Could β-Lactam Antibiotics Block Humoral Immunity?

It has been reported that treatment with β-lactam antibiotics induces leukopenia and candidemia, worsens the clinical response to anticancer immunotherapy and decreases immune response to vaccination. β-lactamases can cleave β-lactam antibiotics by blocking their activity. Two distincts superfamilies of β-lactamases are described, the serine β-lactamases and the zinc ion dependent metallo-β-lactamases. In human, 18 metallo-β-lactamases encoding genes (hMBLs) have been identified. While the physiological role of most of them remains unknown, it is well established that the SNM1A, B and C proteins are involved in DNA repair. The SNM1C/Artemis protein is precisely associated in the V(D)J segments rearrangement, that leads to immunoglobulin (Ig) and T-cell receptor variable regions, which have a crucial role in the immune response. Thus in humans, SNM1C/Artemis mutation is associated with severe combined immunodeficiency characterized by hypogammaglobulinemia deficient cellular immunity and opportunistic infections. While catalytic site of hMBLs and especially that of the SNM1 family is highly conserved, in vitro studies showed that some β-lactam antibiotics, and precisely third generation of cephalosporin and ampicillin, inhibit the metallo-β-lactamase proteins SNM1A & B and the SNM1C/Artemis protein complex. By analogy, the question arises as to whether β-lactam antibiotics can block the SNM1C/Artemis protein in humans inducing transient immunodeficiency. We reviewed here the literature data supporting this hypothesis based on in silico, in vitro and in vivo evidences. Understanding the impact of β-lactam antibiotics on the immune cell will offer new therapeutic clues and new clinical approaches in oncology, immunology, and infectious diseases.


INTRODUCTION
b-lactam antibiotics remain the cornerstone of the antibacterial arsenal, being the most prescribed and the most important class of antibiotics in terms of sales (1)(2)(3). In clinical practice, b-lactam antibiotics are increasingly used to treat pulmonary, urinary, skin, osteo-articular and cardiovascular infections (2,4). However, parenteral use of b-lactam antibiotics was associated, in human, with an increased risk of complications such as candidemia, especially in the presence of catheters (5). In addition, its utilization prior to anticancer immunotherapy was associated with a worse clinical response (6)(7)(8). Nevertheless the direct effect of b-lactam antibiotics on immune cells remains poorly understood (9)(10)(11)(12)(13)(14)(15).
Some b-lactam antibiotics showed direct effect on hMBLs. Cephalosporins were identified as competitive inhibitors of the hMBLs SNM1A, B and C nuclease (32). Ampicillin and cephalosporin, b-lactamin antibiotics, can in vitro inhibit the SNMIC/Artemis protein (33,34). By analogy, this raises the question whether b-lactam antibiotics can induce reversible humoral deficiency in humans by transiently blocking the SNM1C/Artemis nuclease activity. As the conserved catalytic site is highly similar in SNM1A, B and C proteins, we postulate that the b-lactam antibiotics would be capable of inhibiting the SNM1C/Artemis protein leading to a transient immune deficit (humoral) (33,35). Here, after introducing the SNM1C/Artemis protein functionality and implications, we will examine the in silico, in vivo and in vitro evidences to support our hypothesis. We searched in Medline and Google scholar for references with no language restrictions and no time limitations nor status. We performed a docking analysis to evaluate in silico the cephalosporin affinity with the SNM1C/Artemis catalytic site.

THE HMBLS SNM1C/ARTEMIS PROTEIN FUNCTIONALITY, CLINICAL IMPLICATIONS AND CATALYTIC SITE
The SNM1-family proteins, that belongs to the hMBLs, is characterized by a conserved catalytic site with nuclease activity implicated in the maintenance of the genome stability (36). SNM1 is a member of the SNM1 (or PSO2) gene family that encodes proteins involved in DNA processing, DNA metabolism and cell cycle regulation. Whereas SNM1A and B contribute to the intercross linked repair, SNM1C/Artemis contributes to the double-strand break repair. Indeed, SNM1C/Artemis possesses an exonuclease activity in vitro and when complexed and phosphorylated by DNA PKcs, SNM1C/Artemis' specificity switched to an endonuclease activity, which is involved in the V(J) and V(D)J segments rearrangement (37).
The hMBLs SNM1C/Artemis Protein and Its Catalytic Site SNM1C/Artemis, a 78 k-Da protein with 692 amino acids, is a member of the metallo-b-lactamase superfamily of nucleases, characterized by a conserved MBL and b-CASP domain. It belongs to a distinct group of the b-CASP family of nucleases that includes 3 proteins : SNM1A/Pso2 related proteins, SNM1B/ Apollo, SNM1C/Artemis (38). The 3-dimensional structure of the catalytic domain of recombinant human SNM1C/Artemis has been recently solved ( Figure 1A) (35). The SNM1C/Artemis' active catalytic site contains metallo-b-lactamase, b-CASP domain and a cluster of conserved histidine and aspartate residues capable of binding two metals atoms. These particularities of the active catalytic site are similar to the other members of the DNA cross linked repair gene SNM1 family and mRNA 3'end processing endonuclease (35). As observed in the SNM1A protein, SNM1C/Artemis contains only one zinc ion in its catalytic site, the metallo-b-lactamase domain. The b-CASP domain acts as a plug, covering the active site displayed by the MBL domain, and creates the substrate binding pocket, which should confer substrate selectivity (35,38). As particularities, the SNM1C/Artemis nuclease presents a second zinc ion in the b-CASP domain reorienting the putative DNA binding surface and extending the substrate binding pocket to a new pocket, the pocket III ( Figure 1A) (35). This substrate binding site, pocket III, appears to be deeper and wider in the SNM1C/Artemis nuclease than in SNM1A and SNM1B proteins (35). In addition and contrary to the SNM1A and SNM1B proteins, SNM1C/ Artemis exhibits a dominant and extensive positive charge in its substrate-binding surface, probably because of the structurally distinct DNA substrate of this protein ( Figure 1A) (35). Mutation on the His-254 residue disrupts the in vivo and in vitro function of SNM1C/Artemis and results in the radiosensitive severe combined immune deficiency in human (see chapter V) (31,35). His-254 residue is located within the catalytic metallo-b-lactamase domain and has been proposed to be involved at the metal ions coordination. It thus indicates that the unique zinc binding motif in the b-CASP domain of SNM1C/ Artemis, which coordonate His-228, His-245, His 256 and Cys 272 residues, is both structurally and functionally important (35). Interestingly, the His-254 residue is conserved within the SNM1 family (31,38).

Artemis/SNM1C Protein and the V(J) and V(D)J Segments Rearrangement
A central part of the immune system is played by antibodies that recognize and distinguish specific molecular patterns of antigens. Structure of antigens are different and the repertoire is currently estimated to be~10 15 members, for the naïve repertoire, and higher than 10 18 based on the theorical repertoire combination (39). The variable region of immunoglobulins (VH and VL) and T-cell receptors (Va, Vb, Vg, and Vd) are encoded by a modest region of gene segment whose diversity is achieved by a series of programmed DNA rearrangment termed the V(J) and V(D)J rearrangements (40)(41)(42). Immunoglobulin (Ig) or T-cell receptor variable regions are assembled in developping B or T lymphocytes from germline variable (V), diversity (D), and joining (J) gene units (42)(43)(44). While the V(D)J recombination segment is involved in the TCR b and d and in the Ig heavy chain (IgH) variables exons, the V(J) recombination is involved in the TCRa and g and Ig light chain (IgL) variables exons (43). These recombinations could be summarized in 3 different phases ( Figure 2) (42,43,45,46). The first one is lympoid specific and constitutes the initiation of the recombination. Two protein, RAG1 and RAG2, encode by the recombination activated gene 1 and 2 (RAG1 and RAG 2), recognize the recombination signal sequence (

SNM1C/Artemis Mutation in Humans
The Artemis autosomic recessive mutation induces default in the V(J) and V(D)J rearrangement causing humans RS-SCID phenotype characterized by an increase sensitivity to ϒ rays irradiation ( Figure 2). SNM1C/Artemis gene mutation is rare in Europe, < 1/ 500 000 births, but its incidence is much higher in the Athabascanspeaking, Navajo and Apache Native Americans involving 1/2000 births (49). First cases of severe combined immunodeficiency were reported in Athabascan-speaking American Indian Children, who presented B-cells and T-cells deficiency associated with oral and/or genital ulceration (50). Clinical presentations were also characterized by hypogammaglobulinemia and poor antibody response (   inflammation, lymphoproliferative diseases and lymphoma malignancies with later clinical manifestations, in childhood or adulthood (37,53,58,60). While the hMBLs SNM1C/Artemis mutations are associated with immunodeficiency, by analogy, reversible blockage of this protein may induce transient acquired immunodeficiency affecting the humoral response. The impact of cephalosporins on the immune response may have some similarities to the immune deficiencies seen in patients with an innate SNM1C/ Artemis mutation. However, it is assumed that the impact of cephalosporins on the immune response is reversible and less severe than that seen in innate SNM1C/Artemis mutations. In the next chapter we will discuss the evidences arguing for a SNM1C/ Artemis blockage by cephalosporins. In the next chapter we will discuss the evidences arguing for a SNM1C/Artemis blokage by cephalosporins.

b-LACTAM ANTIBIOTICS CAN INHIBIT THE hMBLs SNM1 Fold
Some studies have demonstrated that b-lactam antibiotics inhibit the SNM1 (A, B & C) and MBLAC1. First, in vitro experiment showed that ampicillin inhibit SNM1C/Artemis nuclease activity (33). Li et al. have evaluated the exo-and endo-nuclease activities of the SNM1C/Artemis protein in presence of ampicillin (33). Incubating the singlestranded DNA substrate with SNM1C/Artemis and ampicillin, they observed low levels of SNM1C/Artemis exonuclease and endonuclease products, in an ampicillin dose-dependent manner (33). Furthermore, as the active architecture site of bacterial MBLs and SNM1 proteins are similars, and as b-lactam antibiotics are the target of bacterial MBLs, Lee et al. screened the b-lactam antibiotics candidates inhibitors of SNM1A/B (32). Incubating 20 nucleotides single-stranded DNA substrates modified with fluorophore in the presence of the SNM1A/B proteins and cephalosporins, they observed, analysing the resultant fluorescent subtrate, that ceftriaxone acts as a competitive reversible inhibitors of the SNM1A/B protein (32). While the in vitro inhibitory concentration of ceftriaxone necessary to block SNM1A and SNM1B is 1mg/L, the plasmatic concentration reached in patients treated with ceftriaxone is much higher (32,33). Interestingly and even more recently, a glial-expressed Caenorhabditis elegans gene, swip-10, an ortholog of the mammalian MBLAC1 gene, that encodes for a metallo-b-lactamase domain-containing protein limiting glutamate-dependent changes in dopamine neuron excitability, showed high affinity with ceftriaxone (61). At last, using a real time fluorescence-based nuclease assay, ceftriaxone was identified as inhibiting the nuclease activity of the SNM1/C Artemis with a modest IC50 of 65 mM (34). The ability for b-lactam antibiotics to interact with MBLs, whether in humans or other animals, was thus demonstrated in vitro. We will next review the in silico and in vivo evidences arguing for the possible SNM1C/Artemis inhibition by b-lactam antibiotics ( Table 3).

IN SILICO PROOF OF CONCEPT: CEPHALOSPORIN CAN INTERACT WITH SNM1C/ARTEMIS
In order to test our hypothesis we evaluate in silico the affinity of the SNM1C/Artemis protein with cephalosporin. The analysis showed a strong affinity of SNM1C/Artemis with cephalosporins predicting their possible interaction.

Docking: Artemis Affinity With Cephalosporin
We performed a docking based on the three-dimensional structure of SNM1C/Artemis catalytic domain that was solved and published by Karim et al. (PDB: 6WO0) (35). They also described the pocket III structural feature of the SNM1C/ Artemis as the putative DNA-binding site, that extends the conserved substrate-binding site and is made of mostly positively charged residues on the surface ( Figure 1A). The interaction between the SNM1C/Artemis catalytic domain and the ceftriaxone b-lactam antibiotic could be modeled through molecular docking ( Figures 1B, C) based on the previously mentioned structure (35). In our model, the cephalosporin appeared to preferentially bind within the conserved active site, mostly constituted of negatively charged residues on the surface, as shown with the electrostatic potential surface ( Figure 1C). A binding energy of -3.84 kcal/mol could be predicted using the AutoDockTools software, suggesting promising affinity. This docking analysis is supported by a recent structural proposition, in which ceftriaxone binds to the SNM1C/Artemis protein surface (34).

THE IN VIVO IMPACT OF b-LACTAM ANTIBIOTICS ON IMMUNITY, A REVIEW OF EVIDENCES
Based on a review of the literature, we evaluate the evidences that b-lactam antibiotics have an impact on humoral immunity in animals and humans.

b-Lactam Antibiotics Decrease the Humoral Response in Animals
Reports from veterinarians identified a decrease humoral response in animals treated with b-lactam antibiotics. In pigs, ceftiofur hydrochloride administrated concomitantly to vaccine induced a significant delay in the detection of antibodies in response to pseudorabies virus vaccine and a significant decrease of antibodies levels in response to the swine influenza virus, when compared to controls (65). Furthermore, the immunotoxicity of intravenous cephalosporins (ceftizoxime, ceftezole and DQ2556) was demonstrated in mice, in which an increased spleen weight and enlargement of splenic germinal centres were observed 5 to 7 days after treatment (66). Cephalosporins are described as affecting cellular and humoral response (66). The response to pneumococcal type III (S3) polysaccharides, a T-cell independent immunogen, and the response to sheep red blood cells, a T-cell dependent immunogen, were decreased in mice treated 5 days with cephalosporins (66). In addition, the IgG level in the spleen of mice treated with cephalosporins and immunized with red blood cells was much lower than that of mice treated with penems (67). Finally, BALB/c mice treated with oral ceftriaxone showed a reduction in activated B and T cells in gut-associated lymphoid tissue (GALT), Peyer's patches and mesenteric lymph nodes, as well as a significant reduction in mucus IgG and IgA (68,69). Therefore, animals treated with cephalosporins were not able to mount a sustained protective immune response since a deficient humoral response was observed.
This may result, at least in part, on the blockage of the SNM1C/Artemis protein leading to reduced and delayed production of antibodies.

Clinical and Experimental Reports in Human
In humans, several observations suggest the use of b-lactam antibiotics as a trigger for immunodeficiency, reducing the total number of leukocytes, promoting opportunistic infections and decreasing the response to anticancer therapy with immune checkpoint inhibitors.
In a retrospective analysis of 291 patients treated with antibiotics, leukopenia in addition to neutropenia, thrombopenia and eosinophilia were identified as adverses events of parenteral antibiotics use (70). Interestingly, leukopenia was mainly observed in patients receiving b-lactam (39%) as the only antibiotic (70). Antibiotics may also facilitate opportunistic infections commonly encountered in immunocompromised patients since the use of b-lactam antibiotics has been reported to be significantly associated with an increased risk of candidemia in neonates, infants and adults (Table 3) (5,60,62,63). If the importance of the humoral response in the control of candidiasis has been discussed, and is supported by several in vitro and in vivo observations (71)(72)(73)(74). Anti MP-65 antibodies were related with survival in patients with severe disseminated infection and the administration of human polyvalent IgG mouthwash showed significant reduction of infection in patients with chronic oral candidiasis (62,71).
In cancer patients, the use of antibiotics before treatment with immune checkpoint inhibitors is significantly associated with relapse and worse clinical outcome (8,75,76). Several types of cancer are concerned: renal cell carcinoma, non small cell lung cancer, urothelial carcinoma and melanoma (8,75,76). Interestingly, all authors of these studies agreed that b-lactam antibiotics were the most widely used antibiotics (35 to 45%) followed by quinolones and macrolides (8,75,77). It would have been interesting to explore the lymphocyte pool of cancer's patients receiving ceftriaxone to test our hypothesis.
It should be noted that the direct impact of cephalosporin on Bcells is probably not the only risk factor that could explain the occurrence of candida infection since catheters, microbiota composition and TH17 cell response are clearly involved in these infections (5,63,78,79). Nonetheless, these in vivo observations support the possible immunomodulatory and even immunosuppressive role of b-lactam antibiotics on immune cells favouring the occurrence of opportunistic infections.

DISCUSSION
b-Lactam Antibiotics May Block the SNM1C/Artemis, hMBLs, Activity We postulate that b-lactam antibiotics can induce reversible humoral deficiency in humans by transiently blocking the SNM1C/Artemis nuclease activity and our hypothesis is sustained by in silico, in vitro and in vivo observations ( Figure 3 and Table 3   candidal infections should be actively sought and routinely reported in the event of fever under these antibiotics, the use of catheters should, in addition, be limited.

Further Evidences Are Needed to Confirm This Hypothesis
In vitro experimentations are needed to demonstrate that b-lactam antibiotics, such as cephalosporin, may block the SNM1C/Artemis protein inhibiting the substrates production of this exo and endonuclease. In vivo, testing the ability of mice to produce an effective vaccine response when treated with ceftriaxone, in addition to the evaluation of their B and T cell maturated pool will provide answers to these questions. In humans on cephalosporin, humoral immunity can be studied simply by assessing the vaccine response. Further research evaluating the humoral response, in particular against C. albicans, in patients receiving b-lactam antibiotics, would shed light on our hypothesis. We have focused here on hMBLs but MBLs are present in the world of life, in bacteria, eukaryotes and viruses, so the spectrum of b-lactam may be broader than expected and may also involve some MBLs.

Metallo b-Lactamase in the World of Life
It is important to point out that MBLs are widely distributed in nature. They are found in bacteria, archae, giant viruses, and eukaryotes cells including humans. Thus a landscape of cross enzymatic activities among MBLs enzyme superfamily was reported (31,77,(80)(81)(82)(83). Therefore, MBLs' cross enzymatic activities (hydrolase and nuclease) among different domains of the living were identified, whereas different enzymatic activities observed may be related to divergent host substrates (17). The interspecies common enzymatic activities can be explained by the highly conserved MBLs catalytic site ( Figure 4) (81, 84, 85). Thus, while the catalytic site of MBL is highly conserved, all known enzymatic activities of MBL could possibly be inhibited by b-lactam antibiotics (Figure 3). In humans, blocking the enzymatic activities not only of the SNM1C/Artemis protein but of each of these 18 hMBLs may have a physiological and clinical relevance as yet unknown.
Microbiota and the Indirect Role of b-Lactam on the Immune Response, a Confounding Marker While our hypothesis supports the direct role of b-lactam on immune cells, the impact of these antibiotics via the microbiota on immunityi.e. their indirect impact on the immune response -cannot be neglected. In animals, a reduced antibody vaccinal response was observed in germ-free, antibiotic-induced dysbiosis and antibiotic-treated mice. Interestingly, these antibody  vaccinal responses were re-increased after oral reconstitution of the gut microbiota (9)(10)(11). In humans, the role of the gut microbiota was also highly suspected to modulate the vaccinal response. Bifidobacterium predominance enhanced humoral response to oral and parenteral vaccine (oral polio vaccine, Bacille calmette guerin vaccine, tetanus toxoid) (12,14,15). High abundance of Proteobacteria, Clostridium cluster XI, and Streptococcus bovis and low abundance of Bacteroidetes phylum were associated with an increase serological response to oral rotavirus vaccine (12,14,15). More recently, a reduced H1N1 vaccinal (IgG and IgA) response was reported in patients treated with neomycin, vancomycin and metronidazole (13). Thus, although the impact of b-lactam on vaccine response has not been specifically studied, antibiotics appear to modulate the humoral response to vaccination, and probably, in part, via the gut microbiota.
CONCLUSION b-lactam antibiotics direct impacts on the immune cells should be considered. Since antibiotics can inhibit SNM1A, B and C/ Artemis activities in vitro, and since the catalytic site of hMBL, in particular that of the SNM1 family, is very well conserved, we hypothesized that b-lactam antibiotics could reversibly inhibit the SNM1C/Artemis protein and block the V(D)J recombination. Our hypothesis is sustained by in vitro, in silico and in vivo evidences and needs to be further investigated. The putative interference of beta-lactams with SNM1C/Artemis nuclease activity would become clinically relevant in cases of long-term treatment, such as persistent osteoarticular, cardiovascular or cerebro-meningeal infections. Highlighting the physiological role and clinical impact of the hMBLs as the effect of b-lactam antibiotics on these hMBLs could pave the way for new clinical and therapeutic approaches in immunology, infectious diseases and oncology.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
CM and PP wrote the article. LP performed the in silico analysis. CD, J-LM, and DR supervised the manuscript. All authors contributed to the article and approved the submitted version.