Knockout of CXCR5 increases the population of immature neural cells and decreases proliferation in the hippocampal dentate gyrus

The CXCR5^-/- animals and WT controls for the study were acquired from the Jackson Laboratories (http://​jaxmice.​jax.​org/​strain/​006659.​html). These animals are on a C57/BL6J background. They were housed in approved conditions on a 12-hour light/dark cycle in individually ventilated cages with three to five animals to each cage. Food and water were provided ad libitum. Handling occurred twice per week when each animal was weighed and assessed for general health. All care occurred under the oversight of a veterinarian.

Cohorts of 11 male and female (3 male, 8 female, total n = 11) animals of each genotype were aged to 10 weeks before commencing behavioural studies and were sacrificed by 11 weeks of age for immunohistochemistry (5 female, 1 male, total n = 6) and protein analysis (3 female, 2 male, total n = 5). Cohorts of 10 male animals of each genotype were sacrificed at 6 weeks of age for tissue use in the neurosphere assay, as at this age adult-derived hippocampal neurospheres are maximal [21]. All animals were sacrificed by cervical dislocation by a trained operator.

This study received ethics approval from the University of Adelaide animal ethics committee.

Primary neurospheres from SVZ and hippocampus were separately generated according to previously published protocols [13, 22]. In brief: hippocampal or SVZ tissue was digested by incubation in a mixture containing 0.1% papain (Worthington Biochemical Corporation) and 0.1% DNaseI (Roche Australia) in HBSS (Thermo Scientific) for 16 min at 37°C, titurating twice during the incubation period. Next, the tissue was centrifuged at 750 rpm for 5 min, after which the pellet was resuspended and washed twice in 2 mL of neurosphere growth medium: mouse NeuroCult NSC basal medium containing mouse NeuroCult NSC proliferation supplements (Stem Cell Technologies), 2 μg/mL heparin, 20 ng/mL purified mouse epidermal-like growth factor (BD Biosciences) and 10 ng/mL recombinant bovine fibroblast growth factor02 (Roche). Cells were plated at a density of two hippocampus or SVZ per 96-well plate (BD Biosciences) with 200 μL of neurosphere growth medium per well. For depolarisation KCl was added to half the primary cultures of each hippocampus for a final concentration of 15 mM. SVZ cultures were incubated for 7 days and hippocampal cultures for 10 days in humidified 5% CO₂. Neurospheres were then counted using a standard light microscope with an eyepiece graticule. Small neurospheres (likely derived from NPCs) were defined as ≥50 μm diameter and large neurospheres (likely derived from putative NSCs) were defined as ≥250 μm diameter [21]. Cultures were not passaged further. The operator was blinded to the genotype for counting.

At time of sacrifice brain tissue was collected from six animals (5 females, 1 male) of each genotype (n = 6). These mice were perfused with a 10 mL of 10% formalin solution through the left ventricle followed by retrieval of brain tissue and storage in a 10% formalin solution. Brains were then embedded in paraffin prior to sectioning, with a series of nine 5 μm sections collected at 200 μm intervals throughout the thickness of the hippocampus. Immunostaining was then conducted using antibodies for Ki67 (Abcam), doublecortin (Millipore), Nestin (Abcam)s and IBA-1 (Santa-Cruz). Following de-waxing and dehydration, endogenous peroxidase activity was blocked by incubation with 0.5% hydrogen peroxide in methanol for 30 min. Slides were then washed in 2 × 3 min in phosphate buffered saline (PBS) before antigen retrieval retrieved by heating at close to boiling point for 10 min (TRIS for doublecortin, citrate for others). Once the slides had cooled below 40°C they were washed with PBS before being blocked with 3% normal horse serum in PBS for 30 min. The appropriate primary antibody was applied to the slides which were left to incubate overnight (Ki67 1:2,000, doublecortin 1:8,000, Nestin 1:500, IBA-1 1:1,000). The next day slides were washed in 2 × 3 min of PBS before the appropriate species of IgG biotintylated antibody was added for 30 min (Dako). After a further PBS wash, slides were incubated with streptavidin peroxidase conjugate for 60 min followed by another rinse with PBS. The immunocomplex was then visualised with precipitation of DAB (Sigma D-5637) in the presence of hydrogen peroxide. Slides were washed to remove excess DAB and lightly counterstained with haematoxylin, dehydrated and mounted with DePeX from histolene.

Slides were subsequently digitally scanned using the Hamamatsu NanoZoomer and examined using the associated NDP.view2 software (Hamamatsu). Sequential images were captured of the hippocampus and exported for cell counting within Image J 1.46 (NIH) where the grid and manual cell counter features were utilised. In all instances the hippocampus or dentate gyrus were counted in their entirety, bilaterally. Area of each region was measured in NDP.view2 and results were expressed as cellularity in cells per mm². The operator was blinded to the genotype of the slides during the slide analysis process.

At time of sacrifice blood was collected from five animals of each genotype (3 females, 2 males) by means of cardiac puncture and serum was extracted by spinning blood down at 2,500 rpm for 15 min. Serum was stored at -80°C until analysis. Serum cytokine levels were measured using the BD Cytometric Bead Array (CBA) Mouse Inflammation Kit for the cytokines IL-6, IL-10, MCP-1, IFN-γ, TNF-α and IL-12p70 according to the manufacturer’s instructions.

For the open field and hole-board tests, mice were placed into the same brightly lit square arena, 40 × 40 cm, with clear walls 35 cm high. In the open field test, mice were placed in to the arena for 5 min according to published protocols [23, 24]. The floor was divided into inner and outer zones. Time spent in each zone was measured as an indication of anxiety-like behaviour, and total distance travelled was measured as an indication of baseline locomotor activity.

Mice were trained for the saccharin preference test with a continuous two-bottle choice of water and 0.1% saccharin for 48 h. For this period, mice were moved to individual cages which could hold two bottles and the bottles were switched at 24 h to control for left-right preference. Immediately after this training period, the test was conducted. The test involved mice being moved to individual cages and given the two-bottle choice of water and 0.1% saccharin for 24 h. This protocol has been adapted from several previously published protocols [25, 26]. Preference was calculated as: saccharin preference % = (saccharin cons./(saccharin cons. + water cons.)) × 100.

We investigated whether CXCR5 deficiency might impact upon hippocampal neurogenesis in 11-week-old (adult) mice. In order to assess this, we stained hippocampal sections for the proliferation marker Ki67, Nestin doublecortin (DCX), markers of immature neuronal cells. We counted only positive cells within the subgranular zone for Ki67. Within this subregion Ki67 positive cells are likely to represent proliferative activity of NPCs, and Nestin/DCX the immature cells of neuronal lineage which are generated by the differentiation and migration of those cells [27]. We found that CXCR5 deficiency was permissive of increased density of DCX positive cells in the granular and subgranular zones of the dentage gyrus (505 ± 19.2 vs. 420 ± 25.9; n = 6 per group; P = 0.02). A similar trend was observed in nestin positive cells (331 ± 8.2 vs. 280 ± 7; n = 6 per group; P = 0.0008) (Figure 1). This implies that CXCR5 signaling may impair the maintenance of these populations. Additionally, we demonstrated that CXCR5 deficiency suppressed the number of Ki67 positive cells in the subgranular zone as compared to WT controls (90.0 ± 6.3 vs. 116.3 ± 5.1; n = 6 per group; P = 0.009). This may represent indirect suppression of a compensatory mechanism which would act to maintain DCX and Nestin positive cells under physiological conditions, or a direct effect by which CXCR5 (known to be expressed on NSC/NPC [19]) would support proliferation of these cells. Taken together these data are most suggestive that CXCR5 may exert a detrimental effect on maintenance of Nestin and DCX positive cell populations - for example through a pro-apoptotic effect - requiring enhanced proliferation of subgranular zone cells to compensate. A further possibility is that CXCR5 may support differentiation further down a neuronal lineage, thereby 'draining’ the pools of DCX and Nestin positive cells and requiring an enhancement of subgranular zone cell proliferation (Ki67 positive) to replace the pool. An alternative explanation, in which CXCR5 impairs differentiation of proliferating cells into a neuronal lineage is less likely as this effect would not be expected to simultaneously enhance the proliferation of the subgranular zone precursor/stem cells.

To further interpret the observed in-vivo effects of the CXCR5 deficiency we also analysed the in-vitro activity of several subgroups of NPC/NSCs via the neurosphere culture method. We considered separately the populations of small (≥50 μm) and large (≥250 μm) hippocampus derived neurospheres both under basal culture conditions and the latent populations which are responsive to depolarisation with KCl. Previous studies have described the properties of these populations to include cells of both NPC and in the case of KCl responsive large spheres NSC phenotype [21]. In this study we detected no significant effects of CXCR5 deficiency on any of these hippocampal populations in number or size (n = 10 per group; all P >0.05) (Figure 2). We also detected no significant difference in SVZ derived neurospheres (75.5 ± 12.5 vs. 110.9 ± 15.5; n = 10 per group; P = 0.1) (data not shown). Considered in parallel with the above data, this suggests that the observed reduction in Ki67 staining seen in CXCR5^-/- animals is not due to an absolute loss of NPC/NSC populations as these are still reactive to appropriate in-vitro stimulus. This implies that CXCR5 signalling is not required for maintenance of the proliferative capability of latent NPC/NSC populations.

To investigate the mechanisms by which CXCR5 deficiency may produce a deficit in the number of Ki67 positive cells we investigated several indices of the CNS specific and systemic immune response which are known to influence normal neurogenesis. We investigated the density of IBA-1 positive cells (microglia) in the hippocampus, as depletion of these cells is known to impair neurogenesis [13]. We detected no significant effect of CXCR5 deficiency on density of these cells in this study (50.7 ± 3.1 vs. 54.4 ± 2.3 cells/mm²; n = 6 per group; P = 0.37) (Figure 3). Furthermore, we analysed the percentage of activated (reactive) microglia on the basis of their classically described morphology: large soma, short branches and a 'bushy’ appearance [28]. We detected no difference in the percentage of activated microglia (2.25 ± 0.4 vs. 2.52 ± 0.52%; n = 6 per group; P = 0.69) (Figure 3).

We also assayed the levels of systemic cytokines through use of a cytometric bead assay on serum as several cytokines are known to enhance or impair neurogenesis (for example, Interferon-γ) [10]. There were no significant differences between genotypes in serum levels of the cytokines IFN-γ, IL-6, IL-10, MCP-1, TNF-α or IL-12p70 (all P >0.05) (data not shown). These data suggest that the observed effects of CXCR5 on Ki67, Nestin and DCX cell populations are not mediated by alterations inmicroglial population or gross disruption of soluble immune factors which may ingress from the systemic circulation. Under these non-immunologically challenged experimental conditions, and in the absence of gross disruption of systemic or glial immune function we would not anticipate alteration in CNS-specific cytokine profile.

We next aimed to investigate the behavioural consequences of CXCR5 deficiency. A useful initial test is the open field test which provides information on baseline locomotor activity which may be a significant confounder for further behavioural testing [29]. In our study we found that CXCR5 deficiency was associated with a two-fold increase in baseline locomotor activity as assessed by distance travelled during the test (21.9 ± 0.6 vs. 13.0 ± 0.9; n = 11 per group; P <0.0001) (Figure 4). We also found an increase in time spent in the inner two thirds of the open field by CXCR5 deficient animals which may be interpreted as a reduction in stress responsiveness (55.9 ± 6.2 vs. 34.1 ± 5.4; n = 11 per group; P = 0.015), however such an interpretation should be regarded with caution given the overall increase in locomotor activity (Figure 4A, B). In view of the result on locomotor activity which would make it difficult to interpret the results of conventional hippocampus-dependent learning and memory tasks (for example, Barnes Maze or Morris Water Maze [30]), we have pursued behavioural testing which did not require locomotor activity.

Although the mechanistic relevance of hippocampal neurogenesis to depression or depression-like behaviours remains unclear, it is well established that in many animal models of stress and/or depression-like behaviour these behaviours are correlated with impairments in neurogenesis, and certain antidepressant medications are associated with increases in neurogenesis [5]. We therefore investigated the effects of CXCR5 deficiency on the saccharin preference test which is a well established model of anhedonia-like behaviour and does not require locomotor activity [31]. All animals consistently demonstrated a preference for the saccharin-sweetened solution during training (data not shown), however during testing there was no significant effect of CXCR5 deficiency on saccharin preference as compared to WT controls (72.6 ± 3.5 vs. 80.65 ± 3.0; n = 11 per group; P = 0.1) (Figure 4C). Interestingly, these data suggest a non-significant trend toward an anhedonia-like phenotype of CXCR5 deficient mice - the opposite of what would be expected if depression-like behaviours were associated with impaired neurogenesis.