pathogenesis is not fully understood, in vivo studies support that hypermucoidy enhances virulence by impairing association with host immune cells like macrophages, ultimately promoting invasive infections4,6. Even typically non-mucoid cKp strains present increased virulence upon acquiring the hvKp mucoidy regulator7. In hvKp, mucoidy is regulated by the rmpADC locus, which encodes three components: rmpA, an autoregulator; rmpC, a CPS biosynthesis regulator; and rmpD, the regulator of mucoid phenotype8. RmpD increases mucoidy by directly interacting with Wzc, which controls CPS chain length during biosynthesis4. Environmental factors, such as pH, temperature, and iron and nutrient availability influence mucoidy through transcriptional and post-transcriptional regulation of the rmp locus5,6,9–11. In addition, genetic changes including spontaneous mutations in Wzc and phase-variation within the rmpADC locus further optimize hvKp mucoidy5,12,13. However, molecular mechanisms underlying several of these processes remain poorly defined, and additional regulatory signals are likely to be discovered. Among environmental cues, nutrient availability is a major regulatory signal in the context of infection. Pathogens encounter distinct nutritional environments across different host niches. For example, glucose concentrations in the lung airway surface liquid are 3-20 times lower than those in the plasma14,15. Similarly, the human urinary tract maintains a minimal free glucose concentration under normal conditions. However, disease states, such as hyperglycemia and glycosuria elevate glucose levels in these sites, which concomitantly promotes bacterial growth and persistence14,16,17. The human gastrointestinal (GI) tract represents another nutritionally distinct environment that is both rich and diverse in available carbohydrates18. The GI tract is lined with mucins that serve both as an innate immune defense and as a nutrient source to resident microbes19. Mucins are structurally diverse glycoproteins, commonly composed of sugars like galactose, fucose, N-acetylglucosamine (GlcNAc), N-acetylgalactosamine (GalNAc) and N-acetylneuraminic acid (Neu5Ac). Microbial glycosidases and other carbohydrate-active enzymes liberate sugars from the dietary carbohydrates or host-derived mucins, providing carbon sources for both commensal and pathogenic microbes20,21. In K. pneumoniae, sugar uptake and catabolic pathways are required for gut colonization, highlighting the importance of carbohydrate metabolism in this niche10,22–24. Beyond serving as nutrient substrates, sugars induce broad physiological responses that regulate virulence and fitness traits including capsule production, biofilm formation, antimicrobial resistance, secretion systems and motility25–29. Together, these observations underscore sugars and other nutrients as important regulatory cues governing bacterial virulence and fitness. Though how sugar availability influences hvKp mucoidy in the gut environment remains unknown. K. pneumoniae extra-intestinal infections frequently originate from asymptomatic gut colonization. Hospitalized patients with K. pneumoniae gut colonization have an increased odds ratio (OR) for subsequent infection compared to non-colonized patients (OR 4.0–6.9)30,31. Of those colonized patients who develop a subsequent infection, up to 80% have concordant colonizing and infecting isolates30,31. Thus, GI carriage is widely recognized as the primary reservoir and a key risk factor for K. pneumoniae infections32. However, the factors that shift K. pneumoniae from colonizer to pathogen remain poorly understood. In particular, the influence of K. pneumoniae mucoidy regulation during gut colonization and subsequent dissemination to extra-intestinal sites remains unknown. Our recent work demonstrated that arginine is a key environmental signal that promotes hvKp mucoidy and niche-specific fitness6. Interestingly, we also identified that certain sugars negatively modulate mucoidy, indicating that an opposing nutrient-based regulatory mechanism might exist in hvKp6. Given the carbohydrate-rich nature of the gut environment and the significance of GI carriage, we hypothesized that K. pneumoniae dynamically modulates mucoidy in response to available sugars to optimize its niche-specific fitness. To test this, we employed a genome-wide screen to identify bacterial genes and pathways involved in the regulation of sugar-mediated mucoidy. Here, we demonstrated that host-derived sugars suppressed hvKp mucoidy and decreased CPS chain length uniformity by down-regulating rmpD. Mucoidy suppression required sugar import rather than catabolism and was observed across multiple hvKp isolates. In Gram-negative bacteria, sugar import is coupled to carbon catabolite regulation via cAMP-CRP. Disrupting cAMP-CRP signaling resulted in constitutively hypermucoid hvKp, even in the presence of mucoidy-suppressing sugars. When shifted to a non-mucoid state, hvKp exhibit increased binding to gut mucin and association with epithelial cells. Moreover, the regulation of mucoidy contributes to initial gut colonization. Altogether, we present evidence that sugar import broadly suppresses mucoidy via a cAMP-CRP-dependent mechanism, suggesting that K. pneumoniae mucoidy is regulated in response to niche-specific nutrient signals in vivo and that dynamic control of mucoidy is likely required for successful gut colonization and subsequent pathogenesis. Results Sugars independently impact K. pneumoniae mucoidy and capsular polysaccharide abundance To examine the impact of sugars on K. pneumoniae mucoidy, we selected the model hypervirulent strain KPPR1 and measured its mucoidy and CPS abundance using the standard sedimentation and uronic acid quantification assays in sugar-supplemented media33. Specifically, KPPR1 was cultured in low-iron M9 minimal medium supplemented with 1% casamino acids (CAA) with or without 80 mM sugar (M9+CAA+sugar). Altogether, nine different sugars were tested: D-galactose (Gal), D-glucose (Glc), D-mannose (Man), L-rhamnose (Rha), L-arabinose (Ara), L-fucose (Fuc), D-xylose (Xyl), N-acetyl-D-glucosamine (GlcNAc) and N-acetyl-D-galactosamine (GalNAc). The sugars were selected if they fit one or more of the following parameters: known influence on mucoidy (Glc, Fuc and Ara), metabolism by bacteria (Glc, Fuc, Man, GlcNAc and Xyl), and presence in host glycans and glycoproteins (Gal, Man, GlcNAc, GalNAc) or KPPR1 CPS (Glc and Man)10,34–36. Sugar supplementation (M9+CAA+sugar) significantly suppressed KPPR1 mucoidy, except for GalNAc, which had no significant effect (Fig. 1A). Notably, KPPR1 does not encode agaVWEF, a GalNAc phosphotransferase system (PTS) required for GalNAc transport in other Enterobacteriaceae37. Meanwhile, sugar-specific transporters responsible for importing the other tested sugars are encoded in KPPR1. Furthermore, among the tested sugars, only GalNAc did not support KPPR1 growth when supplied as the sole carbon source, suggesting that sugar-suppressed mucoidy is likely limited to consumable sugars (Supplementary Fig. 1I). Sugar supplementation increases growth medium osmolarity and could exert osmotic stress on the bacterial membrane. As such, osmotic stress could suppress mucoidy. However, equimolar concentrations of GalNAc do not suppress mucoidy, indicating that the observed modulation of mucoidy is not attributable to osmotic stress. Sugar supplementation in minimal growth medium suppresses K. pneumoniae mucoidy independent of capsule abundance. K. pneumoniae strain KPPR1 was cultured in M9+CAA supplemented with (A,C) 80 mM of the respective sugar [D-galactose (Gal), D-glucose (Glc), D-Mannose (Man), L-rhamnose (Rha), L-arabinose (Ara), L-fucose (Fuc), D-xylose (Xyl), N-acetyl-D-glucosamine (GlcNAc) and N-acetyl-D-galactosamine (GalNAc)] or (B,D) varying concentrations of Gal ranging from 0 to 80 mM. A,B Mucoidy was determined by quantifying the supernatant OD600 after sedimenting 0.5 OD600 unit of culture at 1000 x g for 5 min. C,D Uronic acid abundance in crude CPS extracts was quantified for total CPS and supernatant CPS and normalized to OD600. Cell-associated CPS abundance was calculated by subtracting supernatant CPS from the total CPS content. Data presented are the mean, and error bars represent the standard error of the mean. Statistical significance was determined using one-way ANOVA with (A,C) Dunnett’s post-hoc test or (B,D) Šídák correction. Statistical significance was calculated by comparing sugar-supplemented condition to sugar-deficient condition. P-values are displayed above each comparison; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001; # p ≤ 0.0001. Exact p-values: (A) left to right: <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001; (B) left to right: <0.0001, <0.0001, <0.0001, <0.0001, <0.0001; (C) left to right: 0.0425, <0.0001, 0.0144, <0.0001, 0.0016, <0.0001; (D) left to right: 0.0189, 0.0014, 0.0003. Experiments were performed three independent times, each in biological triplicate (n = 9). Source data are provided as a Source Data file. Although supplementing a variety of sugars at 80 mM broadly suppressed mucoidy, we wanted to determine whether the mucoidy suppression was concentration dependent. Therefore, we analyzed KPPR1 mucoidy at different Gal concentrations. Gal was selected over other sugars because it is a common terminal sugar in human mucins at mucosal surfaces and is not a direct constituent of KPPR1 CPS. Furthermore, Gal metabolic pathway has been shown to be required for Klebsiella gut colonization indicating importance of Gal catabolism to Klebsiella pathogenesis24. As low as 5 mM of Gal significantly suppressed KPPR1 mucoidy (Fig. 1B). It is possible that the mucoidy-suppressing effect of Gal at lower concentrations is counteracted by the mucoidy-inducing effect of arginine present in CAA6. Since mucoidy is a CPS-associated property, we also evaluated the cell-associated CPS content of KPPR1 under the same culture conditions (M9+CAA±sugar)4,5. Sugars had effects on KPPR1 CPS abundance ranging from none to significant, depending on the specific sugar (Fig. 1C). Relative to M9+CAA, CPS abundance increased significantly when Xyl, GlcNAc or GalNAc was supplemented but decreased significantly when Gal, Rha or Ara was supplemented (Fig. 1C). When M9+CAA is supplemented with Gal (M9+CAA+Gal), KPPR1 CPS abundance is significantly reduced at some concentrations without a clear dose-dependent response, unlike its effect on mucoidy (Fig. 1B, D). However, when either Gal or Glc were provided as the sole carbon source (M9+Gal and M9+Glc), KPPR1 mucoidy remained significantly suppressed, despite a nearly three-fold increase in CPS abundance compared to M9+CAA (Supplementary Fig. 2A, B). In summary, we generally observed that consumable sugars decrease mucoidy but have varied effects on cell-associated CPS abundance. We interpret these findings to indicate that sugars suppress K. pneumoniae mucoidy independently of their effect on CPS abundance. Sugar import, but not catabolism, is required for sugar-suppressed mucoidy. A–C Mucoidy was determined by quantifying the supernatant OD600 after sedimenting 0.5 OD600 unit of culture at 1000 x g for 5 min. D–F Uronic acid abundance in crude CPS extracts was quantified for total CPS and supernatant CPS and normalized to OD600. Cell-associated CPS abundance was calculated by subtracting supernatant CPS from the total CPS content. In (A) and (D), KPPR1 was cultured in M9+CAA or M9+CAA+sugar [D-galactose (Gal), 2-deoxy-D-galactose (2-D-Gal), D-glucose (Glc), 2-deoxy-D-glucose (2DG)]. Data for Gal and Glc are also shown in Fig. 1 and included here for statistical comparisons. In (B,C,E,F), KPPR1 WT and ΔgalETKM or galP::kan were cultured in M9+CAA±Gal. Data presented are the mean, and error bars represent the standard error of the mean. Statistical significance was determined by using (A,D) one-way ANOVA with Šídák correction and (B,C,E,F) two-way ANOVA with Fisher’s LSD test. Statistical significance was calculated by either comparing sugar-supplemented medium to the sugar-deficient medium or comparing adjacent pairs of bars connected by lines. In (A) and (D), mucoidy and CPS abundance in 2-D-Gal and 2-DG were compared to the mucoidy and CPS abundance in Gal and Glc determined in Fig. 1A, C. P-values are displayed above each comparison; * p ≤ 0.05; ** p ≤ 0.01; # p ≤ 0.0001. Exact p-values: (A) left to right above data bars: <0.0001, <0.0001, <0.0001, <0.0001; connected by line: 0.0011; (B) left to right above data bars: <0.0001, <0.0001; connected by lines: 0.0001, 0.0047; (C) left to right above data bars: <0.0001, 0.0091; connected by lines: <0.0001; (D) left to right above data bars: 0.0142, <0.0001; connected by line: 0.0073; (E) left to right above data bars: <0.0001; connected by line: <0.0001; (F) left to right above data bars: <0.0001; connected by line: <0.0001. Experiments were performed three independent times, each in biological triplicate (n = 9). Source data are provided as a Source Data file. Sugar catabolism is not required to suppress K. pneumoniae mucoidy Since K. pneumoniae mucoidy suppression was limited to consumable sugars, we investigated whether this regulatory process was linked to sugar metabolism or import. To test this, we provided two non-metabolizable sugar analogs, 2-deoxy-D-galactose (2-D-Gal) or 2-deoxy-D-glucose (2DG) in the M9+CAA medium and then measured KPPR1 mucoidy. These sugar analogs are imported via Gal- and Glc-specific transporters38,39. However, neither 2-D-Gal nor 2DG is metabolized, and both sugars are unable to support KPPR1 growth as a sole carbon source (Supplementary Fig. 1A, B). Surprisingly, both 2-D-Gal and 2DG suppressed KPPR1 mucoidy, similar to Gal and Glc, despite being non-metabolizable sugar analogs (Fig. 2A). Although 2-D-Gal supplementation significantly reduced CPS abundance, it remained unchanged following 2DG supplementation, re-affirming that sugar-suppressed mucoidy is not due to reduced CPS abundance (Fig. 2B). To further validate that sugar metabolism is not required for sugar-suppressed mucoidy, we ablated Gal catabolism by deleting the Leloir operon (galETKM). The Leloir pathway is required for the catabolism of Gal, and as expected, the deletion mutant (ΔgalETKM) was unable to utilize Gal as a sole carbon source40. Like WT, culturing ΔgalETKM in M9+CAA+Gal still significantly suppressed mucoidy compared to M9+CAA (Fig. 2B). However, unlike WT, ΔgalETKM exhibited a significant reduction in cell-associated CPS abundance in M9+CAA+Gal, but not in M9+CAA, indicating a conditional effect on CPS synthesis (Fig. 2E). In the Leloir pathway, GalE catalyzes the reversible interconversion between UDP-galactose and UDP-glucose, an essential nucleotide-sugar precursor for CPS biosynthesis. In the presence of exogenous Gal (M9+CAA+Gal), accumulation of Gal metabolic intermediates and imbalance in CPS synthesis precursor pools may have impaired CPS synthesis in ΔgalETKM, specifically in M9+CAA+Gal. Such precursor imbalance-driven CPS biosynthesis regulation has been reported in other CPS synthesis systems41. Finally, we aimed to identify the specific Gal transporter(s) involved in mucoidy suppression. Gal import systems are poorly characterized in K. pneumoniae, so we used E. coli as a reference. In E. coli, Gal is primarily transported via galactose permease (GalP) and ABC-type Mgl transport system (MglBAC)42. In the absence of a Gal-specific PTS, other non-specific PTSs, such as the Man PTS can also facilitate Gal transport. The general PTS components, EI and HPr, encoded by ptsI and ptsH, respectively, are required across all PTS-dependent sugar import43. To investigate these, we measured the growth of the uptake mutants in M9+Gal and M9+CAA+Gal and quantified their mucoidy. No single transporter disruption rendered KPPR1 unable to grow in M9+Gal, indicating that multiple redundant transporters contribute to Gal import in KPPR1 (Supplementary Fig. 3A). Although all mutants reached similar growth density (OD600) at the later stages of stationary growth phase, galP::kan and PTS mutants (ΔptsI and ΔptsH) showed impaired growth rate in M9+CAA+Gal indicated by altered exponential growth phase (Supplementary Fig. 3B). These observations indicate that both GalP and PTS contribute to Gal import. Importantly, only galP loss restored Gal-suppressed mucoidy in CAA+Gal, suggesting GalP is the primary transporter associated with Gal-suppressed mucoidy (Fig. 2C). Furthermore, like ΔgalETKM, galP::kan showed reduced CPS abundance in M9+CAA+Gal, but not in M9+CAA (Fig. 2F). These findings strongly support that Gal-mediated mucoidy suppression occurs upon Gal import and prior to its catabolism. Galactose downregulates the regulator of mucoid phenotype (rmpD) and decreases CPS chain length uniformity. A KPPR1 carrying a GFP reporter vector (Prmp-dasherGFP) or (B–D) unmodified KPPR1 were cultured in M9+CAA±Gal. A KPPR1 rmpADC promoter (Prmp) activity was quantified by measuring the Prmp-driven GFP fluorescence intensity and normalized to OD600. B Relative transcript levels of KPPR1 rmpA, rmpD and rmpC were quantified by qRT-PCR in M9+CAA+Gal and compared to M9+CAA. C Cell-associated CPS was purified and resolved using a gradient SDS-PAGE and stained with 1% Alcian blue followed by silver stain. The presented gel image is representative of three independent experiments. D CPS chain length diversity was quantified by densitometric analysis using ImageJ. Data presented (A,B,D) are the mean, and the error bars represent the standard error of the mean. Statistical significance was determined using (A) unpaired two-tailed t-test and (B,D) multiple unpaired t-tests with multiple-comparison correction using the Benjamini, Krieger and Yekutieli false discovery rate method. Statistical significance was calculated by comparing sugar-supplemented condition to sugar-deficient condition. P-values are displayed above each comparison; ** p ≤ 0.01; # p ≤ 0.0001. Exact p-values: (A) left to right: <0.0001; (B) left to right: 0.0015, 0.0017, <0.0001; (D) left to right: 0.0016, <0.0001. Experiments were performed three independent times, either in biological triplicate (A,B; n = 9) or as a single replicate (C,D; n = 3). Source data are provided as a Source Data file. Galactose downregulates the rmpD mucoidy regulator and decreases capsular polysaccharide chain length uniformity In hvKp, multiple regulatory inputs commonly act via transcriptional regulation of rmpADC locus making it the ‘central’ regulatory locus of mucoidy6,8,10,44,45. Given the primary role of RmpD in regulating mucoidy, we hypothesized that sugars suppress mucoidy by transcriptionally down-regulating rmpD. To test this, we used a Prmp-dasherGFP fluorescent reporter construct to measure rmpADC promoter (Prmp) activity6. We observed significantly reduced Prmp activity in M9+CAA+Gal compared to M9+CAA (Fig. 3A). The ~3.3-fold decrease in promoter activity is consistent with the ~3-fold decrease in mucoidy upon Gal supplementation (Fig. 1A, 3A). Additionally, KPPR1 rmpADC transcript levels were significantly decreased by ~5-fold in M9+CAA+Gal relative to M9+CAA (Fig. 3B). Together, these data suggest that Gal suppresses mucoidy by downregulating the mucoidy regulator, rmpD. Hypermucoidy was conventionally attributed to increased CPS abundance, but recent studies have identified that increased CPS chain uniformity determines mucoidy4,5. CPS chain length can be assessed by analyzing the migration patterns of purified CPS via SDS-PAGE46. KPPR1 typically presents three distinct forms of CPS chains categorized by molecular weight (i.e. chain length) distribution. ‘Form A’ consists of CPS chains with diverse length (low- to high-molecular weight) and appears as a diffuse smear on a gel. ‘Form B’ represents high-molecular-weight CPS chains of uniform chain length and appears as a distinct band. ‘Form C’ corresponds to ultrahigh-molecular-weight CPS chains retained close to loading wells. Hypermucoid strains predominantly synthesize high-molecular weight ‘form B’ CPS chains with uniform length distribution, whereas low- and non-mucoid strains synthesize increased proportions of ‘form A’ CPS chains with diverse length4,5. We purified cell-associated CPS from KPPR1 cultured in M9+CAA and M9+CAA+Gal, and analyzed the CPS chain length distributions by SDS-PAGE. Gal supplementation significantly increased the abundance of ‘form A’ CPS (diverse; mid- to high- molecular weight chains), decreased ‘form B’ (uniform; high-molecular weight chains), and negligibly increased ‘form C’ (ultrahigh-molecular weight chains) compared to M9+CAA (Fig. 3C, D). Decreased CPS chain length uniformity (‘form B’) and increased chain length diversity (‘form A’) are consistent with Gal-suppressed mucoidy, as quantified by sedimentation assay (Fig. 1A, B)4–6. Collectively, our findings indicate that Gal import downregulates rmpD expression leading to the synthesis of CPS chains with decreased chain length uniformity and reduced mucoidy. Importable sugars downregulate rmpADC expression and broadly suppress mucoidy across hypervirulent K. pneumoniae strains Since Gal-suppressed mucoidy was linked to rmpD downregulation, we next investigated whether other mucoidy-suppressing sugars similarly affected rmpADC expression. Using the Prmp-dasherGFP reporter, we measured rmpADC promoter activity in sugar-supplemented conditions. All tested sugars significantly reduced the Prmp activity, whereas GalNAc showed no impact on promoter activity, consistent with its lack of effect on mucoidy (Figs. 1A, 4A). As described earlier, KPPR1 lacks the major GalNAc transporter (GalNAc PTS) and it is unable to support bacterial growth as a sole C-source. These findings indicate supplementation of other host-relevant and importable sugars directly downregulates rmpADC expression, providing a mechanistic basis for sugar-suppressed mucoidy regulation. Sugar import reduces rmpADC promoter activity, and galactose broadly suppresses mucoidy and rmpADC promoter activity in clinical hypervirulent K. pneumoniae strains. A KPPR1 Prmp-dasherGFP was cultured in M9+CAA+sugar and rmpADC promoter (Prmp) activity was quantified based on fluorescent protein expression normalized to OD600. B,C Additional hypervirulent K. pneumoniae strains were cultured in M9+CAA±Gal then mucoidy and Prmp activity were quantified. Laboratory strains tested were KPPR1 and NTUH-K2044, and clinical isolates were Kp4289, Kp4585 and Kp6557. B Mucoidy was determined by quantifying the supernatant OD600 after sedimenting 0.5 OD600 unit of culture at 1000 x g for 5 min. C Prmp activity of hypervirulent strains was quantified as in (A). Data presented are the mean, and the error bars represent the standard error of the mean. Statistical significance was determined using (A) one-way ANOVA with Šídák correction and B,C two-way ANOVA with Šídák correction. Statistical significance was calculated by comparing sugar-supplemented condition to sugar-deficient condition. P-values are displayed above each comparison; # p ≤ 0.0001. Exact p-values: (A) left to right: <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001; (B) left to right: <0.0001, <0.0001, <0.0001, <0.0001, <0.0001; (C) left to right: <0.0001, <0.0001, <0.0001, <0.0001, <0.0001. Experiments were performed three independent times, each in biological triplicate (n = 9). Source data are provided as a Source Data file. To determine whether sugar-suppressed mucoidy is conserved across hvKp strains, we tested the effect of Gal on additional hvKp strains. We selected one laboratory hypervirulent strain (NTUH-K2044) and three clinical isolates (Kp4289, Kp4585 and Kp6557) in addition to KPPR1. Gal significantly suppressed mucoidy in all five hypervirulent strains (Fig. 4B). Furthermore, Gal significantly reduced Prmp activity in all tested hvKp strains when quantified with the Prmp-dasherGFP reporter (Fig. 4C). Notably, the extent of mucoidy suppression in hvKp strains varied by rmp lineage. The rmp1- and rmp2-encoding strains (Kp6557, Kp4289 and Kp4585) presented greater Gal-suppressed mucoidy compared to rmp3 strains (Fig. 4B). This is consistent with a prior report that rmp1 and rmp2 strains more frequently confer the hypermucoid phenotype to classical K. pneumoniae strains compared to rmp37. These lineage-dependent differences in mucoidy regulation likely reflect the intrinsic functional differences in RmpD and Wzc associated with distinct rmp lineages and capsule loci7,47. Quantification of Prmp activity revealed that all hvKp strains had lower baseline Prmp activity than KPPR1 in M9+CAA, yet all strains exhibited a significant reduction in the promoter activity in response to Gal (Fig. 4C). It is important to note that all reporter constructs contained the same rmp3-type Prmp derived from KPPR1 (Fig. 4C). Thus, it is possible that a mismatch between the rmp3 lineage of the reporter construct and the native rmp promoter of the respective strains could have partly contributed to the observed lineage-dependent discrepancies in Prmp activity and mucoidy. Nevertheless, the rmp2 strains Kp4289 and Kp4585 exhibited a greater reduction ( ~ 25-fold) in Prmp activity in response to Gal, which is consistent with their striking Gal-suppressed mucoidy compared to other lineages (Fig. 4B, C). Together, these data indicate that the import of Gal, and likely other sugars, broadly suppresses mucoidy by transcriptionally downregulating rmpADC via a regulatory mechanism conserved across hypervirulent strains, although the magnitude of this regulation is rmp lineage-dependent. Transposon screen identifies genes associated with sugar-suppressed mucoidy To identify bacterial factors involved in sugar-suppressed mucoidy, we screened a K. pneumoniae transposon library for mutants demonstrating increased mucoidy in Gal (Supplementary Fig. 4). We used a previously published, ordered KPPR1 transposon library consisting of 3,733 mutants, representing approximately 72% of all open reading frames (ORFs)5,48. A high-throughput sedimentation assay adapted for a 96-well plate was employed in two successive rounds to screen for mutants with increased mucoidy in M9+CAA+Gal. In the primary screen, 261 mutants were identified as preliminary hits, which were refined to 40 mutants following the secondary screen in M9+CAA+Gal (Supplementary Fig. 4). The hits were validated using a standard, full-scale (3 mL) sedimentation assay, assessing the mutants for increased mucoidy in M9+CAA+Gal compared to WT46. Ultimately, n = 15 transposon mutants were confirmed as final hits that had increased mucoidy in M9+CAA+Gal (Fig. 5A). Targeted gene deletion mutants of each of the 15 hits were generated, and their mucoidy was measured. All but three mutants (ΔrpoN, ΔaroK and ΔyceG) maintained increased mucoidy in M9+CAA+Gal (Fig. 5B). Whole genome sequencing revealed no secondary mutations that could explain the discrepancy in mucoidy between the transposon and deletion mutants of these three genes. Overall, the 15 identified gene hits spanned diverse cellular processes. Most were associated with carbon-responsive global regulation (cyaA, crp, csrD) and nitrogen regulation and signaling (ntrB, ntrC, ptsN, rpoN) (Fig. 5A, B). Additional hits included genes implicated in antimicrobial peptide transport (sapA and sapB), inorganic phosphate uptake (pitA), cell-wall remodeling (yceG), cofactor biosynthesis (epd), pentose phosphate pathway (rpe) and amino acid biosynthesis (aroK) (Fig. 5A, B). Transposon screen identifies K. pneumoniae mutants with increased mucoidy in galactose-supplemented medium. A Transposon (Tn) mutant hits were cultured in M9+CAA±Gal and validated for increased mucoidy in M9+CAA+Gal compared to M9+CAA by quantifying the supernatant OD600 after centrifugation at 1000 x g for 5 min. B Isogenic gene deletion mutants were generated for each KPPR1 Tn mutant identified as positive hits in (A). The mutants were cultured in M9+CAA±Gal, and their mucoidy was determined by quantifying the supernatant OD600 after centrifugation at 1000 x g for 5 min. Data presented are the mean, and the error bars represent the standard error of the mean. A,B Statistical significance was determined using two-way ANOVA with Dunnett’s post-hoc test. Statistical significance was calculated by comparing each mutant to the WT in the respective growth medium. P-values are displayed above each comparison; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001; # p ≤ 0.0001. Exact p-values: (A) left to right: <0.0001, 0.0014, <0.0001, <0.0001, <0.0001, <0.0001, 0.0006, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, 0.0012, <0.0001, <0.0001; (B) left to right: <0.0001, <0.0001, 0.0004, <0.0001, <0.0001, <0.0001, 0.0037, 0.0008, 0.0026, <0.0001, <0.0001, <0.0001, 0.0012, <0.0001, <0.0001, 0.0275, 0.0010, <0.0001, 0.0028, 0.0038. Experiments were performed three independent times, each in biological triplicate (n = 9). † Gene of unknown function; PPP: pentose phosphate pathway. Source data are provided as a Source Data file. A global regulator, cAMP receptor protein (CRP), is required for sugar-suppressed mucoidy CRP is a global carbon catabolite regulator known to play a role in virulence regulation in Klebsiella and other bacterial species49–52. CRP forms a complex with cAMP, the adenylate cyclase (CyaA) product, to upregulate or downregulate various metabolic and virulence-associated genes. cAMP synthesis requires CyaA activation, which is primarily regulated by sugar import through sugar-specific phosphotransferase (PTS) systems. Given these established functions of CRP and the identification of both cyaA::kan and crp::kan as major transposon hits in our screen, we further investigated the role of cAMP-CRP in sugar-mediated mucoidy regulation. We measured mucoidy of KPPR1 ΔcyaA and Δcrp mutants cultured in M9+CAA with or without Gal. As expected, both the mutants lacked intracellular cAMP, restored mucoidy and exhibited significantly increased CPS abundance in M9+CAA+Gal (Fig. 6A–C). Meanwhile, expressing cyaA and crp in trans using native promoters restored Gal-suppressed mucoidy to WT levels in M9+CAA+Gal (Supplementary Fig. 5A). Furthermore, Gal supplementation increased Prmp activity more than ~9.5-fold in both ΔcyaA and Δcrp mutants (Fig. 6C). Interestingly, in M9+CAA, there was no significant change in Prmp activity (Fig. 6C), yet both mutants had significant decreases in mucoidy compared to WT (Fig. 5B). The increased mucoidy and Prmp activity of ΔcyaA and Δcrp in M9+CAA+Gal correlated with an increased abundance of uniform-length CPS chains (‘form B’) and ultrahigh-molecular weight ‘form C’ chains, relative to WT (Fig. 6D, E). In contrast, CPS chains with diverse lengths, represented by ‘form A’, decreased in abundance (Fig. 6D, E). Collectively, these observations implicate cAMP-CRP signaling in sugar-dependent K. pneumoniae mucoidy regulation. Sugar-suppressed mucoidy is linked to the cAMP-CRP global regulator. KPPR1 wildtype (WT), and ΔcyaA and Δcrp mutants were cultured in M9+CAA±Gal. A Cell-associated CPS was extracted and measured for uronic acid content. B Intracellular cAMP levels and (C) Prmp activity were measured using a Prmp and cAMP-CRP-responsive dual fluorescent reporter. Cell-associated CPS were (D) resolved by SDS-PAGE and (E) analyzed for chain length diversity by densitometric analysis in ImageJ. Data presented are the mean, and the error bars represent the standard error of the mean. Statistical significance was determined using two-way ANOVA with Dunnett’s post-hoc test where each mutant was compared to WT in the same growth medium. For (E), log2 fold change (FC) in chain length was compared to a hypothetical value of ‘0’. P-values are displayed above each comparison; * p ≤ 0.05; # p ≤ 0.0001. Exact p-values: (A) < 0.0001; (B) left to right: <0.0001, <0.0001; (C) left to right <0.0001, <0.0001; (E) left to right: 0.0001, 0.0021, <0.0001, <0.0001, <0.0001, <0.0001. CPS staining was performed independently four times with a cumulative total of n = 4 biological replicates. Experiments were performed three independent times in biological triplicate (A–C; n = 9) or four independent times as a single replicate (D,E; n = 4). Source data are provided as a Source Data file. Next, we asked whether cAMP-CRP broadly drives sugar-suppressed mucoidy. We quantified intracellular cAMP levels in KPPR1 cultured in minimal medium supplemented with a panel of sugars. Analogous to Gal, each imported sugar significantly reduced intracellular cAMP levels (Fig. 7A). Sugar import broadly reduces intracellular ATP and cAMP, and suppresses mucoidy via the cAMP-CRP regulatory pathway. KPPR1 wildtype (WT) and/or ΔcyaA were cultured in M9+CAA ± sugar. A Intracellular cAMP levels and (B) ATP levels of WT cultured in different sugars were measured using a cAMP-CRP-responsive fluorescent reporter and BacTiter-Glo Microbial Cell Viability Assay, respectively. C The mucoidy of ΔcyaA cultured in different sugars was determined by quantifying the supernatant OD600 after centrifugation at 1000 x g for 5 min. Data for WT is also shown in Fig. 1 and included here for statistical comparisons. Data presented are the mean, and the error bars represent the standard error of the mean. Statistical significance was determined by comparing (A,B) sugar-added condition to M9+CAA using one-way ANOVA with Šídák correction or (C) mutant to WT using two-way ANOVA with Šídák correction. P-values are displayed above each comparison; * p ≤ 0.05; # p ≤ 0.0001. Exact p-values: (A) left to right: <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001; (B) left to right: <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001; (C) left to right: <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, 0.0112. All experiments were performed three independent times, each in biological triplicate (n = 9). Source data are provided as a Source Data file. Since ATP is the substrate for cAMP synthesis by CyaA, we next measured intracellular ATP of KPPR1 cultured in M9+CAA±sugar. Bacterial cells at the late stationary phase showed significantly lower intracellular ATP compared to M9+CAA (Fig. 7B). Across most sugar-added conditions, ATP levels were consistent with their respective cAMP levels with a notable exception of fucose. Fucose is not readily utilized by KPPR1 and caused no significant changes in ATP level despite a significant reduction in cAMP level (Supplementary Fig. 1F and Fig. 7B). This suggests that cAMP levels can decrease independently of ATP availability. In fact, CyaA activity in E. coli is primarily regulated by the phosphorylation state of EIIAGlc rather than the substrate (ATP) availability53. These data indicate that a reduction in cAMP levels due to sugar supplementation is not necessarily a consequence of changes in ATP levels. Furthermore, differences in ATP levels across sugar-added conditions were not consistent with differences in mucoidy suppression observed among different sugars (Fig. 1A). It is important to note that both cAMP and ATP levels were measured at the late stage of stationary growth, and the levels of these nucleotides could be different during exponential growth. Nonetheless, inhibition of cAMP synthesis by cyaA deletion was sufficient to significantly restore mucoidy in M9+CAA supplemented with each sugar (Fig. 7C). Altogether, our data strongly suggests that the general regulatory effect of sugars on K. pneumoniae mucoidy is mediated by sugar import-driven changes in intracellular cAMP. cAMP-CRP indirectly regulates rmpADC expression To examine whether cAMP-CRP directly binds to the rmpADC promoter, we first used BPROM, a bacterial promoter prediction tool, which identified no putative cAMP -CRP binding site within Prmp. To validate this, we performed a DNA-protein interaction enzyme-linked immunosorbent assay (DPI-ELISA) to assess binding between CRP and the rmpADC promoter. A 338 bp Prmp fragment was incubated with 0.125–5 µg CRP and 0.2 mM cAMP, alongside a known CRP-binding promoter (lacP) as a positive control. cAMP-CRP showed no significant binding within Prmp compared to lacP, indicating that cAMP-CRP does not directly regulate rmpADC (Supplementary Fig. 6A). Therefore, we sought to identify intermediate regulatory factors linking cAMP-CRP to rmpADC regulation. We previously demonstrated that ArgR acts as a direct transcriptional activator of rmpADC expression6. ArgR is a hexameric transcriptional regulator and requires arginine as a cofactor. In M9+CAA+Gal, transcript levels of argR were significantly down-regulated relative to M9+CAA (Supplementary Fig. 6B). However, this reduction was not due to direct transcriptional regulation of argR by cAMP-CRP as DPI-ELISA revealed no significant binding between cAMP-CRP and PargR (Supplementary Fig. 6A). At least one gene of the arginine biosynthesis pathway, argG (arginosuccinate synthetase), has been shown to be directly regulated by cAMP-CRP54. Thus, we propose that cAMP-CRP-dependent changes in arginine metabolism influence the ArgR-mediated regulation of rmpADC. Altogether, these findings suggest that cAMP-CRP acts indirectly as a regulator of rmpADC through other direct regulatory factors. Non-mucoid K. pneumoniae associate at higher frequency with gut mucin and epithelial cells K. pneumoniae is able to invade and translocate across gut epithelial cells via a transcellular route55. Moreover, low- and non-mucoid K. pneumoniae strains typically associate more effectively than hypermucoid strains with host cells, including lung and intestinal epithelial cells48,56,57. However, pathogens must first overcome the mucus barrier (primarily composed of mucin) to reach the epithelial layer, which limits access to the epithelial surface and might even define the dissemination to extra-intestinal sites. We therefore examined whether mucoidy suppression impacts K. pneumoniae interaction with gut mucin and epithelial cells, thereby affecting key steps in gut colonization and eventual dissemination. To test the effect of mucoidy suppression on bacterial mucin binding, we performed a solid-phase mucin binding assay using semi-purified porcine gastric mucin (PGM). A significantly higher proportion ( ~ 5-fold) of KPPR1 WT pre-grown in M9+CAA+Gal was bound to PGM compared to WT grown in M9+CAA (Fig. 8A). Furthermore, a constitutively non-mucoid ΔrmpD mutant, bound to mucin significantly higher ( ~ 7.5-fold) than WT in CAA, regardless of Gal supplementation (Fig. 8A). These findings suggest that modulation of mucoidy by host-derived sugars may influence interactions between K. pneumoniae and mucin. The non-mucoid state enhances K. pneumoniae host interactions in the gut. A,B KPPR1 WT and rmpD were cultured in M9+CAA±Gal prior to quantifying bacterial binding to porcine gastric mucin (PGM) and human intestinal epithelial cells (Caco-2). A A Nunc MaxiSorp plate was coated with crude PGM and treated with ~105 KPPR1 WT and ΔrmpD cells. After treatment, bacterial binding was measured by enumerating colony forming units (CFU) relative to input CFUs. B A monolayer of Caco-2 cells was infected with KPPR1 WT or ΔrmpD for 1.5 and 3 hours at an MOI of 10. Infected cells were washed with PBS and then Caco-2 cells were lysed with Triton X-100. Total bacterial association was quantified by enumerating output CFU relative to input CFU. C To assess relative fitness in a murine gut colonization model, WT KPPR1S and KPPR1-derived mutants were mixed at 1:1 (ΔrmpD) or 1:50 (ΔcyaA and ΔcyaA-ΔrmpD), and 109 CFU were orally gavaged into mice at least three independent times. Large intestines were harvested after 48 h post-infection. CFUs were enumerated and log10 competitive indices (CI; mutant:WT) were calculated using Eq. (1). Each dot represents an individual mouse, where open circles are males and half-filled circles are females. Data are presented as (A,B) mean with standard error of the mean or (C) median with interquartile range. Statistical significance was determined using (A) multiple Mann-Whitney test with Holm-Šídák multiple comparisons correction, (B) two-way ANOVA with Šídák’s multiple comparison test or (C) one-sample two-tailed t test and one-way ANOVA with Holm-Šídák multiple comparisons correction. P-values are displayed above each comparison; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001; # p ≤ 0.0001. In (C), p-value above each bar represents comparison of log10 competitive index of a mutant to a hypothetical value of ‘0’ and p-value above connected lines represents comparison of competitive index between two mutants. Exact p-values: (A) left to right: 0.0003, 0.0007, 0.0172; (B) left to right: 0.0135, 0.0005, <0.0001, 0.0162, <0.0001, <0.0001; (C) left to right: 0.0019, 0.0006, 0.0067. Experiments were performed (A) five independent times in biological triplicate (n = 15), (B) three independent times in biological triplicate (n = 9) and (C) > 2 independent times and a total n per group of 10 (ΔrmpD), 9 (ΔcyaA) and 15 (ΔcyaA-ΔrmpD). Source data are provided as a Source Data file. We next assessed whether mucoidy suppression alters K. pneumoniae association with Caco-2 intestinal epithelial cells. KPPR1 WT grown in M9+CAA+Gal associated with Caco-2 cells more efficiently at 1.5 hours (12-fold) and 3 hours ( > 4-fold) compared to WT in M9+CAA (Fig. 8B). Similarly, WT in M9+CAA+Gal associated with Caco-2 cells at a 3-fold higher rate than WT in M9+CAA after 1.5 hours, which was reduced to a modest 1-fold higher rate at 3 h (Fig. 8B). The constitutively non-mucoid ΔrmpD mutant consistently showed higher association with Caco-2 cells compared to WT, irrespective of Gal (Fig. 8B). These observations align with a previous report where hypermucoid mutants adhered and invaded Caco2 cells at a lower rate than WT57. Together, these results reinforce that sugars suppress mucoidy through a rmpD-dependent mechanism that enhances K. pneumoniae interaction with gut mucin and epithelial cells, with potential consequences in gut colonization and dissemination. Dynamic mucoidy regulation contributes to K. pneumoniae gut colonization To assess the impact of mucoidy regulation on intestinal colonization, we performed competitive infection assays using a murine gut colonization model using mice with intact microbiome, procured from two different vendors to account for gut microbial diversity58,59. This model is essential because K. pneumoniae encodes few mucin-degrading glycosidases and therefore likely depends on commensal mucin degraders, such as Akkermansia muciniphila and Bacteroides thetaiotaomicron, to liberate and access mucin-derived sugars60. Thus, assessing mucoidy in the context of an intact microbiome allows us to evaluate pathobiont-host-microbiome interactions that could shape nutrient availability in the gut. A similar nutrient dependence is demonstrated in Clostridium difficile, which interacts with mucin-degrading commensals to access mucin-derived monosaccharides20. Using the aforementioned model, the ΔcyaA mutant, which is constitutively hypermucoid regardless of sugar availability, and the ΔrmpD mutant, which is constitutively non-mucoid, were competed against the WT (KPPR1s) by oral inoculation of mice (Supplementary Fig. 8A). We also competed the non-mucoid ΔcyaA-ΔrmpD double deletion strain against WT to control for the secondary effects of cyaA deletion. Prior to in vivo competition, ΔcyaA and ΔrmpD strains were competed against WT for in vitro growth in a rich medium, which revealed that ΔcyaA exhibits a significant growth defect compared to WT and ΔrmpD (Supplementary Fig. 8C). Thus, ΔcyaA and ΔcyaA-ΔrmpD strains were competed in vivo against WT using 1:50 (WT: mutant) input ratio, while ΔrmpD and WT were used in 1:1 input ratio (Supplementary Fig. 8D). Bacterial burdens in large intestines were then enumerated after 48 h. All three mutant strains were significantly out-competed by WT after 48 h (ΔrmpD: −1.59; ΔcyaA: −2.92 and ΔrmpD-ΔcyaA: −1.74 log-fold CI) in gut colonization, suggesting that disruption in mucoidy-regulatory pathways can impair K. pneumoniae fitness during gut colonization (Fig. 8C). Furthermore, there was no significant difference between the competitive indices of ΔrmpD or ΔcyaA and the ΔcyaA-ΔrmpD double mutant (Fig. 8C). This indicates that while disruption in mucoidy regulation contributes to gut colonization, high mucoidy cannot overcome the pleiotropic effects of CyaA disruption. It is important to note that several ΔcyaA-ΔrmpD output CFU values were below the detection limit, which may have masked additional differences in competitive fitness of the ΔcyaA-ΔrmpD mutant (Supplementary Fig. 8E). Nevertheless, these data collectively suggest that CyaA contributes to optimal gut colonization, similar to mucoidy regulation via rmpD. Broadly, these observations support a model in which hypervirulent K. pneumoniae mucoidy dynamically modulates mucoidy to achieve optimal gut fitness. Discussion Nutrient signals like amino acids and sugars are known modulators of K. pneumoniae mucoidy, but the mechanism by which sugars influence mucoidy has remained undefined. Here, we investigated how individual sugars influence transcriptional responses and CPS properties that dictate mucoidy. We found that sugars broadly suppress K. pneumoniae mucoidy by downregulating rmpADC and diversifying CPS chain length. This regulation is coupled with sugar import, requiring cAMP-CRP, and is independent of sugar catabolism. Moreover, sugar-suppressed mucoidy is conserved across hypervirulent K. pneumoniae strains and could influence host-pathogen interactions in the gut and other niches. These data support and expand our understanding of how K. pneumoniae metabolic status regulates the mucoid state. We propose a model whereby sugars and amino acids exert opposing effects on K. pneumoniae CPS properties, thereby integrating nutritional cues to fine-tune virulence and fitness within host microenvironments. We systematically tested eleven individual sugars for their effects on K. pneumoniae mucoidy. All, except GalNAc, suppressed KPPR1 mucoidy (Fig. 1A). Notably, GalNAc was the only sugar unable to support growth as a sole carbon source, which is consistent with the absence of GalNAc transporter (ageVWEF) in KPPR1 (Supplementary Fig. 1). These data suggest that consumable sugars broadly suppress K. pneumoniae mucoidy. However, assays using non-metabolizable sugar analogs (2-d-Gal and 2DG) and a galactose catabolic mutant (ΔgalETKM) revealed that mucoidy suppression is triggered by sugar import, not catabolism (Fig. 2A, B). Our data support that sugar-suppressed mucoidy is independent of osmotic stress caused by sugars or pH changes due to sugar catabolism. Because ΔgalETKM cannot metabolize Gal, and WT cannot metabolize 2-d-Gal, the medium pH remains stable, yet we still observe sugar-suppressed mucoidy in these conditions (Supplementary Fig. 7). Similar to our experimental approach, a previous study examined mucoidy in CAA supplemented with glucose, glycerol or fucose10. While this study did not specifically compare mucoidy in sugar-added conditions to M9+CAA, the mucoidy was lower relative to LB, a growth medium that elicits similar mucoidy levels as M9+CAA in our experience. RmpD is a key regulator of mucoidy in hypervirulent K. pneumoniae. The rmp operon, which encodes rmpD, is transcriptionally responsive to diverse environmental cues5,6,9–11. For example, the arginine-responsive ArgR and iron-responsive Fur regulators have been shown to directly activate or repress rmpADC promoter6,11,44. In this study, we present that sugar-suppressed mucoidy is also mediated through transcriptional rmpADC regulation. Gal supplementation markedly reduced rmpADC promoter activity and relative transcript levels of all three genes within the operon (Fig. 3A, B). More broadly, the Prmp activation was consistently reduced across all mucoidy-suppressing sugars (Fig. 4A). Thus, transcriptional regulation of rmpADC could be a general response to sugar availability. Although Gal significantly reduced rmpC expression and moderately reduced CPS abundance, not all sugars had reduced CPS abundance despite a reduction in overall rmpADC promoter activation (Figs. 1C, 4A). We previously observed a similar incongruity between CPS abundance and rmpC expression6. CPS biosynthesis is regulated at multiple layers, and the effect of one regulator could be compensated for by another transcriptional regulator61. Thus, a transcriptional change of one regulator, rmpC, might not translate to a proportional change in CPS abundance due to the activity of a hierarchically dominant regulator. To characterize the genetic basis of sugar-suppressed mucoidy, we screened an ordered KPPR1 transposon library and validated 11 K. pneumoniae genes whose disruption overcame sugar-suppressed mucoidy (Fig. 5A, B). These genes are related to diverse cellular functions, including carbon-responsive global regulation (cyaA, crp, csrD), nitrogen regulation and sensing (ntrB, ntrC, ptsN), peptide transport (sapA, sapB), phosphate uptake (pitA), pentose phosphate pathway (rpe), and one gene of unknown function (3906). While several of these genes have been previously implicated in mucoidy or CPS modulation, they were not studied in the context of sugars as regulatory nutrient cues48,62. Thus, it is important to note that some of the genes could be a general modulator of mucoidy independent of nutrient signal. For instance, sapA and sapB have been previously implicated in mucoidy and CPS regulation, but the underlying mechanism remains unresolved. Nevertheless, the dominance of carbon and nitrogen-associated genes in our screen re-affirms that metabolic status plays a major role in regulating K. pneumoniae mucoidy48. Specifically, shifting the balance of carbon and nitrogen levels could drive K. pneumoniae to a hypo- or hyper-mucoid state. Since amino acids are positive modulators of mucoidy, we speculate that bacterial integration of sugars (carbon-rich) and amino acids (nitrogen-rich) availability in the host environment defines mucoidy levels. This is supported by the observation that sugar-suppressed mucoidy was concentration-dependent in M9+CAA+Gal, where mucoidy-activating arginine in CAA could oppose the mucoidy-suppressing effects of lower sugar concentrations (Fig. 1B). Thus, shifting metabolism in favor of either sugars or amino acids could shift the cell to a hypo- or hyper-mucoid state. These findings collectively support the emerging view that K. pneumoniae mucoidy is controlled by complex and integrated regulatory networks48,61,63. Rather than being regulated as a binary on/off trait, mucoidy is fine-tuned in response to multiple nutrient and environmental cues, providing niche-specific adaptations to the K. pneumoniae cell surface. Although the transposon screen identified several candidates likely involved in sugar-suppressed mucoidy, we focused on cyaA and crp as CyaA activation and cAMP synthesis depend on sugar import. Import of non-PTS or less-preferred sugars activates CyaA to synthesize the second messenger, cAMP64. When complexed with CRP, cAMP-CRP functions as a global transcriptional regulator affecting several cellular processes, including virulence and fitness65–69. Notably, CRP represses K. pneumoniae CPS biosynthesis by acting at multiple CPS biosynthesis promoters52,70. It should be noted that not all sugars consistently changed CPS abundance despite a uniform decrease in intracellular cAMP (Fig. 1C, 7A), suggesting that the regulatory influence of CRP may be context-dependent and modulated by other regulatory factors. Moreover, CRP has also been implicated in mucoidy suppression51,70. Although the precise mechanism remains unclear, cAMP-CRP is proposed to suppress mucoidy by up-regulating the small RNA ArcZ, which inhibits the translation of transcripts associated with MlaA system51. This system maintains outer-membrane asymmetry, which is essential for cell surface CPS retention. Therefore, ArcZ-dependent mucoidy regulation likely occurs independent of rmpD regulation and reflects altered CPS retention rather than chain length regulation. Moreover, the marked increase in rmpADC promoter activity observed in ΔcyaA and Δcrp mutants cannot be explained by ArcZ due to the post-transcriptional mode of regulation of ArcZ. Thus, the sugar-suppressed mucoidy reported here likely acts through a distinct mechanism of mucoidy regulation. CRP-dependent promoters can be activated through several mechanisms depending on CRP binding site configuration and intracellular cAMP concentraiton71. In Escherichia coli, for example, the pck promoter switches from activation to repression depending on cAMP-CRP levels72. At low-[cAMP], CRP binds to the high-affinity binding site located upstream of the transcription start site and activates pck, but at high-[cAMP], it binds to both high- and low-affinity binding sites, causing an overall repression of pck72. In our study, intracellular cAMP levels were elevated in the absence of sugar (M9+CAA), which correlated with increased mucoidy (Fig. 1A, 7A). Conversely, reduced intracellular [cAMP] (M9+CAA+sugar) correlated with decreased mucoidy (Fig. 1A, 7A). However, the linear mucoidy differences in a high- and low-[cAMP] scenarios conflicts with ΔcyaA and Δcrp mutant phenotypes, both of which exhibit increased mucoidy in M9+CAA+Gal, despite undetectable intracellular [cAMP]. This paradox could be explained by considering one or more of the following regulatory scenarios. First, Prmp may contain multiple CRP-binding sites within the promoter of an intermediate regulatory factor with varying affinities to cAMP-CRP. Analogous to the E. coli pck promoter, this could lead to differences in response depending on cAMP-CRP concentrations. Second, CRP may concurrently activate and/or repress multiple Prmp direct regulatory factors. Third, CRP itself may be regulated by additional regulatory inputs that were not identified in this study. However, these proposed regulatory circuits remain speculative and must be validated with further studies. Nevertheless, cAMP-CRP indirectly regulates the rmp promoter, likely in a [cAMP]-dependent manner. K. pneumoniae infections often originate from gut colonization, where it primarily colonizes the small intestine, followed by sustained persistence in the colon over time73. Nutrient availability, such as dietary simple carbohydrates are critical in the colonization and dissemination of K. pneumoniae74. In fact, bacterial carbohydrate uptake and catabolic pathways, particularly galactose, fucose and xylose, are important for K. pneumoniae colonization in murine gut models10,23,24. Based on this knowledge, we examined how sugar-suppressed mucoidy influences K. pneumoniae gut colonization. Within the colon, K. pneumoniae primarily localizes to the outer mucus layer but then, upon reaching a high colonizing density, can invade the inner mucus layer adjacent to the colonic epithelium75. Contact with the epithelial surface is proposed to facilitate dissemination to extra-intestinal sites through transcellular invasion of intestinal epithelial cells55,57,75. Our data suggest that sugar-suppressed mucoidy may enhance the ability of K. pneumoniae to invade and interact with the intestinal mucin, as both K. pneumoniae pre-cultured in Gal and a constitutively non-mucoid strain (ΔrmpD) significantly increased binding to gastric mucin (Fig. 8A). We, however, acknowledge that the type of mucin used in our study does not fully recapitulate the mucin diversity across the entire GI tract. To date, the most well-established function of mucoidy has been observed in limiting bacterial association with host cells ex vivo4–6,8. Previous studies have demonstrated that K. pneumoniae can invade intestinal epithelial cells in vitro, and that hypermucoid strains adhere to and associate with gut epithelial cells less efficiently than WT55,57. Consistent with this, Gal-suppressed mucoidy increased the association of K. pneumoniae to intestinal epithelial cells, suggesting that local sugar availability could suppress cell surface mucoidy and facilitate bacterial interaction with host cells (Fig. 8B). Enhanced interaction with mucin and gut epithelial cells could be consequential for subsequent dissemination from intestinal sites, which was not directly evaluated by our ex vivo and in vivo models. In a mono-colonized murine model, K. pneumoniae could disseminate to mesenteric lymph nodes following initial gut colonization75. Given that carbon sources are a limited resource in the gut environment and that dietary carbohydrates can shape the availability of mucoidy-suppressing sugars, carbohydrate availability likely serves as a key regulatory cue for K. pneumoniae mucoidy during gut colonization and subsequent stages of dissemination. To determine whether mucoidy regulation influences gut fitness in vivo, we performed competitive colonization assays in a murine gut colonization model with an intact microbiome. Constitutively hypermucoid (ΔcyaA), non-mucoid (ΔrmpD) and non-mucoid double mutant (ΔcyaA-ΔrmpD) strains were each competed against WT for gut colonization in a murine model. Both ΔcyaA and ΔrmpD were significantly out-competed by WT, suggesting that alterations in mucoidy-associated pathways can impair gut colonization (Fig. 8C). These findings are consistent with a prior transposon screen in which both hypermucoid mutants (e.g., cyaA, crp, fur) and non-mucoid mutants (e.g., rmpA, rmpD, argR) were recovered at lower frequencies than WT following oral infection, suggesting that a fixed mucoidy state is not optimal for gut colonization76. The significant fitness defect of ΔrmpD, where growth rate differences do not confound interpretation, directly demonstrates that a constitutively non-mucoid state is detrimental to gut colonization. However, the fitness defect of ΔcyaA cannot be attributed solely to hypermucoidy due to the broad regulatory role of cAMP-CRP. In K. pneumoniae, cAMP-CRP controls multiple fitness and virulence determinants including type 3 fimbriae, biofilm formation, antimicrobial resistance, capsule production and metabolism of key nutrients49,52,67,77. To isolate the contribution of hypermucoidy to the ΔcyaA colonization defect, we generated a non-mucoid ΔcyaA-ΔrmpD double mutant (Supplementary Fig. 8A). This double mutant was also significantly outcompeted by WT, and the defect was not significantly different from either ΔrmpD or ΔcyaA single mutant (Fig. 8C). The comparable colonization defect of ΔcyaA and ΔcyaA-ΔrmpD suggests that constitutive hypermucoidy does not meaningfully enhance gut fitness and that the ΔcyaA colonization defect is largely driven by mucoidy-independent functions of cAMP-CRP. This interpretation is consistent with a prior study where a hypermucoid strain showed similar gut colonization as WT but showed increased translocation and virulence57. It is important to note that a larger proportion of ΔcyaA-ΔrmpD were recovered at or below the limit of detection compared to the other mutants, raising the possibility of a more severe colonization defect in the double mutant (Supplementary Fig. 8E). Unfortunately, the sensitivity level of our current models limited our ability to determine whether simultaneous disruption of cAMP-CRP and mucoidy produces an additive colonization defect. Despite these limitations, our results establish that CyaA is required for efficient gut colonization, consistent with the reported importance of cAMP-CRP in gut colonization by classical K. pneumoniae strains78. Mucoidy is unlikely to be the only colonization-relevant factor impacted by cAMP-CRP. This represents a key limitation of using ΔcyaA to interrogate the impact of hypermucoidy and warrants future work to identify the intermediate regulators linking cAMP-CRP to rmpADC to precisely assess the role of sugar-suppressed mucoidy in gut colonization. When considered together, our ex vivo and in vivo data support a model in which hypervirulent K. pneumoniae requires the ability to control mucoidy for effective gut colonization, and a switch between hypermucoid and non-mucoid state may facilitate distinct stages of colonization and dissemination. The metabolic flexibility of K. pneumoniae is consistent with a role for nutrient-responsive signaling pathways in modulating mucoidy and other fitness-associated traits in a niche-dependent manner79. Future studies that map the spatial and temporal regulation of mucoidy during different stages of colonization and dissemination will be critical for understanding how dynamic control of mucoidy contributes to K. pneumoniae infection. While our study focused on how sugars impact CPS-associated features, it is likely that sugar import and as a result, changes in cAMP-CRP alters other bacterial features important in the context of infection26. Thus, sugars may serve as a broad nutrient cue that shapes bacterial adaptation beyond cell surface polysaccharides. Building on this, our findings also raise intriguing questions about how K. pneumoniae integrates multiple and sometimes opposing nutrient cues. The concentration-dependent effects observed here indicate a potential competition between mucoidy-enhancing and mucoidy-suppressing cues in complex nutrient environments. How K. pneumoniae prioritizes between the opposing signals and what determines the dominance of one regulatory pathway over another remains to be studied. Moreover, we did not test all sugars at varying concentrations. Although high sugar concentrations uniformly suppressed mucoidy, whether individual sugars differ in mucoidy-regulating potency at lower concentrations could provide insight into hierarchical effects of individual sugars on suppressing mucoidy. At the mechanistic level, further studies are needed to assess regulation of rmpADC beyond the transcriptional level, as rmpADC promoter activity did not always correspond to mucoidy phenotype, as was observed in ΔcyaA and Δcrp mutants. Although the promoter activation in M9+CAA was similar between WT and the mutants, mucoidy was significantly lower in ΔcyaA and Δcrp than WT (Fig. 5B, 6C). This incongruity between rmp promoter activity and the mucoid phenotype suggests post-transcriptional regulation of rmp transcripts. CsrA is a post-transcriptional global regulator responsive to carbon availability and acid end-product accumulation. In E. coli, it is known to regulate bacterial motility, biofilm formation and metabolism by binding to transcripts and blocking ribosome binding80,81. CsrA activity in E. coli is attenuated by the expression of small regulatory RNAs, csrB and csrC, which in turn are inhibited by CsrD80,82. Notably, csrD is a hypermucoid transposon hit in our screen here and in other mucoidy screens48,62. Thus, post-transcriptional discrepancies between Prmp activity and mucoidy in ΔcyaA and Δcrp could be due to CsrA activity. Interestingly, the cAMP-CRP and Csr cascades cross-regulate expression of one another in other Gram-negative enteric bacteria83–85. Although it remains to be studied, these observations allude to a potential role for CRP and CsrA in maintaining mucoidy homeostasis in K. pneumoniae by acting at different stages of gene expression. Overall, these outstanding questions highlight the need to explore the association of nutrient fluctuations in the gut with the transition of K. pneumoniae from colonizer to invader state. Altogether, we present a model in which cAMP-CRP-dependent regulatory processes broadly fine-tune K. pneumoniae mucoidy, thereby influencing key stages of K. pneumoniae infection, such as intestinal colonization (Fig. 9). Information about sugar import is relayed through changes in intracellular cAMP-CRP levels. Specifically, intracellular cAMP-CRP is reduced following sugar import. Our results support that cAMP-CRP indirectly modulates the rmpADC promoter, leading to altered rmpD transcription. RmpD interaction with Wzc then regulates CPS chain length uniformity and mucoidy. These sugar-driven changes in mucoidy alter bacterial binding to mucin and host epithelial cells, such as those in the GI tract, and the ability to fine-tune mucoidy may facilitate gut colonization. Model of sugar-suppressed mucoidy in hypervirulent K. pneumoniae. K. pneumoniae encounters ecological niches, such as the human gut, where it can import both mucoidy-inducing amino acids and mucoidy-suppressing sugars. In the absence of sugars, EIIAGlc remains phosphorylated, which activates CyaA and induces cAMP synthesis. Sugars import dephosphorylates EIIAGlc, which in an unphosphorylated state, is unable to activate the CyaA adenylate cyclase. cAMP forms a complex with the CRP transcription factor to regulate gene expression. The cAMP-CRP complex indirectly via a yet-to-be identified regulatory factor modulates expression of the mucoidy regulator rmpD. Overall, effect of sugar import on lowering intracellular cAMP levels also reduces rmpADC expression. Reduced cellular levels of RmpD diminishes CPS chain length modulation via Wzc interactions, resulting in less uniform CPS chains on the bacterial cell surface (more Form A and less Form B CPS). This change is accompanied by a negligible increase in ultrahigh-molecular-weight Form C CPS. Decreased CPS chain length uniformity is associated with suppressed mucoidy, which promotes enhanced bacterial binding to gut mucin and intestinal epithelial cells. Solid lines (––) represent experimentally validated relationships and dashed lines (- - -) represent inferred relationships. Created in BioRender. Khadka, S. (2026) https://BioRender.com/jlny5wo. In summary, bacterial pathogens encounter nutritionally diverse environments that critically drive their adaptation for better fitness86. Depending on the niche and stage of infection, the ability to integrate environmental cues and dynamically adjust metabolism, virulence and fitness determinants may be essential for bacterial persistence and successful infection. Our findings identify sugars as key nutrient signals that modulate K. pneumoniae cell surface properties through cAMP-CRP-dependent regulation of mucoidy and CPS chain length. While mucoidy provides protection against phagocytosis by blocking association to macrophages, it can simultaneously impede bacterial adherence to epithelial surfaces, such as those of the gut, where association is likely important for dissemination after initial colonization. In this context, sugar-suppressed mucoidy may represent an adaptive response to host environments, where shifting nutrient profiles influence the bacterial metabolic state to promote colonization or invasive behaviors. Methods All experimental procedures comply with the ethical guidelines established by each institution involved. Institutional Biosafety Committee (IBC) approvals were obtained from the University of Pittsburgh (PROTO202400096) and the University of Toledo (500074-UT). This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals87. Animals were housed in a 12 h light and dark cycle, ambient temperature (~22°C) and humidity of approximately 15–35% . The Indiana University Institutional Animal Care and Use Committee approved this study (protocol 22114; PI: Vornhagen). Bacterial strains and culture conditions All bacterial strains and plasmids used in the study are detailed in Supplementary Data 1. Bacterial strains were cultured in either lysogeny broth (LB) (5 g/L yeast extract, 10 g/L tryptone, 0.5 g/L NaCl) or low-iron minimal medium (6 g/L disodium hydrogen phosphate, 3 g/L monopotassium phosphate, 0.5 g/L sodium chloride, 1 g/L ammonium chloride, 14.7 g/L calcium chloride, 120.3 g/L magnesium sulfate) with different carbon sources as described in the text, and incubated aerobically at 37˚C and 200 rpm shaking, unless otherwise noted. Solid culture medium was prepared by adding 20 g/L bacto-agar to LB. When appropriate, antibiotics were added at the following concentrations: kanamycin (25 µg/mL), chloramphenicol (20 µg/mL E. coli or 80 µg/mL K. pneumoniae), gentamicin (15 µg/mL). Sedimentation assay Mucoviscosity (mucoidy) was quantified by using a low-speed sedimentation assay46,88. Briefly, 1 mL of 0.5 OD600 equivalent bacterial overnight culture in 2 mL microcentrifuge tube was pelleted at 1000 x g for 5 min. The OD600 of the 900 µL supernatant was then quantified and presented as ‘supernatant OD600 per 0.5 OD600 bacterial culture’. Uronic acid quantification Capsular polysaccharide content was measured by quantifying uronic acid content46,89. In brief, capsular polysaccharide was purified by ethanol precipitation from 250 µL of overnight culture. Then, 50 µL 1% Zwittergent 3-14 was added to overnight culture (1:5 ratio) to purify total capsular polysaccharide while 50 µL of ultra-pure water was added to purify cell-free EPS. The mixture was precipitated at 17,000 x g for 5 min and the polysaccharide was collected by precipitating 100 µL of the uppermost supernatant in 400 µL ice-cold ethanol. Then, uronic acid content in the purified polysaccharide was quantified by using a colorimetric assay46. Cell-associated CPS uronic acid level was deduced by subtracting the uronic acid levels in cell-free EPS from the total capsular polysaccharide. Molecular cloning, transformation, and sequencing The oligonucleotides and plasmids used in this study are listed in Supplementary Data 2. K. pneumoniae isogenic knockout mutants were constructed using the λ Red recombineering adapted to K. pneumoniae48,90,91. Fluorescent reporter plasmids were constructed by PCR amplification of the plasmid backbone (pBBR1MCS-5) and specific gene fragments (promoter and fluorescent protein) followed by gel-purification (Monarch, NEB), and assembled using NEBuilder HiFi DNA Assembly Mix (NEB) for 1 h at 50 ˚C. Similarly, cyaA and crp complementation vectors were constructed by ligating PCR-amplified and gel-purified vector backbone (pACYC184) and the respective gene fragments along with their native promoter. The resulting ligated product was then transformed into TOP10 E. coli through heat-shock and confirmed via sequencing. The final plasmid was transformed into K. pneumoniae strains by electroporation91. Plasmid constructs were validated using whole plasmid sequencing (Plasmidsaurus and Eurofins Genomics). Select transposon mutants and all gene deletion mutants were validated using whole genome sequencing (SeqCoast Genomics). Whole genome and plasmid sequence analysis were performed using Benchling [https://benchling.com] and the Bacterial and Viral Bioinformatics Resource Center (BV-BRC)92. Growth curves Bacterial strains were cultured in the respective growth media and incubated aerobically at 37˚C and 200 rpm. Then, a 1:1000 subculture of the overnight culture was prepared in 3 mL of the respective growth medium. One hundred µL of 0.001 OD600 equivalent subculture was then transferred to a sterile 96-well plate. Additionally, 100 µL of sterile water was added to the outermost wells of the plate to prevent evaporation of the samples. Then, OD600 was measured every 30 min for 16 h at 37 °C and continuous orbital shaking (282 rpm) using a microplate reader (EPOCH2-SN, Agilent). For growth media that were non-permissive to bacterial growth, an overnight culture and subsequent subculture were prepared using M9+1% CAA. Fluorescent reporter assay Intracellular cAMP level and rmp promoter (Prmp) activity were measured using one of the two fluorescent reporter constructs (Supplementary Data 1). These reporter constructs were modified from a published fluorescent reporter vector93. For experiments requiring measurement of Prmp only (Fig. 3A, 4C), a single reporter construct expressing Prmp-dasherGFP was used6. For all other experiments requiring measurement of both Prmp and cAMP, a double reporter plasmid expressing Prmp-dasherGFP and lacP1-mScarlet-I was used. In all cases, 1 mL of 0.5 OD600 equivalent culture was centrifuged at 21,000 x g for 10 min to remove culture medium. Following the centrifugation, the supernatant was completely removed, and the pellet was resuspended in 1 mL 1× PBS. Then, 300 µL of the resuspended cells were transferred to a clear-bottom, black 96-well plate. DasherGFP fluorescence intensity was measured using a microplate reader (Synergy HTX, Agilent) at excitation and emission wavelengths of 485 and 528 nm, respectively. Similarly, mScarlet-I fluorescence intensity was measured at excitation and emission wavelengths of 560 and 620 nm, respectively. The fluorescence intensity was normalized to OD600 and reported as an arbitrary fluorescence unit per OD600 (AFU/OD600). Transposon screen in sugar-supplemented growth medium An ordered KPPR1 transposon library and plate-based sedimentation assay were used to screen for transposon mutants with increased mucoidy in galactose-supplemented medium48. Briefly, microplates containing the ordered library (a total of 3733 mutants) were thawed at room temperature and replicated into 100 µL of M9 1% CAA+80 mM gal. The replicated plates were then wrapped with a plastic wrap and incubated at 37˚C with 200 rpm shaking for 18–20 h. Following incubation, the plates were vortexed at low speed for 30 s and measured for OD600 and centrifuged at 1000 x g for 5 min. Then, the upper 50 µL was transferred to a new microplate and the OD600 was measured. Transposon mutants with mucoidy (supernatant OD600 per culture OD600) greater than two times the standard deviation of the plate mean were considered hits of the primary screen. The primary hits were then arrayed in a new microplate for a secondary screen. The newly arrayed plates contained each mutant in three wells as biological triplicates. The secondary screen was performed using a plate-based sedimentation assay, and the hits were identified following an inclusion criterion as described for the primary screen. For validation, KPPR1 WT and the secondary screen hits were cultured in M9+CAA+Gal, quantified for mucoidy using tube-based sedimentation assay and considered a valid hit if its mucoidy was significantly higher than that of WT in M9+CAA+Gal. A complete transposon screen data are available in the Supplementary Data 3. RNA isolation and quantitative RT-PCR Bacterial strain was cultured in 3 mL of low-iron M9 minimal media supplemented with 1% CAA with or without 80 mM galactose and incubated overnight aerobically at 37˚C with 200 rpm shaking. Next day, overnight cultures were sub-cultured 1:100 in the respective medium and incubated aerobically at 37 ˚C until the OD600 reached 0.4–0.5. Then, approximately 1 ×109 colony forming units (CFUs) were mixed with RNAProtect (Qiagen) at 1:2 (sample:RNAProtect) ratio, incubated at room temperature for 5 min and the cells were collected by centrifugation at 5000 x g for 10 min. Then, pelleted cells were lysed with 100 µL lysozyme (15 mg/mL in TE buffer) and treated with 10 µL proteinase K (20 mg/mL in TE buffer). RNA was isolated and purified using RNeasy mini-prep kit (Epoch) following manufacturer’s instruction and eluted with 30 µL RNase-free water and stored at −20˚ C until downstream processing. Genomic DNA was removed from the RNA preparations using ezDNase (ThermoFisher). cDNA was synthesized from the purified RNA using SuperScript IV First Strand Synthesis System (Invitrogen) following the manufacturer’s directions. The resulting cDNA was diluted 1:50 in water and used as template for quantitative real-time PCR (qRT-PCR) with SYBRGreen PowerUp reagent (Invitrogen) in QuantStudio 3 PCR system (Applied Biosystem). Primers used to amplify rmpA, rmpD, rmpC and gap2 (internal control) genes are listed in Supplementary Data 25. The relative fold change of the genes was calculated using the comparative threshold cycle (CT) method94. CPS chain length visualization To analyze chain length, CPS samples were separated by SDS-PAGE and detected using alcian blue staining5,46. Cell-attached CPS were purified from bacteria equivalent to 1.5 OD600. Cells were pelleted at 21,000 × g for 15 min to remove cell-free EPS. The bacterial pellet was resuspended in 1 mL of PBS and centrifuged at 21,000 × g for 15 min. All but 250 µL of the supernatant was removed, mixed with 50 µL 3–14 1% Zwittergent and incubated at 50 ˚C for 20 min. The incubated mixture was then centrifuged at 21,000 × g for 5 min to collect the upper 100 µL of the supernatant. The supernatant was added to a tube containing 400 µL ice-cold ethanol to precipitate polysaccharide. Purified cell-associated CPS was mixed with 4 × SDS gel loading dye (3:1, sample:dye) and resolved using 4–15% TGX stain-free precast gel (Bio-Rad). The electrophoresis was performed for 4.5 h at 300 V on ice at 4˚ C. After electrophoresis, the gel was stained with 0.1% alcian blue (0.1% wt/vol ThermoFisher Alcian Blue 8 Gx in stain base solution) for 1 h. It was then stained with Pierce Silver Stain Kit (ThermoFisher) following the manufacturer’s instructions and imaged on GelDoc Go Imaging System (Bio-Rad). CPS chain length diversity quantification Stained CPS gels were quantified using ImageJ V1.54r for Windows. A gel area of equal size was selected for each lane, including an empty lane for background measurement, and plotted using a one-dimensional electrophoretic gel analysis feature. The one-dimensional gel profile was then plotted, and different CPS forms and baselines were defined in the lane profile. The area of each peak corresponding to a distinct CPS form was quantified. Measurements from an empty control lane were subtracted from each sample lane to adjust for background. The quantification values presented in the figures are log2 fold change relative to wildtype. Comparisons were made between samples and control from the same gel to account for gel-to-gel variability. ATP quantification assay Bacterial intracellular ATP was measured using BacTiter-Glo Microbial Cell Viability Assay kit (Promega; Cat. No. G8230) following the manufacturer’s instructions with slight modifications. One mL of overnight bacterial culture was centrifuged at 21,000 x g for 5 min. After centrifugation, the bacterial pellet was resuspended in 1 mL of sterile water. The cells were washed a total of two times. Then, 50 µL of the washed cells equivalent to 0.5 OD600 were transferred to a black 96-well plate and mixed with 50 µL of BacTiter-Glo reagent. Similarly, 5 two-fold dilutions of 0.4 nanomoles of ATP (Promega; Cat. No. P1132) were included in the assay to generate an ATP standard curve. The plate was incubated at room temperature for minutes in the dark with gentle agitation. Then, luminescence was measured using a microplate reader (Synergy HTX, Agilent) and ATP concentrations were derived based on the standard curve. CRP protein purification Overnight culture of BL21-DE3 E. coli expressing 6xhis-CRP was sub-cultured in LB until OD600 reached ~0.3. Then, 0.5 mM IPTG was added to the sub-culture to induce protein expression at 37 °C for 3 h with 200 rpm shaking. Induced culture was centrifuged and the pellet was frozen overnight at −80 °C. The frozen pellet was resuspended in lysis buffer (50 mM Tris, 300 mM NaCl, 1 M DTT, 25 mM imidazole, Roche Complete EDTA-free protease inhibitor; pH 8.0) and sonicated to lyse the cells. Following lysis, supernatant was collected by centrifugation and His-tagged CRP was purified from the supernatant using HisPur Ni-NTA resin (Thermo Scientific; Cat. No. 88221). Resin-bound 6xhis-CRP was eluted with elution buffer (50 mM Tris, 300 mM NaCl, 1 M DTT, 500 mM imidazole; pH 8.0). Then, the eluted protein was further purified by dialysis at 4 °C using 10 kD MWCO SnakeSkin dialysis tubing (Fisher Scientific; Cat. No. 88243) against dialysis buffer (50 mM Tris, 300 mM NaCl, 1 M DTT, 500 mM imidazole; pH 8.0). Dialysis was repeated a total of four times for at least 3 h with successive imidazole concentration of 500 mM, 150 mM and 50 mM (twice). Final purity of the purified CRP was assessed using BioSafe Coomassie Stain and stored at −80 °C until further use. DNA-protein interaction (DPI)-ELISA DPI-ELISA was performed with slight modifications to the established protocols6,76,95. The targeted promoter sequences (probe) were PCR-amplified using biotinylated primers and purified using a PCR cleanup kit. Two pmol of probe diluted in PBS was added per well to a streptavidin-coated plate (ThermoFisher; Cat. No. 15125) and incubated for 2 h at room temperature to facilitate binding. After incubation, probe-bound wells were washed six times with 250 µL TBS-T and blocked with 200 µL 5% non-fat milk in PBS for 2 h at room temperature. After blocking, wells were once again washed with TBS-T as described earlier, then two-fold serial dilutions of purified His-tagged proteins were added to the respective probe-bound wells and incubated at room temperature for 1 h. In case of 6xHis-CRP, the protein was preincubated with 0.2 mM cAMP at room temperature for 30 min. After protein incubation, the wells were washed four times with 250 µL of TBS-T. Anti-His mouse monoclonal HRP conjugate antibody (Southern Biotech; 4603-05) was diluted 1:1000 in PBS and added to the wells (60 µL/well) and incubated for 1 h at room temperature. Then, the wells were washed three times with 250 µL PBS-T followed by three washes with 250 µL PBS. Finally, 100 µL slow TMB substrate (Thermo Fisher; Cat No. 34024) was added to each well and incubated at room temperature with gentle shaking until visible color was developed which was then stopped by adding 100 µL 0.18 M H2SO4, and A450 was measured using a microplate reader (Synergy HTX, Agilent). Mucin binding assay A Nunc-ImmunoTM MaxiSorpTM 96-well plate (Milipore Sigma; Cat No. M9410-1CS) was coated with 300 µL of 0.5 mg/mL crude porcine gastric mucin (PGM) in 0.1 M acetate buffer (pH 5.0). The coated plate was spun at 250 x g for 3 min and incubated at 37 ˚C for 24 h. Bacterial strains were cultured in 3 mL of low-iron M9 minimal media supplemented with 1% CAA with or without 80 mM galactose and incubated aerobically at 37 ˚C with 200 rpm shaking for 16–18 h. Approximately 105 bacterial cells were added per well in triplicate, spun at 250 x g for 3 min to facilitate binding, and incubated statically at 37 ˚C for 1 h. After incubation, the wells were washed 15 times with 300 µL sterile PBS and treated with 200 µL 0.5% Triton-X100 in PBS for 30 min to release bound bacteria. Input and output bacterial cells were serially diluted and plated on LB agar for enumerating CFU. Epithelial cell association assay Immortalized intestinal epithelial cells derived from a colorectal adenocarcinoma patient (Caco2, Sigma 86010202-1VL) were maintained in DMEM medium with L-glutamine, 4.5 g/L glucose and sodium pyruvate (Corning) supplemented with 20% heat-inactivated fetal bovine serum (Corning), 1% non-essential amino acids (Fisher Sci), 100 U/mL penicillin, and 100 µg/mL streptomycin in an atmosphere of 5% CO2. Bacterial strains were cultured in 3 mL of low-iron M9 minimal media supplemented with 1% CAA with or without 80 mM galactose and incubated aerobically at 37 ˚C with 200 rpm shaking for 16–18 h. Caco-2 cells in a tissue culture-treated petri dish were seeded to a density of 75,000 cells/mL per well to a tissue culture-treated 24-well plate. The seeded cells were incubated for 24–48 h to achieve a 90% confluent monolayer. Confluent Caco-2 cells in 24-well tissue culture dishes were washed with 1 mL of DPBS, then 1 mL bacteria (MOI 10) in additive-free DMEM was added to each well. Samples were spun at 500 rpm (54 × g) for 5 min, then incubated at 37˚ C, 5% CO2 for 1.5 and 3 h. After incubation, samples were washed three times with PBS followed by lysis with 1 mL of 0.2% Triton-X100 in PBS for up to 10 min. Input and cell-associated bacterial counts were determined by serial dilution and CFU enumeration on LB agar. Murine model of infection Six- to 8-week old C57BL6/J (The Jackson Laboratory, strain 000664 RB07) and C57BL/6Crl (Charles River, strain 027 R03) mice were used for the gut colonization studies58,59. Briefly, the KPPR1s, KPPR1 ΔcyaA, KPPR1 ΔrmpD, and KPPR1 ΔcyaA-ΔrmpD strains were cultured overnight from single colonies in LB at 37 °C. Since cyaA deletion impairs growth due to its role in central metabolism, the ΔcyaA and ΔcyaA-ΔrmpD mutants were orally inoculated at a 1:50 (WT:mutant) ratio, while a 1:1 ratio was used for the ΔrmpD mutant (Supplementary Figure 8C). Overall, the inoculum density was adjusted to a final concentration of 4×109 CFU/mL in sterile phosphate-buffered saline (PBS). Exact input ratio was confirmed by differential plating. Mice were inoculated by oral gavage with 109 CFU per mouse. At 48 hours post-inoculation, the mice were euthanized and large intestines were collected in sterile PBS, homogenized, and plated on LB agar containing the appropriate antibiotics to enumerate CFU after overnight incubation at 27 °C. The competitive index (CI) was calculated using the formula shown in Eq. 1. The limit of detection for K. pneumoniae detection in these experiments was 200 CFU/mL. For all samples below the limit of detection, we assumed a bacterial concentration of 199 CFU/mL. Statistics Unless otherwise noted, all replicates represent biological replicates, and the experiments were independently replicated at least three times. Statistical analyses were computed in GraphPad Prism 11 (V11.0.2; GraphPad Software, LLC). For experiments comparing more than two groups on two independent variables, significance was calculated using two-way ANOVA with Dunnett’s post-hoc test or Šídák correction. Meanwhile, one-way ANOVA with either Dunnett’s post-hoc test or Šídák correction was applied to calculate significance between more than two groups on one independent variable. Statistical differences between two groups with independent variables were calculated using student’s t-test. In all instances, results were considered statistically significant if the P-value was less than or equal to 0.05. All data presented are the mean, and the error bars represent the standard error of the mean. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Data 2 Supplementary Data 3 Reporting Summary Transparent Peer Review file Source data Source Data Acknowledgements We thank Drs. Matthew Parsek and Xuhui Zheng for the pBBR1 reporter plasmid backbone. We thank members of the Mike Lab (Brooke E. Ryan, Emily L. Kinney, Grace E. Shepard, Zachary J. Resko, Lindsey R. Krzeminski, Krista Pettee, Jolie G. Lagger), Hunter Coyle (Dr. Lee Harrison Lab, University of Pittsburgh), Department of Medical Microbiology and Immunology at the University of Toledo (Drs. R. Mark Wooten, Robert Blumenthal, Jason Huntley, Peter Andreana) and Program in Microbiology and Immunology/Division of Infectious Diseases at the University of Pittsburgh (Drs. William H. DePas, Matthew Culyba, Anthony Richardson and Patricia Grace) for sharing technical resources, and valuable guidance and feedback. Author contributions S.K. and L.A.M. conceived the project. S.K., G.M.G., D.J.S., K.T. and B.A.A. performed the experiments. S.K., G.M.G. and D.J.S. contributed to the construction of mutants and plasmids. S.K. drafted the manuscript with input from L.A.M., J.V., K.T. and D.J.S. L.A.M. acquired the funding, and supervised and guided the project. All authors reviewed, edited and approved the manuscript. Peer review Peer review information Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available. Funding Research reported in this publication was supported by the University of Toledo College of Medicine and Life Sciences, Department of Medicine at the University of Pittsburgh School of Medicine, American Heart Association [23CDA1056712 (L.A.M.) and 24PRE1197026 (S.K.)] and the National Institutes of Health [K22 AI145849 (L.A.M.) and R35 GM150588 (L.A.M.)]. This content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies. Data availability The sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under accession code PRJNA1466957. Clinical hypervirulent isolates used in this study were sequenced in a previously published study and the associated data are available in the NCBI SRA under the following accession numbers: SRR19095630 (Kp4289), SRR19095615 (Kp4585), and SRR18982353 [https://www.ncbi.nlm.nih.gov/sra/?term=SRR18982352] (Kp6557). All other data generated in this study are provided in the Supplementary Information and Source Data file. Source data are provided with this paper. Competing interests Dr. Vornhagen has consulted for Vedanta Biosciences, Inc. No other authors declare a competing interest. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-75867-z. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Data 2 Supplementary Data 3 Reporting Summary Transparent Peer Review file Source Data Data Availability Statement The sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under accession code PRJNA1466957. Clinical hypervirulent isolates used in this study were sequenced in a previously published study and the associated data are available in the NCBI SRA under the following accession numbers: SRR19095630 (Kp4289), SRR19095615 (Kp4585), and SRR18982353 [https://www.ncbi.nlm.nih.gov/sra/?term=SRR18982352] (Kp6557). All other data generated in this study are provided in the Supplementary Information and Source Data file. Source data are provided with this paper.