Oral and Injected Tamoxifen Alter Adult Hippocampal Neurogenesis in Female and Male Mice

Visual Abstract


Introduction
Adult neurogenesis is a widely conserved process among mammalian species in which resident neural stem cells generate new neurons that integrate into mature circuitry throughout the lifespan. Evidence that neurogenesis in the adult rodent dentate gyrus of the hippocampus supports memory and affect regulation has spurred interest in both its natural functions across species, as well as its potential therapeutic applications in humans (for review, see McAvoy and Sahay, 2017;Miller and Sahay, 2019;Toda et al., 2019).
The study of time-specific and tissue-specific phenomena, such as adult hippocampal neurogenesis, has been accelerated by the introduction of several inducible methods of gene expression manipulation. Tamoxifen (TAM)inducible Cre mouse lines (e.g., CreER T2 ) have proven to be particularly valuable tools for inducible manipulation of gene expression in the adult neurogenic lineage. In these models, Cre recombinase is fused to a mutated estrogen receptor (ER T2 ) which retains Cre in the cytoplasm until the receptor binds its ligand, the selective estrogen receptor modulator TAM (Feil et al., 1997;Danielian et al., 1998). TAM binding allows translocation of the mutated estrogen receptor and its fused Cre enzyme to the nucleus, where Cre can induce recombination of LoxP-flanked genetic sequences. Use of a tissue-specific promoter can further constrain CreER T2 expression to specific cell classes. For example, neural stem and progenitor cells (NSPCs) are commonly targeted using GLAST, Nestin, SOX2, or GFAP promoters (to name a few common versions; Semerci and Maletic-Savatic, 2016).
In adult neurogenesis studies using CreER T2 -mediated LoxP recombination, two dominant approaches to selecting controls have emerged. One approach compares mice of different genotypes (Cre1/À or LoxP1/À), all of which are treated with TAM. A second approach is to compare mice of the same genotype treated with TAM versus those treated with vehicle. This latter comparison is predicated on TAM itself being inert, aside from its ability to induce CreER T2 translocation. Yet, gonadal steroid hormones, including estrogens, can modulate brain processes such as neurogenesis and cell survival (Jorgensen and Wang, 2020).
The effect of TAM on adult neurogenesis is generally understudied and the few existing studies have yielded mixed results. For example, Rotheneichner et al. (2017) found no inherent effects of TAM on cell proliferation and fate in the dentate gyrus of the adult mouse hippocampus, but a more recent study suggested long-lasting suppression of hippocampal cell proliferation in juvenile mice (Lee et al., 2020). Given the widespread use of TAM-inducible models in adult neurogenesis, it is imperative to understand how TAM itself (independent of transgene expression) affects adult neurogenic processes.
Beyond the selection of controls, TAM administration route is another important potential design variable in studies using CreER T2 mouse models. Intraperitoneal injections are by far the most common method of exposing mice to TAM in adult neurogenesis studies to date. However, administration of TAM through voluntary consumption of custom laboratory chow is a promising alternative that might reduce injection-related stress and researcher hands-on time. TAM-infused chows have been used successfully in cardiac research (Kiermayer et al., 2007), and may be applicable to other adult tissues as well (Yoshinobu et al., 2021). It is unknown how TAM chow compares to TAM injection as regards recombination efficiency and appropriate dosing schedules for adult neurogenesis research.
Here, we compare TAM administration by injection and chow, both in terms of the specificity and efficiency of LoxP recombination and the inherent effects of TAM treatment itself on adult neurogenesis. Using a CreER T2 model targeted to NSPCs by a Nestin promoter, our findings suggest that TAM chow can induce LoxP recombination in hippocampal NSPCs with similar specificity as TAM injection, but that TAM chow results in much lower recombination efficiency than injections, likely because of chow avoidance and therefore lower TAM exposure. We also show that both TAM injection and TAM chow paradigms have inherent, although different, effects on adult neurogenesis, underscoring the need to compare TAM-treated experimental animals to genetic controls similarly treated with TAM, regardless of route of administration.

Mice
All mice were group housed with three to five mice per cage in The Ohio State University Psychology building mouse vivarium in standard ventilated cages on a 12/12 h light/dark cycle (lights on 6:30 A.M.) with ad libitum access to food and water. Male and female mice were eight to nine weeks old at the time of the experiment and housed in groups of two to four. All animal use was in accordance with institutional guidelines approved by The Ohio State University Institutional Animal Care and Use Committee.

Experimental design and statistical analysis Study 1: TAM injection versus TAM chow
Seven-to nine-week-old NestinCreER T2 (Lagace et al., 2007;Jackson #016261); R26R-LoxP-STOP-LoxP-EYFP (Srinivas et al., 2001; Jackson #007909) littermates were randomly assigned by whole cage to one of four groups: TAM inject d(day)7 perfuse (n = 10), TAM inject d14 perfuse (n = 11), TAM chow 10 d (n = 7), TAM chow 14 d (n = 11). Mice were bred by crossing wild-type C57Bl/6J mice with mice homozygous for both NestinCreER T2 and R26R-LoxP-STOP-LoxP-EYFP; thus, offspring were heterozygous for both transgenes. TAM-injected mice received daily 180 mg/kg/d intraperitoneal injections of TAM for 5 d. TAM chow mice were first acclimated to standard chow provision in a dish on the cage floor for 3 d before switching to TAM chow. Food was weighed and refreshed every 2-4 d. At the indicated tissue harvest time, mice were anesthetized with an 87.5 mg/kg ketamine, 12.5 mg/kg xylazine mixture, and then transcardially perfused with ice-cold 0.1 M PBS. Body weight data were analyzed by a repeated measures mixed-effects model (day Â TAM) within each sex, followed by Sidak's multiple comparisons within day. This analysis was performed within sex individually to prevent the strong main effect of sex on body weight from masking any more subtle effects of TAM. Food consumption was analyzed by two-way ANOVA (sex Â day) followed by Sidak's multiple comparisons between days within sex. Cell count data were analyzed by oneway ANOVA followed by Tukey's post hoc tests or by two-way ANOVA (hippocampal subregion Â group) followed by Tukey's post hoc multiple comparisons between groups within subregion.

Study 2: TAM versus vehicle injection
Wild-type C57Bl6/J mice were obtained at six weeks of age from The Jackson Laboratory (#000664) and allowed to acclimate for two weeks. Cages of mice were randomly assigned to either TAM (n = 11 mice) or vehicle (sunflower oil, n = 12 mice) injection groups. TAM injections were as described for study 1. On the last 3 d of TAM/vehicle injection, mice also all received daily bromodeoxyuridine (BrdU) injections (150 mg/kg/d, i.p.). Twenty-one days after the last TAM/BrdU injections, mice received a single injection of ethynyldeoxyuridine (EdU; 150 mg/kg, i.p.) and were perfused as in Study 1 2 h later. Body weight data were analyzed by a repeated measures mixed-effects model (day Â TAM) within each sex, followed by Sidak's multiple comparisons within day. BrdU1/EdU1 density was compared by unpaired t test. Density and proportion of BrdU/DCX/NeuN or EdU/GFAP/SOX2 cell subtypes was compared by two-way repeated measures ANOVA followed by post hoc Sidak's multiple comparisons within cell type.

Study 3: TAM versus vehicle chow
Wild-type C57Bl6/J mice were obtained at sixweeks of age from The Jackson Laboratory (#000664) and allowed to acclimate for one week. Cages of mice were randomly assigned to either TAM (n = 11 mice) or vehicle (n = 12 mice) chow groups. Diet was provided as described for Study 1 for 14 d. On the last 3 d of TAM/vehicle chow, mice also all received daily BrdU injections (150 mg/kg/d, i.p.). 21 d after the last TAM/BrdU injections, mice received a single injection of EdU (150 mg/kg, i.p.) and were perfused as in Study 1 2 h later. Food consumption was analyzed by repeated measures two-way ANOVA (chow type Â day) followed by Sidak's multiple comparisons between chow type within days. Remaining analysis for this experiment is similar to that for Study 2.

TAM injection preparation
TAM (Fisher #50-115-2413) was dissolved overnight at 20 mg/ml at 37°C in sterile sunflower oil then stored at 14°C for one week maximum, all while protected from light. Cumulative TAM dose for each mouse was calculated by adding the mg of TAM in each injection over the 5 d of the dosing schedule.

TAM and vehicle chow
TAM chow (500 mg TAM/kg diet, TD.130858, also contains 49.5 g/kg sucrose; 15.4% protein, 55.1% carbohydrate, and 3.4% fat by weight; and 0.25 g/kg of red food color) and matched vehicle chow (15.4% protein, 55.1% carbohydrate, and 3.4% fat by weight; 50 g/kg of sucrose; and 0.25 g/ kg of blue food color) were obtained from Envigo and stored at 14°C in the dark until given to mice. Chow was refreshed every 3 d per manufacturer recommendations.

Tissue processing
Tissue fixation, slicing, and immunofluorescent staining were performed similarly to our previous work (Dause and Kirby, 2020) using the antibodies described in Table 1. In brief, brains were postfixed in 4% paraformaldehyde (Fisher #AC169650010) in 0.1 M phosphate buffer at 14°C for 24 h then equilibrated in 30% sucrose (Fisher S5-3) in 0.1 M PBS (Fisher #BP399-20) at 14°C before slicing on a freezing microtome (Leica) in a 1:12 series of 40mm-thick coronal sections. Sections were stored in cryoprotectant medium at À20°C until processed for immunofluorescent staining. Free-floating sections were rinsed in PBS, blocked in 1% normal donkey serum (Jackson ImmunoResearch #017000121), 0.3% Triton X-100 (Fisher #AC215682500) in PBS and incubated in primary antibody in blocking solution overnight at 4°C on rotation. The next day, sections were rinsed, incubated in secondary antibody in blocking solution for 2 h at room temperature on rotation and then rinsed and counterstained with Hoechst 33342 (1:2000 in PBS, Fisher #H3570) before being mounted on SuperFrost Plus slides (Fisher #12-550-15), dried and coverslipped with Prolong Gold Antifade mounting medium (Fisher #P36934). For BrdU staining, sections were processed for other antibodies first then postfixed in 4% paraformaldehyde for 10 min at room temperature. Sections were then rinsed and incubated in 2N HCl (Fisher A144500) at 37°C for 30 min, followed by rinsing, blocking and primary/secondary incubation as described above. For EdU click labeling, sections were click reacted using a Click&Go EdU 488 imaging kit (Click Chemistry Tools, #1324) according to manufacturer instructions before proceeding with subsequent immunolabeling. Slides were all dried overnight at room temperature in the dark and then stored long-term at 14°C. Other materials used for tissue processing, as well as for animal treatment, are in Table 2.

Cell quantification
Similar to our previous work (Dause and Kirby, 2020), the dentate gyrus of the hippocampus was imaged in 15-mm zstacks at 20Â magnification using a Zeiss Axio Observer Z.1 with apotome digital imaging system and Axiocam 506 monochrome camera (Zeiss). EYFP1 cells were identified based on EYFP1 cytoplasm around a Hoechst1 nucleus. EYFP1 cells were considered radial glia-like neural stem cells (RGL-NSCs) if Research Article: Methods/New Tools they had a SOX21 nucleus in the subgranular zone and colocalized with GFAP in the cytoplasm with an apical morphology. They were considered intermediate progenitor cells (IPCs) if they had a SOX21 nucleus in the subgranular zone but did not have a GFAP1 cytoplasm. They were considered astrocytes if they had a SOX21 nucleus and colocalized with GFAP in the cytoplasm with a stellate morphology. They were considered immature neurons/neuroblasts if they had a Hoechst1 nucleus in the sungranular zone and a cytoplasm that co-localized with DCX. BrdU1 cells were counted in the subgranular zone and granule cell layer throughout the dentate gyrus. Colabeling with DCX and NeuN was assessed in each BrdU1 cell in the zstacks as surrounding DCX1 cytoplasm and/or nuclear NeuN overlap with BrdU. All cell counts were divided by dentate gyrus area sampled to yield a density of cells.

Software
Statistical analyses were performed in GraphPad Prism (version 9.3.1). Graphical abstract was made using BioRender software.

TAM chow results in weight loss and less total TAM exposure compared with TAM injection
To compare TAM-induced LoxP recombination in NSPCs between injection and chow-fed methods, we crossed NestinCreER T2 mice (Lagace et al., 2007) with Rosa26-LoxP-STOP-LoxP-EYFP mice (Srinivas et al., 2001).
At seven to nine weeks of age, NestinCreER T2 1/À;Rosa26-LoxP-STOP-LoxP-EYFP1/À male and female mice were assigned randomly to either injection or chow groups. Injected mice received a standard 5-d injection regimen and were perfused on days 7 or 14 after the start of injections (Fig. 1A). TAM chow-fed mice were given free access to TAM chow for 10 or 14 d before perfusion (Fig. 1A). Both male and female TAM chow-fed mice showed lower body mass compared with TAM-injected mice ( Fig. 1B,C). TAM chow-fed mice consumed very little of the chow for the first ;5 d ( Fig. 1D), a time period which coincided with the greatest divergence in weight in these mice compared with TAM injected mice. Mice appeared active and well-groomed throughout the experiment. By days 6-8, male and female mice significantly increased their apparent chow consumption, coinciding with stabilization of their body weights. Cumulative TAM exposure was 1.23-fold to 1.57-fold higher in injected mice than chowfed mice (Fig. 1E). It should be noted that some chow was crumbled on the cage floor and could not be weighed. These estimates for chow consumption are most likely overestimates. Nonetheless, these data suggest that mice find TAM chow aversive, resulting in less TAM exposure over 14 d than from 5 d of standard injections.

TAM chow and TAM injections show similar recombination specificity
We next quantified the cell phenotype of EYFP1 cells in the DG. At day 7 after TAM injection start, 30.11 6 2.49% of DG EYFP1 cells were phenotypic RGL-NSCs (Fig. 2E). Mice fed TAM chow for 10 d showed a similar fraction of EYFP1 cells that were phenotypic RGL-NSCs (28.48 6 2.48%). By 14 d after TAM injection start, the percent of EYFP1 cells that were RGL-NSCs dropped significantly compared with 7 d to 17.95 6 1.61%. This decrease in proportion of EYFP1 cells that are RGL-NSCs is consistent with differentiation of the recombined, EYFP1 population over time. In the 14-d chow group, 19.28 6 2.47% of EYFP1 cells were phenotypic RGL-NSCs, a fraction that was slightly but significantly lower than the 7-d injection group. The percent of EYFP1 cells showing IPC phenotype were similar at 7 and 14 d after injection start (48.99 6 2.73% and 43.44 6 2.73%; Fig. 2F). However, TAM chow-fed mice showed slightly, although significantly, higher percentages of EYFP1 cells that were phenotypic IPCs than TAM-injected mice (10 d: 54.54 6 3.82%; 14 d: 59.17 6 1.91%). As expected, very few EYFP1 cells showed an astrocytic phenotype in all groups (d7 inject: 4.00 6 1.45%; d14 inject: 0.83 6 0.36%; 10 d chow: 4.40 6 2.76%; 14 d chow: 0.61 6 1.38%; Fig.  2G). We also used co-labeling of doublecortin (DCX) with EYFP to quantify the portion of EYFP1 cells that were phenotypic neuroblasts/immature neurons. DCX co-labeling increased significantly from 7 to 14 d after TAM injection start (28.00 6 2.61% to 54.91 6 2.90% of EYFP1 cells co-labeled for DCX), again consistent with the maturation of the recombined, EYFP1 cell population over time. Both TAM chow-fed groups showed similar percentages of EYFP1 cells co-labeling for DCX as the 7-d injection group (10 d: 24.44 6 1.05%; 14 d: 26.07 6 2.29%). Separate analysis of males and females yielded similar results in each sex ( Fig. 3C-F).
To assess ectopic recombination in non-neurogenic regions of the hippocampus, we quantified EYFP1 cells in CA1 and CA3. In all groups, EYFP1 cells were detectable in CA1 and CA3 but the rates of EYFP expression were low, with .92% of all EYFP1 cells being located in the dentate gyrus (Fig. 2J-K). Groups did not significantly differ from each other in EYFP localization to the dentate gyrus and separate analysis of males and females yielded similar results in each sex (Fig. 3G). All together, these data suggest that both TAM injection and TAM chow drive recombination predominantly in NSPCs with no notable differences in their specificity. They also suggest that 10-14 d of chow exposure leads to a recombined population that is most phenotypically similar to that found 7 d after TAM injection start.
We also quantified cell proliferation of RGL-NSCs and IPCs three weeks after TAM using EdU to label proliferating cells. TAM injection led to a 1.32-fold suppression of total EdU1 cell density in the dentate gyrus compared with vehicle-injected mice (1.31 6 0.09 Â 10 À4 cells/ mm 2 veh vs 0.99 6 0.08 Â 10 À4 cells/mm 2 TAM; Fig. 4H, K). Classification of EdU1 cells as RGL-NSCs, IPCs or neither revealed that this reduction in EdU labeling was driven primarily by loss of EdU1 IPCs (Fig. 4I,K). Quantification of total RGL-NSCs and IPCs similarly showed a significant loss of total IPC density in TAMtreated mice, with a more moderate, nonsignificant decrease in RGL-NSC density (Fig. 4J,K). Similar results were found when males and females were analyzed separately ( Fig. 5D-F). These findings suggest that TAM injection causes a long-term suppression of IPC proliferation that is evident three weeks after TAM has ended.

TAM chow enhances adult neurogenesis acutely but does not suppress cell proliferation long term
To determine whether TAM chow alters adult neurogenesis, we fed adult wild-type C57BL6/J mice TAM or vehicle-matched chow for 14 d, the last three of which were coupled with once per day BrdU injections to label dividing cells (Fig. 6A). Mice were perfused 35 d after the beginning of chow treatment (21 d after the last TAM/BrdU injections), 2 h after a single EdU injection to label acutely proliferating cells. Monitoring of food consumption confirmed that during the first ;3d of chow exposure, mice consumed significantly less TAM chow than vehicle chow (2.88 6 0.10 g veh chow/ms/d vs 1.15 6 0.15 g TAM chow/ms/d), but consumption returned closer to vehicle chow levels shortly thereafter (Figs. 6B, 7A). Total average TAM consumption per mouse was 14.22 6 0.95 mg, similar to that seen in Figure 1E Fig. 6C). TAM and vehiclefed male mice, in contrast, no longer differed in body weight after three weeks of recovery (24.07 6 0.32 g veh chow vs 24.18 6 0.88 g TAM mice; Fig. 6D).
We also quantified cell proliferation of RGL-NSCs and IPCs three weeks after TAM using EdU to label proliferating cells. EdU1 cell density in the DG did not significantly differ between veh and TAM chow-fed mice (1.21 6 0.12 Â 10 À4 cells/mm 2 veh vs 1.06 6 0.07 Â 10 À4 cells/mm 2 TAM; Fig. 6I,L). Classification of EdU1 cells as RGL-NSCs, IPCs or neither similarly showed no difference in EdU1 density of these cell subtypes between chow groups (Fig. 6J,L). Quantification of total RGL-NSCs and IPCs also showed no difference in total RGL-NSC or IPC density (Fig. 6K,L). Similar results were found when males and females were analyzed separately (Fig. 7F-H). These findings suggest that TAM chow feeding does not strongly alter cell proliferation relative to vehicle chow three weeks after chow treatment has ended.
The difference between TAM injection and TAM chow in recombination efficiency is most likely explained by total TAM exposure. Recombination efficiency in TAM-inducible transgenic systems is partly dose dependent (Hayashi and McMahon, 2002), and mice given TAM chow were exposed to overall less TAM than those given TAM injections. Mice  showed strong aversion to the TAM chow, as evidenced by the approximately one-week delay between chow introduction and substantial daily consumption. It is not surprising then to observe substantially less recombination in chow mice than injected mice. The chow was sweetened to increase palatability and delivered on the cage floor in dishes to ease access, so future efforts to overcome this barrier to TAM ingestion may require longer TAM chow treatments. In studies targeting adult neurogenic processes, however, longer chow treatment will result in a continued F (6,18) = 0.9896, p = 0.4613). Mean 6 SEM of n = 4 mice. C, In female mice, body mass in grams 14 and 35 d after diet initiation (two-way repeated measures ANOVA; day Â TAM: F (1,9) = 10.62, p = 0.0099; day: F (1,9) = 42.05, p = 0.0001; TAM: F (1,9) = 0.8454, p = 0.3818; subject: F (9,9) = 0.7591, p = 0.6560). Mean 6 SEM of n = 5-6 mice. D, In male mice, body mass in grams 14 and 35 d after diet initiation (two-way repeated measures ANOVA; day Â TAM: F (1,10) = 18.18, p = 0.0017; day: F (1,10) = 62.45, p , 0.0001; TAM: F (1,10) = 8.540, p = 0.0152; subject: F (10,10) = 1.886, p = 0.1658). Mean 6 SEM of n = 6 mice. E, Density of BrdU1 cells (cells/mm 2 Â 10 4 ) in the dentate gyrus (unpaired t test, t (21) = 1.675, p = 0.1087). Mean 6 SEM of n = 11-12 mice. F, Density of dentate gyrus BrdU1 cells (cells/mm 2 Â 10 4 ) co-labeled with DCX and/or NeuN (two-way repeated measures ANOVA; cell type Â TAM: F (2,42) = 8.654, p = 0.0007; cell type:  more heterogenous population of recombined cells at different stages of differentiation. Temporal precision is often a goal of adult neurogenesis studies, and extending chow treatment times may not always fit with that goal. Another possibility would be to provide TAM via oral gavage. However, oral gavage is notoriously stressful for animals (Brown et al., 2000) and requires expertise on the part of the experimenter to reduce that stress and avoid injury to animals (Arantes-Rodrigues et al., 2012). If high recombination  efficiency and selective targeting of cells of similar maturity are desired, our data suggest that TAM injections are still the preferable method. Despite differences in recombination efficiency between the TAM administration routes tested here, this work also shows that two weeks of TAM chow induces genetic recombination with similar specificity to 5 d of TAM injection. Mice fed TAM chow for two weeks showed a recombined cell population of similar phenotype (primarily NSPCs) as TAM-injected mice 7 d after the start of injections. Thus, TAM-infused chow could be a workable alternative administration route when TAM injection is not feasible, when lower recombination efficiency is desired, or when long-term suppression of cell proliferation would interfere with interpretation of experimental results.
Our findings comparing TAM versus vehicle-injected mice provide several useful insights into the effects of this commonly-used agent on adult neurogenesis. First, we found that neurogenesis derived from cells born during TAM was not substantially affected by TAM injection. These results are similar to those reported in (Rotheneichner et al., 2017), which showed no difference in BrdU1 cell density or phenotype between TAM and vehicle-injected mice 10 d after TAM/BrdU using five-month-old male and female mice. In contrast, using three-to four-week-old (juvenile) male mice, Lee et al. (2020) found that the density of surviving BrdU1 cells approximately one week after TAM/BrdU was suppressed by ;2-fold in TAM-injected versus vehicle-injected mice. Combined, these findings suggest that the effect of TAM injection on production of new DG cells that survive 11 weeks could depend on organismal age. Sex may also be a factor, but this is difficult to discern as previous work either used only males or did not show data for males and females separately. Future research is needed to address the effects of TAM on cell proliferation and survival to better understand how TAM affects neurogenesis across the lifespan and in both sexes.
Second, we show that TAM injection had prolonged effects on progenitor cell proliferation, suppressing progenitor density and proliferation by ;1.3-fold three weeks after TAM. Lee and colleagues similarly reported a ;1.8fold TAM-induced suppression of Mki67-labeled progenitors one week after TAM in juvenile mice. Rotheneichner et al. (2017) reported no difference in PCNA1 progenitor cell number 10 d after TAM in their five-month-old mice, however. Again, this difference may reflect increased susceptibility of juvenile mice to the effects of TAM, a subtle sex difference, or a time point effect.
The mechanism by which TAM suppresses progenitor proliferation weeks after TAM treatment is unclear. One possible mechanism for suppression of cell proliferation associated with TAM treatment in general is brain estrogen receptor modulation by TAM. TAM acts on brain ERa, ERb , and the transmembrane receptor GPR30 (Gonzalez et al., 2016). Specifically, TAM can be an estrogen agonist at GPR30 (Filardo et al., 2000;Vivacqua et al., 2006a,b) and ERb (McDonnell and Wardell, 2010), and it can induce estrogen blockade at ERa (McDonnell and Wardell, 2010). Because ERa agonists can enhance cell proliferation in the adult hippocampus (Mazzucco et al., 2006), it therefore seems possible that TAM blockade of ERa could suppress proliferation. Receptor-independent effects on cell cycle genes, as proposed by Lee et al. (2020), are also a potential mechanism of TAM-induced suppression of proliferation. With either mechanism, however, it is unclear why those effects would persist long after TAM withdrawal. TAM metabolites remain in the brain for ;8 d after administration (Valny et al., 2016), making it unlikely that suppression of proliferation three weeks later is a result of active TAM presence. This long-term persistence could therefore be because initial effects of TAM are slow to arise and/or indelible (e.g., epigenetic changes), or because they reflect effects of TAM withdrawal rather than TAM itself. Future research is needed to parse out these multiple, nonexclusive hypotheses.
We also noted delayed weight gain in female mice that received TAM injection, resulting in higher average body mass of TAM treated females than vehicle treated females three weeks after injection. Because TAM was administered systemically, this delayed weight gain in female mice could be because of TAM effects on multiple central or peripheral tissues. One candidate tissue is the hypothalamus. In mice, hypothalamic ERa-dependent gene expression changes have been linked to common TAM side effects, including decreased movement (Zhang et al., 2021), which could contribute to weight gain.
TAM chow had qualitatively different effects on neurogenesis and cell proliferation than TAM injection. In contrast to TAM injections, TAM chow increased neuronal differentiation of cells born during TAM but had no effect on density of EdU1 cells three weeks after TAM. Caloric restriction and reduced total TAM exposure in chow-fed mice (compared with TAM-injected mice) both likely contribute to this difference between paradigms. Caloric restriction is a particularly important confounding factor to consider when comparing TAM chow-fed to vehicle chow-fed animals. Although Bondolfi and colleagues found no effect of caloric restriction on hippocampal neurogenesis in male mice aged 3-11 months (Bondolfi et al., 2004), others have found that dietary restriction and intermittent fasting enhance neuronal differentiation in the adult hippocampus (Hornsby et al., 2016;Li et al., 2020). Both the caloric restriction during TAM feeding and the return to normal feeding after the end of TAM chow may therefore be affecting neurogenesis in TAMfed mice. Regardless of the reason for differing effects of TAM by administration route, our data further underscore the need for internal TAM-treated controls in experiments using TAM chow, just as with TAM injection.

Limitations and future directions
Because mice often crumble food and spread it across the floor of the cage, we are unable to account for chow that was chewed in this manner but not consumed. Our calculations of the amount of TAM chow consumed are almost certainly overestimations. This is a limitation of using chow that others may want to consider when knowing exact TAM amounts is critical to study design or interpretation.
Because this work examines direct TAM effects on experimental endpoints only in the context of neurogenesis research in adult animals, further testing is needed to determine the suitability of alternate TAM dosing methods for subadult animals and research on other brain processes. Also, our BrdU1 counts reflect both cell proliferation during TAM administration and post-TAM cell survival, making it difficult to disentangle TAM effects on either process independently. Thymidine analogs, such as the BrdU and EdU we used here, mark only a subset of proliferating cells, so it is possible that further or fewer differences would be observed if all proliferating cells were labeled (with a mitotic marker such as PCNA or MCM2) and assessed. It is also possible that any or all of the effects observed in this study could vary throughout the anatomy of the dentate gyrus (dorsal vs ventral, inferior vs superior blades of the subgranular zone). Our counts reflect areas sampled throughout the DG and therefore do not capture this information. In addition, although we provide the parental source of NestinCreER T2 in Extended Data Figure 2-1, we do not have sufficient sample size to draw conclusions about the effect of inheriting maternal versus paternal NestinCreER T2 on any of the outcome measures.
Better understanding of the molecular mechanism by which TAM acts on adult neurogenesis, when provided by injection or chow, is also still needed, as is further assessment of how (if at all) TAM differentially affects reporter-labeled and nonlabeled cell populations. Finally, clarity regarding the actions of TAM on brain estrogen receptors is lacking and could help guide researchers' choice of TAM dosing methods and interpretation of studies where TAM is used.
In conclusion, this work shows that voluntary TAM chow consumption may be a suitable alternative to TAM injections for inducing Cre-lox recombination in some adult neurogenesis studies. This work further identifies several effects of TAM administration protocols, whether by injection or food, on adult neurogenesis endpoints. These effects are separate from genetic recombination effects and are an important confounding variable in experimental designs that rely on comparisons to a vehicletreated control. Thus, we suggest that research using TAM-inducible Cre lines use TAM-treated wild-type (nonrecombination susceptible) littermates as controls, rather than vehicle-treated mice.