Cre Recombinase Driver Mice Reveal Lineage-Dependent and -Independent Expression of Brs3 in the Mouse Brain

Abstract Bombesin receptor subtype-3 (BRS3) is an orphan receptor that regulates energy homeostasis. We compared Brs3 driver mice with constitutive or inducible Cre recombinase activity. The constitutive BRS3-Cre mice show a reporter signal (Cre-dependent tdTomato) in the adult brain because of lineage tracing in the dentate gyrus, striatal patches, and indusium griseum, in addition to sites previously identified in the inducible BRS3-Cre mice (including hypothalamic and amygdala subregions, and parabrachial nucleus). We detected Brs3 reporter expression in the dentate gyrus at day 23 but not at postnatal day 1 or 5 months of age. Hypothalamic sites expressed reporter at all three time points, and striatal patches expressed Brs3 reporter at 1 day but not 5 months. Parabrachial nucleus Brs3 neurons project to the preoptic area, hypothalamus, amygdala, and thalamus. Both Cre recombinase insertions reduced Brs3 mRNA levels and BRS3 function, causing obesity phenotypes of different severity. These results demonstrate that driver mice should be characterized phenotypically and illustrate the need for knock-in strategies with less effect on the endogenous gene.


Introduction
Bombesin receptor subtype-3 (BRS3, Bombesin-like receptor 3, BB3) is a G-protein-coupled receptor critical for the maintenance of energy balance (Jensen et al., 2008;González et al., 2015;Xiao and Reitman, 2016). BRS3 is considered an orphan receptor as its endogenous ligand is not known; specifically, it does not bind the natural ligands (gastrin-releasing peptide and neuromedin B) for the most closely related receptors (Mantey et al., 1997); nor does it bind bombesin, which is the frog ortholog of neuromedin B (Hirooka et al., 2021). Mice lacking BRS3 are hyperphagic, have a reduced resting metabolic rate and body temperature, and consequently become obese (Ohki-Hamazaki et al., 1997b). Brs3 knock-out (KO) mice are also reported to exhibit reduced social responses, a heightened preference for sweetness, and increased aversion to bitterness (Yamada et al., 1999(Yamada et al., , 2000. Although Brs3 is also expressed outside the CNS (Jensen et al., 2008), the metabolic phenotypes of Brs3 KO mice are predominantly mediated by glutamatergic Brs3 neurons, with contributions from neurons expressing MC4R and SIM1 demonstrating the necessity of brain BRS3 (Xiao et al., 2017(Xiao et al., , 2020. Consistent with the phenotypes of BRS3-null mice, administration of a BRS3 agonist reduces food intake and body weight and increases brown adipose tissue (BAT)-induced thermogenesis, heart rate, and blood pressure, while BRS3 antagonists increase food intake and body weight (Guan et al., 2010Nio et al., 2017;Maruyama et al., 2018). Single doses of a BRS3 agonist increased blood pressure in humans (Reitman et al., 2012), likely via increased sympathetic tone (Lateef et al., 2016), reducing interest in central agonism as a human therapeutic for obesity and stimulating interest in studying BRS3 agonists that do not enter the brain (Kiyotsuka et al., 2016).
While pharmacologic and genetic manipulations of BRS3 have revealed the functions of the receptor, BRS3 can also be used as a marker for studying the neural circuitry regulating energy homeostasis. BRS3 is found in discrete brain regions in the mouse (Ohki-Hamazaki et al., 1997a;Yamada et al., 1999;Guan et al., 2010;Zhang et al., 2013;Piñol et al., 2018; see also Allen Mouse Brain Atlas) and other species (Liu et al., 2002;Sano et al., 2004;Zhang et al., 2013;Maruyama et al., 2018). The contributions of hypothalamic Brs3 neurons to energy homeostasis are heterogeneous. Brs3 neurons in the paraventricular hypothalamus (PVH) regulate food intake, but not body temperature or BAT thermogenesis. In contrast, Brs3 neurons in the preoptic area (POA) of the hypothalamus and dorsomedial hypothalamus (DMH) regulate body temperature, energy expenditure, heart rate, and blood pressure, but not appetite (Piñol et al., 2018(Piñol et al., , 2021. POA Brs3 neurons have also been implicated in parental behaviors (Moffitt et al., 2018;Yoshihara et al., 2021).
We previously generated a tamoxifen-dependent BRS3-Cre driver mouse line (hereafter called BRS3-CreER) and used it to elucidate the anatomy, connectivity, and physiology of neurons expressing Brs3 (Piñol et al., 2018). Here we characterize a new, constitutive BRS3-Cre allele (hereafter called BRS3-IRES (internal ribosomal entry site)-Cre) and show that the BRS3-IRES-Cre driver can be used for lineage tracing. In both BRS3-Cre lines, the Cre insertion affects Brs3 expression and function in whole-body physiology. These results provide guidance and have general implications for developing driver mice.

Generation of BRS3-IRES-Cre mice
A cassette encoding IRES-mnCre:GFP was inserted 39 of the termination codon in the last coding exon of the Brs3 gene. The 59 arm (;6 kb with SpeI and SalI sites at 59 and 39 ends, respectively) and 39 arm (;5.7 kb with XhoI and NotI sites at 59 and 39 ends, respectively) of the Brs3 gene were subcloned from a C57BL/6 BAC clone and cloned into polylinkers of a targeting construct that contained IRES-mnCre:GFP, a frt-flanked Sv40Neo gene for positive selection, and HSV thymidine kinase and Pgkdiphtheria toxin A chain genes for negative selection. The IRES-mnCre:GFP cassette has an IRES, a Myc-tag, and nuclear localization signals at the N terminus of Cre recombinase, which is fused to green fluorescent protein followed by an SV40 polyadenylation. The construct was electroporated into G4 ES cells (C57BL/6 Â 129 Sv hybrid), and correct targeting was determined by Southern blot of DNA digested with KpnI using a 32 P-labeled probe upstream of the 59 arm of the targeting construct. Of the 75 clones analyzed, 44 were correctly targeted. One clone that was injected into blastocysts resulted in good chimeras that transmitted the targeted allele through the germline. Progeny were bred with Gt(Rosa)26Sor-FLP recombinase mice to remove the frt-flanked SV-Neo gene. Mice were then continuously backcrossed to C57BL/6 mice. BRS3-IRES-Cre;Ai14 mice showed no evidence of germline recombinase activity in 12 progeny as assessed by lack of generalized tdTomato expression. Routine genotyping is performed with the following three primers: X717 59 CTG CCT CAA GGC AGA GCA GC (Brs3 forward); X718 59 CCT CTT CTT CTC TAC TTG GTG GGC (Brs3 reverse); and X719 59 GCT TCG GCC AGT AAC GTT AGG (IRES reverse). The wild-type allele gives a band of ;350 bp, while the targeted allele gives a band of ; 270 bp after 34 cycles with 20 s annealing at 60°C.

Parabrachial nucleus neuron projection tracing
Male BRS3-IRES-Cre mice were anesthetized with isoflurane and placed on a robotic stereotaxic frame (Neurostar). AAV-1EF1a-DIO-YFP and AAV1-EF1a-DIOsynaptophysin:mCherry were injected bilaterally into the parabrachial nucleus (PBN; 200 nl; AP, À4.8 mm; ML, 61.4 mm; DV, À3.5 mm) at a rate of 0.1 ml/min for 2 min with the needle left in place for the following 5 min. After surgery, mice were allowed to recover for ;3 weeks. Following recovery, mice were anesthetized with phenytoin/pentobarbital and perfused with PBS, pH 7.4, followed by 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer, pH 7.4. Brains were removed and placed in 4% PFA to postfix for 24 h and were subsequently placed in 30% sucrose for several days before being embedded in OCT compound and stored at À80°C. Coronal sections (35 mm) were cut on a cryostat (Leica Microsystems) and collected in cryoprotectant for long-term storage at À20°C . To enhance the signal, sections were washed three times in PBS for 5 min and incubated in a blocking solution [3% normal donkey serum in PBS with Tween 20 (PBST)] for 1 h at room temperature. Sections were incubated overnight at 4°C in PBST with the primary antibodies chicken-anti-GFP (1:10,000; catalog #ab13970, Abcam) and rabbit-anti-DsRed (1:1000; catalog #632496, Takara Bio). The following day, sections were washed three times in PBS and then incubated for 1 h in PBS with the secondary antibodies Alexa Fluor 488 donkey anti-chicken and Alexa Fluor 594 donkey anti-rabbit (1:500; Jackson ImmunoResearch). Sections were then washed three times in PBS, mounted onto glass slides, and coverslipped with Fluoromount-G with DAPI (Southern Biotech). Images were acquired using a Keyence BZ-X700 microscope and an Olympus confocal microscope (model FV-1200, Olympus).

Mouse phenotyping
BRS3-IRES-Cre and littermate control mice were singly housed at 9 weeks of age and placed on a HFD at 11 weeks of age for metabolic studies, as described previously (Xiao et al., 2017). Another cohort of BRS3-IRES-Cre and littermate control mice was singly housed and maintained on chow diet and was used for measuring core body temperature (Tb) by telemetry as described previously (Xiao et al., 2017). Effects of MK-5046 on food intake and Tb were evaluated as reported previously (Xiao alleles. The three exons are numbered with BRS3 coding sequences in black, untranslated regions in white, IRES in blue, T2A in red (with cleavage point indicated by an arrow), Cre in orange, and GFP in green. B, Reporter expression in coronal sections from 5month-old male BRS3-IRES-Cre;Ai14 mice at the indicated level from bregma. Labels in yellow denote reporter detected in BRS3-IRES-Cre but not BRS3-CreER mice and white labels are sites expressing reporter in both lines. Arc, Arcuate hypothalamic nucleus; DMH, dorsomedial hypothalamic nucleus. The EP and SNR fluorescence is from fibers; the rest are from cell bodies. C, D, Striatum (C) and hippocampus (D) at higher magnification. CA1, field CA1; CA3, field CA3. et al., 2020), with the exception that for testing the effect of MK-5046 on food intake mice were fasted for 5 h and dosed 30 min before lights off, and chow intake was measured for the first 2 h after the entry of dark phase.

Single-cell RNA analysis
Single-cell or single-nucleus count matrices were downloaded from GEO and analyzed with R (version 4.0.2). The total number of cells and number of cells with at least one detectable Brs3 transcript were evaluated for each dataset. True single-cell RNA (scRNA) Brs3 positivity rates were estimated from both raw and normalized data by calculating the Poisson mean that would produce the observed ratio of (cells with one Brs3 transcript)/(cells with greater than one Brs3 transcript) and of (cells with one Brs3 transcript)/(cells with two Brs3 transcripts). Brs3 expression in the dentate gyrus was visualized using dataset C [from the Linarsson Lab (http://linnarssonlab.org/dentate/); Hochgerner et al., 2018].
In the Arc-ME dataset GSE93374 (Campbell et al., 2017), of the 13,079 neurons, 552 expressed at least one Brs3 transcript, and these neurons were clustered using Seurat (version 3.2.0; Butler et al., 2018;Stuart et al., 2019). Raw counts were normalized, scaled, and the 2000 most variable genes were used as input for principal component analysis. A resolution of 0.6 and 20 principal components were used for the clustering analysis, which was visualized with t-SNE (t-distributed stochastic neighbor embedding). The resolution (0.4-1.0) and principal components (10-30) were varied to confirm the robustness of the clustering. Neither condition (e.g., diet, sex, fasted state) nor batch were major drivers of the clustering. Differentially expressed genes (DEGs) and cluster marker genes were identified using the Wilcoxon rank-sum test. The top five DEGs for each cluster (based on average log 2 -fold change) were visualized by heatmap. Cluster names were assigned with one or more significantly enriched marker genes (average log 2 -fold change, .1.4; false discovery rate-adjusted p value, ,7.5E-41), with the exception of one "unassigned" cluster in which all DEGs had an average log 2 -fold change of 0.71.

Results
Reporter expression in BRS3-IRES-Cre;Ai14 mice There are currently two BRS3-Cre driver mouse lines available, one constructed by inserting a Cre recombinase sequence just after the stop codon of the Brs3 locus (BRS3-IRES-Cre) and the other with the insertion of T2A-CreERT2 at the stop codon (BRS3-CreER; Fig.  1A). In BRS3-IRES-Cre mice, the Cre recombinase is constitutively active (Palmiter, 2018), whereas in BRS3-CreER mice, tamoxifen administration provides temporal control of the recombinase activity (Piñol et al., 2018). Both drivers should express Cre with the same pattern as wild-type BRS3, and the BRS3-CreER mice express a Cre-dependent reporter with a pattern matching that of Brs3 mRNA (Piñol et al., 2018). The only previous information for BRS3-IRES-Cre mice is that the Cre-dependent reporter is expressed in the PBN (Palmiter, 2018). The GFP in BRS3-IRES-Cre mice was not detected, even using immunohistochemistry, presumably because of very low levels of expression of this receptor.

Lineage tracing in BRS3-IRES-Cre mice
The additional reporter expression in BRS3-IRES-Cre mice could be from the following: (1) Cre expression in regions not normally expressing Brs3 (ectopic sites); (2) higher Cre expression in sites that do express Brs3 (eutopic sites); and/or (3) lineage effects with Cre expression in a precursor cell (or prior expression in the current cell). To distinguish among these possibilities, we injected a virus carrying a Cre-dependent GFP construct into three brain regions of BRS3-IRES-Cre;Ai14 and BRS3-CreER Ai14 mice. Cre activity is required at the time of virus injection/tamoxifen treatment for GFP expression while either prior or current Cre activity will produce a tdTomato signal ( Fig. 2A).
In 3-to 5-month-old BRS3-CreER;Ai14 mice, virus-derived GFP was present in the DMH, but not the hippocampus or striatum (Fig. 2B-D). Virus-derived reporters have a higher copy number and are typically expressed at higher levels than host genome-derived reporters. Thus, lower eutopic Cre activity in BRS3-CreER;Ai14 mice likely does not explain the lack of GFP in the hippocampus and striatum.
BRS3-IRES-Cre;Ai14 mice showed the same viral GFP expression pattern (present in DMH, but not in hippocampus or striatum) as the BRS3-CreER;Ai14 mice. Thus, at the time of viral injection, there was no active Cre recombinase in the hippocampus or striatum. As expected, the BRS3-IRES-Cre;Ai14 mice expressed tdTomato in all three regions. These results suggest that Brs3 is expressed earlier in hippocampal and striatal neuron development, but no longer at 3-5 months of age, consistent with lineage tracing in BRS3-IRES-Cre mice.
Examination of genomic reporter expression in BRS3-IRES-Cre;Ai14 mice on P1 demonstrated expression in the striatum, PVH, and PBN, but not the hippocampus (Fig. 3). By P23, all four of these regions contained tdTomato reporter. Together, the results suggest that striatal Brs3 expression is on at (or before) P1 and turns off sometime before adulthood. Hippocampal Brs3 expression is initially off (P1) and is turned on at (or before) P23.
actual Brs3-positive percentages are 1.1-fold to 3-fold of the nominal percentages, depending on the dataset, normalization, and assumptions used in calculating event rates.
To explore the impact of the Brs3 false negative findings, we studied a dataset of Arc-ME scRNA sequences (Campbell et al., 2017), chosen for its large number of Brs3-positive neurons (4.2% of total). Clustering only the 552 Brs3-positive neurons yielded six clusters, five of which mapped to nine of the clusters reported in the study by Campbell et al. (2017;Fig. 4C,D). Only two of the original clusters had high enough detected Brs3 positivity (46% in n01 and 42% in n02) to indicate that Brs3 might be present in all cluster members. Thus, removing the Brs3 detection bias did not improve the clustering analysis.

Projections of PBN BRS3 neurons
The PBN receives diverse sensory inputs and integrates and transmits this information to many forebrain regions (Palmiter, 2018;Chiang et al., 2019). As a step in characterizing PBN BRS3 neurons, the BRS3-IRES-Cre mouse was used to trace the projections from PBN BRS3 neurons (Fig. 5). Projections were observed to the preoptic area (lateral POA), hypothalamus (PVH, LH, DMH, principal sensory trigeminal nucleus), amygdala (BNST, central nucleus of the amygdala), thalamus [mediodorsal thalamus (MD), ipsilateral MD], and dorsal raphe; the VTA signal is likely caused by fibers of passage because synap- Figure 6. Metabolic phenotype of BRS3-IRES-Cre mice. A, Effect of a high-fat diet in male BRS3-IRES-Cre and littermate control mice (n = 8/group) on body weight, fat mass, lean mass, food intake, and energy expenditure determined by mass balance (Ravussin et al., 2013). Data are the mean 6 SEM; *adjusted p , 0.05 from two-way ANOVA (Extended Data Fig. 6-1, with Sídák's multiple-comparisons test). B-D, Baseline core body temperature (B), effect of MK-5046 (10 mg/kg, i.p., given at onset of the dark cycle in 5 h fasted mice) on food intake (C), and effect of MK-5046 (10 mg/kg, i.p., at 10:00 A.M. in overnight-fasted mice) on body temperature (D). E, Tb changes from baseline (À150 to À30 min) to 60-180 min and 60-120 min after dosing in a crossover design. The p values are from two-way ANOVA with Sídák's multiple-comparisons test, see Extended Data Figure 6-1. Mice in B-E are chow-fed males (n = 5-15/group). In D, data are the mean with the SEM omitted for visual clarity. tophysin expression in the region is very low. Most of these nuclei have a role in regulating energy homeostasis.
A robust thermal phenotype of BRS3-null mice is an increased Tb span (defined as the difference between the 95th and 5th Tb percentiles over integral multiples of 24 h intervals; Xiao et al., 2017). The Tb span was not increased in the BRS3-IRES-Cre mice compared with wild-type littermates, and, as expected, there was no difference in either light-or dark-phase Tb (Fig. 6B).
However, the BRS3-IRES-Cre mice lost the suppression of food intake elicited by treatment with a BRS3 agonist, MK-5046 (Fig. 6C), and the BRS3 agonist effect of increasing light-phase Tb  was blunted in the BRS3-IRES-Cre mice (Fig. 6D,E). These results indicate that the BRS3-IRES-Cre allele is a hypomorph, with a phenotype milder than the complete BRS3-null mice.
In both BRS3-Cre drivers, BRS3 and Cre are encoded in a single mRNA, and, as expected, the Brs3/Cre mRNA ratios were similar in the BRS3-CreER;Ai14 and BRS3-IRES-Cre;Ai14 mice (2.3 and 3.4, respectively). Cre mRNA in the hypothalamus of BRS3-IRES-Cre mice is 18% of the level in BRS3-CreER mice. Interestingly, tdTomato mRNA levels in the hypothalamus of BRS3-IRES-Cre mice were 166% of that in BRS3-CreER mice, suggesting that there are more tdTomato-expressing neurons (and 0 1 2

Brs3
Cre  not more tdTomato mRNA/cell since both mice use the same reporter locus) in BRS3-IRES-Cre;Ai14 mice. Possible mechanistic explanations for increased tdTomato-positive neurons are lineage effects in BRS3-IRES-Cre;Ai14 mice and/or incomplete activation of Cre by tamoxifen in BRS3-CreER;Ai14 mice. In the hippocampus, very low levels of Cre and tdTomato mRNA were detected in both BRS3-CreER; Ai14 and BRS3-IRES-Cre;Ai14 mice. In the striatum, low levels of tdTomato mRNA (BRS3-IRES-Cre . BRS3-CreER) were observed. Apparent differences between mRNA levels and fluorescence signal may be because of translational efficiency and/or protein stability, as protein levels were not quantitated.

Discussion
The design of genomic reporter/driver mice has evolved from random insertion of small-plasmid DNAs to random insertion of larger DNAs (e.g., bacterial artificial chromosomes), and to targeted insertion into the endogenous locus, recently with high efficiency using CRISPR/Cas9 technology. Thus, as with the BRS3-Cre mice, one can now reliably produce driver/reporter alleles that have a high likelihood of correctly tracking the expression pattern of the target gene, a major advance [see (Song and Palmiter, 2018;Luo et al., 2020) for some cautions].
Lineage effects in nonconditional systems are widely recognized (Daigle et al., 2018), but infrequently characterized in detail [exceptions are proopiomelanocortin (POMC; Padilla et al., 2010) and GFAP (Ganat et al., 2006)]. The differences between the two BRS3-Cre drivers illustrate the value of having available both inducible and constitutive recombinases.
The large doses of tamoxifen required for efficient Cre recombinase activity can reduce body weight and cause loss of adipose tissue (Ye et al., 2015). Also, Brs3 expression is sex dimorphic (Xu et al., 2012;Chen et al., 2019), and Brs3 neurons are involved in sex-dimorphic behaviors that may be regulated by estrogen receptors (Moffitt et al., 2018). Thus, the ability to select inducible, tamoxifendependent, and/or constitutive tamoxifen-independent driver mice improves experimental design.
The Brs3 locus is unforgiving for expressing a recombinase/reporter allele since Brs3 is on the X chromosome, undergoes X-inactivation, and is expressed at a low level. Thus, modified alleles must preserve BRS3 function while optimizing reporter expression. Unfortunately, both BRS3-Cre alleles are hypomorphs, with more function in BRS3-CreER than BRS3-IRES-Cre allele (Table 1, summary), and the metabolic effects of the deficiency should be considered when the mice are used.
Each allele produces a single mRNA that encodes both BRS3 and Cre. The T2A sequence in BRS3-CreER causes a ribosome skip, with the downstream protein being produced in similar amounts to the upstream one, although the upstream protein has 17 aa appended, which may affect function (Szymczak-Workman et al., 2012). In the BRS3-IRES-Cre allele, the IRES system results in translation of the downstream Cre at a fraction of the level of the upstream gene (Mizuguchi et al., 2000). While the T2A and IRES sequences differentially affect protein levels, they do not explain the reduced Brs3 mRNA levels, which are presumably because of mRNA instability or reduced transcription caused by the inserted genomic sequences, such as by interfering with enhancer function. It is notable that the BRS3-IRES-Cre allele produces efficient recombination despite a 93% reduction in Brs3 mRNA levels and further reduced Cre protein because of the IRES. The effectiveness of the recombinase means that quantitative differences in Brs3 mRNA levels, such as the sex dimorphism in the BNST and medial amygdala (Xu et al., 2012;Chen et al., 2019), are not detected in reporter mice.
The availability of characterized BRS3-Cre driver alleles facilitates investigation of BRS3 function in pancreatic islets (Feng et al., 2011), neuroendocrine tumors (Sherman et al., 2014), and certain cancers (Moreno et al., 2018;Ramos-Alvarez et al., 2019). The limited number of discrete brain nuclei expressing Brs3 means that the BRS-Cre drivers are particularly valuable for intersectional genetic studies (Madisen et al., 2015) of the neural circuits and networks that control metabolism and other processes.