Functional Divergence of Regulatory and Conventional Bovine Neutrophils following Mycobacterium bovis Infection

Abstract

Introduction

Bovine tuberculosis (bTB), due to Mycobacterium bovis (Mb) infection, is a chronic cattle disease and neglected cause of zoonotic tuberculosis. The role of neutrophils in bTB is overlooked. We recently identified a new neutrophil subset in cattle and mice with a similar morphology to conventional inflammatory neutrophils (N^conv). However, unlike N^conv, these regulatory neutrophils (N^reg) express MHC-II at their surface and can suppress lymphocyte proliferation.

Methods

In this study, we compared the responses of bovine N^conv and N^reg to infection with a virulent Mb strain circulating in France and the attenuated Mb BCG vaccine.

Results

N^reg and N^conv had different transcriptional profiles and were differentially activated by Mb infection. Both N^reg and N^conv efficiently killed Mb, but N^reg had higher levels of phagocytosis activity (2-fold difference) and ROS production (2.3-fold difference). N^reg had higher levels of mitochondrial activity and an ultrastructural organization different from that of N^conv.

Conclusion

Our results provide the first insight into the functional characterization of bovine neutrophil subsets during Mb infection and highlight a new layer of complexity in their functional diversity that must be taken into account to improve our understanding of bTB pathophysiology, which is urgently required to improve the management of this costly disease.

Introduction

Bovine tuberculosis (bTB), due to Mycobacterium bovis (Mb) infection, is a zoonotic disease and one of the most difficult infections to control in cattle. This disease is a major public health issue with a large global economic burden in both developing and developed countries, due to the lack of antituberculosis vaccines for livestock [1], and the lack of reliability of diagnostic techniques [2, 3]. bTB is a notifiable disease dealt with by the culling of entire herds [4], resulting in a large economic, social, ethical and environmental impact. France has been officially bTB-free since 2001, with a prevalence of less than 0.1% in cattle herds [5, 6], but this status is threatened by the many Mb strains circulating among wildlife and livestock in some parts of the country.

The pathophysiology of bTB has been described in less detail than that of human TB, caused principally by Mycobacterium tuberculosis (Mtb). The correlates of protection remain unknown for both human and animal TB, and their identification is a matter of top priority for the World Health Organization and the World Organization for Animal Health. Improvements in our understanding and control of the disease require a deeper understanding of the immune mechanisms involved during infections in cattle. The role of macrophages in the pathophysiology of this disease is well established [7–9], but that of neutrophils – the other main type of phagocytic immune cell – remains largely unknown. However, neutrophils have been shown to play important roles in innate immune resistance to Mtb infection in both humans and mouse models [10]. Indeed, individuals in contact with patients presenting active TB are less likely to be infected with Mtb if they have high peripheral blood neutrophil counts [10]. Conversely, neutrophils constitute the largest cell population in the lungs of patients with active TB [11] and the transcriptional signature of severe TB is driven principally by type I IFN signaling, dominated by neutrophils [12]. This neutrophil-driven IFN signature is also validated at the protein level [13]. In Mtb-resistant mouse models, two different waves of neutrophils are observed after Mtb infection, before and after the onset of adaptive immunity [14]. The neutrophils of the first wave take up mycobacteria by phagocytosis in situ in the lung, whereas the second wave is T cell-dependent, with neutrophils seldom associating with the bacilli, instead establishing close contact with T cells [14], helping form the structured mature granuloma [15, 16]. Early neutrophils generated during the innate phase of the response to infection are therefore key to the containment of the bacilli, either because they kill the bacilli directly, or because they contribute to the formation of the beneficial granuloma that restricts bacterial multiplication. By contrast, uncontrolled and chronic neutrophil influx is also associated with an inflammation burst, tissue damage and a poor clinical outcome in active TB [17, 18].

Neutrophils were long considered to be homogeneous, mostly pro-inflammatory cells capable of reaching sites of infection and rapidly eliminating microorganisms and other dangers. However, this restricted view of neutrophils has evolved considerably over the last decade. It is now well established that neutrophils are heterogeneous [19–21] and form discrete subtypes with a broad range of functions, including anti-inflammatory [22, 23] and regulatory roles [24]. We recently identified a new subset of neutrophils – “regulatory” (N^reg) neutrophils – circulating in both cattle and the mouse model [25]. Like conventional pro-inflammatory neutrophils (N^conv), N^reg have a classical polymorphonucleus and these two types of neutrophils have similar cell-surface markers. However, unlike N^conv, N^reg express major histocompatibility complex class II (MHC-II) on their surface, together with PD-L1 in mice (PD-L1 expression in bovine N^reg cells remains to be determined, as available commercial antibodies did not meet our validation criteria). At steady state, N^reg can suppress T lymphocyte proliferation, thereby acting as key partners in the adaptive immune response. We showed that N^conv exacerbate local inflammation in mouse models of TB, through caspase-1-dependent IL-1β production, accelerating inflammation by maintaining a vicious circle of inflammatory neutrophils and CD4⁺ T cells. By contrast, N^reg act through PD-L1 to block T-cell proliferation and IFN-γ production, thereby reducing inflammation [26].

Our knowledge of N^reg function during infections in cattle remains limited. We recently demonstrated that both the N^reg and N^conv subsets are present in milk from healthy Holstein and Normande cows and that, during mastitis, milk N^reg have a greater bactericidal capacity than their classical counterparts [27]. We also observed a positive correlation between the counts of N^reg and T cells in milk during infection.

Neutrophils play crucial roles in the pathophysiology of TB that differ according to disease stage [28]. We therefore further investigated the functions of the newly described N^reg in controlling Mb at early stages of disease. We used two strains of M. bovis – the virulent Mb3601 strain circulating in France and the attenuated BCG vaccine strain – to confirm that N^conv and N^reg had different transcriptional signatures. We identified key biological pathways that differed markedly between the two neutrophil subsets. We then compared the functions of N^reg and N^conv, including bacterial killing, reactive oxygen species (ROS) production and phagocytosis. Finally, we performed electron microscopy to compare the ultrastructure of the two subsets of neutrophils. These analyses showed that the two neutrophil subsets interact differently with Mb in vitro. Our data pave the way for a reconsideration of the role of neutrophils in the pathophysiology of bTB.

Materials and Methods

Bacterial Strains – Growth Conditions and Inoculum Preparation in a Biosafety Level 3 Laboratory

Mb3601 is a French Mb strain isolated by the French bTB reference laboratory at ANSES from an infected cow in Bourgogne-Franche-Comté. The BCG-WT 1173P2 vaccine strain was obtained from the Pasteur Institute. Recombinant fluorescent BCG mCherry and Mb3601 mCherry strains were obtained by electroporation with the pNIP40-mCherry integrative plasmid and selection on hygromycin B (50 μg/mL, Sigma), as previously described [29]. Bacteria were grown in flasks containing 7H9 broth, 10% ADC, 0.05% Tween 80, and sodium pyruvate (38 mm). At mid-exponential growth phase, the bacteria were harvested, tittered, dispensed into aliquots and stored at −80°C until use. BCG-WT batch titers were determined by plating serial dilutions on 7H11 agar supplemented with 0.5% glycerol and 10% OADC (ADC supplemented with 0.005% oleic acid). Mb3601 batch titers were determined by plating serial dilutions on 7H11 agar supplemented with 10% OADC, 10% calf serum and 0.005% decomplemented sheep’s blood diluted 1:2 in sterile milliQ water. The plates were incubated at 37°C for up to 3 weeks and the colony-forming units (CFUs) were then counted. For each strain and each experiment, inoculum was prepared by thawing one frozen aliquot in 7H9 medium, 10% ADC, 0.05% Tween 80 and incubating it for half a day at 37°C. The aliquot was then centrifuged for 10 min at 1,700 g, and bacterial concentration was adjusted with RPMI medium, as appropriate, for each experiment.

Cow’s Blood Samples

Blood samples were collected from primiparous Prim’Holstein cows from the Orfrasière Animal Physiology Experimental Unit (UE-PAO) at INRAE Centre Val-de-Loire. Blood samples were collected from the jugular vein into 10 mL K2 EDTA Vacutainer tubes and were kept at room temperature until processing (<1 h). All the cows were housed together and our protocol did not interfere with routine herd management. Blood sampling was approved by the Ethics Committee for Animal Experimentation Val-de-Loire (CEEA VdL), registered with the French National Committee for Animal Experimentation (APAFIS no. 202007241825481.V2), and complied with EU regulation 2010/63.

Cell Preparation from Blood Samples

Tubes containing blood were centrifuged at 1,000 g for 30 min at 20°C (acceleration 9/brake 7). The plasma layer and the buffy coat were removed. ACK lysis buffer (Thermo Fisher Scientific, 4 vol buffer/1 vol of blood) was added and the mixture was incubated for 5 min at room temperature to lyse the red blood cells. The remaining cells were washed twice with Dulbecco phosphate-buffered saline (D-PBS) and then resuspended in FACS buffer (D-PBS, 2 mm EDTA, 10% horse serum) for further labeling, or neutroRPMI (RPMI supplemented with 2 mml-glutamine, 10 mm HEPES, and 1 mg/mL BSA with extremely low endotoxin level (≤2.5 endotoxin units/mL, Sigma) for cell culture. Cells were stained with Acridine Orange fluorescent stain (Logos Biosystems) and counted with a Luna-FL™ Automated Fluorescence Cell Counter (Logos Biosystems).

Cell Sorting by Flow Cytometry

All the antibodies and probes used in this study are listed in Table 1, together with the clones, suppliers, and working concentrations or dilutions. Blood cells were labeled by incubation with the primary antibodies (G1 and MHC-II) for 20 min at 4°C. Cells were washed in D-PBS, labeled with the corresponding fluorescently labeled secondary antibody (conjugated with Alexa Fluor 647 and Alexa Fluor 488) and a Fixable Viability Dye (FVD-eFluor 780). Cells were then resuspended in neutroRPMI and live neutrophils (FVD⁻ G1⁺ MHC-II⁻ and FVD⁻ G1⁺ MHC-II⁺ cells) were obtained by sorting on a MoFlo Astrios EQ high-speed cell sorter (Beckman Coulter) according to the protocol described by Rambault et al. [30].

Table 1.

List of antibodies and probes used in this study

Target or name	Isotype	Clone	Working concentration or dilution	Reference	Supplier
Anti-bovine G1	Mouse IgM	CH138A	2 µg/mL	WSC0608B-100	Clinisciences
Anti-feline MHC-II	Mouse IgG1	CAT82A	2 µg/mL	WSC0525F-100	Clinisciences
Anti-mouse IgM Alexa Fluor 647	Goat IgG	Polyclonal	10 µg/mL	A21238	Thermo Fisher Scientific
Anti-mouse IgG1 Alexa Fluor 488	Goat IgG	Polyclonal	10 µg/mL	A21121	Thermo Fisher Scientific
Anti-mouse IgG1 phycoerythrin (PE)	Rat IgG	RMG1-1	1 µg/mL	406608	BioLegend
Fixable Viability Dye (FVD) eFluor 780	–	–	1:10,000	65-0865-14	Thermo Fisher Scientific
Viability dye Zombie Aqua	–	–	1:500	423102	BioLegend
Sulforhodamine-FLICA Polycaspase	–	–	1:250	ICT916	Bio-Rad
CellROX Green	–	–	5 µm	C10444	Thermo Fisher Scientific
MitoTracker Orange	–	–	250 nm	MT7510	Thermo Fisher Scientific
Hoechst 33342	–	–	20 µm	62249	Thermo Fisher Scientific

RNA Extraction

Sorted neutrophils were infected with BCG (BCG-WT at a multiplicity of infection [MOI] of 10) and Mb (Mb3601 at a MOI of 1) strains. The plates were incubated for 2 h at 37°C. The supernatant was then discarded and Tri-Reagent was added to each well. Technical duplicates or triplicates were pooled in an Eppendorf tube and kept at −80°C until RNA extraction. The initial step of RNA extraction consisted of a pre-extraction with chloroform (5:1, vol/vol), followed by vigorous shaking and incubation for 10 min at room temperature. The samples were then centrifuged at 12,000 g for 10 min at 4°C, and the aqueous phase was collected and mixed with isopropanol (1:1, vol/vol). Total RNA was then purified using the NucleoSpin RNA kit (Macherey-Nagel) in accordance with the manufacturer’s instructions, with a DNase treatment step.

RNA Sequencing and Differential Expression Analysis

RNA sequencing was performed by an external service provider (Helixio, France, https://www.helixio.fr/page/rna-seq). RNA integrity was assessed before sequencing (RIN ranging from 5.7 to 9.6 [mean ± scanning electron microscopy [SEM]: 7.70 ± 0.22]). Raw paired-end reads were assessed with FastQC (v0.12.1) [31]. Adapter sequences and low-quality bases were trimmed off with fastp (v0.23.4) [32] in paired-end mode, and only cleaned paired reads were retained. Post-trimming read quality was evaluated with FastQC. Transcripts were quantified by pseudoalignment with the Bos taurus transcriptome (ARS-UCD1.3 assembly, Ensembl cDNA release 113) in Salmon (v1.10.0) [33], using an index built with a k-mer size of 31. The --validateMappings option was enabled, and bias corrections were applied for effective transcript length, sequence composition, and GC content [33]. Transcript-level abundance estimates (quant.sf files) were imported into R (v4.4.2) with tximport (v1.34.0) [34] and summarized at gene level with a transcript-to-gene mapping table derived from the Ensembl annotation, ignoring transcript version suffixes. The resulting gene-level counts were used to construct DESeqDataSet objects with DESeq2 (v1.46.0) [35]. Weakly expressed genes (fewer than 5 normalized counts in at least 6 samples) were filtered out before differential expression analysis. Size factors were estimated with DESeq2, and variance-stabilizing transformation was applied for exploratory analyses. Principal component analysis (PCA) was performed to visualize the global structure of the dataset by neutrophil subtype (N^conv vs. N^reg) and experimental conditions. Differential expression between regulatory (N^reg) and conventional (N^conv) neutrophils was assessed separately for each set of experimental conditions (uninfected, BCG, Mb3601). For each set of experimental conditions, a DESeq2 model with the design formula ∼CellType was fitted, using N^conv as the reference. Log₂ fold-changes were moderated by the apeglm method [36]. Genes were considered differentially expressed if the adjusted p value (Benjamini-Hochberg method) was <0.05 and |log₂FC| ≥ 0.58 [37].

Functional enrichment analyses were then performed separately for each gene set derived from condition-specific and intersecting signatures, distinguishing genes upregulated in N^reg and N^conv. Enrichment analyses were performed with g:Profiler (gprofiler2 package [38, 39]) for B. taurus, applying an overrepresentation analysis framework and querying the Gene Ontology (Biological Process, Molecular Function, Cellular Component [40, 41]), Reactome and KEGG databases. False-discovery rate correction was applied, and terms displaying significant enrichment were visualized on dot plots. The transcriptomic data have been deposited in the European Nucleotide Archive (ENA) under BioProject accession number PRJEB106984.

Cell Survival

Blood cells were labeled with primary antibodies (G1 and MHC-II, see Table 1) for 20 min at 4°C. Cells were washed in D-PBS and labeled with the corresponding fluorescent secondary antibody (conjugated with Alexa Fluor 647 or Alexa Fluor 488) and the Zombie Aqua viability dye. Cells were washed in D-PBS and resuspended in neutroRPMI. Neutrophil apoptosis was evaluated with the Sulforhodamine (SR-)FLICA™ Poly Caspase Kit according to the manufacturer’s instructions. Briefly, cells were infected with BCG-WT (at a MOI of 10) or Mb3601 (at a MOI of 1), with the addition of SR-FLICA™ Poly Caspase reagent at the same time as the bacteria. Staurosporine (STR, working concentration 2 µm) was used as a positive control for this experiment. The plates were incubated at 37°C for various times: 30 min, 2 h, 4 h and overnight. The cells were then fixed by incubation with 4% paraformaldehyde (PFA) for at least 2 h. Fluorescence was measured with an LSR Fortessa™ X-20 Flow Cytometer (BD Biosciences). Flow cytometry data were analyzed with Kaluza Analysis software (2.1 version, Beckman Coulter).

Killing Assay

Sorted neutrophils were infected with Mb (BCG-WT at a MOI of 10 or Mb3601 at a MOI of 1). The plates were incubated for 2 h at 37°C. The supernatant was collected and the cells were lysed in D-PBS supplemented with 0.1% Triton X-100. Both supernatant and lysate were plated at appropriate dilutions on plates containing solid medium of the composition described above. Killing was assessed 2 h after infection, by counting CFUs in the supernatant and cell lysate. The percent killing was determined as CFUs in the supernatant + CFUs in the neutrophil lysate, multiplied by 100 and divided by the total number of bacteria in the inoculum, which was verified for each experiment.

Phagocytosis Assay

Blood cells were labeled by incubation with primary antibodies (G1 and MHC-II) for 20 min at 4°C. The cells were washed in D-PBS and labeled with the corresponding fluorescent secondary antibody (conjugated with Alexa Fluor 647 and Alexa Fluor 488) and the FVD-eFluor 780. The cells were washed in D-PBS and resuspended in neutroRPMI. Labeled cells were then infected with recombinant BCG-mCherry (at a MOI of 10) or recombinant Mb3601-mCherry (at a MOI of 1). At various time points, infected cells were centrifuged at 400 g for 5 min and the supernatant was discarded. We then added 4% PFA to each tube and incubated the tubes overnight. Fluorescence was measured with a LSR Fortessa™ X-20 Flow Cytometer. Flow cytometry data were analyzed with Kaluza Analysis software, 2.1 version.

ROS Production Assay

Blood cells were labeled by incubation with primary antibodies (G1 and MHC-II) for 20 min at 4°C. in the cells were washed in D-PBS and labeled with the corresponding fluorescent secondary antibody (conjugated with Alexa Fluor 647 and phycoerythrin) and the FVD-eFluor 780 viability dye was added. The cells were washed with D-PBS and resuspended in neutroRPMI. ROS production by neutrophils was assessed with the CellROX™ Green Flow Cytometry Assay Kit according to the manufacturer’s instructions. Briefly, cells were infected with BCG-WT (at a MOI of 10) or Mb3601 (at a MOI of 1), with the addition of CellROX Green™ at the same time as the bacteria. The plate was incubated for 2 h at 37°C. The cells were then fixed by incubation with 1% PFA overnight. ROS fluorescence was measured with an LSR Fortessa™ X-20 Flow Cytometer. Flow cytometry data were analyzed with Kaluza Analysis software, 2.1 version.

Mitochondrial Imaging by Confocal Microscopy and Image Analysis

Sorted neutrophils were placed in an 18 well microslide (IBIDI) and incubated with MitoTracker Orange CMTMRos for 30 min and Hoechst 33342 for 10 min at 37°C. The cells were then washed with D-PBS and imaged by confocal laser scanning microscopy (CLSM) (AX NSPARC with WI 60x/1.2 objective, Nikon Europe B.V.). 3D images were analyzed using the GA3 module of the NIS-element software (Advanced research 6.20.02 version, Nikon). Briefly, the cells and their mitochondrial content were segmented in 3D by thresholding. The ratio of the volume occupied by the mitochondria to the total volume of the cell was then calculated.

Scanning Electron Microscopy

BCG- and Mb3601-infected (2 h at a MOI of 1 for both) cells, as well as non-infected cells, were plated on glass slides and fixed by incubating for 24 h in 1% glutaraldehyde, 4% paraformaldehyde (Sigma) in 0.1 m phosphate buffer (pH 7.2). Samples were washed in PBS and were then fully dehydrated in a graded series of ethanol solutions and dried in hexamethyldisilazane. Finally, samples were coated with 40 Å platinum, with a GATAN PECS 682 apparatus (Pleasanton), before observation under a Zeiss Ultra plus FEG-SEM scanning electron microscope (Oberkochen).

Transmission Electron Microscopy

BCG- and Mb3601-infected (2 h at a MOI of 1 for both) cells, as well as non-infected cells, were fixed by incubating for 24 h in 1% glutaraldehyde, 4% PFA (Sigma) in 0.1 m phosphate buffer (pH 7.2). Samples were then washed in PBS and post-fixed by incubation for 1 h with 2% osmium tetroxide (Agar Scientific). Cells were then fully dehydrated in a graded series of ethanol solutions and propylene oxide. They were impregnated with a mixture of (1∶1) propylene oxide/Epon resin (Sigma) and left overnight in pure resin. Samples were then embedded in Epon resin (Sigma), which was allowed to polymerize for 48 h at 60°C. Ultra-thin sections (90 nm) of these blocks were obtained with a Leica EM UC7 ultramicrotome (Wetzlar). Sections were stained with 2% uranyl acetate (Agar Scientific), 5% lead citrate (Sigma) and observed in a transmission electron microscope (JEOL 1011). For quantitative analysis, ultrastructural features were monitored, with NIS-elements D5 software (Nikon). For the quantification of granules neutrophils, cells of comparable size with a clearly visible nucleus and optimal contrast were selected. Granules were manually counted in five individual cells. Granule size measurements were performed on the same micrographs used for granule quantification. Images were recalibrated prior to measurement by defining the scale bar displayed on each micrograph as the reference distance. Approximately 30 granules per cells were measured in five neutrophils.

Statistical Analysis

Statistical analyses were performed with Prism 10.5.0 software (GraphPad), on data from two to eight independent experiments. Depending on the dataset, the following statistical tests were applied: the non-parametric Wilcoxon matched-pairs signed-rank test, the non-parametric Mann-Whitney test, a mixed-effects model fitted using restricted maximum likelihood (REML), or a two-way ANOVA. The p values obtained were interpreted as follows: ns = non-significant; *p < 0.05; **p < 0.01; ***p < 0.001, ****p < 0.0001. Individual data points are shown together with the median or mean, as indicated in the corresponding figure legends.

Results

N^reg and N^conv Have Different Transcriptional Profiles and Activate Different Programs in Response to Mb Infection

We compared the transcriptomes of N^conv and N^reg both at steady state and after Mb infection. We purified the two subsets from blood, infected them by incubation for 2 h with avirulent BCG at a MOI of 10, or virulent Mb3601 at a MOI of 1 and conducted a bulk RNA-seq analysis (Fig. 1a). An unsupervised PCA revealed clear differences between N^reg and N^conv (Fig. 1b) at steady state. The first axis, which accounted for 84% of the variance, separated the two subsets, whereas the second axis, which accounted for only 3% of the variance, did not distinguish clearly between the groups. The strong differential signature between N^conv and N^reg eclipsed the more subtle effect of bacterial infection in the PCA: uninfected, BCG-infected and Mb3601-infected samples were not clearly separated. We further investigated the impact of Mb infection on the transcription program of the two subsets of neutrophils, by separately comparing the impact of each set of infection conditions on N^reg and N^conv in the DESeq2 model. We identified 4,854 genes commonly modulated across conditions, 582 of which were specifically regulated by BCG infection and 406 by Mb infection (Fig. 1c). Using the Gene Ontology knowledgebase, we compiled a list of the top biological pathways in the core set of genes differentially expressed between our two neutrophil subsets (Fig. 1d), and then the core sets of genes modulated by infection with the avirulent BCG (online suppl. Fig. 1a; for all online suppl. material, see https://doi.org/10.1159/jin/adhag002) and virulent Mb3601 strains (online suppl. Fig. 1b). For N^conv, the biological processes identified were mostly related to the immune system response and defense (cell migration, cytokines, killing, autophagy, viral immune responses). For N^reg, the specific biological processes identified were mostly related to metabolic processes and cellular maintenance (cellular biogenesis, protein synthesis, DNA repair and metabolism). Only two pathways were significantly modulated by BCG infection, and both were upregulated in N^reg: mitotic nuclear division and mitochondrial RNA metabolic processes (online suppl. Fig. 1a). By contrast, after Mb3601 infection, 20 pathways were upregulated in N^conv, mostly related to type-I interferon-driven immune responses and T cell activation and proliferation (online suppl. Fig. 1b). These pathways are representative of the neutrophil-driven dysregulation of the immune response that occurs during active TB in humans [12]. The other pathways identified were related to stress-induced apoptotic processes, with additional contributions from metabolic and epithelial transport pathways. Mycobacterial infection, thus, clearly induced different transcriptional signatures and activated different biological processes in N^reg and N^conv, particularly after infection with the virulent Mb strain Mb3601.

Fig. 1.

N^reg and N^conv have different transcriptional profiles. a Schematic representation of the sequence of steps in the experiment. Cattle blood cells were sorted to separate the two subsets of neutrophils. N^reg and N^conv were incubated for 2 h either with medium or Mb strains (BCG-WT at a MOI of 10, Mb3601 at a MOI of 1). RNA was then extracted from these samples and sequenced (RNAseq). b Principal component analysis (PCA) was conducted and the two first dimensions of the PCA plot are shown. c Venn diagram illustrating the number of genes for each set of conditions, including those common to multiple conditions. d Dot plot of the top 20 Gene Ontology Biological Process pathways for each neutrophil subtype, for all conditions considered together (4,854 differentially expressed genes [DEGs]). Pathways are sorted from top to bottom according to their p values, all of which are significant (p < 0.05). Orange dots represent the pathways upregulated in N^reg and downregulated in N^conv, whereas blue dots represent pathways that are upregulated in N^conv and downregulated in N^reg. Dot size is proportional to Gene Count. The data shown are for individual samples of neutrophils from n = 6–11 independent cows, pooled data from 11 independent experiments.

Activation Results in a Shorter Lifespan in N^reg than in N^conv

Neutrophils are known to be short-lived, but infection or other danger signals can considerably modify their lifespan [42]. Lifespan may also differ between subsets. We therefore compared the cell death and apoptosis of uninfected and BCG-infected (at a MOI of 10) or Mb3601-infected (at a MOI of 1) N^reg and N^conv by flow cytometry, with Zombie Aqua as a viability dye and SR-FLICA polycaspase as a marker of apoptosis (Fig. 2a). Staurosporine (STR, 2 µm) was used as a positive control for apoptosis and cell death. N^reg died significantly more rapidly than N^conv, 20% of the cells dying during the first 30 min of incubation (Fig. 2b). Cell death rates increased over time, with approximately 40–70% of N^reg dying overnight (>16 h), versus only 5–35% of N^conv in both the presence and absence of infection. As expected, neutrophils were sensitive to STR, which accelerated cell death, beginning after 2 h of incubation. Interestingly, uninfected and infected Mb3601 neutrophils had similar death kinetics, whereas BCG infection induced significantly higher levels of cell death in both subpopulations. Consistent with the observed cell death kinetics, apoptosis rates were higher for N^reg than for N^conv (Fig. 2c). This difference was already visible at 30 min and remained stable for up to 4 h unless STR was added, as expected. Overnight, apoptosis rates approached 50% for both subsets of neutrophils, except for Mb3601-infected, which had lower apoptosis rates. Nevertheless, even for Mb3601 infection, apoptosis rates remained consistently higher for N^reg than for N^conv (2-fold difference), except for the overnight incubation (1.2-fold difference).

Fig. 2.

N^reg have a higher rate of cell death and die faster than N^conv. Cattle blood cells were centrifuged and granulocytes were isolated by removing the plasma layer and buffy coat and lysing the red blood cells. Granulocytes were then labeled. a Gating strategy for quantifying neutrophil death and apoptosis, with data for 1 representative animal shown. Eosinophils were excluded on the basis of their autofluorescence for BV421. Dead blood N^conv are defined as SSC-A^high FSC-A^high BV421⁻ G1⁺ MHC-II⁻ Zombie Aqua⁺ cells, whereas dead blood N^reg are defined as SSC-A^high FSC-A^high BV421⁻ G1⁺ MHC-II⁺ Zombie Aqua⁺ cells. Blood N^conv undergoing apoptosis are defined as SSC-A^high FSC-A^high BV421⁻ G1⁺ MHC-II⁻ SR-FLICA⁺ cells, whereas blood N^reg undergoing apoptosis are defined as SSC-A^high FSC-A^high BV421⁻ G1⁺ MHC-II⁺ SR-FLICA⁺ cells. Staurosporine (STR) was used to induce cell death as a positive control. b After labeling, cells were infected with BCG at a MOI of 10 or Mb3601 at a MOI of 1 or were treated with STR. Cell death rates were assessed by staining with a viability dye (Zombie Aqua). c After labeling, cells were infected with BCG at a MOI of 10 or Mb3601 at a MOI of 1. Cell apoptosis was assessed with a SR-FLICA polycaspase kit (SR-FLICA) detecting active caspases in neutrophils. b, c The data shown are the median and interquartile range for neutrophils from n = 7–8 independent cows, in two independent experiments. “Time” effect refers to the different timepoints analyzed, “Cond” effect refers to the comparison between N^conv and N^reg under different stimulation conditions. ns, non-significant; **p < 0.01; ****p < 0.0001 in the mixed-effect Restricted Maximum Likelihood Model (REML).

N^reg Have a Similar Level of Microbicidal Activity but Higher Levels of Phagocytosis and ROS Production than N^conv

We investigated the functions of the two neutrophil subsets in greater detail by first comparing their overall killing activity. We sorted N^reg and N^conv by flow cytometry and infected them with BCG at a MOI of 10 or Mb3601 at a MOI of 1. We counted CFU both in the supernatant (viable extracellular bacilli) and within neutrophils (viable intracellular bacilli) to determine overall bacterial killing rates. The two subsets of neutrophils had similar abilities to kill nonvirulent and virulent strains of Mb, for N^reg and N^conv, the respective mean ± SEM killing percent were 82.0 ± 10.1% and 66.0 ± 17.7% for BCG and 57.4 ± 9.5% and 45.6 ± 5.3% for Mb3601 (Fig. 3 and online suppl. Fig. 2). In order to correlate phagocytosis and killing, we infected mixed unsorted neutrophils with recombinant mCherry-BCG strain at a MOI of 10 or mCherry-Mb3601 strain at a MOI of 1 and assessed phagocytosis by flow cytometry at multiple time points (Fig. 4a). Interestingly, at all time points, N^reg had significantly higher levels of phagocytic activity than N^conv, regardless of the virulence of the strain (Fig. 4b, c). Phagocytosis rates were consistently higher for BCG than for Mb, for both subpopulations of neutrophils (1.5-fold difference for BCG and 2-fold difference for Mb3601). Finally, we assessed ROS production by unsorted N^reg and N^conv after infection with BCG at a MOI of 10 and Mb3601 at a MOI of 1 for 2 h. We used the CellROX Green probe to measure hydroxyl radical (HO) and superoxide anion (O₂⁻) levels. Mean fluorescence intensity was quantified by flow cytometry (Fig. 5a) and directly correlated with ROS production. Strikingly, N^reg produced 1.6 times more ROS at baseline than N^conv. This trend was maintained when neutrophils were infected, with significantly higher levels of ROS production observed for N^reg (1.8-fold difference for BCG and 2.3-fold difference for Mb3601 Fig. 5b).

Fig. 3.

N^reg has a capacity to kill BCG and Mb3601 similar to that of N^conv. After cell sorting, cells were infected with BCG at a MOI of 10 and Mb3601 at a MOI of 1 for 2 h. Both supernatants and lysates were plated and the number of colonies was counted to determine the killing rates as a percentage. The data shown are the median and interquartile range for both subsets of neutrophils from n = 3–8 independent cows, in 8 independent experiments. ns, non-significant in the non-parametric Wilcoxon matched-pairs signed-rank test.

Fig. 4.

N^reg performs phagocytosis more effectively than N^conv. Granulocytes from cattle blood were treated and labeled as described in Figure 2. a Gating strategy for the evaluation of phagocytosis; data for one representative animal shown. Eosinophils were excluded on their autofluorescence for BV421. Viable blood N^conv phagocytosing bacteria are defined as SSC-A^high FSC-A^high BV421⁻ FVD⁻ G1⁺ MHC-II⁻ mCherry⁺ cells, whereas viable blood N^reg phagocytosing bacteria are defined as SSC-A^high FSC-A^high BV421⁻ FVD⁻ G1⁺ MHC-II⁺ mCherry⁺ cells. After labeling, cells were infected with BCG-mCherry (b) at a MOI of 10 or Mb3601-mCherry (c) at a MOI of 1. The data shown are the mean values and before-after comparisons for the two subsets of neutrophils from n = 8 independent cows, in 2 independent experiments. *p < 0.05; **p < 0.01 in non-parametric Wilcoxon matched-pairs signed-rank tests.

Fig. 5.

N^reg produce higher levels of reactive oxygen species (ROS) than N^conv. Cattle granulocytes from blood were treated and labeled as described in Figure 2. a Gating strategy for the evaluation of ROS production, with data for 1 representative animal shown. Eosinophils were excluded from whole granulocytes on the basis of their autofluorescence for BV421. Viable blood N^conv are defined as SSC-A^high FSC-A^high BV421⁻ FVD⁻ G1⁺ MHC-II⁻ CellROX Green⁺ cells, whereas viable blood N^reg are defined as SSC-A^high FSC-A^high BV421⁻ FVD⁻ G1⁺ MHC-II⁺ CellROX Green⁺ cells. After cell labeling. b Oxidative stress was measured in both neutrophil subpopulations in the presence and absence of infection, over a period of 2 h, with the CellROX Green probe. This probe reacts with certain types of ROS species, such as the hydroxyl radical (HO) and the superoxide anion (O₂⁻). Mean fluorescence intensity (MFI) was quantified by flow cytometry. The data shown are the mean and before-after comparisons between the two subsets of neutrophils from n = 7–11 independent cows, in 4 independent experiments.

N^reg Have Higher Levels of Mitochondrial Activity than N^conv

RNA-seq identified many pathways related to mitochondrial organization and metabolism (Fig. 1c), and ROS production was markedly higher in N^reg than in N^conv (Fig. 5b). We therefore decided to investigate the mitochondrion-associated signature in the two subsets. We found that genes involved in mitochondrion- and ribosome-related pathways were strongly enriched in N^reg, whereas the N^conv signature was dominated by vesicles and vacuolar contents (Fig. 6a). Using the MitoTracker Orange probe, which binds to active mitochondria, we analyzed and quantified mitochondrial volume by confocal microscopy on sorted neutrophils (Fig. 6b, c). The cellular volume showed no significant difference between the two groups, while both the total mitochondrial volume and the mitochondrial volume normalized to the cellular volume were significantly different between the two groups (Fig. 6c). These results indicate that, despite similar cell sizes, N^reg displayed a significantly higher mitochondrial volume per cell (1.5-fold difference), suggesting a greater mitochondrial content and/or mitochondrial occupancy compared with N^conv. These findings are consistent with our observations for ROS production, as mitochondria can generate hydroxyl radicals and superoxide anions, and the levels of these ROS were markedly higher in N^reg than in N^conv.

Fig. 6.

N^reg have a higher mitochondrial load than N^conv. a Dot plot of the top 20 cellular contents identified for each neutrophil subtype in uninfected conditions (8111 DEG). Pathways are sorted from top to bottom according to their p value, all of which are significant (p < 0.05). Orange dots represent pathways that are upregulated in N^reg and downregulated in N^conv, whereas blue dots represent pathways that are upregulated in N^conv and downregulated in N^reg. Dot size is proportional to Gene Count. b Cattle blood N^conv and N^reg imaged after cell sorting. Nuclei are labeled with Hoechst stain and mitochondria are labeled with Mitotracker Orange CMTMRos, 1 representative animal. c Cell volume and mitochondrial volume were quantified with NIS-Element software and GA3 modules, and mitochondrial volume per cell was then calculated. a The data shown are for individual samples of neutrophils from n = 6–11 independent cows, pooled data from 11 independent experiments. c The data shown are for n = 31–37 neutrophils from 1 cow, 1 experiment. ns, non-significant; *p < 0.05 in non-parametric Mann-Whitney test.

N^reg and N^conv Have Different Membrane and Granule Ultrastructural Organizations

N^reg and N^conv cannot be distinguished on the basis of their phenotypes by conventional microscopy [25]. We therefore studied these two subtypes in more detail by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). On SEM, N^conv were characterized by a predominantly spherical morphology with a smooth, regular cell surface (Fig. 7a). The plasma membrane had a uniformly microfolded surface consisting of short, evenly distributed membrane protrusions, resulting in a finely textured appearance. By contrast, N^reg had a more heterogeneous morphology, with frequent irregularly shaped or ovoid cells (Fig. 7b). The plasma membrane of N^reg had a heterogeneous topography, combining deeply folded regions with prominent, thick membrane ruffles. In addition, N^reg frequently displayed long, thin membrane extensions, such as filopodium-like structures and tether- or nanotube-like projections, some of which established direct connections between neighboring cells (Fig. 7b). Similar ultrastructural differences between N^conv and N^reg are also observed under infected conditions (Fig. 7c–f). Thus, the two subsets differed markedly in terms of cell shape, membrane organization, and the abundance of extended membrane protrusions, indicating an intrinsic morphological difference between N^conv and N^reg, reflected in differences in surface architectures and membrane remodeling states.

Fig. 7.

Phenotypic heterogeneity of the two neutrophil subsets on SEM. Sorted cells were infected with BCG or Mb3601 at the MOI of 1 for 2 h. N^reg and N^conv were fixed and prepared for SEM acquisition. Images of uninfected N^conv (a) and N^reg (b), BCG-infected N^conv (c) and N^reg (e), and Mb3601-infected N^conv (d) and N^reg(f). The data shown are for 5 neutrophils from the same cow, from 1 representative experiment out of 3 experiments.

In TEM analysis, N^conv had a finely textured and densely packed cytoplasm (Fig. 8a). The plasma membrane of N^conv had few ultrastructural irregularities, with only short protrusions and modest membrane ruffling. The cortical cytoplasm appeared relatively uniform, without large membrane invaginations or extended cytoplasmic projections, consistent with the SEM observations. The plasma membrane of N^reg frequently had pronounced ultrastructural irregularities, including membrane ruffles, protrusions, and occasional thin cytoplasmic extensions (Fig. 8b). The peripheral cytoplasm appeared less uniform, with localized thinning and regions suggestive of enhanced membrane remodeling or granules exocytosis, consistent with the SEM findings. In both the N^conv and N^reg subpopulations, and following exposure to BCG (Fig. 8c, d) or Mb3601 (Fig. 8e, f), direct neutrophil-bacterium contacts, phagocytic events, and intracellular bacteria enclosed within membrane-bound vacuoles were observed. Some N^reg had ultrastructural features suggestive of an increase in basal activity, including pronounced membrane remodeling and cytoplasmic or plasma membrane areas consistent with early or partial degranulation.

Fig. 8.

Phenotypic heterogeneity of the two neutrophil subsets in TEM. The sorted cells were infected with BCG or Mb3601 at the MOI of 1 for 2 h. N^reg and N^conv were fixed and prepared for TEM acquisition. Images of uninfected N^conv (a) and N^reg (b), BCG-infected N^conv (c) and N^reg (e), and Mb3601-infected N^conv (d) and N^reg (f). Mycobacteria are indicated by yellow arrows. Quantitative analysis of granule size (g) and granule number (h) for each neutrophil subset in the presence and absence of infection. The data shown are the medians for 5 neutrophils from the same cow, pooled from n = 2 independent cows in 2 independent experiments out of 3 experiments. “Neu” refers to the neutrophil type, “Inf” refers to the infection condition (type of stimulus). ns, non-significant; **p < 0.01; ****p < 0.0001 in two-way ANOVA test.

N^conv displayed an evenly distributed granule population with a markedly heterogeneous morphology: a mixture of round, ovoid, and elongated (elliptical) shapes (Fig. 8a). By contrast, N^reg were characterized by the presence of less numerous and larger cytoplasmic granules, which appeared more homogeneous in shape, most being regular and round (Fig. 8b). However, the granules were unevenly distributed within the cytoplasm, due in part to the presence of large vacuolar or vesicular compartments resulting in areas of relative granule depletion. Quantitative TEM analysis revealed distinct differences in granule size (Fig. 8g) and abundance (Fig. 8h) between N^conv and N^reg, depending on infection status. In the absence of infection, N^reg contained significantly larger granules (1.2-fold difference), but fewer granules per cell compared to N^conv (0.8-fold difference). Following infection, N^conv retained a higher granule count and exhibited a significant increase in mean granule size particularly in response to the virulent Mb3601 strain (1.1-fold difference for BCG, 1.5-fold difference for Mb3601). In contrast, N^reg showed an overall reduction in granule number, while granule size remained relatively large. This divergence suggests differential remodeling of the granular compartment in response to the virulent Mb3601 strain.

Discussion

In humans and mouse models, neutrophils are considered as key cells in the pathophysiology of tuberculosis, exerting either protective or deleterious effects depending on the stage and severity of the disease [28, 43]. Over the last decade, neutrophil heterogeneity has been widely described in homeostatic conditions, cancers, inflammation and infectious diseases. It is now well established that different neutrophil subsets promote pro- or anti-inflammatory responses and modulate both innate and adaptive immunity.

Little is known about the role of bovine neutrophils during Mb infection. As in mice, sequential studies have revealed a prominent neutrophil component of the host response to Mb in early stage lesions [44, 45]. Wang et al. [46] detected a neutrophil gene signature in Mb-infected cattle but in vitro, neutrophils did not appear to be able to eliminate the bacteria [47]. Another study suggested a deleterious role of neutrophils through strong IL-17 production, contributing to the pro-inflammatory response in lung lesions, and the production of arginase, decreasing Mb killing [48].

Following the discovery of the two neutrophil populations, N^conv and N^reg, circulating in cattle blood [25], this study demonstrates that these two populations have very different features, both at steady state and following infection with Mb in vitro, highlighting the need to reconsider the roles of these two bovine neutrophil subsets in the pathophysiology of bTB.

We found that N^conv and N^reg had different transcriptomic signatures. In N^conv, an enrichment was observed in the expression of genes involved in the immune functions expected for classical neutrophils, whereas the N^reg transcriptomic signature was mostly related to metabolic functions and cellular maintenance. Strikingly, the two neutrophil subsets also had different responses to the virulent Mb3601 strain, with only N^conv displaying an upregulation of pathways classically associated with disease activation and poor outcomes in TB, such as the type I interferon pathway [12, 49–51]. Maerzdorf et al. [52] described functional correlations between pathogenesis-driven gene expression signatures in human TB, such as immune inflammatory responses (involving the JAK-STAT pathway), apoptosis, responses to bacterial molecular patterns and the NF-kappa B cascade; all these key pathways are upregulated in N^conv and downregulated in N^reg, suggesting that N^conv is probably the subset responsible for the gene signature classically described in human TB. In an original ex vivo bovine lung explant model, we recently observed the local induction by Mb infection of a strong type I interferon response and a neutrophil recruitment signature [53]. Interestingly, Mb exploits the type I IFN signaling pathway for its own survival in bovine macrophages [54, 55]. It is unknown whether neutrophils contribute to this IFN type I signature during Mb infection in cattle, as demonstrated in human patients with active TB patients [12]. For the mycobacteria specific gene sets, we did not detect different transcriptomic signature of N^conv infected with Mb3601 at different MOIs (data not shown) indicating that the bacterial load does not influence the neutrophil response. Therefore, we believe that the different gene expression profiles of infected neutrophils that we observed between BCG (MOI of 10) and Mb3601 (MOI of 1)-that were selected to preserve cell viability-were driven primarily by the virulence of the strain, although we cannot totally exclude that MOI may have an impact.

The ability of neutrophils to kill Mtb or Mb remains a matter of debate [17]. Some authors have reported an impairment of killing by neutrophils in mice and humans [56, 57] whereas others have reported effective killing by neutrophils [58, 59]. Wang and coworkers investigated the effect of Mb infection on bovine neutrophils obtained from blood by gradient centrifugation [47]. They showed that these neutrophils were able to kill Mycobacterium smegmatis but not a virulent Beijing strain of Mb. Our data show clear bactericidal activity for both N^conv and N^reg against both M. bovis BCG and the French virulent strain Mb3601. This discrepancy may be due to the sorting of neutrophils in our study, a process ensuring a high purity of both subsets, facilitating their study but also activating them. We cannot exclude the possibility that the remaining Mb bacilli present within the neutrophils were able to replicate, increasing CFU counts at later time points.

N^reg share functional similarity to granulocytic myeloid-derived suppressor cells (G-MDSCs), as they are able to suppress T cell proliferation under steady state conditions [25]. G-MDSCs have been described as a permissive niche for Mtb replication, thereby facilitating immune evasion by the pathogen [60]. MDSCs have also been shown to support Mtb replication in the in vitro granuloma model, and to alter the structure of the granuloma [61]. They are also associated with disease progression and greater lethality in TB [62]. By contrast, N^reg display no such behavior, as we showed that they killed BCG and Mb3601 in vitro.

Neutrophils are widely considered to be professional phagocytes, able to engulf various pathogens, foreign particles or even cell debris. In 2021 and 2023, Rambault et al. [25, 27] reported that both neutrophil subsets could engulf E. coli bioparticles and whole E. coli-GFP by phagocytosis, with N^reg making them better phagocytes than N^conv. We also found that both neutrophil subsets were able to take up Mb by phagocytosis; however, N^reg again had a significantly greater phagocytic capacity than N^conv, at 30 min and after up to 4 h of infection. The rapid and enhanced phagocytic activity of N^reg is not, therefore, dependent on the bacterial strain, with this activity instead being a general, non-specialized property of this subtype. Remarkably, RNAseq analysis based on Gene Ontology Biological Process or Cellular Component identified various pathways and organelles associated with phagocytosis in N^conv only, with certain terms: autophagy, endocytosis, lysosome, early and late endosome, and vacuole. These data suggest that, upon initial contact with Mb, N^reg are already primed for phagocytosis, accounting for their greater phagocytic capacity at early time points. Interestingly, we previously reported higher levels of expression of CD11b [25], an important target for mycobacterial entry, on N^reg [63].

Reactive oxygen species (ROS) are able to kill pathogens, but their role in eliminating mycobacteria remains a matter of debate. Some in vitro studies have suggested that ROS production is required for the killing of Mtb [64–66], whereas others have reported resistance to oxidative stress in Mtb in mice [67] or impaired neutrophil killing by ROS in humans [57]. ROS are also known to act as potent immunomodulators in multiple contexts, influencing adaptive immunity through T and B cells [68–71]. N^reg and N^conv displayed similar levels of bacterial killing but N^reg displayed a significantly larger increase in ROS production on infection than N^conv. As N^reg have a unique ability to regulate T cells [25, 26] and ROS are known to be major compounds produced by granulocytic MDSCs to alter the T cell receptor [72], we hypothesize that the higher levels of ROS production in N^reg not only help to kill Mb, but that they also have immunomodulatory effects on other immune cells, such as T cells, especially during the later stages of TB.

N^reg display a significant increase in ROS production accompanied by the activation of numerous pathways related to metabolism and mitochondrial organization. As mitochondria are also a major source of intracellular ROS, we investigated the mitochondrial content of this population to compare N^reg and N^conv further. The mitochondrial compositions of neutrophil subsets remain poorly characterized. Rice et al., [73] identified two subsets of neutrophils in mice with and without cancers: Ly-6G⁺/c-Kit⁺/CXCR2⁻ cells, which are less mature and more oxidative than the other subset, Ly-6G⁺/c-Kit⁻/CXCR2⁺ neutrophils. Interestingly, c-Kit⁺/CXCR2⁻ neutrophils had a mitochondrion-rich phenotype, with enhanced free radical production, especially under glucose-limited conditions. This subtype has several features in common with our N^reg, which have a higher mitochondrial load, as demonstrated by confocal microscopy, and higher levels of ROS production than N^conv. Furthermore, RNAseq analysis revealed an upregulation of the kit gene (+3.075 log₂FC) and a downregulation of the cxcr2 gene (−1.58 log₂FC) in N^reg, consistent with these findings. However, even though c-Kit⁺/CXCR2⁻ neutrophils are considered to be immature, the N^reg we studied here had a highly segmented nuclear morphology similar to that of mature neutrophils [25].

In addition to these observations, the high levels of ROS production observed in N^reg may account for the higher rates of cell death detected in this subset. Indeed, enhanced ROS levels are associated with cell damage and impaired viability, leading to the activation of cell death-related pathways [74, 75]. This study provides evidence of different death dynamics between neutrophil subsets, particularly as concerns apoptosis. Our data indicate that N^reg not only die more rapidly than N^conv, but that they undergo apoptosis at a higher rate in both steady-state and infection conditions, suggesting an intrinsic predisposition toward lower survival. In our RNAseq analysis comparing the two subsets, only N^reg displayed an enrichment in a few significant pathways related to telomeres and telomerase that might explain higher death rates more generally. However, when neutrophils were infected with Mb3601, N^conv displayed an enrichment in pathways related to the up- or downregulation of apoptosis (online suppl. Fig. S1b). No such effect was observed after infection with BCG. This may be due to the virulence of Mb3601, resulting in a greater capacity of this strain to modulate host responses. For four animals, we could infect sorted N^conv with the recombinant BCG-RD1 strain. We analyzed the different apoptosis signaling pathways using ssGSEA enrichment analysis. The transcriptomic profile with BCG-RD1 appeared closer to Mb3601 than to BCG (data not shown). Since the RD1 locus encodes the ESX-1 secretion system, a major virulence determinant involved in host-pathogen interactions and modulation of host-cell death pathways [76, 77], these preliminary observations suggest that RD1-dependent mechanisms may contribute to the apoptotic signatures observed following Mb3601 infection. Neutrophil death occurred in a pathogen-dependent manner: BCG infection accelerated cell death, whereas Mb3601 infection appeared to delay it, highlighting the capacity of mycobacterial strains to influence cell lifespan differentially [78]. Apoptosis is one of the principal pathways contributing to neutrophil death, but it is unlikely to be sufficient to account entirely for the observed differences between neutrophil subsets. Other types of cell death, such as necrosis or NETosis, may also be involved and could have important functional consequences for inflammation and host-pathogen interactions, especially in the context of TB [79, 80]. Indeed, NETosis is also a crucial process in TB, with high NET levels associated with a poor disease outcome [81, 82]. We attempted to investigate NETosis in both N^reg and N^conv by confocal microscopy, after labeling NETs with antibodies directed against myeloperoxidase, neutrophil elastase or citrullinated histone H3. We observed rare NETs in vitro (data not shown), probably not entirely representative of what has been seen in vitro in other models [83] or in vivo [84]. Cell sorting may account for the low frequency of NET formation in vitro, as this procedure can activate neutrophils, altering their capacity to undergo NETosis. This does not necessarily exclude the possibility of such events occurring in vivo. However, our RNAseq analysis with the KEGG database revealed an upregulation of the neutrophil extracellular trap pathway in N^conv (online suppl. Material 1), suggesting that disease exacerbation in vivo may be dependent only on NETs from N^conv. Further investigations of these alternative death pathways are now required to improve our understanding of the fate of neutrophils and the regulatory role of these cells in bTB.

We were unable to identify any morphological differences between N^reg and N^conv on conventional microscopy [25]. However, SEM and TEM revealed remarkable differences in morphology, relating to cell shape, membrane and granule organization, between N^conv and N^reg. These ultrastructural differences may underlie different basal functional states in neutrophils. N^conv have features consistent with a structurally stable quiescent state, whereas N^reg might have a lower activation threshold and greater membrane plasticity, potentially conferring a greater propensity for rapid cell-cell and cell-bacteria interactions. We noted important structural differences between the two subsets, including the presence of thin, elongated filopodium- or nanotube-like structures. Such structures have been described in macrophages [85, 86], but have rarely been described in neutrophils. We cannot confirm their exact nature, but we hypothesize that these protrusions may correspond to tunneling nanotubes or proteoglycofili [87], involved in communication between cells or pathogen neutralization. Differences in granule size and number between the two neutrophil subsets further support their phenotypic divergence. In cattle, neutrophil granules include azurophilic and specific granules, as in mice and humans, but also a third, as yet unnamed, large peroxidase-negative granule that appears to be specific to ruminants [88]. It was not possible to distinguish these types of granules by electron microscopy in our neutrophil subsets, but transcriptomic analysis provided further insight. Reactome pathway analysis (online suppl. Material 1) revealed a significant enrichment in the neutrophil degranulation pathway in N^conv, suggesting a greater antimicrobial effect of degranulation in this subset. This hypothesis was supported by the upregulation of several granule-associated genes in N^conv, including elane, ltf, s100a9, mmp25, and olfm4. However, N^reg also expressed genes related to granules: bpi, cybb, lyz, defb10, defb4, and members of the serpin family (serpinb1, serpinb2, and serpinb6). Together, these findings point to the existence of two different granule-related programs in these two subsets of neutrophils. Further investigations, including granule-focused proteomics analyses in particular, will be required to determine how these differences translate into functional outcomes and whether they are modulated by Mb infection. Neutrophils have been shown to act as a source of antimicrobial agents potentiating the killing of Mtb by macrophages through the delivery of their granule contents [89], but the survival of Mtb within macrophages has been shown to be enhanced following the ingestion of neutrophils containing this bacterium [90]. As N^conv and N^reg have different granule contents, this crosstalk between neutrophils and macrophages should be investigated further, taking the heterogeneity of neutrophils into account.

In conclusion, our study demonstrates that bovine N^reg and N^conv are very different types of neutrophils probably programmed to perform different functions. Following infection with Mb, these two subsets had clearly different transcriptional profiles, but similar bactericidal capacities. N^reg displayed accelerated cell death, enhanced phagocytic activity, an increase in ROS production, and a higher mitochondrial load than N^conv. Electron microscopy-based ultrastructural analyses revealed phenotypic divergences between the two subsets, particularly in terms of cell morphology, membrane organization, and granular content. We hypothesize that the observed morphological differences are a consequence of distinct metabolic states, as suggested by the TEM/SEM observations, rather than being directly associated with the suppressive function of N^reg.

Together, these findings provide the first functional characterization of bovine neutrophil subsets during Mb infection and highlight a new layer of complexity in the functional diversity of neutrophil subsets in TB. Our results are consistent with a hyperactivated, terminally differentiated neutrophil state in N^reg, capable of simultaneously enhancing antimicrobial effector functions while promoting an immunosuppressive environment. In the context of Mb infection in vivo, this combination is biologically plausible and may play an important role in granulomatous disease progression. High ROS production is not only microbicidal but also represents a well-established mechanism by which neutrophils suppress adaptive immunity: N^reg may create a localized redox environment that impairs T-cell activation and function. This phenotype could reflect a host strategy to balance bacterial containment with the limitation of bTB immunopathology. Nevertheless, these neutrophil-Mb interactions relate exclusively to the early stage of the disease, and further in vitro studies on later stage of bTB (granuloma) or in vivo studies are required to gain a full understanding of the role of N^reg in bTB, including their contribution to host immune responses, disease progression and pathological effects on tissues. Elucidation of the specific role of bovine neutrophil subsets in bTB pathophysiology may pave the way for the discovery of new biomarkers, which are urgently needed to improve the management of this costly disease.

Acknowledgments

We warmly thank the UE-PAO experimental unit for blood sampling and animal care. We thank Alix Sausset for her assistance with neutrophil sorting. The Mb3601 strain was obtained from Dr. Maria-Laura Boschiroli (ANSES, French bTB reference laboratory). We thank the Microscopy Facility (US61 ASB) of Tours University and CHRU of Tours (http://microscopies.med.univ-tours.fr) for technical support. We also thank Axelle Guillet and Pauline Courroussé, who performed their Master 1 and professional Bachelor’s degree projects, respectively, in our team and helped with some of the experiments.

Statement of Ethics

Blood sampling was approved by the Ethics Committee for Animal Experimentation Val-de-Loire (CEEA VdL), registered with the French National Committee for Animal Experimentation (APAFIS no. 202007241825481.V2), and complied with EU regulation 2010/63.

Conflict of Interest Statement

The authors declare that this study was conducted in the absence of any commercial or financial relationships that could be construed a potential conflict of interest.

Funding Sources

This work was supported by the ANR Neutro-bTB (Grant ANR-21-CE20-0050). MSV is the beneficiary of a PhD fellowship funded jointly by the Loire Valley Region and the Microbiology Department of INRAE. The academic funding sources (ANR, INRAE, and the Loire Valley Region) had no role in the study design, execution, or analysis, nor in the conception, planning, writing, or decision to publish the manuscript.

Author Contributions

M.S.-V.: conceptualization, formal analysis, investigation, validation, visualization, methodology, and writing – original draft, review, and editing. B.B.: conceptualization, software, data curation, formal analysis, validation, investigation, visualization, methodology, and writing – review and editing. S.E.: methodology, resources, formal analysis, validation, visualization, and writing – review and editing. E.P.: methodology, investigation, formal analysis, validation, and visualization. : methodology, investigation. P.C., Y.L. V., and F.C.: methodology and investigation. R.R.: investigation. J.P.: methodology, formal analysis, and visualization. E.D. -D.: writing – review and editing. P.G.: formal analysis and writing – review and editing. N.W.: conceptualization, and writing – original draft, review, and editing. A.R.: conceptualization, formal analysis, investigation, supervision, funding acquisition, visualization, validation, methodology, project administration, and writing – original draft, review, and editing.

Funding Statement

This work was supported by the ANR Neutro-bTB (Grant ANR-21-CE20-0050). MSV is the beneficiary of a PhD fellowship funded jointly by the Loire Valley Region and the Microbiology Department of INRAE. The academic funding sources (ANR, INRAE, and the Loire Valley Region) had no role in the study design, execution, or analysis, nor in the conception, planning, writing, or decision to publish the manuscript.

Data Availability Statement

The transcriptomic data have been deposited in the European Nucleotide Archive (ENA) under BioProject Accession No. PRJEB106984.