Altered Phenotype and Enhanced Antibody-Producing Ability of Peripheral B Cells in Mice with Cd19-Driven Cre Expression

Given the importance of B lymphocytes in inflammation and immune defense against pathogens, mice transgenic for Cre under the control of Cd19 promoter (Cd19Cre/+ mice) have been widely used to specifically investigate the role of loxP-flanked genes in B cell development/function. However, impacts of expression/insertion of the Cre transgene on the phenotype and function of B cells have not been carefully studied. Here, we show that the number of marginal zone B and B1a cells was selectively reduced in Cd19Cre/+ mice, while B cell development in the bone marrow and total numbers of peripheral B cells were comparable between Cd19Cre/+ and wild type C57BL/6 mice. Notably, humoral responses to both T cell-dependent and independent antigens were significantly increased in Cd19Cre/+ mice. We speculate that these differences are mainly attributable to reduced surface CD19 levels caused by integration of the Cre-expressing cassette that inactivates one Cd19 allele. Moreover, our literature survey showed that expression of Cd19Cre/+ alone may affect the development/progression of inflammatory and anti-infectious responses. Thus, our results have important implications for the design and interpretation of results on gene functions specifically targeted in B cells in the Cd19Cre/+ mouse strain, for instance, in the context of (auto) inflammatory/infectious diseases.


Introduction
The Cre/loxP recombination system has been widely used to edit mammalian genomes in genetic and biomedical studies. Upon recognizing the 34-bp-long loxP motif inserted at defined positions of the genome, the recombinase Cre specifically and efficiently drives the recombination of DNA segments flanked by two loxP recognition sites ('floxed' locus) [1]. Hence, the regulation of Cre expression with inducible or cell/tissue-specific promoters represents an elegant and powerful approach to precisely interrogate the function of genes that have been inactivated or activated in a spatially and temporally specific fashion [2]. In particular, the Cre/loxP-mediated conditional-knockout system circumvents the indispensability of certain genes in embryonic development, and enables the elucidation of their functions in specific organs or cells in vivo.
Despite the widespread usage in experimental studies, the Cre/loxP system has a number of limitations/shortcomings [2,3]. Apart from the variable excision efficiencies of floxed loci, off-target Cre expressions/activities have been reported in several Cre lines, including CD11c-Cre and Thy1-Cre, where a wide range of cells/tissues are targeted beyond expectation [4][5][6]. Moreover, given that the mammalian genome comprises many cryptic/pseudo loxP sites, for instance, at an estimated frequency of 1.2 per megabase in the mouse genome, the mere expression of Cre is potentially toxic to cells and thus may result in reduced proliferation, aberrant DNA recombination, and chromosomal defects [7][8][9][10][11]. For instance, the expression of Cre driven by the promoter of Lck (encoding the protein tyrosine kinase p56) in thymocytes significantly reduces thymic cellularity and promotes the apoptosis of CD4 + CD8 + double-positive T cells [12]. In addition to the toxicity inherent to Cre recombinase, insertion of the Cre transgene into the genome may affect the expression of endogenous genes around the integration site through direct disruption of their sequences, and/or the trans-effect mediated by control elements present in the transgenic vector (i.e., promoters & enhancers, etc.) [13,14]. Interestingly, most Cre knock-in mouse strains seem to tolerate these aforementioned adverse effects, possibly through developmental selection and adaptation processes, and appear to be phenotypically indistinguishable from wildtype (WT) controls as described in the literature [2,3]. Therefore, inclusion of mice or cells expressing the Cre transgene only as a critical control is frequently ignored in the field, which may lead to misinterpretation of the data on gene functions.
The type I transmembrane protein CD19 is first upregulated in the bone marrow (BM) pro-B cells, and thereafter its surface density is maintained at high levels throughout B cell development and maturation stages until the terminally differentiated plasma cells (PCs) arise [15,16]. As such, CD19 is one of the most reliable surface lineage markers for B cells, and transgenic mice with Cre expression under the control of Cd19 promoter represent an excellent model to elucidate the roles of different genes in B cell development, differentiation, and function at steady states and/or in the context of infectious/inflammatory disorders. However, considering the disruption of one Cd19 allele by insertion of the Cre-expressing cassette in the commonly used Cd19 Cre/+ hemizygotes (hereafter referred to as Cd19 Cre/+ mice) originally generated by Rajewsky's group [17,18], and the involvements of CD19 in B cell signaling and function [15], we hypothesized that these Cd19 Cre/+ mice may differ from WT controls as a result of reduced CD19 expression and/or Cre-mediated side effects. In line with this, a few published studies showed that Cd19-driven expression of Cre might reduce BM pre-B cell numbers, alter follicular vs. marginal zone B (MZB) cell development, and hamper the survival of mature B cells in mice with certain genetic backgrounds [19][20][21].
Here we demonstrate that Cd19-driven expression of Cre alters the phenotype and function of peripheral B cells, albeit the comparable early BM B cell development, in the widely used C57BL/6 mice. The integration/expression of Cre transgene reduces the number of MZB and B1a cells, and augments antibody productions both in vivo and in vitro. Moreover, a survey of published studies using Cd19 Cre/+ to delete floxed-sequences in B cells indicated that results in some studies would have been less/more significant if CD19-Cre + , instead of CD19-Cre − , mice had been used as controls, validating the relevance and importance of our observations. As such, Cd19 Cre/+ mice or B cells are critical controls in studies using Cd19-driven Cre to specifically investigate gene functions in B cell biology, especially in infectious or inflammatory conditions involving innate-like MZB and B1 cells as well as antibody productions from conventional B2 cells.

Mice
The Cd19 Cre/+ mice on a C57BL/6 background, with one allele containing a Cre recombinase gene under the control of endogenous Cd19 promoter/enhancer elements, were kindly provided by Prof. Biao Zheng (East China Normal University, Shanghai, China). These mice were bred with WT C57BL/6 mice to obtain Cd19 Cre/+ and WT (Cd19 +/+ ) control littermates. Mice were housed under specific pathogen-free (SPF) conditions and used at 8-12 weeks of age unless otherwise indicated. All experiments were performed in accordance with procedures approved by the Animal Care and Use Committee of Soochow University.

Cell Isolation and Culture
BM cells were harvested by flushing the femurs of mice, and peritoneal cells were isolated by lavaging the peritoneal cavities with 10 mL PBS. Splenocytes were obtained by mechanically dissociating spleens in PBS, followed by passing then through a 70 µM nylon mesh (BD Biosciences, San Jose, CA, USA). Erythrocytes were removed by addition of ammonium chloride lysis buffer. Splenic CD19 + B cells were first enriched by negative selection with the MojoSort Mouse Pan B Cell Isolation Kit II (Biolegend, San Diego, CA, USA), and then were sorted out via a FACS Aria cell sorter III (BD Biosciences) with >95% purity.
For ELISPOT assays, PVDF membranes (MSIPS4510, Millipore, Billerica, MA, USA) were coated with NP 25 -BSA (1.5 µg/100 µL/well) in PBS overnight at 4 • C, and then blocked with culture medium (RPMI-1640 + 10% FBS) for 2 h at 37 • C. Splenocytes (5 × 10 5 /100 µL/well) were added into wells and cultured in an incubator for 6 hr. After washing the PVDF membrane with PBS containing 0.05% Tween-20 3 times, cells were lysed with distilled water for 10 min on ice, followed by washing with PBS twice. Finally, membranes were stained with HRP-coupled goat anti-mouse IgM, IgG1, IgG2c, or IgG3 (1:1000, SouthernBiotech) overnight at 4 • C before coloring with the AEC Substrate Set (BD Biosciences). The membranes were dried in the dark at room temperature, and spots were counted using a dissecting microscope.

Literature Survey
We analyzed 336 publications shown to cite the original article generating/characterizing the Cd19 Cre/+ mice in the PubMed website [18]. Seventy-one hits were discarded, as they were reviews/articles/book chapters that either did not include experiments with Cd19 Cre/+ mice or used Cd19 Cre/Cre homozygotes. Moreover, 9 publications using mice harboring one copy of Cd19 Cre/+ transgene for lineage tracing, imaging, or inducible depletion of B cells were excluded as well. The remaining 256 articles (Supplementary File S1), in which Cd19 Cre/+ was used to delete loxp-flanked sequences, were included in our analysis.

Statistical Analysis
The Mann-Whitney test was used to compare differences among groups by using GraphPad Prism 8.0 software (GraphPad, San Diego, CA, USA) and values at p < 0.05 were considered significant. χ 2 tests were employed to compare the results in the literatures using CD19-Cre − or CD19-Cre + mice as controls.

Comparable B Cell Development in the BM of Cd19 Cre/+ Mice
Although Cd19 Cre/+ mice, originally generated by Rajewsky's group, were phenotypically normal and widely used to specifically delete loxp-flanked sequence in B cells in the last two decades, the Cre expression cassette was inserted into the second coding exon, where it inactivated one allele of Cd19 and thus reduced the latter's surface expression levels [17,18]. Given that CD19 acts as a B-cell receptor (BCR) co-receptor, and may regulate B cell development [22][23][24][25], we performed a detailed comparison on the B cell development and phenotype in WT (Cd19 +/+ ) vs. Cd19 Cre/+ mice.
Therefore, we concluded that the single Cd19 Cre/+ knock in allele has no effect on early B cell development in the BM of mice, at least on the C57BL/6 background analyzed here.

Mildly Disturbed Phenotypes of B Cells in the Periphery of Cd19 Cre/+ Mice
In the periphery, no significant differences in the absolute cell numbers were noted in the blood, spleens, mesenteric lymph nodes, Peyer's patches and peritoneal cavities of WT vs. Cd19 Cre/+ mice. Within the B cell compartment in spleens, a trend of decreased IgM high IgD low B cells, containing transitional T1 cells, B1 and MZB cells, was observed in Cd19 Cre/+ mice (Supplementary Figure S1A,B). Moreover, given that the frequencies of T1 (AA4.1 + IgM high CD23 low ), T2 (AA4.1 + IgM high CD23 high ), and T3 (AA4.1 + IgM low CD23 high ) B cells, discriminated by IgM vs. CD23 expressions [26,27], did not differ significantly between these two groups of mice (Supplementary Figure S1C,D), these data suggested that Cd19 Cre/+ mice may harbor less B1 and/or MZB cells.

Mildly Disturbed Phenotypes of B Cells in the Periphery of Cd19 Cre/+ Mice
In the periphery, no significant differences in the absolute cell numbers were noted in the blood, spleens, mesenteric lymph nodes, Peyer's patches and peritoneal cavities of WT vs. Cd19 Cre/+ mice. Within the B cell compartment in spleens, a trend of decreased IgM high IgD low B cells, containing transitional T1 cells, B1 and MZB cells, was observed in Cd19 Cre/+ mice (Supplementary Figure S1A,B). Moreover, given that the frequencies of T1 (AA4.1 + IgM high CD23 low ), T2 (AA4.1 + IgM high CD23 high ), and T3 (AA4.1 + IgM low CD23 high ) B cells, discriminated by IgM vs. CD23 expressions [26,27], did not differ significantly between these two groups of mice (Supplementary Figure S1C,D), these data suggested that Cd19 Cre/+ mice may harbor less B1 and/or MZB cells.
Indeed, the number of B1 cells tended to decrease in both the spleens and peritoneal cavities of Cd19 Cre/+ mice (Figures 2 and S1E,F). In gated B1 cells, the percentage of CD5 -B1b cells increased at the expense of CD5 + B1a cells in the spleen (Figure 2A,B), peritoneal Indeed, the number of B1 cells tended to decrease in both the spleens and peritoneal cavities of Cd19 Cre/+ mice (Figures 2 and S1E,F). In gated B1 cells, the percentage of CD5 − B1b cells increased at the expense of CD5 + B1a cells in the spleen (Figure 2A,B), peritoneal cavity ( Figure 2C,D), and peripheral blood (data not shown) of Cd19 Cre/+ mice. Moreover, surface levels of CD5 on peritoneal total B1 cells, but not gated B1a or B1b cells, were significantly reduced in Cd19 Cre/+ mice ( Figure 2D), indicating that the attenuated CD5 expressions on peritoneal Cd19 Cre/+ B1 cells mainly resulted from an altered distribution of B1a vs. B1b subsets rather than a general decrease of surface CD5. Notably, only the numbers of B1a, but not B1b, cells were reduced in the spleen and peritoneal cavity of Cd19 Cre/+ mice (Supplementary Figure S1E,F). In addition, numbers of MZB cells were significantly reduced in Cd19 Cre/+ mice (Figures 2A,B and S1E), possibly due to the reduced surface CD19 intensities [23,25,28]. Interestingly, the frequencies of B1 and B1a cells re-turned to normal, while MZB cells remained decreased in older (16 weeks) Cd19 Cre/+ mice (Supplementary Figure S1G). bers of B1a, but not B1b, cells were reduced in the spleen and peritoneal cavity of Cd19 Cre/+ mice (Supplementary Figure S1E,F). In addition, numbers of MZB cells were significantly reduced in Cd19 Cre/+ mice (Figures 2A,B and S1E), possibly due to the reduced surface CD19 intensities [23,25,28]. Interestingly, the frequencies of B1 and B1a cells returned to normal, while MZB cells remained decreased in older (16 weeks) Cd19 Cre/+ mice (Supplementary Figure S1G).
Together, these data show that, despite the normal BM B cell development, Cd19 Cre/+ mice exhibit a mildly decreased generation/maintenance of B1a cells early in life (<3 months old), and a reduced number of MZB cells at least until 16 weeks of age (Figures 1  and 2).

Increased Antibody Levels in Cd19 Cre/+ Mice upon Immunization
Considering the roles of B1 and MZB cells in the production of protecting antibodies, we compared the total or antigen-specific sera antibody levels in Cd19 Cre/+ and control WT littermates before and after immunizations. The baseline antibody levels did not differ Together, these data show that, despite the normal BM B cell development, Cd19 Cre/+ mice exhibit a mildly decreased generation/maintenance of B1a cells early in life (<3 months old), and a reduced number of MZB cells at least until 16 weeks of age (Figures 1 and 2).

Increased Antibody Levels in Cd19 Cre/+ Mice upon Immunization
Considering the roles of B1 and MZB cells in the production of protecting antibodies, we compared the total or antigen-specific sera antibody levels in Cd19 Cre/+ and control WT littermates before and after immunizations. The baseline antibody levels did not differ significantly between these two groups of mice (Supplementary Figure S2A). However, after immunization with the T cell independent type II antigen (TI-II-Ag) NP-Ficoll, Cd19 Cre/+ mice produced significantly increased amounts of NP-specific IgM, IgG1 and IgG3, three major antibody subtypes against soluble protein or carbohydrate antigens in mice [29], at all time-points tested ( Figure 3). Notably, this phenomenon was gender-independent, as similar results were observed in both female ( Figure 3) and male (Supplementary Figure S2B,D) Cd19 Cre/+ mice. The IgM antibodies capable of binding to coated NP 25 -BSA were low and comparable, and the NP-specific IgG was absent in the sera of naive WT and Cd19 Cre/+ mice (Supplementary Figure S2D). significantly between these two groups of mice (Supplementary Figure S2A). However, after immunization with the T cell independent type II antigen (TI-II-Ag) NP-Ficoll, Cd19 Cre/+ mice produced significantly increased amounts of NP-specific IgM, IgG1 and IgG3, three major antibody subtypes against soluble protein or carbohydrate antigens in mice [29], at all time-points tested ( Figure 3). Notably, this phenomenon was gender-independent, as similar results were observed in both female ( Figure 3) and male (Supplementary Figure S2B,D) Cd19 Cre/+ mice. The IgM antibodies capable of binding to coated NP25-BSA were low and comparable, and the NP-specific IgG was absent in the sera of naive WT and Cd19 Cre/+ mice (Supplementary Figure S2D). We next stained cells in the peripheral blood, spleens, or peritoneal cavities with NP-FITC in combination with 7-AAD and antibodies to visualize NP-specific B cells. The sequential gating strategy to define NP-positivity in different B cell subsets is shown in Supplementary Figure S3A-D. The background bindings of NP-FITC to different B cell subsets were minimal and comparable between WT and Cd19 Cre/+ mice (Supplementary Figure  S3C-E). In accordance with higher NP-specific antibody levels, Cd19 Cre/+ mice had more NP + cells on D7 in blood post-NP-Ficoll immunization ( Figure 4A,B). Within the B cell compartment, significantly increased percentages and numbers of NP-specific CD19 + B220 low cells (phenotypically resembling B1 gated in Figure 2A, but containing B1, plasmablasts and pre-plasmablasts in immunized mice) were observed in blood of Cd19 Cre/+ mice on both D7 ( Figure 4A,C) and D14 (Supplementary Figure S2C) after NP-Ficoll injection. Moreover, numbers of NP + B220 high B2 and B220 low cells, as well as CD19 low CD138 high ASCs, were significantly increased in the spleens of Cd19 Cre/+ mice two weeks after NP-Ficoll administration ( Figure 5A-C). ELISPOT analyses confirmed the significantly increased amounts of NP-specific ASCs in the spleens of immunized, but not  Figure S3C-E). In accordance with higher NP-specific antibody levels, Cd19 Cre/+ mice had more NP + cells on D7 in blood post-NP-Ficoll immunization ( Figure 4A,B). Within the B cell compartment, significantly increased percentages and numbers of NP-specific CD19 + B220 low cells (phenotypically resembling B1 gated in Figure 2A, but containing B1, plasmablasts and pre-plasmablasts in immunized mice) were observed in blood of Cd19 Cre/+ mice on both D7 ( Figure 4A,C) and D14 (Supplementary Figure S2C) after NP-Ficoll injection. Moreover, numbers of NP + B220 high B2 and B220 low cells, as well as CD19 low CD138 high ASCs, were significantly increased in the spleens of Cd19 Cre/+ mice two weeks after NP-Ficoll administration ( Figure 5A-C). ELISPOT analyses confirmed the significantly increased amounts of NP-specific ASCs in the spleens of immunized, but not naive, Cd19 Cre/+ mice as well ( Figure 5D,E). Even after 5 weeks, NP + B220 low and CD19 low CD138 high ASCs were still slightly increased in Cd19 Cre/+ mice (Supplementary Figure S4). naive, Cd19 Cre/+ mice as well ( Figure 5D,E). Even after 5 weeks, NP + B220 low and CD19 low CD138 high ASCs were still slightly increased in Cd19 Cre/+ mice ( Supplementary Figure S4). , CD19 + B220 high B2 (B2-NP+), CD19 + B220 low (B220 low -NP+), CD19 + B220 low CD5 + B1a (B1a-NP+), and CD19 + B220 low CD5 -(CD5 − -NP+) B cells in WT vs. Cd19 Cre/+ mice on D7. Gates for CD19 + B220 low /CD19 + B220 low CD5 − cells were equivalent to those for B1/B1b in Supplementary Figure S3A, respectively. (B) A bar graph showing the percentages or numbers (per 40 μL blood) of total NP + cells within gated 7-AADlive cells on D7 and D14. (C) A bar graph showing the percentages of NP + cells within gated B2, B220 low or B1a/B220 low CD5 − cells in the two groups of mice on D7 (n = 9/7 for WT/Cd19 Cre/+ group, respectively). Each symbol represents one single mouse of the indicated genotype, and results are expressed as mean ± SEM (B,C). * p < 0.05; ** p < 0.01.
Likewise, after immunizations with the T cell dependent antigen (TD-Ag) NP19-OVA, the titers of NP-specific IgM and IgG3 were persistently higher at least for 6 weeks in Cd19 Cre/+ mice than those in WT littermate controls, and increased levels of NP-specific IgG2b and IgG2c were detected at later time points in Cd19 Cre/+ mice as well ( Figure 6A). Accordingly, significantly more NP-specific IgM, IgG2c, and IgG3, but not IgG1, ASCs were observed in the spleens of Cd19 Cre/+ mice on D14 post-immunization ( Figure 6E). Frequencies of CD19 low CD138 high ASCs tended to increase in Cd19 Cre/+ mice, while the number and phenotype of germinal center B cells were comparable between these mice ( Figure  6B-D).
Thus, after immunization with TI-II-Ag or TD-Ag, Cd19 Cre/+ mice produce significantly increased levels of antigen-specific antibodies than their WT counterparts, which may be attributed to increased numbers of antigen-specific B/ASCs in vivo.

Increased Antibody-Producing Ability of B Cells from Cd19 Cre/+ Mice upon LPS-Stimulation In Vitro
To investigate whether the increased amounts of antigen-specific B cells in immunized Cd19 Cre/+ mice were attributable to increased survival/proliferation of B cells, we purified B cells from the spleens of control or Cd19 Cre/+ mice, and activated them with anti-IgM or LPS in the absence/presence of B cell survival factor BAFF (B cell activating factor, encoded by Tnfsf13b). Comparable percentages and absolute numbers of B cells were observed in cultures with purified control or Cd19 Cre/+ B cells, albeit the addition of BAFF significantly promoted their survivals as expected (data not shown). No differences on the expression of activation markers CD69/CD25 and costimulatory molecule CD86 were noted as well (data not shown). Moreover, after stimulating B cells with anti-IgM/LPS ± immunization. CD138 + (138 + ) and CD138 ++ (138 ++ ) cells were gated as described in Figure 5B. Each symbol represents a single mouse of the indicated genotype. Results are expressed as mean ± SEM. * p < 0.05; ** p < 0.01; *** p < 0.001.
Thus, after immunization with TI-II-Ag or TD-Ag, Cd19 Cre/+ mice produce significantly increased levels of antigen-specific antibodies than their WT counterparts, which may be attributed to increased numbers of antigen-specific B/ASCs in vivo.

Increased Antibody-Producing Ability of B Cells from Cd19 Cre/+ Mice upon LPS-Stimulation In Vitro
To investigate whether the increased amounts of antigen-specific B cells in immunized Cd19 Cre/+ mice were attributable to increased survival/proliferation of B cells, we purified B cells from the spleens of control or Cd19 Cre/+ mice, and activated them with anti-IgM or LPS in the absence/presence of B cell survival factor BAFF (B cell activating factor, encoded by Tnfsf13b). Comparable percentages and absolute numbers of B cells were observed in cultures with purified control or Cd19 Cre/+ B cells, albeit the addition of BAFF significantly promoted their survivals as expected (data not shown). No differences on the expression of activation markers CD69/CD25 and costimulatory molecule CD86 were noted as well (data not shown). Moreover, after stimulating B cells with anti-IgM/LPS ± IL-4 for 4 days in vitro, the cell division profiles ( Figure 7A) and absolute numbers (data not shown) of B cells did not differ significantly between WT and Cd19 Cre/+ mice.  We next stimulated B cells with high concentrations of LPS with/without IL-4 to induce them to differentiate into PCs and undergo CSR to IgG1 + cells. We observed no significant differences in the percentage of CD138 high PCs or class-switched IgG1 + B cells between WT vs. Cd19 Cre/+ B cells before or after culture ( Figure 7B,C). Nonetheless, Cd19 Cre/+ We next stimulated B cells with high concentrations of LPS with/without IL-4 to induce them to differentiate into PCs and undergo CSR to IgG1 + cells. We observed no significant differences in the percentage of CD138 high PCs or class-switched IgG1 + B cells between WT vs. Cd19 Cre/+ B cells before or after culture ( Figure 7B,C). Nonetheless, Cd19 Cre/+ B cells produced significantly higher levels of IgG2b and IgG3, but not IgM/IgG1, after stimulation with LPS on both D4 and D6 ( Figure 7D), suggesting that the expression of the Cre transgene somehow promotes the CSR to IgG2b and IgG3 in LPS-stimulated B cells.
In sum, although B cells in Cd19 Cre/+ mice do not display an enhanced survival, proliferation, PC differentiation, or CSR to IgG1 + cells in our culture systems, they do produce more IgG2b and IgG3 upon LPS stimulation in vitro.

Survey of Published Literature Using Mice Containing Cd19 Cre/+ Transgene Reveals That the Phenotype Is Confounded by Different Controls Used
As shown above, despite little effect on early B cell development in the BM, the single Cd19 Cre/+ knock in allele mildly alters the frequencies of mature B cell subsets (MZB and B1a) and elevates antibody productions both in vivo and in vitro (Figures 1-7). We thus performed a survey of published articles using Cd19 Cre/+ gene fl/fl mice to study B cell-intrinsic functions of floxed genes, as we reasoned that the choice of controls (i.e., CD19-Cre − vs. CD19-Cre + ) may have biased the results/conclusions obtained.
We analyzed 256 articles where the Cd19 Cre/+ transgene was used to delete loxp-flanked sequence in B cells (Supplementary File S1), among which 136 studies contained the necessary information allowing for comparisons between the control (CD19-Cre − or CD19-Cre + ) and Cd19 Cre/+ gene fl/+ or Cd19 Cre/+ gene fl/fl mice in at least one of the following parameters: percentages/numbers of MZB and B1a cells, or antibody levels in vivo/vitro (Supplementary Figure S5A). Similar to the phenotypes of Cd19 Cre/+ -transgenic mice described in this manuscript, 50% (28/56) of studies reported that CD19-Cre conditional-knock out mice/cells exhibited at least one of the following features in comparison with their CD19-Cre − counterparts: reduced MZB/B1a cells or higher antibody levels in sera/culture supernatants. By contrast, only 29% (23/80) of studies using CD19-Cre + controls were classified as similar. Thus, information on the functions of floxed genes deleted/overexpressed by the Cd19-driven Cre recombinase is significantly related to the choice of control mice used (χ 2 test, p = 0.019, Supplementary Figure S5A), and it is likely that differences in some studies would have been less/more significant if CD19-Cre + , instead of CD19-Cre − , mice had been used as controls.

Discussion
Data presented in this study show that expression of Cre under the control of Cd19 promoter reduces the number of splenic MZB cells, alters the frequency of B1a vs. B1b cells in the periphery, and potentiates antibody productions both in vivo and in vitro after immunization/stimulation. Thus, Cd19 Cre/+ mice or B cells are critical controls in studies using Cd19-driven Cre to specifically inactivate or activate genes in B cells, as lacking these controls may lead to misinterpretation of the data and biased conclusions on gene functions in B cell biology.
Expression of the Cre recombinase alone may be toxic, possibly via recognizing cryptic/pseudo loxP sites in the genome, and thus significantly impacts the survival and function of mammalian cells, including T lymphocytes [7][8][9]12,46,47]. However, our results, showing the undisturbed BM B cell development, comparable total B cell numbers in the periphery, intact survival of B cells in vitro, and augmented humoral responses to both TD-and TI-II Ags in vivo, rule out a general and significant toxicity of Cre recombinase in Cd19 Cre/+ mice on the widely used C57BL/6 background.
Moreover, potentially altered endogenous gene expressions resulting from inserting/integrating of the Cre transgene represent another confounding factor in Cre-mediated gene editing systems. The Cd19 Cre/+ mice used in our study were originally generated by Rickert et al., in which the Cre-expression cassette was inserted into the second exon of Cd19, thereby disrupting the latter's coding sequence [17,18]. Thus, the Cre-expressing heterozygotes (Cd19 Cre/+ )/homozygotes (Cd19 Cre/Cre ) are equivalent to the respective Cd19 heterozygous (Cd19 +/− )/homozygous (Cd19 −/− ) knock-out mice in terms of CD19 expression [17,18]. Although CD19 is dispensable for early B cell development in the BM, its deficiency results in a near complete loss of MZB and B1a cells in the periphery of mice [17,[23][24][25]28,[48][49][50]. Hence, the reduced number of MZB and B1a cells in the periphery of Cd19 Cre/+ mice may relate to their diminished surface CD19 expressions (Figures 1 and 2). A detailed side-byside comparison among WT, Cd19 Cre/+ , and Cd19 Cre/Cre mice would give more insights on the effect of CD19 levels in these phenotypes.
As a type I transmembrane protein, CD19 functions as a dominant signaling component of a multimolecular complex on the surface of B cells. It acts as an essential co-receptor for BCR signal transduction by recruiting and amplifying the activation of Src-family protein tyrosine kinases Lyn and Fyn, and/or through the activation of PI3K and downstream Akt kinases [15]. Moreover, the CD19/21 complex colligates with BCR and thereby enhances B cell activation induced by antigen-bearing complement [15]. Given that MZB cells/precursors proliferate more than follicular B cells [51,52], and persist longer in the periphery without the influx of BM cells [53], it is conceivable that the CD21 high MZB cells rely more on CD19-propagated tonic BCR signaling for development, survival, and homeostatic proliferation in the periphery [24]. Likewise, it has been proposed that B1a, but not B1b, precursors reside almost exclusively in the fetal and neonatal liver, hence their numbers in adult mice are maintained primarily through BCR-dependent self-renewal of pre-existing cells rather than the replenishment from progenitors [54]. Surface levels of CD19 positively correlate with both the development and self-renewal of B1a cells [22,48]. As such, we speculate that the reduced MZB and B1a cells in Cd19 Cre/+ mice may be mainly attributed to attenuated CD19 expression/signaling. Intriguingly, the effect of Cd19 Cre/+ on B1/B1a cells appears to be age-dependent, as their frequencies revert to normal in mice beyond 3 months (Supplementary Figure S1G) [44], suggesting that these cells adopt compensatory mechanisms to overcome the effect of Cd19 Cre/+ later in vivo. Nonetheless, the potential toxic effect of Cre recombinase could not be excluded, as these in vivo long-persisting and highly proliferative B cells, especially MZB cells, might express higher levels of Cre at certain stages and/or be more sensitive to Cre-mediated DNA-damages.
In addition to development/phenotypes, B cells in Cd19 Cre/+ mice functionally resemble those with disrupted/reduced CD19 levels in terms of augmented responses to TI-II-Ags in vivo (Figure 3), indicating the presence of a similar cause-outcome relationship in these mice [50,55,56]. Apart from amplifying BCR signaling, CD19 may deliver a negative signal that inhibits B cell proliferation [56,57]. Therefore, it is possible that CD19low expressing B cells in Cd19 Cre/+ mice are less suppressible, relative to CD19 high WT B cells, to the inhibitory signal afforded by CD19 upon binding to unknown ligands in vivo, thereby proliferating and differentiating more robustly following TI-II-Ag immunization ( Figures 4 and 5). Alternatively, it was proposed by Sato et al. that the lack of surface CD19 may upregulate the threshold of negative selection of B cells in the BM, and thereby allows a larger number of B cells with low-affinity receptors for Ficoll to mature and enter the circulation, resulting in enhanced peripheral responses in CD19 −/− mice [55]. However, it is unlikely that naive Cd19 Cre/+ mice possess more Ficoll-or NP-specific B cells, as their frequencies are comparable to those in unimmunized WT mice (Figures 5E and S3E). These data also suggest that the specific increase (proliferation, selection and/or differentiation) of NP-specific B cells in immunized Cd19 Cre/+ mice is not due to differences in the pre-immune BCR repertoire [58].
In contrast, CD19 seems to be required for B cell responses to TD-Ags, as CD19 −/− mice are profoundly deficient in producing antibodies against this type of Ag [17,55,59]. Of note, the number of conventional B cells is reduced approximately by half in peripheral lymphoid tissues of CD19 −/− mice [49,59]. Restitution of CD19 −/− mice with one copy of human CD19 almost completely restores TD-Ag-induced antibody responses to WT levels without reverting the defects of B cell numbers, indicating that the expression of one allele of human Cd19-transgene enhances the responses of B cells against TD-Ags in vivo on a per cell basis [59]. Hence, it is conceivable that CD19, albeit being critical, fine tunes TD-Ag-induced antibody responses in vivo as well, and akin to the situations in TI-II-Ag immunized mice, the enhanced antibody responses against NP-OVA in Cd19 Cre/+ mice might also be attributed, at least partially, to the attenuated surface CD19 intensities on B cells ( Figure 6). Moreover, it has been shown that Cd19-deficiency selectively compromises Th2-dependent isotype switching of B cells, possibly owing to the impaired Th2-B cell interaction in vivo [60,61]. The increased levels of NP-specific IgG1 in NP-Ficoll-administered, but not NP-OVA-immunized, Cd19 Cre/+ mice thus might relate to the greater impact of CD19 on Th2-coordinated B cell differentiations in vivo [60,61]. In addition, a few studies showed that CD19 on B cells blunts T cell responses by promoting the generation of regulatory B cells [62][63][64]; thus, reduced surface CD19 on Cd19 Cre/+ B cells may promote antibody productions indirectly via enhancing T cell responses in vivo. Nevertheless, the possibility that integration of the Cre-expressing cassette may render B cells hyperresponsive to both TI-II-and TD-Ags in vivo, for instance, via the trans-effect of control elements it contains, cannot be excluded. The increased amounts of (NP-specific) ASCs, but not germinal center B cells, in NP-OVA-immunized Cd19 Cre/+ mice suggest that the integration of Cre cassette and/or low levels of CD19 preferentially promote B cell differentiation after TD-Ag immunization.
Cd19 Cre/+ B cells appear to have selectively increased ability to produce IgG2b and IgG3 upon LPS activation in vitro, because their proliferation, PC differentiation, IL-4induced CSR to IgG1, and secretions of IgM and IgG1 are all comparable to those of WT B cells (Figure 7). In line with our observations, data from a recent publication showed that LPS-induced CSR to IgG3, but not LPS/IL-4-triggered CSR to IgG1, is enhanced in Cd19 Cre/+ B cells [45]. Although the underlying mechanisms remain unknown, we hypothesize that expression of the Cre recombinase might somehow promote CSR to IgG2b and IgG3 in B cells under certain conditions, for instance, via inducing DNA recombination in adjacent regions. Thereafter, these antigen-specific IgG3 antibodies might promote other IgG productions by complexing and facilitating the transportation of TD-Ag to splenic follicles [65].
Notably, our findings described here indicate that Cd19 Cre/+ mice are critical controls in studies where Cre driven by the Cd19 promoter is used to ablate sequences in B cells, as the lack of this control may affect the interpretation of data and thus bias the conclusions. Indeed, the literature survey we performed to compare results from different studies using CD19-Cre − or CD19-Cre + mice as controls indicated that conclusions were confounded by the types of control mice used (Supplementary Figure S5A). In addition to parameters measured in this article, results in CD19-Cre + -uncontrolled studies investigating the roles of genes in B cells in other contexts, such as infections, autoimmune diseases or inflammatory disorders, should also be considered with caution. For instance, studies have shown that Cd19 Cre/+ mice have a faster weight recovery post influenza infection, exhibit an exaggerated lupus development/mortality, or live slightly shorter when compared to their CD19-Cre − counterparts [21,42,43].

Conclusions
We have demonstrated a mildly perturbed phenotype and significantly augmented antibody-secreting ability of peripheral B cells in Cd19 Cre/+ mice on the widely used C57BL/6 background. Thus, Cd19 Cre/+ mice are critical controls in studies using Cd19 Cre/+ to investigate gene functions in B cells, especially in the context of inflammation and infection with the involvement of innate-like MZB/B1 cells as well as antibody productions from conventional B2 cells.