A genome-wide genetic screen identifies CYRI-B as a negative regulator of CEACAM3-mediated phagocytosis

Phagocytosis is a critical cellular defense mediated by professional phagocytes including granulocytes, macrophages and dendritic cells. Dedicated phagocytic receptors, such as Fc-receptors, complement receptors, scavenger receptors, CEACAM3 and Dectin-1 (also known as CLEC7A) play an important initiating role in this process (Janeway and Medzhitov, 2002). Whereas Fc-receptors and complement receptors require prior opsonization, CEACAM3 and Dectin-1 directly bind to specific surface-exposed features of their target particles (Underhill and Ozinsky, 2002; Goodridge et al., 2012). In contrast to Dectin-1, which recognizes conserved glycan moieties exposed on the yeast cell wall, CEACAM3 detects a particular functionality of various human-restricted bacterial pathogens – the possession of CEACAM-binding adhesins (Pils et al., 2008). Indeed, the ability to engage members of the human CEACAM family has evolved independently in multiple bacterial lineages, as evidenced by the structurally diverse CEACAM-binding adhesins expressed on the surface of the respective pathogens (Tchoupa et al., 2014). Most of these CEACAM-binding adhesins recognize homologs of CEACAM3 abundantly expressed on epithelia, such as CEACAM1 and CEA (also known as CEACAM5) (Kuespert et al., 2006; Beauchemin and Arabzadeh, 2013). In contrast, the product of the CEACAM3 gene is only present on granulocytes, and bacterial adhesins seem to be selected to avoid recognition by this receptor (Bonsignore et al., 2019). Accordingly, CEACAM3 is involved in a molecular arms race with specialized human-restricted bacterial pathogens such as Neisseria gonorrhoeae (Adrian et al., 2019; Bonsignore et al., 2019). Similar to other CEACAM family members, the extracellular IgV-like domain of CEACAM3 is responsible for binding microbial adhesins, including the Opa proteins of Neisseria gonorrhoeae (Schmitter et al., 2004; Roth et al., 2013; Hill and Virji, 2003; Tchoupa et al., 2015; Koniger et al., 2016). However, the functionality of CEACAM3 as a phagocytic receptor depends on a short amino acid sequence in its cytoplasmic domain reminiscent of an immunoreceptor tyrosine-based activation motif (ITAM). Phosphorylation of this motif by Src family protein tyrosine kinases (PTKs) allows recruitment of the SH2-domain containing Rac guanine nucleotide exchange factor Vav (McCaw et al., 2003; Schmitter et al., 2004; Schmitter et al., 2007). As a consequence, CEACAM3 engagement results in a rapid increase in GTP-loaded Rac, activation of p21-activated kinase (PAK), and WAVE-complex-dependent cytoskeletal rearrangements driving the formation of prominent lamellipodia (Hauck et al., 1998; Billker et al., 2002; Schmitter et al., 2004; Pils et al., 2012; Buntru et al., 2012). Stimulation of CEACAM3 is also linked to the production of reactive oxygen species and release of cytokines by granulocytes (Sarantis and Gray-Owen, 2007; Buntru et al., 2011; Sarantis and Gray-Owen, 2012; Heinrich et al., 2016). Therefore, CEACAM3-triggered signaling has to be tightly controlled to prevent excessive inflammation, but cellular regulators of this signaling pathway are poorly characterized.

To identify potential regulatory elements of CEACAM3-initiated phagocytosis, we performed a genome-wide CRISPR/Cas9 screen in human phagocytes. Concentrating on genes that consistently produced a gain-of-function phenotype when depleted by three independent sgRNAs, we identified the Rac-GTP-binding protein CYRI-B as a prominent negative regulator of CEACAM3-driven signaling events. CYRI-B sequesters GTP-bound Rac, thereby limiting Rac-driven downstream processes such as activation of PAK kinases and lamellipodia formation (Fort et al., 2018; Shang et al., 2018; Yuki et al., 2019; Le et al., 2021). We found that knockout of CYRI-B allowed increased phagocytosis of CEACAM3-binding bacteria, which was reverted by re-expression of wild-type, but not Rac-binding-deficient, CYRI-B. These findings not only validate the genome-wide screening approach, but also reveal an unexpected additional layer of Rac regulation in myeloid cells, which could serve as a target to control excessive granulocyte activation.

As a pre-requisite to a genome-wide lentivirus-based screen, the transduction efficiency and phenotypic knockout kinetics of the used cell model have to be evaluated and optimized. Therefore, we screened several human myeloid cell lines commonly employed as granulocyte surrogates for their susceptibility to lentiviral transduction using a GFP-encoding recombinant lentivirus. Although 40–50% of NB4 cells and JOSK-M cells (a subline of U937 cells) became GFP-positive 3 days after lentiviral transduction, the transduction efficiency seen with HL60 cells was >90% (Fig. S1A). Furthermore, the expression of GFP from the lentiviral construct was several-fold higher in HL60 cells, as evidenced by the elevated mean fluorescence intensity (MFI) of the transduced HL60 cell population (Fig. S1A). Accordingly, we turned our attention to HL60 cells and generated a stable HL60 cell line expressing the blue-fluorescent protein Cerulean (HL60 Cer) (Fig. S1B). These cells were transduced with a lentivirus encoding the Cas9 enzyme, a sgRNA targeting the Cerulean coding sequence, as well as a puromycin resistance cassette. At 3 days after viral infection, the cells showed unaltered Cerulean fluorescence (Fig. S1B). At that point, the infected cell population was incubated with puromycin to select for stably transduced cells (Fig. S1B). After 3 days of puromycin selection, a small population of Cerulean-negative cells was observed, which after a further 2 days of puromycin selection amounted to ∼20% of the cell population (Fig. S1B,C). HL60 cells express only low amounts of CEACAM3 and express additional CEACAM family members, including CEACAM1, CEACAM4, CEACAM6 and CEACAM8 (Skubitz et al., 1996; Delgado Tascon et al., 2015; Goob et al., 2022). To maximize CEACAM3-mediated phagocytic events, we therefore stably introduced a CEACAM3–mKate fusion protein (Fig. 1A). The resulting HL60 CEACAM3 cells exhibited rapid opsonin-independent phagocytosis of CEACAM3-binding Neisseria gonorrhoeae (Ngo Opa₅₂) as indicated by the shift in cell-associated CFSE-fluorescence derived from internalized CFSE-labeled bacteria (Fig. 1B). In contrast, CEACAM1-binding (Ngo Opa₅₄) or non-CEACAM-binding (non-opaque; Opa⁻) gonococci (Ngo Opa⁻) were hardly internalized under these conditions (Fig. 1B). At a low multiplicity of infection (MOI), when phagocytosis was not saturated, more than 40% of HL60 CEACAM3 cells had phagocytosed CEACAM3-binding bacteria within 15 min of infection (Fig. 1B; Fig. S1D). Furthermore, HL60 CEACAM3 cells infected with Opa₅₂-expressing gonococci, but not with non-opaque (Opa⁻) or Opa₅₄-expressing bacteria, exhibited prominent lamellipodia, as reported for CEACAM3-mediated uptake by granulocytes (Schmitter et al., 2004; Goob et al., 2022) (Fig. 1C). Accordingly, we employed HL60 CEACAM3 cells together with a genome-wide library of single-guide RNAs (sgRNAs) in the lentiviral vector lentiCRISPR v2 (Sanjana et al., 2014) to screen for factors modifying the opsonin-independent phagocytosis via CEACAM3. The layout of the screen is depicted in Fig. 1D. To this end, the HL60 CEACAM3 cells were transduced with the Human CRISPR Knockout Pooled Library (GeCKov2; part A) targeting 19,000 human genes with three different sgRNAs each. After 3 days, the virus-infected cells were selected with 0.3 µg puromycin for 5 days, and an aliquot of the puromycin-resistant HL60 CEACAM3 cell population was set aside (‘Input’). Next, cells were infected at an MOI of 10 for 15 min with CSFE-labeled Opa₅₂-expressing N. gonorrhoeae and, upon gating on infected cells, a high-fluorescence population (‘High’) with an above-average phagocytosis and a corresponding low-fluorescence sample (‘Low’) with a below-average uptake of CEACAM3-binding bacteria was collected by cell sorting. Genomic DNA from the Input, High and Low samples was used to amplify the integrated sgRNA and to subject the amplicons to next generation sequencing. Comparison of the relative abundance of specific sgRNAs in the high or low population compared to the input sample was conducted to reveal genes that influence CEACAM3-mediated phagocytosis.

Following the procedure outlined in Fig. 1D, we infected the lentivirally transduced and puromycin-treated HL60 CEACAM3 cell population with an MOI of 10 CSFE-labeled Opa-negative gonococci (Ngo Opa⁻), Opa₅₂-expressing N. gonorrhoeae (Ngo Opa₅₂), or left them uninfected (Fig. 2A). As expected, Opa₅₂-expressing N. gonorrhoeae were readily phagocytosed, whereas Opa-negative bacteria showed no increased signal compared to the uninfected control, indicating that phagocytosis was strictly mediated by the Opa–CEACAM interaction (Fig. 2A). After 15 min of phagocytosis, we collected a total of 3×10⁶ cells from a Low population representing the cells within the lowest 20% percentile of fluorescence intensity (Fig. 2B). Furthermore, 2×10⁶ cells of a High population were collected, representing the cells within the highest 20% percentile of fluorescence intensity, which indicates above average phagocytosis (Fig. 2B). Re-analysis of the sorted cell populations indicated that the Low population on average contained only ∼50% of phagocytosed bacteria, while the High population on average phagocytosed ∼130% of CFSE-labeled Opa₅₂-expressing gonococci compared to the input population (Fig. 2B). Sequencing of the sgRNA repertoire encoded by cells of the Input, Low and High populations (>16,000,000 reads/sample) allowed us to obtain a quantitative estimate of the relative distribution of individual sgRNAs in the low and high populations compared to the input population. There were 14 sgRNA targets that showed a >2-fold enrichment in the Low population and a depletion (<1) in the High population relative to the Input for all three sgRNAs targeting the same gene. These sgRNA target genes could encode proteins with a positive contribution to CEACAM3-mediated phagocytosis. The ten top hits from this list are depicted in Fig. 2C and encode proteins with a role in vesicle trafficking (RABGGTA), chromatin regulation (SMARCA2) and nucleotide synthesis (PRPS1L1), as well as transmembrane proteins, such as TMEM185A, prominin 1, also known as CD133 (PROM1), the Golgi protein GORASP1 and the zinc transporter SLC39A6. Similar to the signal peptidase (SPCS3), these proteins might have an indirect role in facilitating CEACAM3 processing or transport to the membrane and thereby impact CEACAM3-mediated phagocytosis. A direct connection to CEACAM3-mediated phagocytosis is plausible for proteins such as spectrin A1 (SPTAN1) or the Rac guanine nucleotide exchange factor Dock 3 (DOCK3), which might be involved in the cytoskeletal re-arrangements accompanying bacterial uptake by CEACAM3.

We also applied the same algorithm to identify sgRNAs that showed a >2-fold enrichment in the High population and a depletion (<1) in the Low population for all three sgRNAs targeting the same gene. Thereby, we identified 29 sgRNA targets, which could have a negative impact on CEACAM3-mediated phagocytosis and whose depletion should lead to a gain-of-function phenotype. The ten top hits from this list are depicted in Fig. 2C and encode proteins with a role in receptor tyrosine kinase signaling (fibroblast growth factor receptor like 1; FGFRL1) and inhibition (LRIG2), lipid signaling (S1PR4), maturation of secretory vesicles (NBEAL2), lysosome positioning (BLOC-1-related complex subunit 8; MEF2BNB), and regulation of the GTPase Rac (CYRI-B). In these cases, a link to activation of myeloid cells and phagocytosis can be easily envisioned. On the other hand, a potential negative impact of DNAJC19 (protein import in mitochondria), TRIM61 (a ubiquitin E3-ligase), trophinin (TRO; cell–cell adhesion protein), and hyaluronan and proteoglycan link protein 2 (HAPLN2; extracellular matrix assembly) on CEACAM3-initiated processes is not easy to reconcile. Accordingly, it is instrumental to confirm and investigate the connection of these putative factors to CEACAM3-mediated phagocytosis in more detail.

To get a first idea of whether the genome-wide screen yielded meaningful hits, we concentrated on CYRI-B (also termed FAM49B). CYRI-B, the CYFIP-related Rac interactor B, bears similarity to the WAVE-complex protein CYFIP, and binds GTP-loaded Rac. Thereby, CYRI-B sequesters GTP-Rac and impedes Rac-dependent cellular processes (Fort et al., 2018; Shang et al., 2018). To this end, we produced clonal CYRI-B knockout lines derived from HL60 CEACAM3 cells. Cells were transduced with lentivirus encoding Cas9 and sgRNAs targeting the critical exon of CYRI-B in the 3′-vicinity of the ATG start codon (Fig. 3A). Transduced cells were selected for 7 days in puromycin and single-cell-derived clonal lines were expanded. Several independent cell lines showed either partial or complete loss of CYRI-B protein expression by western blotting (Fig. 3B). Importantly, clonal line C6 with a complete knockout of CYRI-B retained equivalent total expression and surface localization of CEACAM3 compared to the control transfected cells (Fig. S2). Next, HL60 CEACAM3 cells and HL60 CEACAM3 CYRI-B knockout clone 6 cells (CYRI-B ko) were infected for 15 min with CFSE- and biotin-labeled gonococci, and phagocytosis was evaluated by differential staining of intracellular and extracellular bacteria followed by immunofluorescence microscopy (Fig. 3C). As expected, both HL60 cell lines did not internalize non-opaque bacteria (Fig. 3C). However, Opa₅₂-expressing gonococci were taken up by HL60 CEACAM3 cells, and disruption of CYRI-B expression in CYRI-B ko cells resulted in a more than doubling of the number of internalized bacteria (Fig. 3C,D). This striking gain-of-function was verified by flow cytometry-based analysis of phagocytosis (Fig. 3E). Again, non-opaque gonococci were hardly found inside the cells, whereas Opa₅₂-expressing bacteria were taken up by HL60 CEACAM3 cells and were internalized in even greater numbers by HL60 CEACAM3 CYRI-B ko cells (Fig. 3E). To investigate whether the lack of CYRI-B also affects opsonin-dependent phagocytic uptake, besides the opsonin-independent clearance via CEACAM3, we tested Fc-receptor-mediated phagocytosis by wild-type and CYRI-B ko HL60 cells. To this end, non-opaque gonococci were opsonized with rabbit anti-N.gonorrhoeae antibodies prior to infection. The non-opaque IgG-opsonized bacteria were readily internalized by HL60 CEACAM3 cells (Fig. 3F). Importantly, HL60 CEACAM3 CYRI-B ko cells showed a significantly increased uptake of the opsonized microbes, reflecting more than a doubling of bacterial internalization compared to that for the HL60 CEACAM3 cells (Fig. 3F). The uptake of these non-opaque bacteria was clearly mediated by Fc-receptors, as non-opsonized non-opaque gonococci were not internalized (Fig. 3F). These results indicate that lack of CYRI-B can enhance uptake by multiple phagocytic receptors, whereas receptor-independent engulfment of non-opsonized non-opaque gonococci, which neither engage with CEACAM3 nor Fc-receptors, is not affected (Fig. 3F).

We wondered about the consequences of CYRI-B deficiency for Rac-dependent processes in phagocytes. Indeed, Rac activation in response to CEACAM3 engagement has been shown to direct the cytoskeletal rearrangements required for lamellipodia formation and efficient engulfment of particles (Schmitter et al., 2004; Schmitter et al., 2007). In line with a role of CYRI-B in dampening Rac signaling, CYRI-B ko cells showed elevated levels of GTP-bound Rac even in the absence of bacterial infection (Fig. 4A). As a consequence of the elevated Rac-GTP loading, downstream effectors of Rac, such as the p21-activated kinases PAK1 and PAK2 (hereafter collectively referred to as PAK) showed increased phosphorylation in HL60 CEACAM3 CYRI-B ko cells (Fig. 4B). PAK kinases have been implied in regulating F-actin dynamics via modulating LIM-kinase activity (Edwards et al., 1999), phosphorylation of Arp2/3 complex subunits (Vadlamudi et al., 2004) and inhibition of the guanine nucleotide exchange factor β-PIX (also known as ARHGEF7) (Davidson et al., 2021) suggesting that enhanced PAK activity could contribute to the phenotype of HL60 CEACAM3 CYRI-B ko cells. Accordingly, we used pharmacological inhibition of PAK with IPA-3 to interfere with PAK-dependent cytoskeletal dynamics. Indeed, pharmacological inhibition of PAK with IPA-3 severely diminished internalization of Opa₅₂-expressing N. gonorrhoeae by HL60 CEACAM3 and the HL60 CEACAM3 CYRI-B ko cells, suggesting that PAK is an important effector downstream of Rac-GTP (Fig. 4C). Moreover, de-regulated Rac activity in CYRI-B-deficient HL60 cells was accompanied by elevated F-actin levels (Fig. 4D). These findings imply that the enhanced Rac activity resulting from CYRI-B deficiency is responsible for the gain-of-function phenotype of CEACAM3-expressing phagocytes.

To verify that the phenotypes observed in CYRI-B-deficient cells are a direct consequence of the lack of this protein, we re-established CYRI-B expression in HL60 CEACAM3 CYRI-B ko cells. First, we introduced silent mutations in the sgRNA target sequence of the human CYRI-B cDNA (Fig. 3A). The resulting sgRNA-resistant version of CYRI-B wild-type (CYRI-B WT), fused to mCherry via a self-cleavable 2A peptide, was then stably re-introduced into HL60 CEACAM3 CYRI-B ko cells (Fig. 5A). Whole-cell lysates prepared from HL60 CEACAM3 CYRI-B ko CYRI-B wild-type (WT) re-expressing cells showed a ∼63 kDa band corresponding to the mCherry–CYRI-B fusion protein and a band at ∼37 kDa co-migrating with the endogenous CYRI-B observed in HL60 CEACAM3 cells (Fig. 5A). This is in line with previous reports indicating various degrees of self-cleaving activity by viral 2A peptides (Kim et al., 2011). Importantly, when these cells were infected with CFSE-labeled Opa₅₂-expressing N. gonorrhoeae, the elevated phagocytosis seen for the CYRI-B ko cells was strongly reduced, clearly demonstrating that the gain-of-function phenotype was indeed resulting from the lack of CYRI-B expression (Fig. 5B). Furthermore, although PAK phosphorylation was increased in HL60 CEACAM3 cells upon infection with Opa₅₂-expressing gonococci and PAK phosphorylation was even further elevated in HL60 CEACAM3 CYRI-B ko cells, the re-expression of CYRI-B in the CYRI-B ko cells suppressed the exaggerated PAK phosphorylation (Fig. 5C,D). In line with the increased levels of GTP-loaded Rac and phosphorylated PAK, HL60 CEACAM3 CYRI-B ko cells showed spontaneous formation of large lamellipodia when observed by scanning electron microscopy (Fig. 5E). Again, this gain-of-function phenotype was suppressed upon re-introduction of CYRI-B WT (Fig. 5E). These results clearly demonstrate that the intensified Rac-driven cytoskeletal processes in the HL60 CEACAM3 CYRI-B ko cells are a direct consequence of CYRI-B deficiency.

The CYRI-B-deficient HL60 CEACAM3 cells also afforded the possibility to investigate which functional domains are critical for the negative regulatory role of CYRI-B. To this end, we generated missense mutations in the sgRNA-resistant CYRI-B cDNA to disrupt the N-terminal myristoylation signal (CYRI-B G2A), to compromise the Rac-GTP binding interface (CYRI-B R160D/R161D) or a combination of both (CYRI-B G2A R160D/R161D) (Fig. 6A). These constructs or the mCherry-encoding empty virus were then introduced into the HL60 CEACAM3 CYRI-B ko cells and stable clonal cell lines were selected. Clones that re-expressed similar levels of the different CYRI-B variants were selected (Fig. 6B). Importantly, when these cells were infected with CEACAM-binding gonococci, re-expression of CYRI-B WT, but not introduction of mCherry, reduced the elevated internalization of bacteria back to the levels observed in HL60 CEACAM3 cells (Fig. 6C). The successful complementation of the HL60 CEACAM3 CYRI-B ko cells by the re-expression of WT CYRI-B confirmed the previous analysis and demonstrated that the gain-of-function phenotype of the CYRI-B-knockout cells is not based on an off-target effect of the CRISPR/Cas9 treatment. Moreover, expression of the Rac-binding-deficient CYRI-B R160D/R161D mutant was not able to revert the phenotype of the HL60 CEACAM3 CYRI-B ko cells, demonstrating that the effect of CYRI-B on CEACAM3-mediated phagocytosis depends on its ability to sequester GTP-bound Rac (Fig. 6C). In contrast, the myristoylation of CYRI-B was not required, as CYRI-B G2A was perfectly fine in diminishing the exaggerated phagocytosis of CEACAM3-binding gonococci (Fig. 6C). Importantly, these functional data with regard to phagocytosis were mirrored by the PAK phosphorylation levels in the different cell lines (Fig. 6D). Re-expression of CYRI-B WT and CYRI-B G2A reduced the elevated PAK phosphorylation levels back to background levels, whereas re-expression of mCherry alone, CYRI-B–mCherry R160D/R161D, or CYRI-B-mCherry G2A R160D/R161D did not revert the phenotype (Fig. 6D). Together, these findings demonstrate that CYRI-B limits CEACAM3-initiated responses by sequestering the small GTPase Rac, thereby controlling the extent of CEACAM3-driven downstream processes.

CEACAM3 is a specialized phagocytic receptor, enabling primate granulocytes to recognize and eliminate a restricted set of host-adapted bacterial pathogens (Bonsignore et al., 2019). To detect novel cellular regulators of CEACAM3-mediated phagocytosis, we performed an unbiased, genome-wide genetic screen in a human myeloid cell line. Although loss-of-function effects were seen for several genes, including components of the cytoskeleton, we focused our attention on genes whose disruption consistently resulted in a gain-of-function with regard to the uptake of CEACAM3-engaging microbes. We predicted that such a phenotype should arise from the deletion of factors with a negative regulatory role in CEACAM3-mediated phagocytosis.

In agreement with this prediction, abrogating the expression of the CYFIP-related Rac interactor B (CYRI-B), one of the factors identified in the screen, led to elevated internalization of CEACAM-binding Neisseria gonorrhoeae. Owing to the Rac-GTP sequestering activity of CYRI-B, it has been previously observed that CYRI-B diminishes Rac-GTP-initiated downstream signaling and thereby interferes with Rac-dependent actin dynamics (Fort et al., 2018; Shang et al., 2018). This function of CYRI-B can directly explain the phenotype with regard to CEACAM3-mediated phagocytosis observed for the CYRI-B-deficient cells. Indeed, compromising the Rac-binding interface of CYRI-B at residues R160 and R161 also abrogates the effect of CYRI-B on CEACAM3. The recent crystal structure (PDB: 7AJK) of CYRI-B in complex with Rac nicely reveals why CYRI-B selectively acts on Rac, but not on other related Rho family GTPases, such as Cdc42 (Yelland et al., 2021). Therefore, the surfacing of CYRI-B as a negative regulator of CEACAM3-mediated phagocytosis in this unbiased screening approach highlights the central role played by Rac in CEACAM3-initiated cellular responses. Although earlier studies employing CEACAM3-expressing HeLa cells indicated a contribution of the Rho family GTPase Cdc42 for CEACAM3-dependent bacterial internalization (Billker et al., 2002), investigations in myeloid cell lines and primary human granulocytes pointed to a specific and exclusive function of Rac, but not Cdc42, downstream of CEACAM3 (Hauck et al., 1998; Schmitter et al., 2004). This tight connection between CEACAM3 and Rac is not only further emphasized by the involvement of the WAVE/Scar complex in regulating the cytoskeletal rearrangements necessary for CEACAM3-driven phagocytosis (Pils et al., 2012), but is now underscored by our finding of a prominent negative-regulatory role of CYRI-B. In this respect, CYRI-B structurally mimics the CYFIP component of the WAVE complex, functioning as a competitive inhibitor of Rac-GTP-dependent stimulation of Scar/WAVE and its Arp2/3-complex-mediated actin nucleating activity (Yelland et al., 2021; Kaplan et al., 2020). In line with the idea that Rac is the critical Rho family GTPase in CEACAM3-initiated phagocytosis, our screen not only uncovers, with CYRI-B, a selective negative regulator of Rac, but also indicates that the Rac-specific guanine nucleotide exchange factor Dock3 could play a positive role in this process.

Exaggerated activation of Rac can lead to massive cytoskeletal re-organization and, via the activation of the NADPH oxidase, to the production of reactive oxygen species. Therefore, this process needs to be constrained in its magnitude, but also needs to be spatially confined to maximize the formation of productive lamellipodial protrusions. Accordingly, CYRI-B could provide local and/or temporal regulation of Rac to prevent excessive phagocyte responses during infection, which could damage host tissues. Recently, CYRI-B has also been shown to protect animals from infections by facultative intracellular pathogens, such as Salmonella, Mycobacterium and Listeria, which enter and multiply in phagocytes (Yuki et al., 2019). In this scenario, the reduction in phagocytosis afforded by CYRI-B seems to reduce the chances of these microbes for intracellular multiplication and systemic spread via macrophages (Yuki et al., 2019).

How CYRI-B itself is regulated, is currently unclear. Based on the crystal structure, Yelland et al. have suggested the existence of an autoinhibited CYRI-B dimer, which is not able to bind the small GTPase (Yelland et al., 2021). By a rachet-like movement of a C-terminal α-helical bundle, CYRI-B would need to first attain the proper conformation for associating with Rac-GTP (Yelland et al., 2021). Accordingly, binding interactions or post-translational modifications of the CYRI-B C-terminal portion might trigger the resolution of the CYRI-B dimer and allow CYRI-B binding to Rac-GTP. Indeed, several lysine residues in the C-terminus of CYRI-B are the target of ubiquitylation (Uniprot entry Q9NUQ9; https://www.phosphosite.org/proteinAction.action?id=19888&showAllSites=true), a modification, which could guide the large conformational changes in CYRI-B. Interestingly, our screen for modulators of CEACAM3-mediated phagocytosis also identified enzymes involved in mono- or poly-ubiquitylation such as TRIM61. Although their actual contribution is currently unclear, it will be interesting to unravel a potential functional connection between such enzymes and CYRI-B modifications in CEACAM3-mediated phagocytosis.

In conclusion, a genome-wide genetic screen to unravel factors influencing CEACAM3-mediated phagocytosis has not been conducted previously. The screening results could be verified by identifying CYRI-B as a negative regulator of CEACAM3-initiated processes and yielded additional leads for further investigation. Our results, therefore, provide an initial framework to comprehensively describe the processes and individual proteins that orchestrate this innate immune defense.

The following primary and secondary antibodies were used. Primary antibodies were against the following proteins: CEACAM3/CEA [mouse monoclonal, clone 308/3-3 from ImmunoTools, Friesoythe, Germany; western blotting (WB): 1:2000, flow cytometry (FC): 1:200]; GFP (mouse monoclonal, clone JL8, lot A5033481 from Clontech, Mountain View, CA, WB: 1:5000); RFP (used to detect mCherry, mouse monoclonal, clone 6G6, lot 51020014AB-01 from Chromo Tek, Martinsried, Germany, WB: 1:2000); tubulin (mouse monoclonal, clone E7 from Developmental Studies Hybridoma Bank, University of Iowa, IA, WB: 1:1000); Rac (mouse monoclonal #PA5-17519 from Thermo Fisher Scientific, WB: 1:1000); CYRI-B (mouse monoclonal, clone D8, lot H2817, from Santa Cruz Biotechnology, WB: 1:500); phospho-PAK1-(Thr423)/phospho-PAK2-(Thr402) (rabbit polyclonal #2601, lot 18 from Cell Signaling Technology, WB: 1:1000); PAK1 (rabbit polyclonal #2602 from Cell Signaling Technology, WB: 1:1000); anti-N. gonorrhoeae/anti-N.meningitidis (rabbit polyclonal; IG-511, custom made by Immunoglobe, Himmelstadt, Germany; Kuespert et al., 2011). Secondary antibodies were horseradish peroxidase (HRP)-conjugated goat anti-mouse-IgG (WB: 1:10,000) and HRP-conjugated goat anti-rabbit-IgG (WB: 1:5000) were from Jackson ImmunoResearch (Baltimore, MD).

The plasmid pLL3.7 encoding a CMV-promoter driven expression cassette of GFP was Addgene plasmid #11795, deposited by Luk Parijs (Rubinson et al., 2003). The plasmid plentiCRISPR v2 sgRNA-Cerulean containing a Cerulean targeting sgRNA has been described previously (Grimm et al., 2020). Human GeCKOv2 CRISPR knockout pooled library was Addgene plasmid #1000000048 (deposited by Feng Zhang) and the A part of the pooled library was used in the screen (Sanjana et al., 2014). The CYRI-B-targeting plasmid pLentiCRISPRv2-sgCYRI-B was generated by digesting pLentiCRISPRv2 (Addgene plasmid #52961, deposited by from Feng Zhang) with BsmBI and insertion of a CYRI-B targeting oligonucleotide (using the annealed sense oligonucleotide 5′-CACCGGACATGCACAGACCTTGAGC-3′ and antisense oligonucleotide 5′-AAACGCTCAAGGTCTGTGCATGTCC-3′).

To re-express CYRI-B in CYRI-B ko cells the CYRI-B ORF was amplified by PCR from HsCD00369072 (PlasmID, Harvard Medical School repository) with primers (forward, 5'-ACTCCTCCCCCGCCATGGGGAATCTTCTTAAAGTTTTGAC-3′ and reverse, 5'-CCCCACTAACCCGTTGCAGCATGGATTTAATTTGC-3′) containing sequences compatible with ligase independent cloning (LIC) into the vector pDNR-Dual-LIC (Adrian et al., 2019). To render CYRI-B insensitive to the CRISPR gRNA used to create the knockout cell line, site-directed mutagenesis was performed with overlapping primers (forward, 5′-CTTCTTAAAGTTTTGACCTGTACCGATTTGGAACAGGGGCCAAATTTTTTCC-3′ and reverse, 5′-GGAAAAAATTTGGCCCCTGTTCCAAATCGGTACAGGTCAAAACTTTAAGAAG-3′). The resulting construct contains five silent mutations in the CRISPR gRNA-binding region in CYRI-B. This Cas9-insensitive mutant (CYRI-wt) was used as a template for site-directed mutagenesis to create the G2A mutant (forward, 5′-CTCCTCCCCCGCCATGGCCAATCTTCTTAAAGTTTTG-3′ and reverse, 5′-CAAAACTTTAAGAAGATTGGCCATGGCGGGGGAGGAG-3′) and the R160D/R161D mutant (forward, 5′-GAATGATTTCAGCTATTATGATGACACATTGAGTCGTATGAGG-3′ and reverse, 5'-CCTCATACGACTCAATGTGTCATCATAATAGCTGAAATCATTC-3′). To transfer the Cas9-resistant versions of CYRI-B WT and R160D/R161D into a lentiviral expression vector, the ORFs were amplified by PCR (forward, 5′-GATACCCGGGATGGGGAATCTTCTTAAAGTTTTGAC-3′ and reverse, 5'-GAGTGAATTCTCATTGCAGCATGGATTTAATTTGC-3′) as were CYRI-G2A and CYRI-G2A/R160D/R161D (Forward, 5′-GAATCCCGGGATGGCCAATCTTCTTAAAGTTTTGAC and reverse: 5'-GAGTGAATTCTCATTGCAGCATGGATTTAATTTGC-3′) using primers with SmaI and EcoRI restriction sites. The PCR products were ligated in SmaI and EcoRI of the lentiviral vector pUltra-Hot (Addgene plasmid #24130, deposited by Malcolm Moore).

HL60 cells (CCL-240, ATCC, Manassas, Virginia) and the derived HL60 CEACAM3 mKate cells (Adrian et al., 2019) were cultured in RPMI 1640 medium (Gennaxxon, C4095) supplemented with 10% fetal calf serum (FCS; Gibco, 10279-106, lot no. 2166469), maintained at 37°C and 5% CO₂ and were split every 2–3 days. HL60 Cerulean cells were generated by lentiviral transduction of HL60 cells with a Cerulean-encoding lentivirus (Goob et al., 2022). At 5 days after transduction, single Cerulean-positive cells were sorted on a FACS AriaIII (BD Biosciences) into 96-well plates and a single-cell-derived HL60 Cerulean clone was expanded.

Human embryonic kidney 293 T cells (293 T; American Type Culture Collection CRL-3216) were grown in DMEM (Gennaxon, C4036) supplemented with 10% calf serum (Biowest, S0750, batch S00O510006). For transient transfection, 293 T cells were seeded at 25% confluence the day before and transfected using calcium phosphate precipitation.

The cell lines were regularly tested by PCR according to Uphoff and Drexler (2005) for the presence of mycoplasma contamination and confirmed to be negative. They were not further authenticated.

Lentiviral particles were produced by transient transfection of 293 T cells with the plasmids pMD2.G (Addgene #12259) (3.5 μg), psPAX2 (Addgene #12260) (5 μg), and the respective lentiviral vector (6.5 µg). At 3 days after transfection, the virus-containing supernatant was collected, sterile filtered (pore size 0.45 μm) and added to the recipient cells, which also received 8 μg/ml hexadimethrine bromide. Virus-infected HL60 cells were centrifuged at 800 g for 1 h to spin the viral particles onto the cells (spinfection). pLentiCRISPRv2 transduced cells were selected with 0.3 µg/ml puromycin for 5 days before bacterial uptake and sorting by FACS.

10⁶ cells were taken up in FACS buffer (PBS, 5% FCS) and transferred to Eppendorf tubes. Cells were centrifuged at 650 g for 2 min at 4°C and incubated with the primary antibody in FACS buffer for 1 h on ice. Samples were washed three times with FACS buffer and then incubated with the fluorescently labeled secondary antibody in FACS buffer for 30 min at 4°C in the dark. Cells were washed as before, suspended in 600 µl FACS buffer, and transferred to FACS tubes. Finally, cells were analyzed by flow cytometry (BD LSRII, FACSDiva software; BD Biosciences) and obtained data were evaluated using the FlowJo software package.

Cells were lysed in RIPA buffer (0.1% w/v SDS, 1% sodium deoxycholic acid, 1% Triton X-100, 50 mM Hepes, 150 mM NaCl, 10% glycerol, 1.5 mM MgCl₂, 1 mM EGTA, 10 mM sodium pyrophosphate and 100 mM NaF) or NP40 lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% NP40, 1 mM EDTA, 50 mM NaF) in the presence of protease inhibitors (aprotinin from Roth, Karlsruhe, Germany, and leupeptin and PMSF from Applichem, Darmstadt, Germany), 1 mM sodium orthovanadate and 10 mM of para-nitrophenol phosphate. After centrifugation at 18,000 g and 4°C for 30 min, cleared whole-cell lysates were taken up in SDS buffer and boiled at 95°C for 5–10 min. Proteins were separated by SDS-PAGE and transferred to PVDF membranes. Subsequently, membranes were blocked in blocking buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.05% Tween 20 and 2% BSA) for 1 h at room temperature (RT) and then incubated with primary antibodies (diluted in blocking buffer) overnight at 4°C. The next day, membranes were washed three times with TBS-T (50 mM Tris-HCl pH 7.5, 150 mM NaCl and 0.05% Tween 20) and incubated with the HRP-conjugated secondary antibody (in TBS-T) for 1 h at RT, before ECL detection on a Chemidoc Touch Imaging System (Bio-Rad, CA). Full-size scans of all western blots used to generate the cropped panels depicted in the main figures are compiled in Fig. S3.

N. gonorrhoeae MS11-B2.1 strains (non-piliated Opa₅₂-expressing strain N309 and non-piliated, non-opaque strain N302) (Kupsch et al., 1993) were grown on GC agar plates supplemented with vitamins. Bacteria were cultured at 37°C and 5% CO₂ and freshly streaked on plates 16–20 h before experiments.

10⁶ CEACAM3–mKate2-expressing HL60 cells were suspended in 1 ml phagocytosis buffer (PBS, 0.9 mM CaCl₂, 0.5 mM MgCl₂, 5 mM glucose, 1% heat-inactivated FCS). Opa₅₂-expressing N. gonorrhoeae (Ngo Opa₅₂) or non-opaque gonococci (Ngo Opa-) were re-suspended in 1 ml PBS, stained by addition of 2 μg/ml 5/6-carboxyfluorescein succinimidyl ester (CFSE; Molecular Probes, Karlsruhe, Germany) for 20 min with constant shaking in the dark, and then washed three times with PBS (4 min, 1700 g). For opsonization with IgG, CFSE-stained Ngo Opa⁻ were incubated 30 min at 37°C with 20% heat-inactivated rabbit anti-N.gonorrhoeae serum (IG-511) and subsequently washed two times with PBS. Cells were infected for the indicated time at the indicated MOI under gentle rotation at 37°C. To stop bacterial uptake, samples were washed with ice-cold PBS, taken up in ice-cold phagocytosis buffer and mixed with Trypan Blue (final concentration 0.2 mg/ml) to quench the CFSE signal derived from extracellular bacteria (Pils et al., 2006). Bacterial uptake was analyzed by flow cytometry using a FACS Fortessa (BD Biosciences), and raw data were analyzed using FlowJo software.

To discriminate between extracellular and internalized bacteria, FITC–biotin–avidin staining was employed as described (Agerer et al., 2004). To this end, HL60 cells were seeded onto poly-L-lysine-coated coverslips and allowed to adhere overnight in RPMI with 1% FCS. Opa₅₂-expressing N. gonorrhoeae were washed with PBS, stained with 2 μg/ml 5/6-carboxyfluorescein succinimidyl ester (CFSE; Molecular Probes, Karlsruhe, Germany) and simultaneously surface biotinylated with Sulfo-NHS–LC–biotin (Perbio Science, Bonn, Germany). After washing the bacteria three times with PBS, the labeled bacteria were used to infect the adherent cells for 15 min at a MOI 30. Infected samples were washed with cold PBS and fixed with 4% PFA in PBS for 20 min and washed three times with PBS. Extracellular bacteria were counter-stained with streptavidin–Alexa-Fluor-647 (10 µg/ml, Molecular Probes, S3235) for 1 h at RT in the dark. Samples were washed three times with PBS and mounted on glass slides using Dako Mounting Medium (Dako). Finally, samples were imaged using a Leica SP5 confocal microscope to detect intracellular bacteria (CFSE staining only) and extracellular bacteria (CFSE and Alexa-Fluor-647 staining).

Genomic DNA was isolated from the input cell population and sorted cell populations with a genomic DNA isolation kit (PureLink) and used as template (125 ng per reaction) in multiple PCRs (Phusion polymerase, HF buffer, Thermo Fisher Scientific) using primers flanking the sgRNA sequences (forward primer 5′-CTTGAAAGTATTTCGATTTCTTGGC-3′ and reverse primer 5′-CCTTTCAAGACCTAGCTAGCG-3′). After initial denaturation (90 s, 98°C), 30 cycles of amplification were performed (25 s, 57°C; 7 s, 72°C; 10 s, 98°C). PCR products were isolated from an agarose gel (1.5%) and extracted using a gel extraction kit (Genaxxon). Eluates were sent for sequencing on an Illumina sequencing platform (2×150 PE, Eurofins). The sequencing data have been deposited at the NCBI Sequence Read Archive under the BioProject ID PRJNA957296, containing three separate Biosamples SAMN34249396 (input population), SAMN34249397 (low phagocytosis) and SAMN34249398 (high phagocytosis). Raw paired-end reads were merged using FLASH (Magoc and Salzberg, 2011). Read abundances per gene were determined in each sample and normalized for the total number of reads (normalized read abundance). Finally, the normalized read abundance in the ‘low’ and ‘high’ phagocytic samples were divided by the normalized read abundance of the ‘input population’ sample to obtain the enrichment ratio.

Opa₅₂-expressing N. gonorrhoeae were fixed in 4% paraformaldehyde-containing PBS for 20 min and then washed three times with PBS. Fixed bacteria were added to HL60 cell lysates and rotated at 4°C for 3 h. Bacteria-bound CEACAMs were precipitated at 2650 g for 5 min, washed twice with PBS, and boiled in 2× Laemmli loading buffer.

HL60 cells were seeded in triplicate into poly-L-lysine-coated 12-well plates at 3×10⁵ cells/well in serum-free medium and infected for 15 min at an MOI of 30 with Opa₅₂-expressing N. gonorrhoeae (Opa₅₂) or N. gonorrhoeae that do not express Opa proteins (Opa⁻). Extracellular bacteria were killed by a 45-min incubation in culture medium containing 50 μg/ml gentamicin. Infected cells were lysed with 0.5% (w/v) saponin in PBS for 15 min, and released bacteria were plated on GC agar plates at various dilutions. The re-isolated colony forming unit (cfu) values were determined and statistical analysis of three independent experiments was performed using a one-way ANOVA followed by a Dunnett's post test.

Rac activity was determined with a pulldown assay using the recombinant PAK p21-binding domain (PBD) fused to GST (Schmitter et al., 2004). For each pulldown, 1×10⁷ HL60 cells were lysed (50 mM Tris, pH 7.5, 200 mM NaCl, 5 mM MgCl₂, 1 mM DTT, 10% glycerol, 1% NP40, 1 mM PMSF, 10 µg/ml aprotinin and 10 µg/ml leupeptin) on ice for 20 min and centrifuged for 10 min at 4°C at 20,000 g. Supernatant containing equal amounts of protein were transferred to a new tube, input samples were taken and 25 µl Sepharose–PBD beads (10 µg GST–PBD) were added. After binding for 1 h at 4°C on a rotator (15 rpm). Beads were washed three times with PBD wash buffer (25 mM Tris-HCl pH 7.5, 40 mM NaCl, 30 mM MgCl₂, 1 mM DTT, 1%, 1 mM PMSF, 10 µg/ml aprotinin and 10 µg/ml leupeptin) and subsequently boiled in 2× Laemmli loading buffer. Input samples and PBD pulldown samples were run on the same SDS-PAGE gel followed by western blotting.

HL60 cells (0.5×10⁶) were fixed in 4% PFA for 30 min at 37°C on a rotator (15 rpm), washed with PBS and subsequently permeablized for 3 min at RT (PBS with 0.1% TritonX-100 and 5% BSA). F-actin was stained with phalloidin-conjugated Alexa-Fluor-488 (Molecular Probes, 1:50 in PBS with 0.5% BSA) for 45 min at RT. Cells were washed twice with PBS and resuspended in FACS buffer. The mean fluorescence intensity (MFI) of 10,000 cells per sample was determined by FACS analysis.

HL60 cells were seeded at 4×10⁴ cells/well in phagocytosis buffer on 12 mm glass coverslips coated with poly-L-lysine (10 mg/ml). Cells were infected for 30 min with Opa₅₂-expressing N. gonorrhoeae at an MOI of 30. Samples were fixed at 4°C for a total of 60 min in EM fixative (0.1 M HEPES, pH 7.2, 3% formaldehyde, 2% glutaraldehyde, 0.09 M sucrose, 0.01 M CaCl₂ and 0.01 M MgCl₂) that was renewed one time. Fixed samples were washed three times with washing buffer (0.1 M HEPES, pH 7.2, 0.09 M sucrose, 0.01 M CaCl₂ and 0.01 M MgCl₂) and dehydrated in an ethanol gradient, critical point dried with liquid CO₂ (BAL-TEC SCD 030 Critical Point Dryer) and sputter-coated with 8 mm gold-palladium. Samples were imaged at 5 kV in a field emission scanning electron microscope (Auriga; Carl Zeiss AG, Jena, Germany).

Data are presented as mean±s.d. as indicated in the figure legend. Statistical significance was determined with either an unpaired two-tailed Student's t-test or a one-way ANOVA followed by a Dunnett's post test, using Prism5 (GraphPad). The significance level was set at P<0.05 and is indicated by *P<0.05, **P<0.01, ***P<0.001, or ns, not significant. Further statistical details are provided in the figures and figure legends.

We thank S. Feindler-Boeckh and C. Hentschel for expert technical assistance. We are grateful to the Flow Cytometry facility of the Department of Biology, Konstanz University (FlowKon) and Michael Laumann (Electron Microscopy Service of the Department of Biology, University of Konstanz) for support and sample analysis.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.260771.reviewer-comments.pdf.