Adherence Enables Neisseria gonorrhoeae to Overcome Zinc Limitation Imposed by Nutritional Immunity Proteins

ABSTRACT

Neisseria gonorrhoeae (Gc) must overcome the limitation of metals such as zinc to colonize mucosal surfaces in its obligate human host. While the zinc-binding nutritional immunity proteins calprotectin (S100A8/A9) and psoriasin (S100A7) are abundant in human cervicovaginal lavage fluid, Gc possesses TonB-dependent transporters TdfH and TdfJ that bind and extract zinc from the human version of these proteins, respectively. Here we investigated the contribution of zinc acquisition to Gc infection of epithelial cells of the female genital tract. We found that TdfH and TdfJ were dispensable for survival of strain FA1090 Gc that was associated with Ect1 human immortalized epithelial cells, when zinc was limited by calprotectin and psoriasin. In contrast, suspension-grown bacteria declined in viability under the same conditions. Exposure to murine calprotectin, which Gc cannot use as a zinc source, similarly reduced survival of suspension-grown Gc, but not Ect1-associated Gc. We ruled out epithelial cells as a contributor to the enhanced growth of cell-associated Gc under zinc limitation. Instead, we found that attachment to glass was sufficient to enhance bacterial growth when zinc was sequestered. We compared the transcriptional profiles of WT Gc adherent to glass coverslips or in suspension, when zinc was sequestered with murine calprotectin or provided in excess, from which we identified open reading frames that were increased by zinc sequestration in adherent Gc. One of these, ZnuA, was necessary but not sufficient for survival of Gc under zinc-limiting conditions. These results show that adherence protects Gc from zinc-dependent growth restriction by host nutritional immunity proteins.

INTRODUCTION

Neisseria gonorrhoeae (Gc) is the causative agent of the common sexually transmitted infection gonorrhea and a pathogen of global concern, causing an estimated 87 million cases of disease worldwide per year (1). Increasing rates of antibiotic resistance have led to limited therapeutic options, culminating in only one recommended empirical treatment: the third-generation cephalosporin, ceftriaxone (2). Furthermore, previous infection is not protective against future disease, and there is no protective vaccine (2). In men, urogenital gonorrhea can cause urethritis and a characteristic purulent discharge, but can also lead to ascending infection, including the risk of infertility due to epididymitis (3). Gonorrhea in women has the potential to cause lasting damage, especially because the initial symptoms of cervical infection are frequently nonspecific such that individuals are often unaware that they are infected (2). When bacteria ascend from the cervix to the upper reproductive tract, inflammation can result in chronic pelvic pain (pelvic inflammatory disease) and fallopian tube scarring that may lead to infertility or ectopic pregnancy (4). It is therefore especially important to understand the ways in which Gc evade the initial immune response to establish infection at the cervix.

One important host innate immune defense mechanism is nutritional immunity, by which the host sequesters nutrients from invading microbes. For transition metals such as zinc and iron, which are essential for almost all organisms (5, 6), mammals have evolved to produce proteins that bind transition metal ions with high affinity and prevent their acquisition by pathogens (7). For example, the S100 family proteins calprotectin (S100A8/A9) and psoriasin (S100A7) have the ability to bind zinc with high affinity in a calcium-dependent manner (7). Calprotectin is a 25 kDa heterodimer of S100A8 and S100A9, and the dimer possesses two asymmetric zinc binding sites. Site I can bind zinc, copper, or manganese with high affinity, while site II can only bind zinc or copper (8). Calprotectin has been shown to restrict the growth of some bacterial and fungal pathogens, in most cases through zinc or manganese sequestration (9–12). The related S100 protein psoriasin is a 22 kDa homodimer of S100A7. It possesses two identical sites that have been shown to bind zinc, but not iron, manganese, nickel, or copper (13). Psoriasin has also been shown to restrict the growth of some bacterial and fungal pathogens, potentially through zinc sequestration (13–15).

Neisseria species have the remarkable ability to overcome nutritional immunity by expressing TonB-dependent transporters (TdTs) that directly bind to and extract transition metals from human metal-binding proteins, ultimately transporting the metal into the cytoplasm for incorporation into biomolecules (16, 17). The prototype TdT is TbpA, which binds and extracts iron from human transferrin (18). The importance of this system to gonococcal pathogenesis is reflected by the fact that bacteria lacking TbpA and its lipoprotein partner TbpB fail to establish urethral infection in men experimentally inoculated with N. gonorrhoeae (19). Because they are surface-expressed, not phase-variable, and important for infection, TdTs have been proposed as potential targets for a vaccine against gonorrhea (20, 21). Other characterized TdTs include LbpA, which binds to human lactoferrin to acquire iron (22), HpuB, which acquires heme-iron from hemoglobin (23), and FetA, which acquires iron from xenosiderophores such as enterobactin (24). More recently, we have shown that TdfH and TdfJ acquire zinc from the human proteins calprotectin and psoriasin, respectively (25, 26). The remaining TdTs, TdfF and TdfG, are iron-regulated, but their metal acquisition phenotypes remain uncharacterized (21).

We have shown that TdfH binds human calprotectin, while TdfJ binds human psoriasin. Both proteins extract and internalize the zinc using a process that is dependent on TonB (25–27). Gc is able to replicate in chemically-defined medium where the sole zinc source is zinc-loaded calprotectin in a TdfH-dependent manner, and with zinc-loaded psoriasin in a TdfJ-dependent manner. TdfH and TdfJ are expressed by Gc during infection of symptomatic women, and TdfJ is more highly expressed in Gc infecting women than men (28, 29). Neutrophils are both the first responders to Gc infection and a major source of calprotectin in vivo (30, 31). We showed that ΔtdfH Gc had reduced survival in neutrophil extracellular traps, which contain abundant calprotectin (25).

In conditions of zinc abundance, zinc ions likely freely diffuse through Gc outer membrane porins (32, 33) and are captured in the periplasm by binding to the zinc shuttle protein ZnuA (34). ZnuA then delivers zinc to ZnuB and ZnuC, which transport the ion into the cytoplasm (34). Zinc acquisition is regulated in part by the zinc uptake repressor, Zur, a zinc-responsive transcriptional repressor (35) (initially called PerR by Wu et al. [36]). In conditions of high cytoplasmic zinc concentration, Zur binds to zinc. This increases the affinity of Zur for a palindromic DNA motif called the Zur box, which is found in the promoter region of Zur-regulated genes, thereby repressing their transcription (35). Conversely, when cytoplasmic zinc is low, zinc ions dissociate from Zur, which loses its affinity to DNA, leading to de-repression of gene expression. Genes that are subject to Zur repression include those involved in zinc uptake such as TdfH, TdfJ, and the ZnuABC operon, as well as genes required to cope with low-zinc conditions, such as zinc-independent ribosomal proteins (36).

Colonization of the cervical epithelium is an important step in both asymptomatic infection by Gc and symptomatic gonorrhea in women (37). Gc forms microcolonies on the cervical epithelium in a pilus-dependent manner (38), and subsequently invades the transition zone and endocervical epithelium in a manner that is influenced by expression of Opa adhesion proteins (39). Colonization puts the bacteria in close proximity to human cells and is anticipated to change the physiology of the bacteria (40). Calprotectin and psoriasin are found in the ectocervix, as assessed by histology and transcript abundance (15, 41–44). Moreover, calprotectin and psoriasin are among the most abundant proteins detected in the cervical secretions of healthy women, and increase in abundance during inflammation (45–47). The effect of zinc sequestration by calprotectin and psoriasin on the ability of Gc to colonize human cervical epithelium is unknown.

In this work, we tested the hypothesis that Gc requires TdfH and TdfJ to acquire zinc and effectively colonize the surface of Ect1 human immortalized ectocervical cells (48) in the presence of the zinc-sequestering proteins calprotectin and psoriasin. Gc lacking tdfH and tdfJ had a > 3 log growth defect in medium containing calprotectin and psoriasin, which was rescued with excess zinc. To our surprise, ΔtdfH ΔtdfJ Gc that was adherent to Ect1 cells in the same medium did not exhibit this zinc- and S100 protein-dependent reduction in viability. Similarly, wild-type Gc that were adherent to Ect1 cells had significantly better survival over time in a medium containing murine calprotectin, which Gc cannot use as a zinc source, compared with bacteria grown planktonically (27). We ruled out the possibility that Ect1 cells were providing zinc to adherent Gc. Instead, we found that attachment of Gc to abiotic surfaces was sufficient to protect Gc from this zinc sequestration. RNAseq analysis revealed a unique cohort of genes that are statistically significantly differentially expressed by bacteria that are both attached and zinc-limited. These findings indicate that Gc in an adherent state is better adapted to conditions in which zinc is sequestered, enhancing bacterial survival when confronted with nutritional immunity.

RESULTS

TdfH and TdfJ are required for Gc to grow in media in which zinc is sequestered by human calprotectin and psoriasin.

We previously showed that Gc uses TdfH to grow in a chemically defined medium where zinc-loaded calprotectin is the only source of zinc, and uses TdfJ for growth with zinc-loaded psoriasin (25–27). Here, we sought to assess how TdfH and TdfJ contributed to the ability of Gc to overcome nutritional immunity when sufficient zinc for growth is sequestered by calprotectin and psoriasin. These studies used keratinocyte serum-free medium (KSFM), which contains 1.98 μM zinc as measured by ICP-MS in the presence of growth factors, or 0.48 μM zinc without growth factors (Table S1 in the supplemental material). Complete KSFM containing growth factors was left untreated, or was supplemented with 1.4 μM (each) recombinant human apo-calprotectin or apo-psoriasin to sequester zinc, with or without an additional 3 μM ZnSO₄ to overcome zinc restriction by these proteins (calprotectin and psoriasin each contain two transition metal binding sites). Calprotectin and psoriasin at 1.4 μM were experimentally determined to be sufficient to achieve maximal growth restriction of ΔtdfH and ΔtdfJ Gc of strain FA1090 in KSFM containing growth factors, respectively (Fig. S1).

Wild-type (WT) Gc grew over time in KSFM, with CFU increasing ∼60-fold over the 10-h time course (Fig. 1A). As expected, WT Gc grew similarly in KSFM, KSFM with added calprotectin or psoriasin, and KSFM with excess zinc, with or without calprotectin or psoriasin (Fig. 1A). These results indicate (i) the concentration of zinc in KSFM is sufficient to support WT Gc growth over time, (ii) WT Gc can acquire metals from human nutritional immunity proteins in KSFM, and (iii) the concentration of zinc added was not toxic for Gc. ΔtdfH, ΔtdfJ, and ΔtdfH ΔtdfJ Gc grew identically to WT in KSFM without psoriasin or calprotectin or with excess zinc (Fig. 1B to D). The CFU of ΔtdfH Gc decreased ∼10-fold relative to the inoculum after 10 h in KSFM containing calprotectin, but grew similarly to WT bacteria in medium without calprotectin or in KSFM containing the TdfJ ligand psoriasin (Fig. 1B). Compared with conditions where the tdfH mutant could acquire zinc, ΔtdfH Gc in calprotectin-containing KSFM had a significant decrease in growth at 6, 8, and 10 h of incubation, with an ∼2.8 log difference between these conditions at 10 h (Fig. 1B). Growth of ΔtdfH Gc in the presence of calprotectin was rescued when excess zinc was added (Fig. 1B). Likewise, CFU of ΔtdfJ Gc significantly decreased over time in KSFM containing psoriasin, but not calprotectin or untreated medium, and psoriasin-restricted growth of ΔtdfJ Gc was rescued by adding excess zinc (Fig. 1C). Moreover, compared with untreated KSFM, CFU of double mutant ΔtdfH ΔtdfJ Gc significantly declined in the presence of calprotectin or psoriasin, and was rescued in each case with the addition of excess zinc (Fig. 1D). These results are consistent with the hypothesis that Gc uses TdfH to overcome zinc sequestration by calprotectin, and TdfJ to acquire zinc sequestered by psoriasin. Growth of ΔtdfH Gc was not impaired in the presence of psoriasin, and ΔtdfJ Gc was not impaired in the presence of calprotectin, which supports the concept that each TdT-ligand interaction is unique and specific. Calprotectin is known to bind metals other than zinc, specifically manganese and copper (49–51). However, the reduction in ΔtdfH ΔtdfJ Gc CFU in KSFM containing calprotectin and psoriasin was only rescued with the addition of zinc, not manganese or copper (Fig. 1E).

FIG 1.

TdfH and TdfJ are required for Gc to grow in media in which zinc is sequestered by human calprotectin and psoriasin. (A–D) Gc of the indicated genotype was inoculated in KSFM with growth factors (KSFM) alone (gray), or containing 1.4 μM human calprotectin (Cp; orange) or 1.4 μM human psoriasin (Ps; blue), with or without 3 μM ZnSO₄ (Zn) (solid lines, no Zn; dotted lines, +Zn). Genotypes are A, WT; B, ΔtdfH; C, ΔtdfJ; D, ΔtdfH ΔtdfJ. Data are presented as the log₁₀ transformation of CFU recovered at the indicated time points. Error bars represent standard deviation. Asterisks indicate adjusted P values from an ordinary two-way ANOVA with Tukey’s multiple-comparison test, comparing the KSFM condition versus +Cp (blue *) and/or +Ps (orange *). (E) ΔtdfH ΔtdfJ Gc was inoculated in KSFM with growth factors containing 1.4 μM Cp and 1.4 μM Ps (+Cp +Ps), alone or with 6 μM ZnSO₄ (+Zn), 6 μM MnCl₂ (+Mn), or 6 μM CuCl₂ (+Cu). Dotted line indicates the concentration of bacteria in the inoculum. CFU were enumerated 8 h later. Each color indicates one biological replicate. Asterisks indicate adjusted P values from a one-way ANOVA with Šídák’s multiple-comparison test. Data are from the following number of biological replicates: A, 3-5; B, 4-5; C, 4-7; D, 4-5; E, 3. A-E: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Taken together, these data show that Gc cultured in liquid medium depends upon TdfH and TdfJ to overcome calprotectin and psoriasin-mediated sequestration of zinc, respectively.

ΔtdfH ΔtdfJ Gc in contact with Ect1 cells is protected from calprotectin- and psoriasin-mediated zinc sequestration.

We next tested the hypothesis that Gc infection of lower female genital tract epithelial cells relies upon TdfH and TdfJ to acquire zinc from calprotectin and psoriasin. To do so, the immortalized ectocervical cell line Ect1 was infected with Gc (48). Ect1 cells were selected because Gc has been shown to adhere to the apical aspect of Ect1 cells and grow in association with them (52), and cells of the human ectocervix, but not endocervix, express calprotectin and psoriasin (15, 41, 43). However, Ect1 cells did not produce or secrete calprotectin or psoriasin to levels that would sequester the concentration of zinc that we measured in KSFM (Fig. S2), and the concentration of these proteins released by Ect1 cells into conditioned medium was 1.5–2.5 logs lower than what has been reported in human cervicovaginal lavage fluid (15, 53). Addition of Gc or the proinflammatory cytokine IL-1α, which has been reported to increase production of calprotectin and psoriasin (54, 55), did not increase secreted levels of these proteins (Fig. S2).

To model the presence of calprotectin and psoriasin in human cervicovaginal secretions, we supplemented KSFM with calprotectin and psoriasin as in the in vitro growth assays above. Ect1 cells were infected with WT or ΔtdfH ΔtdfJ Gc in this medium for 2 h. Cells were washed to remove unbound bacteria, and CFU were enumerated from saponin lysates. There was no difference in the level of association between WT and ΔtdfH ΔtdfJ Gc after 2 h (Fig. S3A). To examine if the Gc was internalized by Ect1 cells, Gc were added to Ect1 cells for 6 h, then lysed to enumerate cell-associated CFU (adherent) or treated with gentamicin to kill extracellular Gc for an additional hour. At this time, on average 0.25% of cell-associated Gc were gentamicin-resistant (invaded) (Fig. S3B, S3C), and there was no difference in the number of adherent or invaded CFU between WT and ΔtdfH ΔtdfJ Gc (Fig. S3B, S3C). Thus, Gc colonizes but does not efficiently invade Ect1 cells; for this reason, cell-associated CFU will be referred to as “adherent.”

To examine how TdfH and TdfJ affected the ability of Gc to infect epithelial cells under zinc-limited conditions, Ect1 cells were infected with WT or ΔtdfH ΔtdfJ Gc in one of three conditions: in KSFM, KSFM with calprotectin and psoriasin, or KSFM with calprotectin, psoriasin, and ZnSO₄, using the conditions established for Gc in suspension in Fig. 1. After 2 h, cells were washed to remove non-adherent bacteria and replaced in medium with the same additives. The adherent Gc at 2 h served as the baseline against which outgrowth and growth decline were compared. After 9 h infection (2 h initial adherence plus 7 h of incubation), the supernatant was collected (detached Gc) and the cells were lysed with saponin (adherent Gc), and CFU were enumerated from both. The sum of detached and adherent bacteria was termed “cell-associated Gc,” since the detached bacteria originated from the adherent bacteria at 2 h. The same Gc cultures were incubated in these media conditions without Ect1 cells, and CFU were enumerated after 9 h.

After 9 h infection, 5-fold fewer cell-associated ΔtdfH ΔtdfJ Gc were recovered from Ect1 cells in the presence of calprotectin and psoriasin compared to infection in KSFM or with excess zinc, a statistically significant decrease (Fig. 2A). However, the numbers of ΔtdfH ΔtdfJ Gc were still 15-fold higher than after 2 h adherence (dotted lines, Fig. 2A). The minimal decline of ΔtdfH ΔtdfJ Gc recovered from Ect1 cells in the presence of calprotectin and psoriasin occurred in both detached (Fig. 2B) and cell-adherent Gc (Fig. 2C). In comparison, the CFU of ΔtdfH ΔtdfJ Gc grown in suspension in KSFM containing calprotectin and psoriasin declined 70-fold compared to the inoculum, indicating overall death in the bacterial population under these zinc-limited conditions (Fig. 2D). Numbers of ΔtdfH ΔtdfJ Gc in KSFM containing calprotectin and psoriasin were 3.8 logs less than in KSFM or in KSFM with excess zinc, a statistically significant decrease (Fig. 2D). WT Gc grew similarly well in all conditions tested, regardless of association with Ect1 cells (Fig. 2E and F). Thus, contrary to expectations, infection of Ect1 cells enhanced ΔtdfH ΔtdfJ Gc survival under conditions where zinc was sequestered by calprotectin and psoriasin.

FIG 2.

ΔtdfH ΔtdfJ Gc in contact with human Ect1 immortalized ectocervical cells are protected from human calprotectin- and psoriasin-mediated zinc sequestration. (A–D) ΔtdfH ΔtdfJ Gc was inoculated into KSFM with growth factors alone, or with 1.4 μM human calprotectin (Cp) and 1.4 μM human psoriasin (Ps) with or without additional 3 μM ZnSO₄ (Zn). For A–C, Gc (2.5 × 10⁴ CFU) was added to confluent Ect1 cells (4 × 10⁵). After 2 h, unbound bacteria were washed away. For one set of wells, cells were lysed with saponin and CFU were enumerated from the lysate (dotted line). The remainder of the wells were replenished with fresh experimental media. After an additional 7 h, the media were collected (detached; B), the cells were lysed with saponin (adherent; C), and CFU were enumerated from both. Cell-associated (A) Gc is the sum of the detached and adherent Gc per experimental replicate. Each color represents an individual biological replicate completed on separate days. (D). The same starting culture of ΔtdfH ΔtdfJ Gc was inoculated into the same batch of media in wells without Ect1 cells, where the bacteria do not attach (media-grown). Each color indicates one biological replicate, and dotted line represents the inoculum for each replicate. (E, F) The same conditions as in A and D were carried out, except using WT Gc. (A-F) Asterisks indicate adjusted P values from a one-way ANOVA with Tukey’s multiple-comparison test; *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

Adherence alone, without live Ect1 cells, is sufficient to protect ΔtdfH ΔtdfJ Gc from zinc sequestration by calprotectin and psoriasin.

We investigated whether Ect1 cells played an active role in contributing to the increased survival of ΔtdfH ΔtdfJ Gc in the presence of psoriasin and calprotectin. To test if Ect1 cells altered the medium in such a way that supported Gc zinc-dependent growth, KSFM containing calprotectin and psoriasin was incubated at 37°C for 7 h on Ect1 cells with or without ΔtdfH ΔtdfJ Gc infection, or without Ect1 cells for comparison. Medium without calprotectin and psoriasin incubated under these conditions was used as a control. Freshly grown ΔtdfH ΔtdfJ Gc was then inoculated into the cell-free media and CFU were enumerated at 7 h. ΔtdfH ΔtdfJ Gc significantly declined in all media conditions containing calprotectin and psoriasin, regardless of exposure to Ect1 cells, and was restored in unmodified media (Fig. 3A). WT Gc grew in all conditions over 7 h (Fig. S4A).

FIG 3.

Protection of Gc from zinc sequestration requires Ect1 cell contact and is not replicated by culturing in Ect1-conditioned medium. (A) KSFM with or without 1.4 μM human calprotectin and 1.4 μM human psoriasin (+Cp +Ps) was incubated in wells without Ect1 cells (control), with confluent Ect1 cells (Ect1), or with confluent Ect1 cells infected with ΔtdfH ΔtdfJ Gc at an MOI of 0.5 (Ect1+Gc). After 7 h at 37°C, 5% CO₂, the media were collected and sterile filtered. ΔtdfH ΔtdfJ Gc (10⁵ CFU/mL) was inoculated into each of the conditioned media (dotted line indicate inoculum), and CFU were enumerated after 7 h. Each color indicates one biological replicate. Asterisks indicate adjusted P values from a one-way ANOVA with Šídák’s multiple-comparison test. (B–D) KSFM with or without 1.4 μM human calprotectin and 1.4 μM human psoriasin (+Cp +Ps) was added to wells containing confluent Ect1 cells (B, C) or no cells (D). 0.4 μm pore Transwell filter inserts were added to each well. ΔtdfH ΔtdfJ Gc (10⁵ CFU/mL) was inoculated into the lower (B) or upper (C, D) chamber of the filter insert. See diagram inserted in each graph that depicts the incubation conditions. For B, wells were washed and replaced with fresh media, using the conditions in Fig. 2 CFU were enumerated 7 h later. Colors correspond to matched individual biological replicates, and dotted lines indicate the CFU present after 2 h (B) or the inoculum (C, D) for each replicate. Asterisks represent P values from unpaired Student’s t tests. (E–G) ΔtdfH ΔtdfJ (10⁵ CFU/mL) was added to Ect1 cells that were alive (F) or killed with 4% paraformaldehyde (E). Cells were washed and replaced with fresh media after 2 h; dotted line indicates adherent CFU at 2 h. The media were KSFM alone, or with 1.4 μM human calprotectin and 1.4 μM human psoriasin (+Cp +Ps) with or without 3 μM ZnSO₄ (+Zn). For G, ΔtdfH ΔtdfJ (10⁵ CFU/mL) was inoculated directly into the experimental media with no Ect1 cells; dotted line indicates the inoculum in the well. CFU were enumerated from cell lysates (E, F) or from the well (G) after 7 h. Colors correspond to matched individual biological replicates. Asterisks represent adjusted P values from one-way ANOVA with Tukey’s multiple-comparison test. (A–G) *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

To further test the effect of Ect1 cells on the calprotectin- and psoriasin-mediated growth restriction of ΔtdfH ΔtdfJ Gc, Ect1 cells were grown in the lower reservoir of tissue culture wells containing 0.4-μm-pore Transwell filter inserts, and ΔtdfH ΔtdfJ Gc was inoculated in the upper reservoir. While direct contact with Ect1 cells protected ΔtdfH ΔtdfJ Gc from declining in CFU (Fig. 3B; compare to Fig. 2A), ΔtdfH ΔtdfJ Gc separated from Ect1 cells by the semi-permeable membrane had a greater than 3-log reduction in CFU in the presence of calprotectin and psoriasin (Fig. 3C). These results were nearly identical to ΔtdfH ΔtdfJ Gc in Transwell filter inserts when no Ect1 cells were present (Fig. 3D). WT Gc was unaffected by the presence of calprotectin or psoriasin in each of these conditions, as expected (Fig. S4B-D). Taken together with the results with conditioned KSFM, these findings indicate that the medium from Ect1 cells is not sufficient to support growth of ΔtdfH ΔtdfJ Gc under zinc-limiting conditions imposed by calprotectin and psoriasin.

Based on these results, we hypothesized that contact with Ect1 cells is important for ΔtdfH ΔtdfJ Gc to overcome zinc limitation by calprotectin and psoriasin. To test whether the epithelial cells played an active role in this process, Ect1 cells were killed with paraformaldehyde and washed before infection with ΔtdfH ΔtdfJ Gc in KSFM. After 2 h of attachment, the cells were washed to remove unbound bacteria as in Fig. 2, and KSFM containing calprotectin and psoriasin, with or without excess zinc, or KSFM alone as a control was added for an additional 7 h. Gc adherent to paraformaldehyde-killed Ect1 cells did not decline in CFU over time in the presence of calprotectin and psoriasin (Fig. 3E), which was similar to what was measured for the bacteria associated with live Ect1 (Fig. 3F). In contrast, Gc in the same medium with calprotectin and psoriasin but without Ect1 cells declined over 2 logs in CFU compared with KSFM that was untreated or contained excess zinc (Fig. 3G). Together with the conditioned medium and Transwell experiments, we conclude that protection of ΔtdfH ΔtdfJ Gc from zinc limitation does not require an active contribution from the epithelial cells.

Adherent wild-type Gc is protected from zinc sequestration mediated by murine calprotectin, and attachment to glass recapitulates the protective effect of Ect1 cell adherence.

The above results led us to propose a new hypothesis: that adherence enhanced the ability of Gc to survive in zinc-limiting conditions. We sought to test this hypothesis in a WT bacterial background in which potential mechanisms could be more easily screened than in ΔtdfH ΔtdfJ Gc with addition of human calprotectin and psoriasin. Previous work from our group showed that TdfH+ Gc can acquire zinc from human but not murine calprotectin (27). Therefore, we measured the effect of adding murine calprotectin on growth of WT Gc. The viability of WT Gc significantly declined over time when grown in suspension in KSFM in the presence of murine calprotectin compared to medium that was unmodified, or contained murine calprotectin and excess zinc, reaching a 4 log difference at 10 h postinoculation (Fig. 4A). The reduction in CFU imposed by murine calprotectin was rescued by the addition of excess zinc, but not manganese or copper (Fig. 4A and B). We then tested whether WT Gc were protected from zinc limitation imposed by murine calprotectin during infection of Ect1 cells. Here, Ect1 cells were infected with WT Gc as described in Fig. 2 (2 h infection, wash, 7 h chase), except that the KSFM contained murine calprotectin with or without excess zinc. WT Gc in the presence of murine calprotectin survived no differently after adherence to Ect1 cells than bacteria without calprotectin, or when excess zinc was added (Fig. 4C), while the same bacteria grown in medium containing murine calprotectin but without Ect1 cells showed an almost 2-log decrease in CFU (Fig. 4D). Therefore, murine calprotectin restricts growth of WT Gc in a manner that phenocopies ΔtdfH ΔtdfJ Gc incubated with human calprotectin and psoriasin, when grown in suspension as well as when adherent to epithelial cells.

FIG 4.

Murine calprotectin restricts the zinc-dependent growth of wild-type Gc in suspension, which is overcome by adherence to Ect1 cells and to glass. (A) WT Gc (10⁶ CFU/mL) was grown in KSFM containing growth factors, alone or with addition of 2 μM murine calprotectin (+mCp) and/or 6 μM ZnSO₄ (+Zn). Data are presented as the log₁₀ transformation of CFU recovered at the indicated time point. Data points represent the mean of three independent experimental replicates. Asterisks represent adjusted P values from a two-way ANOVA with Tukey’s multiple-comparison test. KSFM condition with KSFM +mCp. (B) WT Gc was grown as in A with or without 2 μM murine calprotectin (+mCp), 6 μM ZnSO₄ (+Zn), 6 μM MnCl₂ (+Mn), or 6 μM CuCl₂ (+Cu). Data points indicate the log₁₀ CFU recovered after 8 h. Dotted line, inoculum. Asterisks represent adjusted P values by one-way ANOVA with Šídák’s multiple-comparison test. (C and E) WT Gc (10⁶ CFU/mL) in KSFM was inoculated onto confluent Ect1 cells (C) or acid-washed glass coverslips (E) for 2 h. Dotted lines indicate the CFU adherent after 2 h. Nonadherent Gc was washed away, and Gc was incubated an additional 7 h in experimental media: KSFM with growth factors, alone or with addition of 3 μM murine calprotectin (+mCp) and/or 6 μM ZnSO₄ (+Zn). D, Gc was inoculated into experimental media without adherence and incubated for 7 h. For D, dotted lines indicate inoculum CFU for each well. Each color indicates one biological replicate. Asterisks represent P values from one-way ANOVA with Tukey’s multiple-comparison test. (A–D) ****, P < 0.0001.

To directly test whether the attachment-mediated protection of Gc from zinc limitation required epithelial cells, we measured the survival of WT Gc in the presence of murine calprotectin when the bacteria were adherent to acid-washed glass coverslips (56). Here, WT Gc were allowed to adhere for 2 h to acid-washed glass coverslips, and nonadherent bacteria were washed away. After a further 7 h, detached and adherent CFU were enumerated. WT Gc adherent to glass coverslips grew in the presence of murine calprotectin, and were equivalent in CFU to coverslip-associated bacteria in KSFM that was unmodified, or contained murine calprotectin and excess zinc (Fig. 4E). Zinc concentration in the medium did not increase with the addition of glass coverslips, suggesting that the glass was not a source of free zinc (by ICP-OES: 0.86 μM zinc when incubated in empty wells, 0.88 μM zinc when incubated with acid-washed glass coverslips). ΔtdfH ΔtdfJ Gc also grew when adherent to glass coverslips in the presence of human calprotectin and psoriasin (Fig. S5). Taken together, these results show that Gc attached to surfaces, whether biotic (Ect1 cells) or abiotic (coverslips), do not demonstrate the reduction in viability that is imposed by zinc-binding proteins for bacteria grown in suspension.

Adherent Gc express a distinct pattern of genes under conditions of zinc sequestration.

To understand how attachment enables Gc growth under zinc-limited conditions, we took the unbiased approach of RNA-sequencing (Fig. 5A). RNA was collected from WT Gc that were adherent to glass coverslips (adh) or maintained in suspension (sus), using KSFM containing murine calprotectin (sequestered; seq) or KSFM containing murine calprotectin and excess zinc (excess). Glass coverslips were used instead of Ect1 cells so that there was no human RNA present. Suspension bacteria were inoculated in the experimental media and collected after 4 h. Adherent bacteria were allowed to attach to acid-washed glass coverslips for 2 h in unmodified KSFM. Coverslips were washed once with sterile PBS to remove nonadherent bacteria and the experimental media were added. After 4 h, adherent bacteria were collected and cell pellets were processed to purify RNA and processed for Illumina sequencing (see Materials and Methods). For each experiment, we confirmed that Gc exposed to murine calprotectin exhibited zinc-dependent growth decline over time, but CFU at the time of collection for RNA harvest was equivalent (Fig. S6).

FIG 5.

Zinc- and adherence-dependent transcriptional responses of Gc. (A) Schematic of RNAseq experimental conditions and pairwise comparisons. WT Gc was inoculated into wells containing glass coverslips and allowed to adhere for 2 h (Gc adherent to glass; adh). Nonadherent CFU were washed away, and KSFM containing 3 μM murine calprotectin (zinc sequestered; seq), or 3 μM murine calprotectin with 6 μM ZnSO₄ (zinc excess; excess) was added. Alternatively, WT Gc was inoculated directly into wells without coverslips (Gc in suspension; sus) in the same media with sequestered and excess zinc. After 4 h, bacteria were collected, and RNA was extracted from all four conditions and subjected to RNAseq for each of 3 biological replicates. Comparison A indicates the 549 ORFs differentially expressed in adherent sequestered versus suspension sequestered (blue). Comparison B indicates the 53 ORFs differentially expressed in adherent sequestered versus adherent excess (yellow). Comparison C indicates the 90 ORFs differentially expressed in suspension sequestered versus suspension excess (red). (B) Diagram representing the ORFs that are differentially expressed in the same direction between the pairwise comparisons schematically represented in (A). ORFs that were not differentially expressed, or were differentially expressed in different directions between pairwise comparisons, are not represented.

Over 500 transcripts were differentially abundant between adherent and suspension-grown Gc in the presence of murine calprotectin (Fig. 5B, Data set S1). From these transcripts, we identified the ORFs that were associated with adherence as well as with zinc sequestration (Table 1). The differential expression of these ORFs was examined for three experimental comparisons: the effect of zinc limitation on adherent Gc (adh seq versus adh excess); the effect of zinc limitation on suspension Gc (sus seq versus sus excess); and the effect of adherence in zinc-limited conditions (adh seq versus sus seq). Four ORFs were more abundant in all three comparisons. Three of these ORFs are known to be zinc-repressed and regulated by Zur: the periplasmic zinc shuttle protein ZnuA (NGO_0168), a zinc-regulated hypothetical periplasmic protein (NGO_1049), and a zinc-independent ribosomal subunit RpmE2 (NGO_0930) (35, 36). One gene was reduced in abundance in all of the aforementioned comparisons, AdhP (NGO_1442), which has been previously described as induced under high zinc concentrations and in the presence of Zur (35, 36). Transcripts for an additional six ORFs were higher with adherence under zinc sequestered conditions (adh seq versus sus seq) and with zinc sequestration when adherent (adh seq versus adh excess), but not with zinc sequestration in suspension (sus seq versus sus excess). None of these genes have previously been reported to be zinc regulated (35, 36). Finally, transcripts for 12 ORFs were increased with zinc sequestration when adherent (adh seq versus adh excess) and when in suspension (sus seq versus sus excess), but not significantly changed with adherence under zinc sequestered conditions (adh seq versus sus seq). Three of these ORFs were previously identified as Zur-repressed in Gc (36): tdfJ (NGO_1205), tdfH (NGO_0952), and hmcD (NGO_0166). We had anticipated that all of the genes previously described as Zur-regulated would be differentially abundant in at least one of the zinc sequestered versus excess conditions (36). However, three ORFs previously described as Zur-repressed in Gc were not differentially expressed in sequestered versus excess zinc: znuB (NGO_0169), znuC (NGO_0170), and rpmJ (NGO_0931). Additionally, three ORFs previously described as zinc-repressed in N. meningitidis were also not differentially abundant when zinc was sequestered in these conditions: queC (NGO_0129), queF (NGO_1684), and NGO_1685. However, queF and NGO_1685 were increased with adherence under zinc sequestered conditions (adh seq versus sus seq) (Data set S1). These results indicate that the transcriptional response of Gc to zinc sequestration is distinct between suspension-grown and adherent bacteria, involving genes whose products are important for zinc acquisition as well as differential expression of metabolic genes not previously linked to zinc homeostasis. We next focused on investigating the effect of genes that were differentially expressed in adherent versus suspension conditions when zinc was sequestered, and that were previously known to be involved in zinc acquisition. The primary candidate gene that fit these parameters was ZnuA.

TABLE 1.

Annotated list of Gc open reading frames responsive to zinc and/or adherence as analyzed by RNA-seq

			L2FCd in A	L2FC in B	L2FC in C
Ensembl genea	Symbolb	Zn repressedc	(adh seq vs. sus seq)	(adh seq vs. adh excess)	(sus seq vs. sus excess)	Definitione
Increased in A, B, and C
NGO_0930	RpmE2	(+)	0.93	3.34	2.26	50S ribosomal protein L31
NGO_1049		(+)	0.60	2.33	1.46	hypothetical protein
NGO_0168	ZnuA	(+)	0.50	1.33	0.73	zinc ABC transporter substrate-binding protein
NGO_0217	FbpA		0.81	0.88	0.64	iron ABC transporter substrate-binding protein
Decreased in A, B, and C
NGO_1442	AdhP	(−)	−0.81	−0.78	−1.42	alcohol dehydrogenase
Increased in A and B
NGO_0340	CysK		1.97	0.55	ns	cysteine synthase
NGO_0374	GlnQ		1.47	0.62	ns	amino acid ABC transporter ATP-binding protein
NGO_1318	HemO		0.75	0.74	ns	heme oxygenase
NGO_0554			0.68	1.08	ns	hypothetical protein
NGO_1368	MtrF		1.53	0.52	ns	aminobenzoyl-glutamate transporter
NGO_0639	LldD		0.79	0.72	ns	L-lactate dehydrogenase
Increased in B and C
NGO_1370			ns	2.51	1.68	membrane protein
NGO_1205	TdfJ	(+)	ns	2.03	1.53	TonB-dependent receptor protein
NGO_0166	HmcD	(+)	ns	1.42	1.16	hypothetical protein (possible periplasmic protein)
NGO_1378	ExbB		ns	1.05	0.94	biopolymer transporter ExbB
NGO_1496	TbpB		ns	1.00	0.78	transferrin-binding protein B
NGO_1495	TbpA		ns	0.82	0.63	transferrin-binding protein A
NGO_0322			ns	0.73	1.35	hypothetical protein
NGO_0704	RibB		ns	0.66	0.63	3,4-dihydroxy-2-butanone 4-phosphate synthase
NGO_1029	FumC		ns	0.64	0.72	fumarate hydratase
NGO_0753	NarX		ns	0.60	0.54	two-component system sensor kinase NarX/NarQ
NGO_1688	OmpU		ns	0.56	0.54	hypothetical protein
NGO_0952	TdfH	(+)	ns	0.88	0.49 f	Ton-B dependent receptor protein
Not significantly changed in A, B, or C
NGO_0129	QueC	(+)	ns	ns	ns	7-cyano-7-deazaguanine synthase
NGO_0169	ZnuB	(+)	ns	ns	ns	zinc ABC transporter permease
NGO_0170	ZnuC	(+)	ns	ns	ns	zinc ABC transporter ATP-binding protein
NGO_0931	RpmJ	(+)	ns	ns	ns	50S ribosomal protein L36

^aBy alignment to ASM684v1.51.

^bGene symbols from ASM684v1.51, (98), and literature search.

^cAccording to (36) or (35). (+) indicates Zn or Zur repressed, (–) indicates Zn or Zur induced.

^dL2FC, Log₂ fold change. Positive values: higher in the first than the second condition. Negative values: lower in the first condition. ns, not significantly changed (adjusted p > 0.05 or |L2FC|<0.5).

^e(99) using genome: g2d_mirror/Neisseria_gonorrhoeae_FA_1090/ASM684v1_genomic.

^fTdfH had a L2FC less than the 0.5 cutoff in sus seq versus sus excess.

The periplasmic zinc shuttle protein ZnuA is necessary but not sufficient for Gc to acquire zinc under zinc-limiting conditions.

Transport of zinc into the cytoplasm in Gc is mediated by the tripartite ZnuABC system (34). By RNA-seq, ZnuA was upregulated in both suspension and adherent Gc when zinc was sequestered, and was significantly higher in adherent than suspension zinc-sequestered bacteria. These findings were further investigated by qPCR using RNA harvested from Ect1 cells that were infected for 4 h with WT Gc in the presence of human calprotectin and psoriasin. Transcript abundance of znuA, normalized to 5S rRNA, was significantly greater in adherent relative to media-grown Gc (Fig. 6A). Since ZnuA or ZnuABC has been implicated in other pathogens as a primary means of overcoming calprotectin-mediated zinc sequestration (9, 57, 58), these observations led us to hypothesize that adherent Gc increases expression of ZnuA as a way to overcome zinc limitation.

FIG 6.

Expression of ZnuA is necessary for survival of adherent Gc under zinc-limiting conditions, but overexpression of ZnuA is not sufficient for Gc in suspension to overcome zinc-restricted growth. (A) WT Gc was inoculated into media for 4 h (media grown, blue), or added to Ect1 monolayers for 2 h, washed, and further incubated for 4 h as in Fig. 2 (adherent; orange). The media used were KSFM alone, or with 1.4 μM human calprotectin and 1.4 μM human psoriasin (+Cp + Ps), with or without 3 μM ZnSO₄ (+Zn). RNA was extracted and qRT-PCR was conducted using primers specific for znuA and 5S rRNA. ZnuA expression was calculated as 2^-ΔΔCT, normalized to 5S rRNA, and is expressed relative to KSFM suspension-grown bacteria. Asterisks represent adjusted P values from an ordinary two-way ANOVA with Šídák’s multiple-comparison test. (B–C) ΔtdfH ΔtdfJ Gc with an IPTG-inducible znuA complement (ΔtdfH ΔtdfJ znuA^C) (B) and ΔtdfH ΔtdfJ Gc (C) were inoculated into KSFM alone, or with 1.4 μM human calprotectin and 1.4 μM human psoriasin (+Cp + Ps), 3 μM ZnSO₄ (+Zn), and/or 1 mM IPTG (+IPTG) as indicated, and CFU were enumerated after 9 h. Dotted lines indicate the inoculum of each biological replicate, indicated by a different color; limit of detection (LOD) is indicated by the black dashed line. Asterisks represent adjusted P values from an ordinary one-way ANOVA with Tukey’s multiple-comparison test. (D–E) ΔznuA, ΔtdfH ΔtdfJ, or ΔtdfH ΔtdfJ znuA^C Gc was grown for 18 h on GCB agar plates with the specified additions (TPEN = 10 μM TPEN, IPTG = 1 mM IPTG, Zn = 25 μM ZnSO₄). Lysates of Gc were separated by 4–20% gradient SDS-PAGE, transferred to nitrocellulose, and Western blotted for ZnuA with Zwf as loading control. (D) One representative of three biological replicates. (E) Band intensity for each replicate was quantified, and displayed as the relative intensity (ZnuA/Zwf) for each condition. Asterisks represent P values from ordinary one-way ANOVA with Šídák’s multiple-comparison test. (F–G) ΔtdfH ΔznuA Gc (10⁵ CFU/mL) was added to Ect1 cells in KSFM (F) or inoculated in experimental media into wells without Ect1 cells (G). Experimental media were KSFM alone, or with 1.4 μM human calprotectin (+Cp) with or without 12 μM or 24 μM ZnSO₄ as indicated (Zn). For F, cells were washed and experimental media were added after 2 h; dotted line indicates adherent CFU at 2 h. For G, dotted line indicates the inoculum in the well. CFU were enumerated from cell lysates (F) or from the well (G) after 7 h. Colors correspond to matched individual biological replicates. Asterisks represent adjusted P values from an ordinary one-way ANOVA with Tukey’s multiple-comparison test. For simplicity, only significant comparisons within each condition with or without calprotectin are depicted. (A-G) *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

To determine if upregulation of ZnuA is sufficient to protect Gc from zinc limitation imposed by human calprotectin and psoriasin, an IPTG-inducible znuA complement construct (znuA^C) was introduced into the ΔtdfH ΔtdfJ Gc background (see Materials and Methods). This strain retained its native znuA gene in addition to the inducible construct. ΔtdfH ΔtdfJ znuA^c Gc was grown in suspension in medium containing psoriasin and calprotectin, with or without IPTG to induce znuA expression (Fig. 6B). CFU were enumerated after 8 h. As a control, ΔtdfH ΔtdfJ Gc without the inducible ZnuA construct was grown in the same media conditions (Fig. 6C). Overexpression of znuA did not rescue suspension-grown ΔtdfH ΔtdfJ znuA^C Gc from growth restriction by calprotectin and psoriasin (Fig. 6B). Induction of the complementation construct increased ZnuA protein levels approximately 3-fold in Gc treated with the zinc chelator TPEN (N,N,N’,N’-tetrakis(2pyridinylmethyl)-1,2,ethanediamine) (Fig. 6D, quantified in E). Overexpression of ZnuA in a WT Gc background using the same complement construct did not affect bacterial zinc-dependent growth (Fig. S7). Since overexpression of ZnuA, at least to the levels achievable in this inducible complementation system, is not sufficient to rescue the survival defect of suspension-grown Gc under zinc-limited conditions, this suggests that the upregulation of znuA in adherent Gc does not fully explain bacterial protection from zinc limitation.

Although ZnuA overexpression did not rescue the survival defect of zinc-sequestered Gc in suspension, we sought to determine whether ZnuA was necessary to protect attached Gc under zinc-restricted conditions. To test this, a znuA deletion was introduced into ΔtdfH Gc, and the resulting double mutant was inoculated onto Ect1 cells or into KSFM without adherence, as described in Fig. 2. ΔznuA ΔtdfH Gc declined >3 logs over the inoculum in KSFM without excess zinc, even in the absence of calprotectin (Fig. 6G). This is consistent with previous experience with the znuA mutant, which requires additional zinc to grow even in rich medium (26). Adherence to Ect1 cells did not rescue the growth defect of ΔznuA ΔtdfH Gc and appeared similar to the double mutant in suspension (Fig. 6F and G). Full restoration of ΔznuA ΔtdfH Gc growth required addition of 24 μM ZnSO₄, compared to 6 μM that was used in experiments with bacteria carrying an intact ZnuA (Fig. 2, Fig. 3, Fig. 4). The addition of human calprotectin did not significantly affect the recovered CFU in either suspension-grown or adherent ΔznuA ΔtdfH Gc (Fig. 6G-H). We conclude that while overexpression of ZnuA is not sufficient to rescue suspension-grown Gc from zinc limitation, ZnuA is necessary for Gc growth in any conditions other than highly abundant zinc, whether the bacteria are adherent or in suspension.

DISCUSSION

Gc is able to overcome nutritional immunity by expressing specific transporters that extract transition metals from human metal-sequestering proteins. Zinc is required for Gc growth in vitro, and our previous work showed that TdfH and TdfJ are required for Gc to acquire zinc from calprotectin and psoriasin, respectively (25–27). The goal of this project was to determine the contribution of TdfH and TdfJ to Gc infection of mucosal epithelial cells of the female genital tract, specifically from the ectocervix, where calprotectin and psoriasin are detected (15, 41). To our surprise, we found ΔtdfH ΔtdfJ Gc that was adherent to ectocervical cells was able to grow in the presence of human calprotectin and psoriasin, conditions that restricted the zinc-dependent growth of Gc in suspension. We further determined that this increased survival could be replicated in a reductionist system using WT Gc adherent to glass coverslips in the absence of ectocervical cells, and in the presence of murine calprotectin, which Gc cannot use as a zinc source. Comparative analysis of the transcriptional profiles of adherent and suspension-grown Gc under conditions of low available zinc indicated that adherent bacteria express relatively more of certain genes implicated in the response of Neisseria to low zinc concentrations. These findings reveal a new way in which adherence is critical to Gc pathogenesis, by enhancing bacterial survival when confronted with limited amounts of the essential micronutrient transition metal, zinc.

This study was initiated to examine the role of TdfH and TdfJ in mucosal infection by Gc. Because calprotectin and psoriasin are present in the female genital tract and expressed by ectocervical but not endocervical cells (15, 41, 43), we used the immortalized human ectocervical cell line Ect1 E6/E7 to study the adherence and invasion of Gc. However, Ect1 did not secrete abundant calprotectin and psoriasin; thus, additional psoriasin and calprotectin was added to sequester the zinc in the growth medium, as would be found in female genital secretions. Intriguingly, TdfJ was found to be 4.86-fold more highly expressed in Gc in the female genital tract than in the male urethra (29). This finding may be connected to the observation that male seminal fluid contains approximately 1,000-fold more zinc than the 0.9–1.5 μM zinc reported in female cervical secretions (59–63). The concentrations of calprotectin and psoriasin used in this study to sequester the low μM zinc concentrations in KSFM are in keeping with the estimates of these S100 proteins in female genital secretions. Calprotectin and psoriasin are among the most abundant proteins detected by multiple unbiased proteomics studies on cervicovaginal lavage fluid or cervical mucus from human patients (45–47). One study on undiluted cervical mucus reported a mean calprotectin concentration of 996 nM (53), and a study of cervicovaginal lavage fluid reported mean psoriasin concentrations of 62 nM in 2 mL saline lavage (15). The presence of these proteins extracellularly in the female genital tract mucosa supports our use of exogenous recombinant protein in the absence of sufficient calprotectin and psoriasin production by cultured Ect1 cells, especially since Gc was predominantly extracellular in this Ect1 model.

Zinc-loaded calprotectin and psoriasin can serve as zinc sources for Gc in chemically defined medium, in a TdfH- and TdfJ-dependent manner, respectively (25–27). Here, we found that apo-calprotectin and psoriasin, when added to zinc-containing medium, limited the growth of Gc in a TdfH- and TdfJ-dependent manner, and could be overcome by adding excess zinc. These findings are important because they demonstrate that these S100 proteins can sequester zinc and limit growth of Gc if the bacteria’s dedicated outer membrane transporters are absent. The decline in CFU for Gc lacking TdfH and/or TdfJ under zinc-limiting conditions began to manifest after 4 h of incubation (see Fig. 1 and Fig. S7). While the zinc-storage capacity for Gc has not been described, these findings suggest depletion of Gc zinc reserves takes several hours.

The most surprising finding in this study was that ΔtdfH ΔtdfJ Gc survived and replicated in association with Ect1 cells in medium containing human calprotectin and psoriasin in a zinc-dependent manner, conditions that led to overall death for the same Gc in suspension. Initially we considered that Ect1 cells released zinc (free or bound to proteins) into the medium, or that the cells were internalizing or degrading calprotectin and psoriasin to increase extracellular availability of free zinc. However, by using conditioned media and filter supports to physically separate Gc from Ect1 cells, the mechanism of protection was instead found to be contact-dependent. We then considered that Gc might acquire zinc directly from the cell surface, but this possibility was ruled out when paraformaldehyde-killed and washed Ect1 cells were found to support growth of ΔtdfH ΔtdfJ Gc in zinc-limited conditions. The final piece of data suggesting that Ect1 cells were not necessary for Gc to be protected from zinc limitation was that adherence of Gc to the abiotic surface of acid-washed glass was sufficient to support bacterial enhanced survival under zinc-limited conditions. We conclude that adherence alone is sufficient to enhance survival of Gc under zinc-restricted conditions. We envision at least three nonexclusive ways in which Gc accomplishes this feat: it is more efficient at zinc uptake, it is more efficient at retaining zinc intracellularly, or it physiologically adapts to require less zinc. Findings presented here with znuA as well as other genes differentially expressed by RNA-seq provide preliminary support for each of these possibilities. Future studies that measure zinc uptake and release by Gc are needed to determine how Gc adapts to zinc limitation when adherent to epithelial cells.

We were surprised to find that the protection of adherent Gc from zinc sequestration was not directly related to the ability of TdfH and TdfJ to acquire zinc from human calprotectin and psoriasin. In fact, addition of murine calprotectin, which Gc cannot use as a zinc source (27), restricted the zinc-dependent growth of WT bacteria. The use of murine calprotectin was preferable to the small molecule TPEN (N,N,N’,N’-tetrakis(2pyridinylmethyl)-1,2,ethanediamine), which is commonly used to chelate zinc in biological systems (64). TPEN is membrane-permeable and not completely selective for zinc (65–67). We found that KSFM containing TPEN killed Ect1 cells (not shown), in keeping with reports that TPEN causes cell death and DNA damage in cell culture (67, 68). The ability of adherent Gc to grow in the presence of calprotectin or psoriasin, in the absence of any specific TonB-dependent transporters, may be due to occasional release of zinc from these S100 proteins, which have a relatively lower affinity for zinc compared to that of transferrin for iron (K_d ∼ 10 pM for calprotectin and 400 pM for psoriasin, versus 10⁻²⁰ M for transferrin) (11, 13, 69).

Neisseria and many other bacteria respond to changes in zinc availability through the zinc uptake regulator, Zur (NGO_0542), which senses cytoplasmic zinc concentrations and, when zinc is abundant, represses expression genes involved in zinc uptake (35, 36). Zur was not differentially expressed as a result of adherence or zinc sequestration in this study (Data set S1). Six of the ten genes reported to compose the Gc Zur regulon were upregulated in Gc under zinc sequestration by murine calprotectin. Of the six, transcripts for three were more abundant in adherent Gc than suspension-grown Gc when zinc was sequestered: znuA, ngo1049, and rpmE2. These three ORFs may represent genes involved in the response to zinc limitation that are also modulated by adherence. ZnuA has been implicated in other pathogens as a primary means of overcoming calprotectin-mediated zinc sequestration, by trapping any zinc that enters the periplasm and shuttling it to the ZnuBC machinery for import into the cytoplasm (9, 57, 58, 70). Here we found that ZnuA was necessary for adherent Gc to survive in the presence of calprotectin, underscoring its importance to zinc acquisition in Gc. However, ZnuA overexpression was not sufficient to enable growth of suspension-grown Gc when zinc was limited (Fig. 6). NGO_1049 is bioinformatically predicted to be localized to the periplasm and contain a metal-binding motif (I. Liyayi and A.K.C., unpublished results), while RpmE2 is a zinc-independent ribosomal protein that replaces its zinc-dependent paralogue RpmE (NGO_2126) under conditions of zinc starvation (71). The remaining transcript that was increased by both zinc sequestration and adherence was FbpA (NGO_0217), which has been reported to be regulated by iron but not zinc (72). FbpA is a periplasmic protein, which binds iron and transfers it to inner membrane proteins for transport into the cytoplasm (73). Future studies will examine the contribution of these gene products to the increased survival of adherent Gc under zinc-limited conditions.

A related set of genes of interest were those whose transcripts were increased with zinc sequestration that was specific to the adherent bacteria (increased in adherent sequestered versus excess Gc and adherent versus suspension zinc sequestered, but not in suspension sequestered versus excess conditions). None of these genes have been previously described as zinc- or Zur-regulated, but each has been reported to be either repressed in anaerobic conditions or induced in the presence of hydrogen peroxide, and many of these genes were previously described as iron-repressed. For instance, CysK (NGO_0340) was induced in strain MS11 Gc by adherence to cervical cells (40) and contributes to epithelial colonization by N. meningitidis (74), and LldD is a L-lactate dehydrogenase that is important for intracellular survival of Gc under microaerobic conditions (75). GlnQ (NGO_0374), a predicted amino acid transporter, is upregulated in a hyperpiliated pilT mutant (76). The increase in expression of these genes suggests a metabolic shift in adherent, zinc-sequestered Gc that either results from or facilitates bacterial survival. We did not find a role for expression of type IV pili or opacity-associated (Opa) proteins in the enhanced survival of zinc-limited, adherent Gc (Fig. S8), suggesting a more general way in which Gc “senses” adherence to induce this protective gene expression profile. Notably, the adherent, zinc-sequestered Gc regulon did not overlap with the regulon of strain 1291 Gc in mature biofilms established on glass under flow conditions (77, 78). We did not directly test for biofilm components, and previous studies of Gc biofilms required at least 24 h to allow biofilm formation (79, 80). However, over the 4 h of adherent growth prior to RNA collection, it is likely that Gc organized into microcolonies (80). While these bacteria are transcriptionally distinct from Gc in canonical biofilms, interbacterial interaction and colony structure may play a role in the protection of adherent Gc and will be investigated in the future.

The remaining three genes in the Zur regulon were induced by zinc sequestration, but were unaffected by adherence (35, 36): tdfJ, hmcD, and tdfH (Table 1). tdfH fell just below the 0.5-fold change cutoff in suspension sequestered versus excess (0.49), but this change was statistically significant, so it has been included in this subset of genes. Since these genes were not significantly differentially expressed between adherent and suspension Gc under zinc sequestration, they are unlikely to be responsible for the increased survival of adherent Gc under zinc-limited conditions. However, they may work with the other differentially expressed genes described above to enable enhanced Gc survival, and may indicate a more general cross talk between zinc and iron regulons for Gc during infection.

Four genes that are repressed in excess zinc in N. meningitidis were not differentially expressed in any of the comparisons examined: znuB (NGO_0169), znuC (NGO_0170), rpmJ (NGO_0931), and queC (NGO_0129) (35, 36). It is particularly surprising that znuC and znuB were unchanged, because they are the first two ORFs of the operon containing znuA. This observation suggests additional regulatory mechanisms for the znuCBA operon that are relevant in the context of bacterial adherence, which will be a subject of future studies. The inability of zinc sequestration by murine calprotectin in KSFM to derepress the full Zur operon could indicate that the concentration of murine calprotectin that was used is not sufficient to fully sequester zinc, that the concentration of excess zinc used does not fully saturate Zur, and/or a more nuanced regulation of zinc-regulated genes based on the similarity of each promoter’s Zur binding sites to the canonical Zur binding motif, as was reported in N. meningitidis by Pawlik et al. (35). In addition to the transcriptional studies reported here, posttranscriptional gene regulatory mechanisms may also be responsible for the increased survival of adherent Gc under zinc-limited conditions, a topic for future investigation.

There is a continuous interplay between the need for a pathogen to acquire zinc during infection and the restriction of zinc bioavailability by the infected and inflamed host (47, 55, 81). The healthy female genital tract possesses high levels of calprotectin and psoriasin, which we predict would increase during the neutrophilic inflammation that characterizes Gc infection. Other S100 proteins that are known to bind zinc have been detected in the female genital tract (7, 45, 46, 82). Additionally, epithelial cells import zinc using ZIP transporters and sequester zinc internally using metallothioneins (83), which could create a local environment of lower zinc near cell surfaces. Therefore, we anticipate that the ability to acquire zinc from human proteins, as well as the ability to survive despite zinc sequestration by proteins Gc cannot directly use, are both important for Gc to effectively colonize human epithelia. In this light, the ability of Gc to proliferate when adherent when zinc is sequestered suggests that Gc has an adaptive defense against zinc limitation during infection. Targeting the necessity for zinc acquisition in Gc during infection is a potential therapeutic approach for drug-resistant gonorrhea, for instance by inhibiting the binding of zinc by ZnuA or by developing vaccines against TonB-dependent transporters that are expressed in infected individuals (21, 28). Taken together, the multifaceted ways in which Gc resists human nutritional immunity exemplify how Gc has evolved to evade and circumvent the human immune response to infection in order to become the urgent public health threat it is today.

MATERIALS AND METHODS

Bacterial strain construction.

Wild type (WT) for these studies is piliated Opaless FA1090 with locked-on OpaD (OpaD+, also called OpaD⁺_nv) (84). The Opa and PilE genes of this strain are not phase variable, which removed potential confounding effects from variability of these important bacterial adhesins. ΔtdfH Gc was generated by spot transforming wt Gc with genomic DNA from strain MCV955 (Opaless FA1090 tdfH::kan) (25) as described in (85) and selection on GCB (BD Difco 228920) containing 30 μg/mL kanamycin. ΔtdfJ Gc was generated by spot transforming WT with genomic DNA from strain MCV928 (FA19 tdfJ::Ω) (73), with selection on GCB containing 50 μg/mL spectinomycin, followed by two additional rounds of backcrossing into WT. ΔtdfH ΔtdfJ Gc was generated by transforming gDNA from ΔtdfH into ΔtdfJ and selecting on plates containing both 50 μg/mL spectinomycin and 30 μg/mL kanamycin. ΔznuA was generated by transforming WT with gDNA from MCV951 (FA19 znuA::Ω) on GCB plates supplemented with 25 μM ZnSO₄, 25 μM MnCl₂, 12 μM Fe(NO₃)₃, and 5 mM d-mannitol, with selection on GCB plates with these supplements and 50 μg/mL spectinomycin, followed by backcrossing twice into WT (26). All mutants were confirmed by PCR and/or sequencing (Table S2). znuA^c was generated by transforming WT with gDNA from MCV951 transformed with pVCU234 containing znuA^c, selecting on 0.5 μg/mL chloramphenicol, and backcrossing twice into WT Gc (26). ΔtdfH ΔtdfJ znuA^C was generated by transforming ΔtdfH ΔtdfJ Gc with gDNA from znuA^c. In both cases, overexpression of ZnuA was confirmed by Western blotting with a ZnuA-specific antibody (see Fig. 6). Opaless and OpaD⁺_nv ΔpilE have been previously described (84, 86).

Unless otherwise noted, Gc was prepared for experiments by inoculating −80°C stocks onto GCB and culturing for 24 h at 37°C, 5% CO₂. Gc was then passaged to GCB containing 10 μM zinc chelator TPEN (N,N,N’,N’-tetrakis(2pyridinylmethyl)-1,2,ethanediamine) (Sigma P4413) and grown for 18 h at 37°C, 5% CO₂. This concentration of TPEN induced robust TdfJ expression by Western blotting without restricting growth of Gc (data not shown). ΔznuA Gc was grown on GCB supplemented with 25 μM ZnSO₄, because this mutant does not grow on GCB without supplemental zinc (data not shown).

Recombinant S100 protein preparation.

Human and murine calprotectin were produced as previously described (27, 87). Human psoriasin was sub-cloned into a pET22b vector and purified as described (26) using a protocol developed in the Grötzinger laboratory (88). In all cases, the molarity of the recombinant proteins is presented based upon the molecular weight of the dimer (calprotectin: S100A8/S100A9 heterodimer; psoriasin: S100A7 homodimer).

Media growth assay.

Keratinocyte serum free medium (KSFM) was obtained from Gibco (17005042). KSFM with growth factors was prepared by adding bovine pituitary extract to 0.05 mg/mL and human recombinant EGF to 0.1 ng/mL final concentrations (both provided with KSFM by the manufacturer), and CaCl₂ (Fisher M-5133) to a final concentration of 400 μM. KSFM with growth factors was incubated for 20 min at 37°C with increasing concentrations of S100 proteins, with or without 3 μM ZnSO₄ (Sigma-Aldrich 204986). Based on these results (Fig. S1), the subsequent experiments used 1.4 μM human calprotectin, 1.4 μM human psoriasin, or 2 μM murine calprotectin. WT or tdf mutant Gc (1 × 10⁶ CFU/mL) was inoculated into the experimental media in round-bottom 96-well plates (Sarstedt 82.1582.001). No adherence of Gc to the wells was observed in any experiment. At the experiment’s start (time zero h) and the indicated time points, wells were mixed by pipetting, and 20 μL of the sample was removed for serial dilution and plating on GCB. CFU were enumerated 20–24 h later. Results are presented as the log₁₀ transformation of CFU recovered at a given time point. Statistics are by two-way ANOVA with Tukey’s multiple-comparison test.

Metal repletion experiments.

Media growth experiments were performed as above but with the following changes. KSFM with growth factors was incubated for 20 min at 37°C with 2 μM murine calprotectin, or 1.4 μM human calprotectin and/or 1.4 μM human psoriasin, with or without 6 μM ZnSO₄, 6 μM MnCl₂ (Sigma M3634), or 6 μM CuCl₂ (Sigma C3279). Statistics are adjusted P values from a one-way ANOVA with Šídák’s multiple-comparison test. The comparisons reported are for each S100 protein with and without additional metal.

Ect1 infection.

Ect1 E6/E7 immortalized ectocervical cells (ATCC CRL-2614) were acquired from ATCC and maintained in KSFM with growth factors plus antibiotics (1× antibiotic-antimycotic Gibco 15240062) at 37°C in 5% CO₂. For experiments, cells were seeded at 5 × 10⁴ cells/well in 24-well flat-bottom tissue-culture treated plates and grown to confluence (approximately 4 × 10⁵ cells/well). Two days before the experiment, cells were changed to medium without antibiotics. On the day of the experiment, KSFM with growth factors was preincubated with 1.4 μM human calprotectin and 1.4 μM human psoriasin for 20 min, with or without 3 μM ZnSO₄. WT or ΔtdfH ΔtdfJ Gc (2.5 × 10⁴ CFU) were inoculated on Ect1 cells for an MOI of ∼0.6 CFU/cell. After 2 h, cells were washed to remove non-adherent Gc. One set of cells was lysed with 1% saponin for 10 min at 37°C (Fluka analytical 47036), diluted to 0.2% saponin with GCBL, serially diluted, and plated to enumerate cell-associated CFU. After 9 h total (7 h after wash), supernatants were collected, diluted, and plated for CFU to enumerate detached bacteria. Cell-associated bacteria were collected by treating the wells with 1% saponin to lyse the Ect1 cells, and then diluting and plating for CFU as above. Media-grown bacteria were inoculated at 10⁵ CFU/mL and grown on the same day and in the same KSFM conditions as the adherent Gc, for 9 h total. For each adherent Gc experiment, the CFU associated with Ect1 cells at 2 h is plotted as horizontal lines in colors corresponding to experimental replicates. For the media-grown condition, the measured inoculum CFU is plotted as horizontal lines. CFU at the experimental endpoint is plotted as the log₁₀ transformation of recovered CFU. Statistics are from one-way ANOVA with Tukey’s multiple-comparison test.

Conditioned media.

KSFM with 400 μM CaCl₂ and lacking growth factors was incubated for 20 min at 37°C with or without 1.4 μM human calprotectin and 1.4 μM human psoriasin. These media were added to wells of a 24-well plate containing confluent Ect1 cells, Ect1 cells infected with ΔtdfH ΔtdfJ Gc following the protocol described above, or no Ect1 cells, and incubated at 37°C in 5% CO₂. After 7 h, media were removed, sterile filtered, and stored at 4°C overnight. The next day, the media were warmed to 37°C and used for growth assays using ΔtdfH ΔtdfJ or WT Gc as described in the media growth assay methods. After 7 h, samples were diluted and plated on GCB to enumerate CFU. Results are presented as log₁₀ transformed CFU, and statistics were calculated by ordinary one-way ANOVA with Šídák’s multiple-comparison test.

Separation of bacteria and Ect1 cells with filter inserts.

Ect1 cells were maintained and seeded as described above. Three conditions were employed, using KSFM without growth factors with or without 1.4 μM human calprotectin and 1.4 μM human psoriasin (see Fig. 3 for schematics). (i) Contact-dependent infection: 10⁵ CFU/mL ΔtdfH ΔtdfJ or WT Gc was inoculated into wells containing Ect1 cells and incubated for 2 h in KSFM without growth factors, washed to remove nonadherent bacteria, and fresh media replaced (KSFM +/− calprotectin and psoriasin) for an additional 7 h as described above, except that a Transwell filter insert (6.5 mm, 0.4 μm pore polyester; Corning Costar) was also added to the well. Dotted lines indicate the CFU adherent at 2 h in each experimental replicate. (ii) Exposure without Ect1 contact: Ect1 cells were incubated for 2 h in KSFM without Gc and containing Transwell filter inserts, then washed. At the time of medium replacement, ΔtdfH ΔtdfJ or WT Gc was inoculated onto the Transwell filter, which separated the bacteria from the Ect1 cells that were seeded in the wells, and incubated an additional 7 h. Dotted lines indicate the inoculum for each experimental replicate. (iii) No Ect1 cells: ΔtdfH ΔtdfJ or WT Gc was inoculated onto Transwell filter inserts in wells that did not have Ect1 cells for 7 h. Dotted lines indicate the inoculum for each experimental replicate. Results are presented as the log₁₀ transformed CFU for each condition. Statistics were calculated by unpaired Student's t test within each experimental condition.

Bacterial growth after adherence to live versus dead Ect1 cells.

ΔtdfH ΔtdfJ Gc (10⁵ CFU/ml) was inoculated into wells containing Ect1 cells that were alive, or killed by treatment with 4% paraformaldehyde (Electron Microscopy Sciences 15714) for 20 min at 37°C (cells were washed four times to remove fixative prior to addition of Gc). Infection was allowed to proceed as described above for Ect1 infection, with cells washed at 2 h and experimental media added. Media conditions were KSFM without growth factors alone, or containing 1.4 μM human calprotectin and 1.4 μM human psoriasin, with or without 6 μM ZnSO_4. One set of cells was lysed at 2 h after washing, and the other at 9 h total, from which CFU were enumerated. Gc inoculated into wells containing these media conditions without Ect1 cells served as a control. Results are presented as log₁₀ transformed CFU, and statistics were calculated by ordinary one-way ANOVA and Tukey’s multiple-comparison test.

Bacterial attachment to coverslips.

Round glass coverslips (12-mm diameter) (Fisher Scientific 12–545-80P) were acid-washed by heating in 1 N HCL at 60°C for 4 h, followed by successive washes of distilled water, 70% ethanol, and 100% ethanol. Gc (10⁶ CFU/mL) was inoculated onto coverslips in wells of a 24-well plate in 250 μL KSFM without growth factors. After 2 h, the medium was removed and coverslips were washed once with sterile PBS+ 5 mM MgSO₄. CFU were enumerated from one set of wells that were resuspended in KSFM by vigorous pipetting and serial dilution. Experimental media were KSFM alone or containing 2 μM murine calprotectin, with or without 6 μM ZnSO₄. Experimental media were added to the other set of wells and incubated for an additional 7 h, at which time detached and adherent Gc were collected and CFU were enumerated. For each experimental replicate, CFU were simultaneously collected from infection of Ect1 cells or inoculation into wells lacking coverslips (suspension-grown) as described above. Data are presented as the log₁₀ transformation of the CFU recovered at the experiment endpoint, and asterisks represent significant differences by ordinary one-way ANOVA with Tukey’s multiple-comparison test.

RNA extraction and qPCR of Gc from Ect1 infection.

Ect1 cells were infected as described above, in KSFM without growth factors, using a higher inoculum (5 × 10⁷ CFU/mL) and for a shorter time (2 h + 4 h after wash). For comparison, Gc were grown for 4 h under the same conditions, in wells without Ect1. At the experimental endpoint, the supernatant was removed and adherent Gc were collected in 200 μL per well of RNAprotect cell reagent (Qiagen 76526). Cells from 4 replicate wells per condition were pooled, pelleted, resuspended in 200 μL RNAprotect cell reagent, and stored at −80°C until processed to purify RNA. Four wells of uninfected Ect1 cells in RNAprotect cell reagent were spiked into the bacteria grown without Ect1.

Samples were lysed according to the “Enzymatic Lysis and Proteinase K Digestion of Bacteria” protocol from Qiagen before proceeding with RNeasy Plus minikit (Qiagen 74134) per manufacturer’s instructions. qPCR was conducted using a Thermo Fisher QuantStudio3 instrument, Power SYBR green PCR master mix (Thermo Fisher Scientific 4368577), RNase inhibitor (Life Technologies N8080119), and Multiscribe Reverse Transcriptase (Life Technologies 4311235) and qRT-PCR primers (Table S2). Data were normalized to a 5S rRNA reference control. Data are expressed as relative expression (2^-ΔΔCT) relative to the KSFM medium-grown condition. Statistics were calculated by ordinary two-way ANOVA with Šídák’s multiple-comparison test.

Generation of guinea pig anti-ZnuA antiserum.

FA1090 ZnuA peptide Ac-CSYAEATKGIQPLKAEE-amide was synthesized by Vivitide (formerly New England Peptide) and confirmed by HPLC (>85% purity by percent area) and mass spectral analysis (calculated mass within 0.1% of MW: 1879 Da). Three mg of peptide was fused to keyhole limpet hemocyanin and utilized in four immunizations of two guinea pigs, which were ultimately exsanguinated yielding ∼15mL serum per guinea pig. Optimal dilution factor for antiserum was determined by Western analysis of whole cell lysates of GC overexpressing or not expressing ZnuA. To remove nonspecific reactivity, antiserum was preincubated with a nitrocellulose membrane onto which a lysate from ΔznuA Gc had been transferred.

ZnuA overexpression and bacterial growth in suspension.

KSFM without growth factors was incubated for 20 min at 37°C with no additives, with 1 mM IPTG (for induction of znuA complement expression), or with 1.4 μM human calprotectin and 1.4 μM human psoriasin, with or without 1 mM IPTG or 3 μM ZnSO₄. Gc (10⁵ CFU/mL) was inoculated into each medium condition and incubated at 37°C in 5% CO₂. CFU were enumerated from the inoculum and after 9 h of incubation as described above for media growth. Data are presented as Log₁₀ CFU recovered, and statistics represent P values from ordinary one-way ANOVA with Tukey’s multiple-comparison test.

To prepare lysates for Western blot, Gc of the indicated strains was cultured for 24 h on GCB at 37°C, 5% CO₂. Gc was then passaged to the following GCB plates: +Zn: 25 μM ZnSO₄; +TPEN: 10 μM TPEN; +IPTG: 1 mM IPTG; +TPEN+IPTG: 10 μM TPEN, and 1 mM IPTG. After 18 h, bacteria were collected using polyester swabs (Fisher Scientific 25–806) and resuspended in PBS + 5 mM MgSO₄. Samples were normalized to equal optical densities and resuspended in sample buffer containing 12 mM Tris pH 6.8, 5% glycerol, 0.4% SDS, 1% β-mercaptoethanol, and 0.02% bromophenol blue. Samples were boiled for 5 min and sheared with a syringe and 22-gauge needle (BD 305156).

Western blot.

Bands were resolved using a 4–20% Criterion TGX precast gel (5671094, Bio-Rad) and transferred in tris-glycine methanol buffer (0.3% Tris-HCl, 1.4% glycine, 20% methanol) to nitrocellulose. Blots were blocked with 2.5% BSA (Fisher BP1600) in TBST (0.24% Tris-HCl, 1% NaCl, 0.01% Tween 20, pH 7.6) for 1 h, and primary and secondary antibodies were diluted in TBST + 2.5% BSA. Blots were incubated with guinea pig anti-ZnuA antiserum (1:1,000), followed by donkey anti-guinea pig 800CW (1:15,000). Rabbit anti-Zwf antiserum (1:10,000) (89) and goat anti-rabbit 680H+L (1:15,000) were used as loading control. The blot was imaged on a LI-COR Odyssey CL-X, and relative band intensities were quantified using Image Studio software. ZnuA protein levels were normalized to the Zwf loading control and statistics were calculated by ordinary one-way ANOVA with Šídák’s multiple-comparison test. Blot image was cropped and rearranged for ease of interpretation, but not otherwise altered.

RNA extraction for RNAseq from Gc attached to coverslips or in suspension.

Adherent Gc: WT Gc (10⁸ CFU/mL in 500 μL KSFM without growth factors) were grown as described above and allowed to adhere to 18-mm diameter acid-washed glass coverslips (12-545-100P) for 2 h, then washed to remove nonadherent Gc. The experimental media (KSFM without growth factors containing 3 μM mCp, with or without 6 μM ZnSO₄) were added and the infection was allowed to proceed for another 4 h at 37°C in 5% CO₂, at which time the supernatant was removed and 1 volume (500 μL) KSFM without growth factors and 2 volumes (1 mL) RNAprotect bacteria reagent (Qiagen 76506) were added. Gc was collected by scraping, and cells from 8 wells were pooled together in a 15-ml conical tube, incubated at room temperature for 5 min, and pelleted at 4,000 × g for 11 min. Suspension-grown Gc: WT Gc was prepared in the same experimental media as described above, inoculated at 1 × 10⁸ CFU/mL into 96-well round-bottom plates, and grown for 4 h at 37°C in 5% CO₂. Gc was harvested by pooling 2 ml of culture per condition, adding 2 volumes RNAprotect bacteria reagent to the pooled cultures and centrifuging as above. In all growth conditions, pellets were processed and RNA was purified as described for qPCR. DNase treatment was completed using a TURBO DNAfree kit (Invitrogen AM1907).

Illumina RNA sequencing was conducted by the Genomics Resource Center of the Institute of Genome Sciences at the University of Maryland. rRNA reduction was completed using NEBNext rRNA Depletion Kit (Bacteria) (NEB E7850), and RNA library prep was done using NEBNext Ultra II Directional RNA Library Prep Kit (NEB E7760). Sequencing was performed using NovaSeq S2 on the NovaSeq6000 platform, with >5M 150 bp paired-end read pairs per sample.

Reads were checked for quality using FASTQC (90) (v0.11.8), trimmed using BBMAP (91) (v3.8.16b), and aligned to the Neisseria gonorrhoeae FA1090 genome with Ensembl annotations (ASM684v1.51) using STAR (92) (v2.7.1a). Transcripts per million calculations were performed by RSEM (93) (v1.3.1), the results of which were imported into R (v4.0.2) (94) and Bioconductor (95) (v3.12) using tximport (96) (v1.18.0). Significant genes were called using DESeq2 (97), using fold change cutoffs and pvalue cutoffs of 0.5 and 0.05, respectively. RNA-seq data are deposited at Gene Expression Omnibus (GEO) database under accession number GSE191020.

ICP-MS and ICP-OES.

Quantification of zinc and other metals was conducted by the Center for Applied Isotope Studies at the University of Georgia. KSFM with and without growth factors was acidified to a final concentration of 2% nitric acid, diluted 10-fold, and analyzed by ICP-MS. KSFM without growth factors was incubated for 7 h in wells of a 96-well round-bottom plate (incubated in empty wells) or with acid-washed glass coverslips in wells of a 24-well plate (incubated with glass coverslips). Media were removed from wells, like wells were pooled, and samples were acidified to 2% nitric acid and analyzed by ICP-OES.

Figure generation and statistical analysis.

All graphical figures and statistical analyses were generated using GraphPad Prism 9 for macOS Version 9.2.0, GraphPad Software, San Diego, CA, USA, www.graphpad.com. Specific statistical tests used are noted for each figure.