Salmonella enterica serovar Typhimurium (STm) strains used in this study are shown in Table 1 and are derived from the type strain 14028s (ATCC, Manassas, VA). Bacterial culturing was performed as described (6, 10). Unless otherwise stated, 5 mL cultures of Luria-Bertani (LB) medium were inoculated with a single isolated colony from a freshly streaked LB agar plate and grown overnight (~16 h) at 37°C with aeration (220 rpm) in a MaxQ 4000 benchtop incubator (ThermoFisher Scientific, Waltham, MA). Overnight cultures were subcultured 1:50 in LB and grown to mid-log phase (OD₆₀₀ of ~0.7) prior to experimentation. Culture optical density at 600 nm (OD₆₀₀) was monitored using a GENESYS 10S UV-Visible spectrophotometer (ThermoFisher Scientific). When appropriate, media was supplemented with kanamycin (50 µg/mL), carbenicillin (100 µg/mL), gentamicin (10 µg/mL), and/or 5-bromo-4-chloro-3-indolyl-beta-D-galacto-pyranoside (X-gal; 40 µg/mL). L-arabinose was added to a final concentration of 0.4% to induce P_BAD promoters. Strains carrying temperature-sensitive plasmids were maintained at 30°C (17).

The Sal4 B cell hybridoma cell line secreting monoclonal polymeric IgA was maintained as described (15, 21). Chimeric Sal4 IgG₁ was provided by MappBio, Inc (San Diego, CA) and has been previously characterized in vitro and in vivo (5).

The oligonucleotide primers used to construct plasmids and STm mutant strains are listed in Table S1 . λ Red recombination was utilized to generate strains with the target gene replaced by an antibiotic resistance cassette as previously described (17). Kanamycin and gentamicin resistance cassettes were amplified from pKD13 (17) or pUC18-mini-Tn7T-Gm-lacZ (19), respectively. The resulting PCR products were purified using a DNA Clean & Concentrator-5 kit according to the manufacturer’s instructions (Zymo Research, Irvine, CA). Electrocompetent STm carrying the pKD46 plasmid (SL94) was grown in the presence of 0.4% arabinose (to induce expression of the λ Red recombinase genes encoded by pKD46), washed with 10% glycerol, and transformed with the concentrated PCR product, then recovered in LB for 1 h at 30°C with aeration. Recovered cells were plated onto LB agar containing kanamycin or gentamicin and incubated overnight at 37°C. Recombinants were verified using colony PCR for antibiotic resistance cassette insertion at the correct genomic location (17). The pKD46 plasmid was cured by incubating the recombineered strains for 4 h at 42°C with aeration in the absence of antibiotic selection.

A constitutive lacZ cassette was stably integrated into the STm 14028s chromosome via Tn7 transposition, as described (18). To accomplish this, the pUC18-R6k-mtn7-kanR TN7 recombination vector was first modified to carry the lacZ ORF regulated by the constitutive PA1/04/03 promoter. The lacZ ORF and PA1/04/03 promoter were first amplified from pUC18-mtn7-lacZ-gmR and pUC18-R6k-mtn7-ecfp-gmR (19), respectively, using tailed primers listed in Table S1 . The PCR products were then purified, assembled into pUC18-R6k-mtn7-kanR in-between the SacI and XmaI restriction sites with the NEBuilder HiFi DNA Assembly kit, and subsequently transformed into chemically competent E. coli DH5α λpiR (NEB, Ipswich, MA). The resulting lacZ Tn7-integration plasmid (pGW167) was electrotransformed into a STm 14028s strain harboring pTn7 (SL172) that had been cultured to mid-log-phase in LB supplemented with carbenicillin and 0.4% arabinose (18). Following 2 h recovery at 30°C with aeration, recombinants were then selected on LB agar supplemented with kanamycin (50 µg/mL) and X-gal (40 µg/mL). The Tn7 transposase plasmid (pTn7) was then cured from blue colonies that arose through subsequent culturing in LB at 42°C with aeration in the absence of antibiotic selection.

To generate the yhjH overexpressing plasmid pYhjH, the coding sequence of yhjH from the STm 14028s genome was PCR amplified with yhjH-specific primers containing built-in restriction sites (listed in Table S1 ) and cloned into the XbaI and HindIII sites of a pBAD24 vector (20). E. coli DH5α competent cells were transformed with the resulting yhjH construct (NEB). Transformants were selected on LB agar plates containing ampicillin and the sequence of the selected clone was verified by nucleotide sequencing. To construct a WT pYhjH strain, the plasmid was isolated using a QIAprep Spin Miniprep Kit (QIAGEN, Germantown, MD) and transformed into GGW377 by electroporation. Transformants were selected on LB agar containing carbenicillin.

Overnight cultures of STm were sub-cultured 1:50 in fresh media and grown at 37°C with aeration to mid-log phase (OD₆₀₀ of ~0.7). Cells were then collected via centrifugation (6000 x g) for 4 min and washed with sterile PBS (pH 7.4). This step was repeated for a total of two wash steps. Cultures were adjusted to an OD₆₀₀ of 1.0 and transferred into a borosilicate glass tube (16 x 25 mm), resulting in ~2.0 x 10⁸ CFU/mL in 5 mL total volume, prior to Sal4 treatment. For mixture experiments, the indicated strains were prepared as described above, combined in a 1:1 ratio, and gently inverted to thoroughly mix prior to antibody treatment. Sal4-mediated agglutination was recorded using an iPhone 6s (Apple, Cupertino, CA) with the ‘TimestampCamera’ application. Timelapse videos were annotated using Canva (Sydney, Australia). To obtain colony forming units (CFUs), 150 µL was taken from the very top of the culture tube at the air/liquid interface, serial diluted five-fold in sterile PBS pH 7.4, and 100 µL from two consecutive dilutions was plated on LB agar and spread using glass plating beads. Plates were incubated at 37°C overnight in a Heratherm IMH60 incubator (ThermoFisher Scientific) overnight and counted the following day using an eCount Colony Counter (Heathrow Scientific, Vernon Hills, IL). CFU counts from the dilution plates were averaged to calculate a CFU/mL value for each tested condition. Plates with >300 CFUs or <30 CFUs (per 100 μL) were considered too numerous or too few to count, respectively, and were excluded from the final data set.

Bacterial cultures were prepared as described above and the OD₆₀₀ value of the washed cells was standardized to a value of 1.0 before addition of 100 µL per well to Immulon 4HBX plates (ThermoFisher Scientific). The plate was covered with a plastic lid and incubated at 4°C overnight (~18 h). The next morning, 200 µL of blocking solution (2% goat’s serum in PBS containing Tween-20 [0.1% v/v]) was added to each well and the plate was incubated on a plate rocker (VWR, Radnor, PA) for 2 h at room temperature. The plate was washed three times with PBS-T prior to addition of Sal4 antibodies diluted in blocking solution. The plate was incubated for 1 h at room temperature on a plate rocker and washed again three times. Goat-anti-mouse IgA-HRP secondary antibody (Sigma-Aldrich, St. Louis, MO) was diluted in blocking solution at 1:2000 and 100 µL was added to each well. The plate was incubated for one h at room temperature on a plate rocker and then washed three times prior to development with SureBlue TMB 1-Component Microwell Peroxidase Substrate (SeraCare, Milford, MA). The peroxidase reaction was stopped using 1M phosphoric acid and the absorbance at 450 nm (Abs₄₅₀) was read by a SpectraMax iD3 plate reader (Molecular Devices, San Jose, CA).

Motility assays were performed essentially as described (21). Liquid LB media with and without antibodies was combined with an equal volume of liquified 0.6% LB agar, poured into a 100 x 15 mm square grid petri dish (ThermoFisher Scientific), and allowed to solidify at room temperature. Agar plates were stab inoculated with 1 µL of an STm overnight culture and then incubated at 37°C until bacterial growth approached the border of a 2x2 square grid. Plates were imaged using a Gel Doc XR Gel Documentation System (Bio-Rad, Hercules, CA) and the bacterial migration diameter was measured using Fiji software version 2.9.0 (22).

STm colonies from freshly streaked agar plates were used to inoculate individual wells of a 96-well microtiter plate each containing 200 µL LB media (CELLTREAT Scientific Products, Pepperell, MA). Plates were incubated for 3 h at 37°C and 220 rpm, then 3 µL from each well was spotted onto a nitrocellulose membrane (Bio-Rad) and allowed to dry for at least 30 minutes in a fume hood at room temperature. The membrane was then submerged in blocking solution and incubated on a plate rocker overnight at 4°C. The membrane was washed 3 times in 0.1% PBS-T for 10 min prior to addition of 10 µg/mL Sal4 diluted in blocking solution. The membrane was incubated for 90 min at room temperature on a plate rocker and washed again three times in PBS-T. The membrane was incubated with goat-anti-mouse IgA-HRP secondary antibody (Sigma-Aldrich) diluted in blocking solution at 1:2000 for 1 h and then washed five times in PBS-T. Finally, SureBlue TMB 1-Component Microwell Peroxidase Substrate (SeraCare, Milford, MA) was applied to the membrane to detect Sal4 binding. The membrane was then imaged using a Gel Doc XR Gel Documentation System (Bio-Rad).

Statistical analysis was performed using GraphPad Prism 9 software (San Diego, CA). The schematic from Figure 1 was designed using BioRender.com (Toronto, ON, Canada).

To examine the interaction between STm and Sal4, we developed a method to visualize and quantitate classical IgA-mediated agglutination of STm in real-time, which we refer to as the “snow globe” assay ( Figure 1 ). In this assay, STm is grown to mid log phase, harvested, washed, and then resuspended in PBS to reach a final optical density (at 600 nm) of 1.0. Cultures are then transferred to a borosilicate glass tube, treated with Sal4 IgA at 15 µg/mL, and incubated at room temperature for 2 h. Bacterial aggregate formation is then captured with timelapse videography.

In the absence of Sal4 IgA, wild type (WT) STm remained in suspension for the entire 2 h experimental period ( Figure 2A ; Movie S1 ). The cell suspension was uniformly turbid and free of visible sedimentation at the bottom of the culture tubes. Conversely, the Sal4 IgA-treated cultures appeared to flocculate (“snow”) and aggregate at the bottom of the culture tube within 30 min. Agglutination appeared to occur beginning at the top of the air-liquid interface, with the upper phase (~1-2 cm) of the liquid culture becoming notably more transparent as compared to the untreated control over time. To investigate whether the visible clearing was associated with reduction in colony forming units (CFUs), we removed 150 µL from just below the air-liquid interface of the control and Sal4 IgA-treated cultures and plated on LB agar. In the control cultures, the number of CFU/mL remained relatively constant over the 2 h period, demonstrating that STm does not auto-aggregate or passively settle due to gravity in the absence of Sal4. In contrast, there was a time-dependent reduction in CFUs/mL for the Sal4 IgA-treated cultures that achieved statistical significance within 1 h ( Figure 2B ). At 2 h, there was a ~10-fold reduction in total CFU/mL recovered from the air-liquid interface of cultures treated with Sal4 IgA relative to the control cultures. Nevertheless, ~10⁷ CFUs/mL remained in solution, indicating that either STm cells were not sufficiently coated with Sal4 IgA to promote agglutination of the entire population, or that the cell density dipped (as a result of agglutination) below a threshold required for classical agglutination to effectively occur. Theoretically, classical agglutination is a self-limiting process in situations where aggregates are cleared through settling (as in the snow globe assay) or removed via peristalsis (as would be expected to occur in the gut).

We next titrated Sal4 to determine the effect of antibody concentration on the dynamics of STm agglutination. In the snow globe assay, 7.5, 15, and 30 µg/mL of Sal4 IgA each caused a significant reduction in recovered CFU/mL relative to the untreated control ( Figure 2C ; Movie S2 ). These results demonstrate that STm agglutination by Sal4 IgA is a dose-dependent process in the snow globe assay, consistent with previous reports from our lab (6). To determine the impact of antibody isotype on STm agglutination in the snow globe assay, we utilized a chimeric human IgG₁ derivative of Sal4 that has been previously evaluated in vitro and in vivo for effects on STm invasion and agglutination (5, 6). We observed that Sal4 IgG was as effective as Sal4 IgA at reducing CFUs relative in culture supernatants, compared to the untreated control at 2 h ( Figure 2D ). In addition, Sal4 IgG and IgA displayed identical STm flocculation kinetics ( Movie S3 ). However, the resulting bacterial sediments were visually distinct; Sal4 IgA-treated aggregates were compact and tightly settled at the bottom of the culture tube, whereas the Sal4 IgG-treated aggregates were less dense and distributed loosely along the walls of the tube ( Figure 2E ). This observation is consistent with previous light microscopic observations that Sal4 IgA induces more densely packed and larger diameter STm aggregates compared to equal concentrations of Sal4 IgG (6). Thus, while Sal4 IgG was as effective as Sal4 IgA at reducing the number STm in suspension, the nature of the aggregates is different between isotypes.

Finally, we confirmed that Sal4 IgA-mediated agglutination of STm in the snow globe assay is dependent on the presence of the O5 antigen. We and others have reported that Sal4 IgA fails to recognize STm strains with deletions in oafA, which encodes the O-acetyl transferase that confers the O5 serotype through acetylation of the 2-hydroxyl group of the abequose residue (5, 14, 23, 24). Indeed, we confirmed that Sal4 IgA does not recognize an STm oafA mutant via dot blot and ELISA ( Figures S1 , S3A ). In the snow globe assay, the ΔoafA mutant remained in solution and was free of visible flocculation following 2 h of treatment with Sal4 IgA ( Figure 3 ; Movie S4 ). Collectively, our results demonstrate that the snow globe assay is a reproducible and quantitative method to monitor STm agglutination in real time.

We have previously noted that the process of agglutination of STm by Sal4 IgA is reminiscent of the early stages of biofilm formation (6, 10, 11). Bacterial biofilm development involves a complex interplay between flagellar motility, second messenger c-di-GMP signaling, and cell-cell adhesion (25). To explore these relationships in the snow globe assay, we tested the ability of Sal4 IgA to agglutinate three different motility-deficient STm mutants: ΔmotB, lacking a component of the flagellar stator complex; ΔflhC, absent of the master transcriptional activator of the flagella operons; and ΔfljBΔfliC, devoid of both biphasic flagellar filaments (26–29). We confirmed that the three mutants were indeed non-motile in a soft agar motility assay ( Figure S2 ). Moreover, all three strains reacted with Sal4 IgA by dot blot analysis, confirming that they express the O5 antigen ( Figure S3A ).

The motility mutants were then evaluated in the snow globe assay in the presence and absence of Sal4 IgA. Over the course of the 2 h experiment, the ΔmotB and ΔflhC mutants remained in solution following Sal4 IgA treatment, with the antibody-treated and untreated cultures indistinguishable from each other ( Figure 4A ; Movie S5 ). The STm ΔfljBΔfliC mutant also remained in solution following Sal4 IgA-treatment, although we noted that this culture was somewhat clear at the air-liquid interface and there was a corresponding slight (albeit non-significant) reduction in CFUs recovered relative to the untreated control ( Figure 4B ). None of the untreated cultures displayed any agglutination over the course of 2 h, indicating that these three mutant strains do not passively settle out of solution, regardless of flagellar expression or function (or lack thereof). On the other hand, overnight incubation of these three strains with Sal4 IgA did eventually result in agglutination and pellet formation at the bottom of the culture tube (data not shown), suggesting that the non-motile mutants are not fully resistant to the effects of Sal4 IgA. Rather, we postulate that flagellar-based motility is required for the rapid (<2 h) onset of antibody-mediated agglutination, possibly by increasing the frequency of cell-cell collisions that would nucleate the agglutination process. To test this, we incubated the non-motile ΔmotB and ΔfljBΔfliC mutant strains both statically and with gentle agitation (200 rpm) in the presence and absence of Sal4 IgA for 2 h. We observed that both non-motile mutants moderately agglutinated when combined with Sal4 treatment and agitation, exclusively ( Figure S4 ). Taken together, these results indicate that bacterial cell collisions drive antibody-mediated agglutination at high cell density.

We therefore predicted that a non-motile mutant would be prone to agglutination in the presence of a nucleating factor, such as motile WT cells. To test this possibility, we performed a modified snow globe assay in which the non-motile ΔfljBΔfliC mutant (lacZ ^-) was mixed 1:1 with an WT strain of STm (lacZ ⁺) that constitutively expresses β-galactosidase. The formation of bacterial aggregates was monitored by videography and the proportion of each strain present at the top of the culture was determined by plating CFUs on LB agar containing X-gal. To validate this modified assay, we first compared a lacZ ^- WT strain mixed with the lacZ ⁺ WT strain. The WT : WT mixture behaved as expected, in that visible agglutination occurred following Sal4 IgA treatment and the ratio of the two strains (lacZ ^-; lacZ ⁺) in the supernatants after 2 h was roughly 50:50 ( Figures 4C, D ; Movie S6 ). We then examined a 1:1 mixture of STm ΔfljBΔfliC (lacZ ^-) with WT STm (lacZ ⁺) and observed that the mixed culture agglutinated following Sal4 IgA treatment and there was a statistically significant reduction in CFUs in the culture supernatants relative to the untreated mixed control. Blue-white screening of the resulting colonies indicated that the WT and STm ΔfljBΔfliC strains were in equal proportion ( Figures 4C, D ), revealing that a non-motile mutant is effectively agglutinated by Sal4 IgA in the presence of a motile counterpart.

We envisioned two models to explain Sal4 IgA-mediated agglutination of STm ΔfljBΔfliC in the presence of WT STm. The first is “bystander catch”, in which WT cells that aggregate following Sal4 IgA exposure entrap non-motile bystander cells in close proximity. We previously reported that STm produces an exopolysaccharide (EPS)-like substance in response to Sal4 IgA treatment, and this could enhance cell-cell adhesion in addition to antibody cross-linking (10, 11, 30). Alternatively, agglutination is driven by the forceful collision of two cells that share Sal4 IgA’s epitope, thereby enabling antibody-mediated intercellular crosslinking or cell-cell bridging to occur, as proposed by Hoces and colleagues (1). To distinguish between these two models, we performed a modified snow globe assay in which the wild type STm (lacZ ⁺) was mixed 1:1 with a STm ΔoafA (lacZ ^-) mutant. In this experiment, the WT:ΔoafA mixture exhibited a partial (or mild) agglutination phenotype and resulted in a markedly less thick/dense sediment at the bottom of the culture tube, as compared to the WT : WT and WT:ΔfljBΔfliC mixtures ( Movie S6 ). When we examined CFUs, there was no significant reduction in cell numbers from the culture supernatants when compared to the untreated condition, confirming that agglutination of the WT strain was markedly attenuated in the presence of the STm ΔoafA (lacZ ^-) strain ( Figure 4C ). This was further supported by the observation that the STm ΔoafA (lacZ ^-) mutant outnumbered the WT strain in the cell suspension by ~4 fold with Sal4 treatment ( Figure 4D ). While these results favor the cell collision model noted above, they also suggest that bystander catch does indeed occur at high cell densities.

We have previously reported that Sal4 IgA triggers STm EPS production and biofilm formation through activation of one or more diguanylate cyclases (DGC) that modulate c-di-GMP signaling in STm (10). Considering the link between c-di-GMP signaling, motility, and biofilm formation, we hypothesized that Sal4 IgA-mediated agglutination in the snow globe assay may be influenced by DGCs previously implicated in motility arrest, including STM14_2408, STM14_3275, and STM14_5467 (referred to hereafter as their LT2 annotations STM1987, STM2672, and STM4551, respectively) (31, 32). To test this hypothesis, we engineered STm strains carrying deletions in STM1987, STM2672, and STM4551, as described in the Materials and Methods, and confirmed reactivity with Sal4 IgA by dot blot ( Figure S3B ). In the snow globe assay, all three mutants were each readily agglutinated by Sal4 IgA (15 μg/mL) with kinetics and changes in CFUs that were virtually identical to that of the WT strain ( Figures 5A, B ; Movie S7 ). Thus, the absence of these DGCs previously implicated in motility arrest does not impact Sal4 IgA-mediated agglutination of STm under the conditions of the snow globe assay.

As the c-di-GMP pool in STm is regulated by a network of multiple DGCs, phosphodiesterases (PDEs), and proteins with both catalytic functions, it is possible that a defect caused by knocking out a single DGC is masked by the activity of the other enzymes. With this in mind, we hypothesized that overexpression of a DGC or PDE might influence Sal4-mediated agglutination of STm. To test this, we utilized previously constructed expression plasmids that encode the DGC yeaJ gene or the PDE gene yhjH under control of an arabinose-inducible promoter (10). The STm strains carrying these plasmids were confirmed to be reactive with Sal4 IgA via dot blot ( Figure S3C ). In the snow globe assay, the strains overexpressing YeaJ or YhjH were agglutinated by Sal4 in a manner that was indistinguishable from the WT strain carrying the empty vector ( Figures 5C, D ; Movie S8 ). Our results suggest that cyclic-di-GMP signaling does not play a direct role in antibody-mediated agglutination at high cell densities under these experimental conditions.