The contribution of pO₂ on human naive B cell differentiation was investigated in B cell differentiation cultures while maintaining constant pCO₂. Resting naive B cells (CD19⁺IgD⁺CD27^-IgG^-IgA^-) from human peripheral blood were stimulated under atmospheric (pO₂ = 21%), normoxic (pO₂ = 3%) or hypoxic (pO₂ = 1%) conditions, using a combination of CD40L-expressing 3T3 cells (12), IL-4 and IL-21, known to facilitate class switch recombination, memory B cell and antibody secreting cell formation ( Figure 1A ) (12, 41, 42). Alterations in pO₂ did not affect CD40L expression levels on 3T3 cells used for stimulation ( Figure S2I , subtype high). Up until day 7 B cell survival was similar among pO₂ conditions, but at day 11 hypoxic cultures contained fewer cells ( Figures S2A,B ). In all cultures a rapid upregulation of activation marker CD80 was observed ( Figure S2C ). Remarkably, naive B cell differentiation, as determined by the formation of antibody secreting cells (CD27⁺CD38⁺) and CD27⁺ (CD27⁺CD38^-) B cells, was dramatically altered by differential pO₂. Hypoxia induced a 5 to 10-fold induction of CD27⁺ B cells as compared to atmospheric (p<0.0001) and normoxic cultures (p<0.0001) at day 7 ( Figures 1B, C ), and absolute cell no. in Figure S2D ), at the expense of antibody secreting cell formation at day 11. In addition, in normoxic cultures, the relative numbers of CD138⁺ antibody secreting cells were 2 to 4-fold increased relative to atmospheric (p<0.001) and hypoxic cultures (p<0.0001, Figure 1D ), in line with increased concentration of secreted IgM and IgG in normoxic culture supernatant at day 11 ( Figure 1E ). Naive B cells poorly survived and differentiated when cultured under extreme hypoxic conditions (pO₂ = 0.5%, Figure S2E, F ). Furthermore, using 3T3 cells with higher or lower expression of CD40L resulted in reduced proliferation, especially at reduced oxygen pressures (resp. ‘VH’, and ‘Low’, vs ‘High’; Figures S2H, I ). The addition of anti-IgM F(ab’)₂s to the cultures, as additional BCR stimulation, did not alter in vitro B cell survival and differentiation at differential pO₂s ( Figure S2J ). Next, we assessed the effect of pO₂ on antibody class switching and Ig production. Protein expression of DNA-editing enzyme AID –indispensable for class switch recombination and somatic hypermutation– were substantially elevated over time regardless of oxygen pressure, but significantly lower at low pO₂ at day 7 and 11 ( Figure 1F ), in accordance with previous studies in mice (32), reflecting the overall larger fraction of differentiated cells under these conditions ( Figures 1B, C , S2K ). In line with AID expression levels, IgG⁺ B cell frequencies were higher in atmospheric cultures (p<0.001, Figure 1F ) with minimal switch to IgA⁺ overall, as was expected by the IgG promoting culture conditions (CD40L + IL-21 + IL-4, Figure S2L ). Altogether, these data illustrate a profound regulation of human naive B cell differentiation by oxygen levels and show that normoxia promotes differentiation into potent antibody secreting cells while hypoxia promotes differentiation into CD27⁺ B cells.

Given that cellular metabolism has been reported to be linked to gene expression and B cell differentiation (43), we assessed the metabolic status of B cells that were cultured at different pO₂s. Using flow cytometry, HIF1α levels were increased throughout hypoxic cultures compared to normoxic and atmospheric cultures ( Figure 2A ). After 7 days of in vitro stimulation B cells showed similar increase in cell size in all conditions compared to unstimulated cells ( Figure S3A ). Dividing lymphocytes typically rely on aerobic glycolysis, fermenting imported glucose into lactate rather than oxidizing it in the mitochondria for energy. Glycolytic activity was assessed by uptake of 2-NBDG, a fluorescent analog of glucose, and production of lactate as an end product of glycolysis. Uptake of 2-NBDG and lactate production were not significantly different between different pO₂ culture conditions, but overall higher compared to resting naive B cells suggesting that indeed stimulated B cells require more glucose but demands did not change considerably upon differential pO₂ environments ( Figure 2B ). Also expression of GLUT1, a glucose transporter, varied minimally between culture conditions ( Figure S3B ). To asses mitochondrial oxidative metabolism we analyzed mitochondrial mass, potential and formation of reactive oxygen species (ROS) by flow cytometry at day 7. Mitochondrial mass, potential and mitochondrial ROS were higher in B cells cultured at hypoxic conditions ( Figure 2C ). Uptake of a fluorescent fatty acid probe was reduced in B cells under hypoxic conditions ( Figure 2D ), suggesting these cells rely less on oxidation of fatty acids. At day 11 metabolic markers yielded similar result among different pO₂ conditions, except for ROS levels that were significantly higher in hypoxic cultures compared to atmospheric cultures ( Figure S3C ). Taken together, these results indicate that highly proliferative B cells adopt glycolysis independent of local pO₂ levels and mitochondrial-associated ROS levels are elevated in hypoxic cultures.

The amplitude of B cell proliferation and differentiation is regulated by a complex interplay of signaling pathways induced by antigen recognition, CD40 ligation and reception of Tfh-derived cytokines IL-4 and IL-21 (summarized in Figure S4A ) (8, 44). Upregulation of BCL6 as well as IRF4 expression suggests differentiation of cells with a GC-like phenotype ( Figures 3A, B , respectively). In line, PNA positivity, indicative for expression of GC-specific glycans, and CD95 expression was observed early on in culture and decreased prior to the generation of CD27⁺ B cells and antibody secreting cells ( Figures 3C, D ). PNA binding was reduced for B cells cultured at hypoxic pO₂ at day 3 and 5 potentially due to the accelerated CD27⁺ differentiation in hypoxic cultures. Expression of CD86, an activation marker highly expressed on light zone B cells (45, 46), increased rapidly during culture and remained significantly higher expressed on B cells cultured at hypoxic pO₂ ( Figure 3E ). CD40 signaling primarily shapes the magnitude of B cell expansion and survival through induction of NFκB p65 and c-Myc (21–23). Moreover, NFκB p65 and c-Myc are reported to maintain cellular GC commitment (47). B cells cultured at hypoxic pO₂ showed a reduction in NFκB p65 ( Figure 3F ) and c-Myc ( Figure 3G ) expression from day 5 onwards, supporting the concomitant differentiation into CD27⁺ cells, and to a lesser extent antibody secreting cells, at day 7 and reduced B cells numbers observed in hypoxic cultures thereafter ( Figure S2B ).

Next, the effect of pO₂ on the IL-4R/IL-21R signaling pathways to regulate B cell and antibody secreting cell transcriptional programs was analyzed. Increased phosphorylation of STAT6 (pSTAT6), a signaling protein downstream of the IL-4R, was observed in B cells cultured at hypoxic pO₂, peaking at day 3 and 5 (p<0.01 at day 5) ( Figure 3H , S4B ). This coincided with higher expression of XBP-1s, a transcription factor (TF) downstream of pSTAT6, at day 7 ( Figure 3J , right panel), and was not the result of variable IL-4R expression levels, determined by RT-qPCR ( Figure S4C ). Peak levels of pSTAT3 on day 3 and 5, with minor differences when comparing the different pO₂ conditions ( Figure 3I , S4D ), coincided with upregulation of BLIMP1, directly downstream of pSTAT3, and repression of PAX5, most prominently in B cells cultured at hypoxic pO₂ ( Figure 3J , S4E ). PAX5 supports B cell identity. Consistent with this activity, B cells cultured at atmospheric pO₂ retained highest PAX5 levels throughout culture, and thus overall a lower fraction of differentiated cells under this conditions ( Figure 3I , S4D ) (48). Expression of BLIMP1 and XBP-1s was higher under hypoxic conditions ( Figure 3J ), and did not differ within the CD27^-CD38^- population at varying pO₂ at day 7 ( Figure S4F ), suggesting elevated BLIMP1 and XBP-1s expression in B cells cultured at hypoxic pO₂ arises from the expanded CD27⁺ compartment ( Figure 1C ). This CD27⁺ population did not secrete more Ig at day 7 compared to cells cultured at normoxic or atmospheric pO₂ despite elevated XBP-1s expression, which is essential for the unfolded protein response in antibody secreting cells ( Figure S4G ). At day 11, naive B cells differentiation into antibody secreting cells in normoxic cultures outperformed atmospheric cultures in which B cells retained higher PAX5 levels and repression of BLIMP1 and XBP-1s ( Figure S4H ). Furthermore, cells in normoxic cultures had highest XBP-1s levels in line with higher Ig production observed in these cultures (p<0.05) ( Figure 1G ). Overall, B cell culture at different pO₂ affects CD40-, IL-21R- and IL-4R-dependent signaling pathways essential for efficient B cell survival and differentiation. Expression dynamics of B differentiation-associated TFs PAX5, BLIMP1 and XBP1 were influenced by local pO₂, skewing the formation of CD27⁺ and/or antibody secreting cell populations at hypoxic or normoxic pO₂, respectively.

Culturing B cells at hypoxic pO₂ resulted in a remarkable increase in CD27⁺ B cells which emerged earlier as compared to atmospheric and normoxic cultures ( Figure 1C ). Interestingly, there was not only enlargement of this compartment but also formation of a population of CD27⁺⁺ cells (4.4% of CD19+ population), absent in normoxic and atmospheric cultures at day 7 (p<0.05) ( Figure 4A , S5 ) To assess the phenotype of these CD27⁺⁺ B cells, transcription factor profiles directing B cell differentiation were compared between CD27^-, CD27^+/-, CD27^+/+ and CD27⁺⁺CD38⁺⁺ populations that developed under hypoxic pO₂ at day 7 ( Figure 4B ). PAX5 expression was lower and BLIMP1 expression higher in CD27^+/++ cells compared to CD27^-CD38^- B cells, with PAX5 and BLIMP1 expression being similar between CD27⁺⁺ and CD27⁺⁺CD38⁺⁺ populations ( Figure 4C , upper and middle panel). Expression of XBP-1s increased as differentiation progressed, being highest in the CD27⁺⁺CD38⁺⁺ population ( Figure 4C , lower panel). Moreover, a rise in the frequency of IgG⁺ B cells coincided with increasing CD27 expression, being highest in the CD27⁺⁺ population ( Figure 4D ).

As the CD27⁺⁺ showed a distinct transcriptional profile compared to CD27^+/- and CD27^- B cells, we assessed whether this translated into functional differences with regard to antibody secreting cell differentiation and Ig production. To this end naive B cells were cultured for 7 days under hypoxia, after which different CD27 expressing populations were sorted and subsequently cultured at antibody secreting cell-promoting normoxic pO₂ for an additional 4 or 7 days ( Figure 4E ). CD27⁺⁺ cells showed a significant increase in antibody secreting cell differentiation and antibody production as compared to the CD27^+/- and CD27^- subsets (p<0.05) ( Figure 4F ), indicating that with the same level of stimulation the CD27⁺⁺ cells had enhanced capacity to differentiate towards functional antibody secreting cells. In summary, a population of CD27⁺⁺ B cells emerged exclusively in hypoxic cultures, with the potential to rapidly differentiate into antibody secreting cells and produce Ig after restimulation.

Given that during an in vivo GC reaction B cells cycle between the hypoxic light zone and normoxic dark zone, we assessed the effect of a pO₂ transition on B cell differentiation and class switch recombination in vitro. To this end, naive B cells were cultured at either hypoxic or normoxic pO₂ and transferred at stated time points to normoxic or hypoxic pO_2, respectively. Independent of timing, a transition from hypoxic to normoxic pO₂ increased antibody secreting cell differentiation concomitant with decreased CD27⁺ cell formation ( Figures 5A, B ). Transition from hypoxic to normoxic pO₂ at day 3 of culture enhanced antibody secreting cell differentiation at day 11 compared to continued culture at hypoxic pO₂ and was similar to continuous culture under normoxic pO₂ ( Figures 5A, B ), with BLIMP1 expression lower compared to hypoxic cultures and similar to levels observed in normoxic cultures ( Figure 5C ). A transition from 1 to 3% pO₂ at day 5 led to a moderate generation of CD27⁺ B cells and reduced antibody secreting cells formation in line with continuous hypoxic cultures. When hypoxic cultures were transitioned to normoxic pO₂ at day 7, this induced an efficient formation of antibody secreting cells at day 11. Different from cultures at hypoxic pO₂ on day 3 and 5, at day 7 hypoxic cultures consist for a significant proportion (~20%) of CD27^+/++ B cells, which efficiently differentiate to antibody secreting cells upon restimulation and transition to normoxic pO₂ as was shown in Figure 4F . Overall this suggests time-dependent transitions in pO₂ steer B cell differentiation, where early (day 3) transitions from hypoxic to normoxic pO₂ lead to antibody secreting cell formation comparable to continuous normoxic cultures whereas a pO₂ transition later on during culture (day 7) induces antibody secreting cell formation - in part - via the CD27⁺⁺ compartment. These data show that in vitro B cell differentiation can occur via distinct trajectories steered by differential pO₂, where hypoxia drives differentiation via a CD27⁺⁺ population and atmospheric and normoxic pO₂ via CD38⁺ or directly towards antibody secreting cells ( Figure S6A ). Remarkably, transitions from hypoxic to normoxic pO₂ at day 3 and 5 enhanced class switch recombination to IgG⁺ B cells concomitant with a reduction in IgM⁺ B cells ( Figure 5D ). Even though the percentage of CD138⁺ antibody secreting cells was similar ( Figure 5E ) and lower compared to continuous normoxic cultures ( Figure 1D ), transition of hypoxic to normoxic pO₂ at day 3 and 5 resulted in higher quantities of secreted IgM and IgG in culture supernatants ( Figure 5F ). Enhanced IgG class switch recombination and Ig production were no longer observed when the pO₂ transition was made at day 7.

Vice versa, transition from normoxic to hypoxic pO₂ at day 3 and 5 enhanced CD27⁺ B cell formation and BLIMP1 expression compared to continuous normoxic cultures and was more similar to continuous hypoxic cultures ( Figures S6B–D ). A transition at day 7 from normoxic to hypoxic pO₂ was deleterious for both CD27⁺ B cell and antibody secreting cell differentiation. class switch recombination to IgG⁺ B cells and Ig production levels ranged between those found in continuous normoxic and hypoxic cultures ( Figures S6E–G ).

Overall, time-dependent transitions between hypoxic and normoxic pO₂ during culture govern different human B cell differentiation trajectories and IgG class switch recombination in vitro.

The GC reaction is at the heart of the humoral immune response. Both extrafollicular and GC B cell responses take place within human tissues where oxygen pressure may range from 0.5-6% pO₂ with hypoxic regions (0.5-1% pO₂) described within GC light zone regions. Preciously little information is available about the potential role of local pO₂ on human B cell fate decision during their cycling between dark zone and light zone, typically characterized by dissimilar oxygen pressures. Here we studied the effect of atmospheric (21%), tissue associated (normoxia, 3%) and hypoxic (1%) pO₂ on human B cell differentiation. We observe profound differences in B cell differentiation, dynamics of emergence of various cell populations, class switch, and Ig production under these different oxygen pressures. Figure 6 summarizes our key findings on the contribution of pO₂ on human naive B cell differentiation. In both hypoxic and normoxic cultures, naive B cells differentiated into a GC-like phenotype as evidenced by the expression of BCL6 and IRF4, as well as increased expression of CD95 and PNA binding. Similar to light zone B cell phenotype and function in vivo (8), B cells cultured at hypoxic pO₂ showed high expression of CD86 and differentiated rapidly into CD27⁺ B cells. Moreover, culture at hypoxic pO₂ generated a unique population of CD27⁺⁺ B cells efficient in antibody secreting cell differentiation and Ig production upon restimulation. In normoxic cultures, naive B cells predominantly differentiated into antibody secreting cells, which occurred less rapidly compared to CD27⁺ B cell formation at hypoxic pO₂ in line with time-dependent waves of memory B cell and antibody secreting cell formation in vivo (2). Finally, time-dependent transitions from hypoxic to normoxic pO₂ during culture changes B cell differentiation trajectories and promotes IgG class switch recombination and Ig production. This indicates during in vivo B cell cycling between GC light zone and GC dark zone, the dynamic variation in pO₂ is also likely to form another regulatory layer for human B cell differentiation. Overall, our results identify oxygen as a critical factor in dictating human B cell differentiation and demonstrate the necessity of incorporating GC-like physiological pO₂ variations rather than continuous atmospheric pO₂ to study human B cell responses in vitro.

In vivo, antigen-specific triggering of the BCR is required for the generation of an efficient GC response. We did not observe an effect on B cell survival or differentiation by addition of a BCR stimulus to the B cell cultures at different pO₂, in line with previous reports showing limited additional effects of BCR triggers in in vitro cultures (12). Despite the absence of a BCR trigger, we saw a clear upregulation of c-Myc expression which is normally upregulated upon positive selection in the GC, in line with previous reports indicating CD40L ligation to be sufficient to induce c-Myc expression (25). We observed a reduction of c-Myc and NFκB expression in cells cultured at hypoxic pO₂ after an initial surge upon culture initiation. c-Myc and NFκB signaling play a major role in transcriptional regulation of GC B cell proliferation and cell survival, thereby correlating with our observations regarding reduced cell numbers towards the end of cultures at hypoxic pO₂. Naive B cell survival was abrogated upon culture at 0.5% pO₂ and upon higher CD40L co-stimulation (subtype VH) at low pO₂, but not atmospheric pO₂, due to impaired proliferation. Hence, the level of CD40L co-stimulation impacts B cell survival at physiological pO₂.

Highly proliferative lymphocytes have been described to adopt glycolysis (49), which -although not comprehensively studied- is in line with the increase in glycolysis-associated metabolic traits observed in our cultures independent of in vitro pO₂. B cells cultured under hypoxic pO₂ carried out mitochondrial oxidative metabolism, in addition to glycolysis. The difference in B cell differentiation trajectories and expansion of the CD27⁺ compartment in hypoxic cultures may explain the expanded metabolic program towards mitochondrial oxidative metabolism at day 7. Much of the research on crosstalk between immune responses and pO₂ has focused on the TF HIF1α. We observe upregulation of HIF1α expression throughout all culture conditions not only at hypoxic pO₂. HIF1α can also be induced by signaling through CD40L, BCRs as well as TLR ligands, probably explaining the increase in HIF1α also in non-hypoxic cultures (50). Furthermore, a recent study could not identify expression of hypoxia-induced genes in ex vivo analyzed mouse GC B cells (40), indicating HIF1α is not the sole regulator of hypoxia-associated responses.

Clear differences were observed in the expansion of CD27⁺ B cell and/or antibody secreting cell compartments in human naive B cells cultures at hypoxic and normoxic pO₂, respectively, driven by differences in the underlying molecular pathways regulating B cell survival and differentiation. It has been shown that there is a time-dependent developmental switch in the output of the GCs, such that it first dedicates itself to memory B cell generation and later switches to mainly producing antibody secreting cells (2). During culture of naive B cells at hypoxic pO₂ a hypoxia-dependent CD27⁺⁺ population emerged early during culture. The question arises whether these CD27⁺⁺ cells may resemble memory B cell generated early within a GC response. Albeit unanswered in this study, it did become clear that these CD27⁺⁺ cells did not secrete Igs, whereas upon restimulation they swiftly differentiated into antibody secreting cells and exhibited efficient Ig secretion, in line with functionalities of memory B cells (51). Nevertheless, these cells clearly expressed a transcription factor signature of increased BLIMP1 and XBP-1s expression concomitant with a reduction of the B cell identity TF PAX5, similar to antibody secreting cells and different from memory B cells (16, 19). In line with varying pO₂ during a GC response, where B cells cycle between hypoxic light zone to normoxic dark zone, we observed time-dependent effects of pO₂ transitions on human B cell differentiation trajectories. Taken together, hypoxic pO₂ primes B cells for efficient antibody secreting cell differentiation from a CD27⁺ phenotype that, in order to occur, favors a transition to a normoxic environment. Besides B cell differentiation trajectories, also IgG class switch recombination was influenced by pO₂ transitions. When cultured at continuous pO₂, atmospheric pO₂ yielded the highest proportion of IgG⁺ B cells (albeit not corroborated with equally elevated levels of IgG antibody production). Similar numbers of IgG⁺ B cells could be generated in cultures that transitioned from hypoxic to normoxic pO₂ at day 3 or 5. Overall, our in vitro studies demonstrate the necessity of incorporating GC-like physiological pO₂ variations rather than continuous atmospheric pO₂ to study human B cell responses in vitro, as it more closely resembles the in vivo GC microenvironment and recapitulates time-dependent regulation of B cell fate decisions.

Besides local pO₂, other factors such as availability of nutrients, migratory and physical signals such as shear stress are also likely to contribute to B cell fate decision during both extrafollicular and GC B cell responses, warranting the need for 3D culture systems or organoids that more closely resemble the GC microenvironment to facilitate human in vitro B cell differentiation studies. In addition, several pathological conditions have been described to be related to hypoxia, such as myocardial ischemia (52), metabolic diseases (53), chronic heart and kidney diseases (54), and reproductive diseases (55). Moreover, pO₂ has been measured and found to be significantly reduced in the majority of tumors, and sensitivity to chemotherapeutic agents changed dramatically under hypoxic conditions (56–58), highlighting the importance to adopt hypoxic cultures also for other avenues such as cancer-related drug testing and in vitro disease models.

In summary, we demonstrate that oxygen is an important regulator of human naive B cell differentiation by promoting oxygen-dependent differentiation trajectories and IgG class switch recombination.

3T3 mouse fibroblast cells expressing human CD40L (subtype: high), described previously (12), were harvested and irradiated with 30 Gy and seeded 10x10³ cells/well on 96-well plates for overnight adherence in B cell culture medium (RPMI medium supplemented with FCS (5%, Bodinco), penicillin (100 U/mL, Invitrogen), streptomycin (100 μg/mL, Invitrogen), β-mercaptoethanol (50 μM, Sigma-Aldrich), L-glutamine (2mM, Invitrogen), human apo-transferrin (20 μg/mL, Sigma-Aldrich) depleted for IgG using protein A sepharose (GE Healthcare). The next day, naive B cells were sorted and 250 naive B cells/well were co-cultured with the seeded 3T3-CD40L expressing cells in the presence of IL-4 (25ng/ml; Peprotech) and IL-21 (50ng/ml; Peprotech) for 3 – 11 days to determine proliferation, isotype switching, GC, memory and plasma cell formation and Ig production. Cultures were maintained at 37°C in an atmosphere with 5% pCO₂ and 21%, 3% or 1% pO₂. For some cultures after 3, 5 or 7 days cultures were moved from 1 to 3% pO₂ and vice versa.

IgM, IgA, IgG, and IgG1 and IgG4 expression levels were measured in culture supernatants at day 7, 11 and 14. 96-well maxisorb plates were coated with monoclonal mouse anti-human IgM (2μg/ml, MH15-1), anti-IgA (1μg/ml, MH14-01), anti-IgG (2μg/ml, MH16-1), anti-IgG1 (1μg/ml, MH161-01), and IgG4 (1μg/ml, MH164-01) in PBS all provided by Sanquin Reagents. Culture supernatants were incubated for 1h and secreted Igs were detected using 1μg/ml HRP-conjugated mouse anti-human IgM, IgA, IgG, IgG1, or IgG4 (Sanquin Reagents) in HPE (Sanquin Reagents). The ELISA was developed using TMB substrate, stopped by addition of 2M H₂SO₄ and absorbance was measured at 450 and 540 nm. OD values were normalized to values of a titration curve of a serum pool that was included in each plate.

L(+)-lactate production was measured in B cell culture supernatant using L-lactate Assay Kit (ab65331, Abcam) according to manufacturer’s instructions. Lactate production was corrected for cell counts as measured by flow cytometry.

RNA isolation was performed as described elsewhere (60). Primers were developed to span exon–intron junctions and then validated (IL-4R: 5’- CCCTGAAGTCTGGGATTTCCT -3’). Gene expression levels were measured in duplicate reactions for each sample in StepOnePlus (Applied Biosystems, Foster City, CA, USA) using the SYBR green method (Applied Biosystems, Foster City, CA, USA). Expression of housekeeping gene 16S was used for normalization.

Statistical analysis was performed using GraphPad Prism 8. Data were analyzed using Repeated Measures one-way ANOVA with Tukey’s multiple comparison test, Repeated Measures two-way ANOVA with Sidak’s multiple comparison test or mixed-effects analysis with Dunnett’s multiple comparison test where appropriate. Results were considered significant at p <0.05. Significance was depicted as *p <0.05 or **p <0.01, ***p <0.001 or ****p <0.0001. To show significance between specific datasets within a graph containing multiple datasets, the * shows the significance between the dataset matching the color of * and the dataset closest to the *.