The widely used Ucp1-Cre^Evdr transgene elicits complex developmental and metabolic phenotypes

Summary:

Bacterial artificial chromosome transgenic models, including most Cre-recombinases, enable potent interrogation of gene function in vivo but require rigorous validation as limitations emerge. Due to its high relevance to metabolic studies, we performed comprehensive analysis of the Ucp1-Cre^Evdr line which is widely used for brown fat research. Hemizygotes exhibited major brown and white fat transcriptomic dysregulation, indicating potential altered tissue function. Ucp1-Cre^Evdr homozygotes also show high mortality, growth defects, and craniofacial abnormalities. Mapping the transgene insertion site revealed insertion in chromosome 1 accompanied by large genomic alterations disrupting several genes expressed in a range of tissues. Notably, Ucp1-Cre^Evdr transgene retains an extra Ucp1 gene copy that may be highly expressed under high thermogenic burden. Our multi-faceted analysis highlights a complex phenotype arising from the presence of the Ucp1-Cre^Evdr transgene independently of the intended genetic manipulations. Overall, comprehensive validation of transgenic mice is imperative to maximize discovery while mitigating unexpected, off-target effects.

Introduction:

Mouse transgenic models such as overexpressors, reporters, and Cre-recombinases empower spatial and temporal genetic manipulation enabling unparalleled interrogation of gene function in vivo. These models have driven discoveries across biological scales, from molecular processes to whole body physiology and have become indispensable for elucidating the foundations of health and disease.

Most transgenic mouse alleles in use today have been generated via bacterial artificial chromosome (BAC) technology^1,2. This involves inserting sequences of interest (e.g. Cre-recombinase) into BAC plasmids (–150–350 kb) containing regulatory elements that confer spatiotemporal expression. For instance, insertion after a promoter sequence permits cell type- or stage-specific Cre-recombinase expression. The modified BAC, including contextual sequences, is then randomly integrated into the host genome in a nonspecific, stochastic manner, typically forming a multicopy concatemer³.

While revolutionary, Cre-driver lines generated by BAC transgenesis carry potential limitations that are rarely investigated. Validation of Cre-drivers is usually restricted to verification of specific expression in the targeted cell type. Initiatives like the CrePortal^4,5 have been invaluable for collating Cre expression data and provide a valuable resource to guide appropriate use of Cre-drivers. Yet other limitations associated with BAC transgenesis are rarely examined: (1) The insertion site of transgenes are mostly unknown; only 5.03% and 3.40% of all transgenic and Cre alleles, respectively, have mapped integration sites collated in Mouse Genome Informatics [Figure S1A–B]; (2) insertion can result in large genomic abnormalities that are not routinely inspected^6,7,8 and additionally the insertion may directly influence the phenotypes observed by different mechanisms^9–12; (3) passenger sequences are virtually never reported but may lead to unintended phenotypes¹³; (4) Cre transgenes are largely used in hemizygosity masking phenotypes that would otherwise be evident^14,15; (5) the common absence of Cre-only control groups precludes assessment of perturbations directly attributable to the presence of the Cre transgene or protein itself.

BAC transgenics have been instrumental for generating adipose-specific Cre driver lines to dissect the biology of the highly thermogenic brown adipose tissue (BAT) and the energy storing white adipose tissue (WAT)^16–18. Cre-recombinase lines utilizing the adiponectin promoter enabled targeting of all adipocytes^19–21. Promoter elements from UnCoupling Protein 1 (Ucp1) have conferred selective Cre-recombinase expression in brown adipocytes. Although other BAT Cre-targeting tools existed²², and other tools to target brown adipocytes are used²³, two Ucp1-Cre drivers dominate the literature currently: the constitutive Ucp1-Cre^Evdr line from the Rosen lab²⁴ and the tamoxifen inducible Ucp1-CreERT2^Biat line from the Wolfrum lab²⁵. Both lines show remarkable specificity, full penetrance, and robust activity on brown adipocytes^26–32. Among the two, the Ucp1-Cre^Evdr line has been more widely adopted, featuring in 78.85% of manuscripts in the Mouse Genome Informatics records. This skewed utilization may be due to greater accessibility in open repositories and concerns about tamoxifen effects on adipocytes^33,34.

Despite the extensive use of the constitutive Ucp1-Cre^Evdr allele, comprehensive validation of this driver line is lacking. Here, we perform an in-depth characterization of Ucp1-Cre^Evdr using genetic, genomic and physiologic approaches. Transcriptomic analysis showed substantial gene expression changes in both brown and white adipose tissues of hemizygous Ucp1-Cre^Evdr mice compared to wild-type littermate controls. Ucp1-Cre^Evdr homozygotes show high mortality, craniofacial abnormalities, and growth retardation. Molecular characterization of the Ucp1-Cre^Evdr transgene insertion site demonstrated substantial genomic alterations including disruptions of several genes at the integration locus in chromosome 1. Notably, the Ucp1-Cre^Evdr transgene retains an additional Ucp1 gene that may exhibit strong expression under high thermogenic burden. These effects suggest unintended consequence on brown adipose tissue function by Ucp1-Cre^Evdr. More broadly, our study highlights the critical need for extensive validation of BAC transgenic drivers.

Results:

UCP1-Cre^Evdr Homozygosity Induces Lethality, Growth Impairment, and Craniofacial Abnormalities.

While attempting to generate a Ucp1-Cre^Evdr mediated deletion of a Ucp1 floxed allele, we were unable to identify mice homozygous for the Ucp1 floxed allele and simultaneously harboring the Ucp1-Cre^Evdr transgene through standard genotyping (see below). Given that germline Ucp1 knockout mice are viable, embryonic lethality due to Ucp1 deficiency does not explain this bias in genotyping ratios. Thus, we ponder the question on whether the Ucp1-Cre^Evdr transgene itself may underlie the observed effects.

To rigorously evaluate the Ucp1-Cre^Evdr model for its use in discovery-based research, we generated control, hemizygous, and homozygous littermates by crossing Ucp1-Cre^Evdr hemizygous [Figure 1A]. We designated them as controls, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice, respectively. We reasoned that this strategy would enable comprehensive assessment of potential developmental, physiological, and molecular perturbations arising from this widely utilized Cre driver line.

Figure 1:

(A) Experimental strategy for the generation of control, Ucp1-Cre^Evdr hemizygous (1xUcp1-Cre^Evdr) and Ucp1-Cre^Evdr homozygous (2xUcp1-Cre^Evdr) mice.

(B) Expected and observed offspring genotypes obtained from 1xUcp1-Cre^Evdr to 1xUcp1-Cre^Evdr crosses separated by sex. N=251 pups from 46 litters. Statistical significance was calculated using Chi-square test.

(C) Kaplan-Meier survival plot of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice from 3 to 6 weeks of age. n= 60 controls, 152 1xUcp1-Cre^Evdr, 23 2xUcp1-Cre^Evdr. Statistical significance was calculated using Log-rank (Mantel-Cox) test.

(D) Representative photographs of alizarin red S stained skulls of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice. n=2 females and 2 males.

(E) Skull shape index (ration between condylobasal length and the interorbital constriction length) of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice. n=2 females and 2 males.

(F) Body weights of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr females. n= 7 controls, 9 1xUcp1-Cre^Evdr, 7 2xUcp1-Cre^Evdr.

(G) BAT weights of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr females. n= 7 controls, 9 1xUcp1-Cre^Evdr, 5 2xUcp1-Cre^Evdr.

(H) WAT weights of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr females. n= 7 controls, 9 1xUcp1-Cre^Evdr, 5 2xUcp1-Cre^Evdr.

(I) Liver weight of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr females. n= 7 controls, 9 1xUcp1-Cre^Evdr, 5 2xUcp1-Cre^Evdr.

(J) Skeletal muscles weight of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr females. n= 7 controls, 9 1xUcp1-Cre^Evdr, 5 2xUcp1-Cre^Evdr.

(K) Other organs weight of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr females. n= 7 controls, 9 1xUcp1-Cre^Evdr, 5 2xUcp1-Cre^Evdr.

(L) Representative H&E images of fat depots and liver from control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr females. n= 4 per genotype.

Unless otherwise noted, data are mean + SEM and statistical significance was calculated using one-way ANOVA followed by Tukey’s multiple comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001.

To unambiguously discriminate transgene copy number, we developed a quantitative copy number assay to detect Cre in genomic DNA rather than relying on endpoint PCR genotyping [Figure S1C]. At three weeks of age, we find significantly fewer 2xUcp1-Cre^Evdr mice than the expected Mendelian ratio of 25% [Figure 1B]. Specifically, only 14.04% of females and 16.06% of males are homozygous across 251 pups from 46 litters [Figure 1B]. Analysis of both sexes together reveals that 2xUcp1-Cre^Evdr comprises just 15.14% of the offspring, reflecting approximately 60% survival [Figure S1D]. Sex distribution is unaffected, indicating no differential penetrance between sexes [Figure S1E]. Moreover, over 40% of 2xUcp1-Cre^Evdr die spontaneously from 3 to 6 weeks of age, while 1xUcp1-Cre^Evdr and controls show no mortality [Figure 1C]. This mortality phenotype occurs with no indication of malaise in 2xUcp1-Cre^Evdr mice. The dramatic reduction in viability and high postnatal lethality in 2xUcp1-Cre^Evdr mice suggests profound biological perturbations by the Ucp1-Cre^Evdr transgene.

To understand the potential effects of Ucp1-Cre^Evdr, we next examined body weights of controls, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice from three to six weeks old. 1xUcp1-Cre^Evdr female and male mice are indistinguishable from controls [Figure S1F–G]. However, female and male 2xUcp1-Cre^Evdr mice display 15% and 15–19%, respectively, lower body weights from 3–6 weeks of age compared to controls and hemizygotes [Figure S1F–G]. Additionally, 2xUcp1-Cre^Evdr appeared to have calvarial defects. To more carefully characterize this dysmorphology, we dissected the heads of six-week-old controls, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice and performed Alizarin Red staining. As suspected, 2xUcp1-Cre^Evdr mice have a more domed, less elongated skull [Figure 1D]. Specifically, the frontal bones appear reduced, while the parietal bones appear increased in size. This dysmorphology resulted in a significantly reduced condylobasal to interorbital constriction length in 2xUcp1-Cre^Evdr mice [Figure 1E]. Since frontal bones are neural crest derived and parietal bones are mesodermally-derived, this may indicate a differential effect of the homozygosity of the Ucp1-Cre^Evdr transgene in the development or cross-communication of these two populations^35,36. Together, these data unveil craniofacial dysmorphologies and growth retardation in 2xUcp1-Cre^Evdr mice.

To better understand the effects of the Ucp1-Cre^Evdr transgene on mouse growth, we performed comprehensive tissue dissections at 6 weeks. Despite lower total body mass in 2xUcp1-Cre^Evdr females [Figure 1F], dissection of individual fat and lean tissues show that body weight reduction is surprisingly not due to a homogeneously global reduction of weight of each independent tissue. In particular, 2xUcp1-Cre^Evdr females show no change in BAT depots weights, including interscapular (iBAT), subscapular (sBAT) and cervical (cBAT) compared to control littermates [Figure 1G]. However, posterior subcutaneous or inguinal (psWAT), retroperitoneal (rWAT) and perigonadal (pgWAT) WAT depots are severely impacted, with 39%, 53% and 60% decrease in weight respectively [Figure 1H]. Beyond WAT, only quadriceps mass differs in 2xUcp1-Cre^Evdr females compared to controls [Figure 1I–K]. Male homozygotes also exhibit similar decrease in body weight and dramatic WAT depletion along with reductions in liver, quadriceps and gastrocnemius mass [Figure S1H–M]. Thus, tissue-specific effects underlie the global growth retardation in 2xUcp1-Cre^Evdr.

Histological analysis of iBAT and psWAT reveals no major changes between genotypes in either females or males [Figure 1L, S1P]. However, adipocytes in pgWAT of 2xUcp1-Cre^Evdr females and males appear to be smaller in size [Figure 1L, S1P]. This suggests that the changes in psWAT and pgWAT weights may be due to different mechanisms involving the generation of adipocytes or control of their size. We next investigated whether aberrant Cre expression could explain the dramatic WAT defects in 2xUcp1-Cre^Evdr. As expected, Cre mRNA is undetectable in all fat depots of control mice [Figure S1N–O]. In female iBAT, Cre expression correlates with transgene copy number, with 3.16-fold higher levels in 2xUcp1-Cre^Evdr than 1xUcp1-Cre^Evdr. However, in psWAT and pgWAT of 2xUcp1-Cre^Evdr females, Cre levels remain hardly detectable and unchanged compared to 1xUcp1-Cre^Evdr [Figure S1N]. Analysis of male fat depots show similar results [Figure S1O]. This tissue distribution expression argues against Cre misexpression driving WAT perturbations in 2xUcp1-Cre^Evdr mice.

The UCP1-Cre^Evdr Transgene is Inserted in Chromosome 1, Disrupts Genomic Integrity, and Harbors an Extra Ucp1 Gene Copy.

To date, the genomic integration site and structure of the Ucp1-Cre^Evdr transgene are unknown. To elucidate the integration site of the Ucp1-Cre^Evdr transgene, we performed targeted locus amplification (TLA, Cergentis) in a hemizygous male. TLA is an unbiased, genome-wide method that utilizes sequence-specific inverse PCR of a circularized genomic DNA library following NlaIII fragmentation and crosslinking. Subsequent deep sequencing of PCR products enables mapping of the transgene insertion locus³⁷.

TLA using Cre-specific primers reveals the Ucp1-Cre^Evdr transgene integrated into chromosome 1, cytoband A5 [Figure S2A, 2A]. As expected, Cre primers also detects homology near the endogenous Ucp1 locus in chromosome 8, indicating inclusion of surrounding Ucp1 genomic sequences in the transgene [Figure 2A]. Primer pairs surrounding Ucp1 produces high signal levels at the Ucp1 locus [Figure S2A–B]. TLA maps the concatemer insertion site of Ucp1-Cre^Evdr to chr1:20,962,125–21,016,858 [Figure 2B]. Integration induces a –54 kb deletion flanking the insertion sites, along with a 3’ –280 kb inversion [Figure 2B]. This directly deletes or inverts the entirety or large portions of 4 genes (Paqr8, Efhc1, Tram2, Tmem14a) [Figure 2B]. Additionally, the concatemer localizes in close proximity to other 7 genes (Il17a, Il17f, Mcm3, Gsta3, Khdc1a, Khdc1c, Khdc1b). Several noncoding sequences within or in close proximity to the concatemer may also be affected [Figure 2B]. The majority of the genes directly affected or in close proximity to the Ucp1-Cre^Evdr concatemer exhibit low expression in adipose tissues, but they are selectively highly expressed in an array of other tissues [Figure 2C]. Knockout mouse models have not been generated for each coding gene affected by the Ucp1-Cre^Evdr transgene [Figure 2D]. However, out those generated, only the Mcm3 knockout mice show prenatal lethality with complete penetrance³⁸ [Figure 2D]. Although this is quite distinct from what we observe in 2xUcp1-Cre^Evdr mice [Figure 1B–C, S1D–E], the genomic disruption induced by the Ucp1-Cre^Evdr concatemer may contribute to the above identified perturbations.

Figure 2:

(A) Whole genome TLA mapping analysis of 1xUcp1-Cre^Evdr genome using primers specifics for the sequences of Cre.

(B) Schematic representation of the identified integration site of the Ucp1-Cre^Evdr transgene.

(C) Gene expression of coding genes surrounding the Ucp1-Cre^Evdr transgene from the Mouse ENCODE transcriptome data.

(D) Knockout mortality phenotype association to each coding gene surrounding the Ucp1-Cre^Evdr transgene in MGI.

(E) Coverage of BAC 148M1 inserted within the Ucp1-Cre^Evdr transgene as determined by TLA.

(F) de novo reconstruction of the CRE-proximal part of Ucp1-Cre^Evdr transgene mRNA from iBAT RNA-seq data. UTR: untranslated region; NLS: nuclear location signal; CDS: coding sequence; Unk: unknown.

(G) Copy number assay of Cre of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice. n= 3 per genotype.

(H) Copy number assay of Ucp1 of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice. n= 8 per genotype.

(I) Absolute copy number by ddPCR of Cre of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice. n= 3 per genotype.

Next, we explored the Ucp1-Cre^Evdr concatemer structure. To do this, we first analyzed the TLA data. We find that –75% of the original BAC used to generate the Ucp1-Cre^Evdr mice (BAC 148M1), which covers –230Mb of chromosome 8 surrounding the Ucp1 gene, is incorporated with the transgene in chromosome 1 [Figure 2E]. Notably, this includes an entire extra copy of the Ucp1 gene. To further elucidate transgene structure, we performed de novo assembly of Cre-proximal transgene sequence of iBAT RNA-seq reads in Ucp1-Cre^Evdr mice using the Cre-recombinase coding sequence as bait. Upstream of the Cre coding sequence, we identified the proximal Ucp1 5’UTR sequence followed by the start codon and SV40 nuclear localization signal [Figure 2F]. The Cre coding sequence is followed at 3’ by a 3’UTR and a short sequence of unknown function [Figure 2F]. The length of sequencing fragments limits the extend of the transgenic Ucp1 gene we can detect as part of the Ucp1-Cre^Evdr transgene unambiguously against the endogenous copy; however, we find Cre transgene bound to Ucp1 exons 1 and part of exon 2 [Figure 2F]. The presence of Ucp1 mRNA within the Ucp1-Cre^Evdr transgene transcript suggests that the extra copy of Ucp1 gene may be expressed.

TLA cannot discern the number of repetitions occurring within the concatemer. To clearly determine the copy number of Ucp1 and Cre genes within the Ucp1-Cre^Evdr concatemer, we employed two quantitative PCR-based techniques. First, we developed copy number assays against the Ucp1 intron 3 to assess the number of copies of Ucp1 gene in genomic DNA of controls, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice. Control littermates were used as reference for two copies of Ucp1 gene [Figure 2G–H]. In contrast, 1xUcp1-Cre^Evdr mice have three copies and 2xUcp1-Cre^Evdr mice have four copies of the Ucp1 gene [Figure 2G–H]. However, this assay requires calibration with reference samples, limiting its ability to discern Cre copy number within the transgene concatemer. To solve this, we used digital droplet PCR (ddPCR)³⁹ to quantify the absolute copy number of Cre in HaeIII-digested genomic DNA. As anticipated, control mice contained no copies of Cre. In contrast, 1xUcp1-Cre^Evdr mice harbored a single Cre copy, while 2xUcp1-Cre^Evdr mice contained two copies per genome [Figure 2I]. Taken together, these complementary assays indicate that the Ucp1-Cre^Evdr transgene structure comprises one additional copy of the Ucp1 gene and a single copy of Cre.

UCP1-Cre^Evdr Transgene Induces Profound Effects in BAT and WAT Biology.

Because Ucp1-Cre^Evdr directly affects fat size, we next examined if the Ucp1-Cre^Evdr transgene directly impacts fat biology. First, we performed unbiased whole genome expression profiling of iBAT and psWAT from control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr female mice. Strikingly, the presence of the transgene profoundly alters the transcriptomic landscape of both fat depots. In 1xUcp1-Cre^Evdr iBAT, 1012 genes are upregulated and 905 downregulated compared to controls [Figure 3A]. Even more dramatic effects are evident in 1xUcp1-Cre^Evdr psWAT, with 3742 genes upregulated and 3130 downregulated despite barely detectable transgene expression [Figure 3B, S1N]. Comparisons between 2xUcp1-Cre^Evdr and control females reveal similar transcriptomic perturbations, with over 10-fold more altered genes in psWAT (8313) than in iBAT (760) [Figure 3C–D]. A lower number of genes are significantly different in iBAT and psWAT when comparing 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr [Figure S3A–B]. Remarkably, this suggest that the major effect appears from having just one copy of the Ucp1-Cre^Evdr transgene. The dramatic transcriptomic effects in Ucp1-Cre^Evdr fat depots, especially the psWAT, suggest either potent inter-tissue consequences or major effects from transgene insertion. In summary, unbiased transcriptional profiling indicates that the Ucp1-Cre^Evdr transgene may profoundly impact the molecular state of both brown and white adipose tissues.

Figure 3:

(A) RNA-seq comparing female control and 1xUcp1-Cre^Evdr iBAT gene expression (left). Each dot represents one gene. Corresponding GO analysis (right). Genes and pathways significantly enriched in controls are labeled in orange and those enriched in 1xUcp1-Cre^Evdr are labeled in red.

(B) RNA-seq comparing female control and 1xUcp1-Cre^Evdr psWAT gene expression (left). Each dot represents one gene. Corresponding GO analysis (right). Genes and pathways significantly enriched in controls are labeled in orange and those enriched in 1xUcp1-Cre^Evdr are labeled in red.

(C) RNA-seq comparing female control and 2xUcp1-Cre^Evdr iBAT gene expression (left). Each dot represents one gene. Corresponding GO analysis (right). Genes and pathways significantly enriched in controls are labeled in orange and those enriched in 2xUcp1-Cre^Evdr are labeled in brown.

(D) RNA-seq comparing female control and 2xUcp1-Cre^Evdr psWAT gene expression (left). Each dot represents one gene. Corresponding GO analysis (right). Genes and pathways significantly enriched in controls are labeled in orange and those enriched in 2xUcp1-Cre^Evdr are labeled in brown.

(E) qPCR analysis of iBAT of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr females at 6 weeks of age. n= 6.

(F) qPCR analysis of psWAT of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr females at 6 weeks of age. n= 6.

(G) qPCR analysis of pgWAT of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr females at 6 weeks of age. n= 6.

(H) Rectal temperature of control and 1xUcp1-Cre^Evdr females undergoing acute cold challenge. n= 3 controls, 4 1xUcp1-Cre^Evdr. Statistical significance was calculated using unpaired t-test within timepoint.

(I) Tail temperature of control and 1xUcp1-Cre^Evdr females undergoing acute cold challenge. n= 3 controls, 4 1xUcp1-Cre^Evdr. Statistical significance was calculated using unpaired t-test within timepoint.

(J) BAT temperature of control and 1xUcp1-Cre^Evdr females undergoing acute cold challenge. n= 3 controls, 4 1xUcp1-Cre^Evdr. Statistical significance was calculated using unpaired t-test within timepoint.

(K) qPCR analysis of iBAT of control and 1xUcp1-Cre^Evdr females after cold challenge or maintained at room temperature. n= 3. Statistical significance was calculated using unpaired t-test between room temperature (RT) and cold samples.

(L) qPCR analysis of psWAT of control and 1xUcp1-Cre^Evdr females after cold challenge or maintained at room temperature. n= 3. Statistical significance was calculated using unpaired t-test between room temperature (RT) and cold samples.

(M) qPCR analysis of pgWAT of control and 1xUcp1-Cre^Evdr females after cold challenge or maintained at room temperature. n= 3. Statistical significance was calculated using unpaired t-test between room temperature (RT) and cold samples.

Unless otherwise noted, data are mean + SEM and statistical significance was calculated using one-way ANOVA followed by Tukey’s multiple comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001. For RNA-seq, differential genes were selected by false discovery rate (FDR) < 0.05 with no fold-change cut-off.

Pathway analysis of the significantly altered genes reveals downregulation of mitochondrial activity pathways (e.g., electron transport chain, respiratory chain, energy generation) in 1xUcp1-Cre^Evdr psWAT [Figure 3B]. Conversely, mRNA biology pathways are upregulated in 1xUcp1-Cre^Evdr psWAT [Figure 3B]. 1xUcp1-Cre^Evdr iBAT display irregularities in diverse pathways not directly related to energy metabolism [Figure 3A]. Similar patterns are observed in 2xUcp1-Cre^Evdr mice, with psWAT exhibiting suppressed energy generation pathways and elevated mRNA biology [Figure 3D]. In contrast to 1xUcp1-Cre^Evdr iBAT, 2xUcp1-Cre^Evdr iBAT uniquely show upregulation of lipid metabolism including very long chain fatty acid metabolism [Figure 3C]. iBAT of 2xUcp1-Cre^Evdr appears to be enriched in genes related to multiple terms of fatty acid metabolism while psWAT of 2xUcp1-Cre^Evdr is depleted of them [Figure S3A–B]. Collectively, these results poised a scenario in which a single copy of the Ucp1-Cre^Evdr transgene may affect energy metabolism and thermogenesis pathways in iBAT and psWAT and these effects are heightened in 2xUcp1-Cre^Evdr mice.

We next used qPCR to verify whether thermogenic gene expression is affected by the Ucp1-Cre^Evdr transgene. In iBAT, key markers of the classic thermogenesis pathway are largely unchanged across control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr females, apart from increased Elovl3 in 2xUcp1-Cre^Evdr [Figure 3E]. In contrast, psWAT of 1xUcp1-Cre^Evdr females display an approximately 50% reduction in thermogenic genes (i.e., Ucp1, Prdm16, Ppgc1a, Cidea, Cox7a), but not Elovl3 [Figure 3F]. These suppressions are amplified in 2xUcp1-Cre^Evdr psWAT with reductions of 98% in Ucp1, 69% in Prdm16 and 94% in Cidea [Figure 3F]. Similar thermogenic depletion is evident in 2xUcp1-Cre^Evdr pgWAT of females, including 96% lower Ucp1 expression on average [Figure 3G]. Male iBAT and psWAT, but not pgWAT, show similar results to females suggesting a gender specific effect on pgWAT [Figure S3C–E]. Collectively, these results indicate that the presence of the Ucp1-Cre^Evdr transgene impairs expression of thermogenic genes in psWAT.

Given the profound effects of Ucp1-Cre^Evdr on thermogenic gene expression, we hypothesized that the transgene alone, without additional genetic manipulation, would impact cold response. To test this, room temperature-acclimated control and 1xUcp1-Cre^Evdr female mice were exposed to 4°C for 6 hours. Intriguingly, 1xUcp1-Cre^Evdr mice exhibit elevated core body temperature, measured by a rectal thermometer, before cold exposure which normalized to control levels during cold [Figure 3H]. Under room temperature conditions, 1xUcp1-Cre^Evdr mice display lower tail but higher iBAT surface temperatures compared to controls [Figure 3I–J]. This suggests a scenario in which 1xUcp1-Cre^Evdr shows tail vasoconstriction and elevated iBAT thermogenesis as a physiological mechanism to increase body temperature. Upon cold exposure, 1xUcp1-Cre^Evdr tail temperature normalized while iBAT remain hyperactive [Figure 3I–J]. However, this is insufficient to maintain core temperature. Together, these data reveal dysfunctional thermogenic regulation and body temperature control in mice harboring just one copy of the Ucp1-Cre^Evdr transgene.

We next investigated the effects of acute cold exposure on control and 1xUcp1-Cre^Evdr mice. After 6 hours of cold exposure, 1xUcp1-Cre^Evdr mice exhibit slightly greater body weight loss compared to controls [Figure S3F]. iBAT weight is unchanged between genotypes; however, psWAT and pgWAT are smaller in 1xUcp1-Cre^Evdr mice [Figure S3F], indicating increased lipid utilization to maintain body temperature. Cold-induced thermogenic gene expression is largely similar between control and 1xUcp1-Cre^Evdr iBAT, with comparable upregulation of Ucp1 (–2.8-fold), and Elovl3 (–4.2-fold) [Figure 3K, S3H]. However, psWAT displayed blunted activation, with Ucp1 increasing –52-fold in controls but only –16-fold in 1xUcp1-Cre^Evdr by cold together with reduced upregulation of other markers such as Elovl3 [Figure 3L, S3I]. pgWAT shows no differences after cold treatment between controls and 1xUcp1-Cre^Evdr mice [Figure 3M, S3J]. Histological analysis aligns with the gene expression data, revealing fewer multilocular adipocytes in 1xUcp1-Cre^Evdr psWAT after cold exposure, while iBAT and pgWAT appeared unaffected [Figure S3G]. Together, these data suggest an scenario of impaired psWAT thermogenic activation in response to acute cold stress in mice harboring the Ucp1-Cre^Evdr transgene.

UCP1-Cre^Evdr transgene has the potential to express high levels of UCP1.

The discovery of an additional transgenic Ucp1 gene within the Ucp1-Cre^Evdr transgene raised the question of its potential functionality. We hypothesized that this transgene derived Ucp1 could contribute to overall UCP1 levels. However, the transgenic Ucp1 sequence is identical to endogenous C57Bl6/J Ucp1, precluding its discrimination from the native genes. To overcome this limitation, we adopted a tissue-specific approach utilizing Ucp1-floxed mice (Ucp1tm1a(EUCOMM)Hmgu, EUCOMM)⁴⁰ harboring LoxP sites flanking exon 2 [Figure S4A–B]. By crossing Ucp1-floxed mice with Ucp1-Cre^Evdr animals, we would selectively ablate endogenous Ucp1 while preserving the transgenic variant. This would allow us to assess the functional impact of the transgenic Ucp1 gene within the Ucp1-Cre^Evdr concatemer.

However, standard endpoint PCR genotyping failed to identify any mice homozygous for Ucp1-floxed (Ucp1-fl/fl) and positive for Ucp1-Cre^Evdr (aka. Ucp1-fl/fl^Ucp1-CreEvdr mice)[Fig 4A]. Concurrently, an excess proportion of heterozygous Ucp1-floxed (Ucp1-fl/+) carriers of Ucp1-Cre^Evdr (aka. Ucp1-fl/+^Ucp1-CreEvdr mice) was observed [Fig 4A]. As explained above, embryonic lethality due to Ucp1 deficiency is not expected.

Figure 4:

(A) Expected and observed offspring genotypes obtained from end-point PCR genotyping and FRT copy number assay. n=131 pups. Statistical significance was calculated using Chi-square test.

(B) Copy number assay of FRT. n= 4.

(C) BAT weights of control, Ucp1-fl/+^Ucp1-CreEvdr and Ucp1-fl/fl^Ucp1-CreEvdr males. n= 14 control, 13 Ucp1-fl/+Ucp1-CreEvdr, 9 Ucp1-fl/flUcp1-CreEvdr.

(D) WAT weights of control, Ucp1-fl/+^Ucp1-CreEvdr and Ucp1-fl/fl^Ucp1-CreEvdr males. n= 14 control, 13 Ucp1-fl/+Ucp1-CreEvdr, 9 Ucp1-fl/flUcp1-CreEvdr.

(E) Representative H&E images of fat depots from control, Ucp1-fl/+^Ucp1-CreEvdr and Ucp1-fl/fl^Ucp1-CreEvdr males. n= 4. Scale bar, 50μm.

(F) qPCR analysis of Ucp1 in adipose tissue depots of control, Ucp1-fl/+^Ucp1-CreEvdr and Ucp1-fl/fl^Ucp1-CreEvdr males. Values are relative to those of iBAT control. n= 8.

(G) Western blot of iBAT protein lysates of control, Ucp1-fl/+^Ucp1-CreEvdr and Ucp1-fl/fl^Ucp1-CreEvdr males.

(H) Body of control and Ucp1-fl/fl^{Ucp1-CreERT2Biat} males. n= 7 control, 6 Ucp1-fl/fl^{Ucp1-CreERT2Biat}.

(I) BAT weights of control and Ucp1-fl/fl^{Ucp1-CreERT2Biat} males. n= 7 control, 6 Ucp1-fl/fl^{Ucp1-CreERT2Biat.}

(J) WAT weights of control and Ucp1-fl/fl^{Ucp1-CreERT2Biat} males. n= 7 control, 6 Ucp1-fl/fl^{Ucp1-CreERT2Biat}.

(K) Western blot of iBAT protein lysates of control and Ucp1-fl/fl^{Ucp1-CreERT2Biat} males.

(L) BAT temperature of control and Ucp1-fl/fl^Ucp1-CreEvdr males undergoing acute cold challenge. n= 6 controls, 3 Ucp1-fl/fl^Ucp1-CreEvdr. Statistical significance was calculated using unpaired t-test within timepoint.

(M) Tail temperature of control and Ucp1-fl/fl^Ucp1-CreEvdr males undergoing acute cold challenge. n= 6 controls, 3 Ucp1-fl/fl^Ucp1-CreEvdr. Statistical significance was calculated using unpaired t-test within timepoint.

(N) Rectal temperature of control and Ucp1-fl/fl^Ucp1-CreEvdr males undergoing acute cold challenge. n= 6 controls, 3 Ucp1-fl/fl^Ucp1-CreEvdr. Statistical significance was calculated using unpaired t-test within timepoint.

(O) Rectal temperature of control and Ucp1-fl/fl^{Ucp1-CreERT2Biat} males undergoing acute cold challenge. n= 3 controls, 3 Ucp1-fl/fl^Ucp1-CreEvdr. Statistical significance was calculated using unpaired t-test within timepoint.

Unless otherwise noted, data are mean + SEM. Statistical significance was calculated using unpaired t-test or one-way ANOVA followed by Tukey’s multiple comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001.

Thus, we ponder the hypothesis that the transgenic Ucp1 gene within the Ucp1-Cre^Evdr concatemer [Fig 2E] would mask the floxed status of the endogenous Ucp1 alleles by yielding a wildtype band in endpoint PCR genotyping. To overcome this, we obtained the sequences surrounding the FRT site present in floxed allele (tm1c) of the Ucp1tm1a(EUCOMM)Hmgu mice [Figure S4A–B]. Next, we designed a copy number assay specific to detect this FRT sequence [Figure S4A–B]. We used wildtype, Ucp1-fl/+ and Ucp1-fl/fl mice to calibrate our assay to zero, one and two copies of FRT [Figure 4B]. Using this approach, we readily distinguished Ucp1-fl/+^Ucp1-CreEvdr mice harboring one FRT copy from Ucp1-fl/fl^Ucp1-CreEvdr mice with two FRT copies [Figure 4B]. Genotyping with the FRT copy number assay revealed expected Mendelian ratios of control, Ucp1-fl/+^Ucp1-CreEvdr, and Ucp1-fl/fl^Ucp1-CreEvdr progeny [Figure 4A], proving that this strategy overcomes the confounding effects from the transgenic Ucp1 sequence.

At 6 weeks of age, total body weight is equivalent between control, Ucp1-fl/+^Ucp1-CreEvdr, and Ucp1-fl/fl^Ucp1-CreEvdr mice [Figure S4C]. BAT depots weights are similar between controls and Ucp1-fl/+^Ucp1-CreEvdr [Figure 4C]. However, Ucp1-fl/fl^Ucp1-CreEvdr mice display marked –2-fold increase in weight in all BAT [Figure 4C]. WAT, liver or muscle tissue weights are unchanged across genotypes [Figure 4D, S4D–F]. In summary, targeted BAT-specific Ucp1 ablation elicit pronounced BAT growth without impacting body and WAT weight.

Histological examination reveals the iBAT hypertrophy in Ucp1-fl/fl^Ucp1-CreEvdr mice is attributable to uniformly enlarged brown adipocytes engorged with excessive lipid [Figure 4E]. In contrast, WAT depots morphology is largely unaffected by genotype [Figure 4E]. This shows a lack of morphological change compensation in WAT by the targeted deletion of Ucp1 in BAT.

qPCR analysis reveals a –50% and –70% reduction in iBAT Ucp1 mRNA in Ucp1-fl/+^Ucp1-CreEvdr and Ucp1-fl/fl^Ucp1-CreEvdr mice, respectively, compared to controls [Figure 4F]. Cre mRNA is equal in iBAT of Ucp1-fl/+^Ucp1-CreEvdr and Ucp1-fl/fl^Ucp1-CreEvdr mice [Figure S4G]. iBAT also shows a compensatory increased expression of classic thermogenic key markers Ppargc1a, Cox7a, and Elovl3 [Figure S4J]. In psWAT, Ucp1 expression is unaltered; however Ucp1-fl/fl^Ucp1-CreEvdr mice display elevated expression of Cidea, Cox7a, and Elovl3 [Figure 4F, S4K]. pgWAT gene expression is largely unchanged [Figure 4F, S4L]. Cre mRNA expression is slightly upregulated in psWAT and pgWAT of Ucp1-fl/fl^Ucp1-CreEvdr mice partially recapitulating a possible compensation in psWAT, but with levels still much lower than those found in iBAT [Figure S4H–I]. Thus, BAT targeted Ucp1 ablation induces depots-specific effects, including a rather unique selective compensatory thermogenic activation in iBAT and psWAT.

The high residual Ucp1 mRNA expression in iBAT of Ucp1-fl/fl^Ucp1-CreEvdr mice was unexpected. This is because, first, Ucp1 is restricted to mature brown adipocytes and second, because Ucp1-Cre^Evdr drives recombination in essentially all brown adipocytes, as we and others have previously shown^28,41. To further investigate this, we examined UCP1 protein levels by western blot. Remarkably, iBAT lysates of Ucp1-fl/fl^Ucp1-CreEvdr mice retained variable but high UCP1 protein levels, averaging approximately 70% of control [Figure 4G]. Given the broad UCP1-Cre^Evdr-mediated excision in brown adipocytes, the substantial UCP1 retained seems unlikely to be derived from endogenous Ucp1 genes. This paradoxical preservation of UCP1 protein suggests functionally significant expression from the transgenic Ucp1 within the UCP1-Cre^Evdr concatemer.

As a control, we crossed Ucp1-floxed mice with the tamoxifen-inducible Ucp1-CreERT2^Biat allele. Tamoxifen treatment of Ucp1-fl/fl^{Ucp1-CreERT2Biat} mice does not change body weight but increases BAT weights with a slight enlargement of pgWAT and liver, while other fats and tissues are unchanged [Figure 4H–J, S4M–O]. Critically, tamoxifen treatment leads to highly efficient ablation of UCP1 protein in iBAT [Figure 4K]. This confirms that the Ucp1-floxed allele can be efficiently deleted and reinforce the hypothesis that the lack of UCP1 deletion in Ucp1-fl/fl^Ucp1-CreEvdr may arise from ectopic transgene expression.

To test for functionality, we performed acute cold challenges by exposing mice at 6°C for 6 hours. Intriguingly, Ucp1-fl/fl^Ucp1-CreEvdr mice exhibit rather elevated core and BAT temperatures before starting cold exposure, compared to littermate controls [Figure 4L–N]. Unexpectedly, Ucp1-fl/fl^Ucp1-CreEvdr mice are proficient in maintaining core body temperature during cold exposure at the same level than littermate controls [Figure 4N]. In stark contrast, Ucp1-fl/fl^{Ucp1-CreERT2Biat} mice rapidly become hypothermic at 6°C, reflecting the efficient ablation of UCP1 in BAT [Figure 4O]. Together, these results indicate that the UCP1 protein present in Ucp1-fl/fl^Ucp1-CreEvdr BAT, which may be bestowed by the Ucp1-Cre^Evdr allele, may confer sufficient thermogenic capacity to preserve body temperature.

Discussion

The Cre-Lox system is invaluable for spatial and temporal dissection gene function, tracing lineages, and labeling cells. However, the validation of Cre-recombinase transgenes in the literature is usually incomplete. In particular, the UCP1-Cre^Evdr transgene transformed BAT and metabolism research. However, our findings reveal several key unexpected caveats of this widely used line including (1) increased mortality, growth defects, and craniofacial alterations in homozygosity; (2) substantial genomic disruptions; (3) profound impacts on BAT and psWAT function in both hemizygosity and homozygosity; and (4) potential misexpression of Ucp1 itself under high thermogenic burden. These findings highlight the importance of carefully considering the validation of transgenic lines before embarking on experimental studies.

BAC Cre-drivers are usually validated only by examining spatiotemporal recombination, fueling comprehensive databases that guide selection among a large and growing number of Cre-recombinases available^4,5. Causes for unexpected transgene expression include insertion effects on local regulation or integration of sequences leading to ectopic expression. Alternatively, unanticipated activity may reflect previously unknown endogenous gene expression. Adipocyte-targeting Cre lines exemplify these issues. The promoter of the fatty acid binding protein 4 gene, also known as adipocyte protein 2 (aP2), was used to generate two independent aP2-Cre mouse models with the intention to specifically target mature adipocytes^42,43. However, these aP2-Cre models were found to inefficiently target mature adipocytes while exhibiting broad recombination in the brain, endothelial cell in adipose tissues, macrophages, adipocyte progenitors, and elsewhere^{5,16,20,44–47}. This prompted development of Cre lines driven by the promoter of adiponectin^19,20. However, Ucp1-Cre^Evdr also exhibits widespread brain activity^41,48, including regions controlling feeding and non-shivering thermogenesis^49–51. Importantly, very low endogenous Ucp1 expression partially overlaps with these brain areas^41,48, suggesting Ucp1-Cre^Evdr partially recaptures native Ucp1 regulation. However, if this reflects endogenous Ucp1 expression or expression of the Ucp1 gene found within the UCP1-Cre^Evdr transgene, that we find here, is not known at this point.

The recent close examination of some BAC transgenic lines, beyond cellular expression, has shown that the transgenes themselves can result in phenotypes leading to potential misinterpretations of the intended genetic modifications^52–54. Random BAC transgene insertion is frequently associated with substantial genomic alterations, often disrupting gene coding sequences and creating small or large rearrangements^6,7,8. As expected, Ucp1-Cre^Evdr mice exhibit major structural variations at the insertion site including relatively large deletions and inversions. Historically, the position in the genome and the genomic alterations induced by the random insertion of a transgene have not been systematically examined. This is partially due to previous low-resolution techniques like FISH or linkage mapping which poorly defined insertion sites and structures. However, new sequencing approaches such as whole genome sequencing and TLA enable fine mapping of insertion locus, disruption effects, and integrated sequences³⁷. For instance, whole genome sequencing and TLA both revealed Adipoq-Cre^Evdr transgene inserted into the Tbx18 gene on chromosome 9^19,55,56, perturbing Tbx18 expression and adding passenger gene copies with possible widespread effects⁵⁶. Additionally, BAC transgenes normally integrate as concatemers leading to multiple full or partial copies of the transgene^3,57,58. Using ddPCR, we find that Cre coding sequence is present in a single copy in Ucp1-Cre^Evdr mice. This was also the case for the Adipoq-Cre^Evdr transgene⁵⁶. However, ddPCR analysis is limited to a small specific sequence via specific primers; thus, if partial coding or non-coding sequences are present, they may not be detected by ddPCR³⁹. TLA is also not capable of defining the order or number of copies within the inserted concatemer in a transgenic line. Defining insertion sites and the full structure including inserted sequences may only be possible by new long-range genome sequencing^7,59,60. Additionally, transgenic strains are bred for generations, allowing accrual of modifications in transgenic sequences and the genetically linked endogenous sequences over time. Furthermore, Cre-drivers with initially robust expression can become leaky or lose activity. Monitoring integrated sequences over mouse line generations could enhance integrity of lines maintained and avoid misinterpretations.

The genetic alterations induced by random transgene insertion and the passenger sequences inserted can lead to confounding genetic effects that are dependent on the transgene rather than the intended genetic alteration. In Ucp1-Cre^Evdr mice, we observe drastic transcriptional dysregulation in iBAT and psWAT, suggesting major changes in tissue function. Moreover, interactions between transgene effects and specific genetic manipulations (e.g., deletions) are unpredictable. While linking fat transcriptomics to whole body physiology is challenging, these data indicate that Ucp1-Cre^Evdr the transgene has the potential to profoundly perturb adipose function. In this sense, 1xUcp1-Cre^Evdr mice show altered body temperature dynamics and distinct cold reactions when compared to controls. Compared to Adipoq-Cre^Evdr mice, which show minimal adipose tissues gene expression changes⁵⁶, the Ucp1-Cre^Evdr effects are considerable. Beyond targeted tissues, transgenes can reprogram untargeted tissues due to genomic disruption and passenger genes. For Ucp1-Cre^Evdr, the extra Ucp1 gene seems highly expressed under high thermogenic demand, exemplifying how passenger genes can have unexpected impact. Overall, transgene insertion effects are diverse and context dependent. Thus, thorough characterization of each line is essential to parse transgene-specific artifacts from intended genetic effects.

Homozygosity often reveals phenotypes undetectable in heterozygotes, as most loss-of-function mutations display recessive inheritance. Thus, generating homozygous BAC transgenic models can uncover cryptic transgene-dependent effects. For example, crosses of hemizygous Adipoq-Cre^Evdr mice have not produced homozygous mice suggesting lethality, although this experiment may have been underpowered⁵⁵. The unexpected high mortality and other major effects caused by homozygosity of Ucp1-Cre^Evdr imply impacts beyond adipose tissue. The multiple genes that are directly affected by the insertion site and that are expressed in a range of tissues may be responsible for these phenotypes. Moving forward, evaluating homozygous transgenic models, despite logistic challenges, may be a strong paradigm to ensure detection of subtle artifacts.

A limitation across published studies finding unexpected transgenic effects is that they usually lack mechanistic resolution. For example, the underlying causes of 2xUcp1-Cre^Evdr mortality and 1xUcp1-Cre^Evdr fat transcriptional changes remain unresolved. However, this knowledge gap in the literature is reasonable given the manifold possibilities. Effects could arise from chromosomal rearrangements, long-distance interactions, 3D conformation changes, or passenger sequences within the BAC, among other potential mechanisms. Moreover, unraveling any single mechanism may provide limited core insights into normal and pathogenic biology, as insertion effects likely stem from complex interactions only arising in the context of the transgenic line analyzed. While mechanistic details are invaluable, delineating the precise causes underlying insertion artifacts would require substantial efforts unlikely to significantly produce advancements. As such, the mechanistic ambiguity in these studies is expected.

New techniques for generating transgenic mice can mitigate issues with random BAC insertion, bolster rigor and drive discovery. CRISPR-directed insertion at known safe harbor loci provides control over transgene placement. Additionally, rational design of regulatory sequences enables precise spatiotemporal expression, avoiding complications from passenger DNA in large BAC constructs. As these and other innovations become widespread, they will complement and enhance previous data obtained with BAC transgenics.

The effects we report here may justify the reevaluation of some prior work using Ucp1-Cre^Evdr mice to clarify confounding outcomes. However, such assessments will be complicated by the unpredictable interactions between the Ucp1-Cre^Evdr effects and intended genetic changes. The strength and specificity of Ucp1-Cre^Evdr transgene in brown adipocytes ensures that this line remains useful in verified contexts. In future studies, control groups with only the Ucp1-Cre^Evdr transgene could be incorporated to parse its specific effects. Orthogonal tools like additional Cre lines (e.g., inducible Ucp1-CreERT2^Biat allele) can also confirm results. Finally, new Cre-recombinase drivers may be generated, using methodologies described above, to confirm previous results. Though challenging, careful experimental design and layered validation can distinguish between effects dependent on the transgene, gene manipulation, or the interaction between the two.

In conclusion, validation of research tools is a requirement of several funding agencies, yet standards for doing so remain opaque. While BAC transgenics have revolutionized basic and biomedical research, limitations have become increasingly apparent. Overall, transparent validation, cautious interpretation, and technological innovations will maximize scientific rigor of BAC transgenics as future tools to catalyze discovery.

STAR METHODS

Lead contact and material availability

Further information and request for resources and reagents should be directed to and will be fulfilled by the Lead contact, Joan Sanchez-Gurmaches (juan.sanchezgurmaches@cchmc.org).

Experimental model and subject details

Mice and mice housing

All mice used in this study were in C57Bl6/J background. UCP1-Cre (JAX stock 024670), R26R-mTmG mice (JAX stock 007676), Ucp1-CreER mice were described before²⁵. Ucp1 flox mice were obtained from the EUCOMM program (C57BL/6N-Ucp1^{tm1a(EUCOMM)Hmgu}/Ieg) after removal of the LacZ and neomycin cassette by Flippase.

Unless noted otherwise, mice were housed in the CCHMC Animal Medicine Facility in a clean room set at 22°C and 45% humidity on a daily 12h light/dark cycle, and kept in ventilated racks fed ad libitum with a standard chow diet, with bedding changed every two weeks. See figure legend for specific age and number of mice used. All animal experiments were approved by the CCHMC IACUC.

For long term temperature acclimation experiments, mice were housed in rodent incubators (7001–33 series, Caron) in pairs within the facilities of animal medicine of CCHMC. Room temperature group mice were co-housed in the same facility as the mice in rodent incubators. Mouse cages were changed weekly using components pre-adjusted to temperature. No cage enrichment was used in this set of experiments.

Method Details

MGI alleles and publications search

MGI database was searched with search word “transgenic” on 12/16/2022. Out of the 10,166 results, mice with Transgenic allele type were selected. Out of these, mouse models with location indicated as “unknown” were assigned as the location of the transgene unidentified group. The remainder mouse models were automatically assigned to transgene with known location group. To analyze Cre-driver mouse models the symbol of each mouse model was search for containing “Cre”. Out of the 1,968 mouse models found, location was assigned as above. Number of publications assigned to specific transgenic mice were found in MGI on July 2023.

GWAS associated traits and mouse mortality phenotype search.

Mouse mortality phenotype associated to gene knockouts was searched in the MGI database. GWAS associated traits to specific genes were search in the Phenotype-Genotype Integrator of the NCBI.

Growth and Survival

Body weight and survival of mice was followed starting at week 3 of age as tail snips were taken for genotyping. Body weights and survival was recorded weekly after that until 6 weeks of age.

Tamoxifen treatment

Tamoxifen was dissolved in corn oil/ethanol (9:1 vol/vol) at 2mg/mL by shaking at 4°C overnight. 6 week old mice were injected with 2 mg/day/mouse for 5 times in a period of seven days. A subgroup of mice was additionally injected for four days during the third week after first injection. Mice were sacrificed three weeks after first injection.

Tissue dissection

Tissues were carefully dissected to avoid surrounding tissue contamination. Adipose tissue notation used here was described previously⁶¹. Mice were dissected at early morning without fasting or any other alteration, unless noted in the figure legend.

Acute cold exposure

Mice were placed at 4°C early in the morning of the experiment in overnight pre-chilled caging with free access to pre-chilled water with or without food.

Temperature measurements

Internal temperature was recorded by using a rectal thermometer probe (RET-3, Braintree Scientific Inc.). BAT and tail temperatures were obtained using an infrared thermal camera (FLIR T530 24°) in lightly anesthetized mice and analyzed with FLIR tools.

Tissue histology

Tissue pieces were fixed in 10% formalin. Embedding, sectioning and Hematoxylin and Eosin (H&E) staining was done by the CCHMC Pathology Core facility.

Craniofacial morphometric analysis

Samples were incubated in 0.005% Alizarin Red S (Sigma-Aldrich, A5533) in 1% KOH for 24 hours at room temperature and cleared in 1% KOH for 72 hours. Once cleared, samples were incubated in Glycerol:KOH 1% (50:50) solution. For imaging and long-term storage, samples were kept in 100% glycerol. Stained skulls were imaged using a Leica M165 FC stereo microscope system for measurements. The condylobasal length and the interorbital constriction length were measured in ImageJ. The ratio between them was used as skull shape defining factor.

qPCR analysis

Tissues were homogenized in a FASTPREP-24 (MP Biomedicals) using Qiazol (Qiagen). Total RNA was isolated using RNeasy kit (Qiagen), retrotranscription was done using High Capacity cDNA reverse transcription kit (#4368813, Applied Biosystems) and analyzed in a QuanStudio 3 real-time PCR machine (Thermofisher). Primer sequences are shown in Table S1.

Western blot analysis

Tissues were homogenized in a FASTPREP-24 (MP Biomedicals) and lysed in RIPA buffer (150 mM NaCl, 50 mM Hepes at pH 7.4, 0.1% SDS, 1% Triton X-100, 2 mM EDTA, 0.5% Na-deoxycholate) containing a protease and phosphatase inhibitor cocktail. Protein lysates (typically 10mg per lane) were mixed with 5X SDS sample buffer and run in SDS acrylamide/bis-acrylamide gels (typically 10 or 12%), transferred to PVDF membranes and detected with specific antibodies as specified in Table S2.

Copy number assays

Genomic DNA was isolated from tails or liver using the DNeasy Blood and Tissue kit spin columns (Qiagen) and diluted to 1 ng/μL. Copy number assays were done using Taqman ^® copy number assays (ThermoFisher) using predesigned oligonucleotides assays using Tfrc as reference gene (ThermoFisher)(Table S4). qPCR was performed on a QuantStudio 6Flex real-time PCR system using the following protocol: 95°C for 10min followed by 40 cycles of 95°C for 15 s and 60°C for 60 s with manual Ct threshold at 0.2 and Autobaseline on. Results were analyzed on Copy Caller 2.0 software (ThermoFisher).

Digital droplet PCR

Reaction mixture was composed of 10 uL 2x ddPCR Supermix (without dUTPs; Bio-Rad, Hercules CA), 1 uL each of the proves against housekeeping and target gene (Bio-Rad)(Table S4), 1uL of HaeIII (NEB, R0108), 50 ng of DNA template, and adjusted to a final volume of 20 uL. Droplets were generated in a 96-well polypropylene plate (Bio-Rad) using the QX200 droplet generator (Bio-Rad). The plate containing the water-in-oil emulsions was sealed with foil using a PX1 PCR Plate Sealer (Bio-Rad) and placed in a C1000 Touch Thermal Cycler (Bio-Rad). The following conditions were used for amplification: 95°C for 10 minutes, 94°C for 30 seconds and 60°C for 1 minute (40 cycles 2°C/sec ramp rate), a 10-minute hold at 98°C, and a final hold at 4°C. The plate was processed using the QX200 droplet reader (Bio-Rad). Results were analyzed using QuantaSoft Analysis Pro software version 1.0.596.

Targeted Locus Amplification

UCP1-Cre transgene location and genetic rearrangements associated were approximated by Targeted Locus Amplification (TLA)(Cergentis B.V.) in splenocytes. Splenocytes were isolated from 8 weeks old UCP1-Cre hemizygous mice by pushing the spleen through a 40 mm mesh and collecting in 10% fetal bovine serum in PBS. After red blood cell lysis and washes with 10% fetal bovine serum in PBS, around 10 million spleen cells were aliquoted in cryovials in freezing media (10% fetal bovine serum, 10% DMSO, in PBS). TLA analysis was performed by Cergentis B.V. as previously reported³⁷ with six independent pairs of primers (Table S3) using mouse mm10 genome as host reference.

Whole genome gene expression profile.

RNA-seq reads were aligned to UCSC mouse genome 10 mm using STAR aligner⁶². Only uniquely aligned reads were used for downstream analysis. Raw read counts for each gene were measured using FeatureCounts in the subread package⁶³ with an option, “-s 2 -O --fracOverlap 0.8”. Differential gene expression analysis was performed using EdgeR⁶⁴. Genes with Fold-change > 1.5 and FDR < 0.05 were selected as differentially expressed genes. Gene ontology analysis was performed using Enrichr⁶⁵.

De novo assembly.

To build a Cre transgene sequence, we performed incremental alignment and de novo assembly. Initially, we built a STAR reference combining mm10 and the known Cre CDS sequence. Read 1 and 2 from each RNA-seq sample were aligned to the combined reference separately, where we selected read pairs for de novo assembly if at least one out of the pair is aligned to the Cre reference. And then, we pooled the selected read pairs and performed de novo assembly using Trinity⁶⁶. Given the design of Cre transgene, we anticipated that the assembled sequence should fully cover Cre CDS and span from Cre CDS sequence to the 5’ half and 3’half of the Ucp1 exon #1. We observed that the very first assembly results fully cover the Cre CDS and connect between Ucp1 5’ half and Cre CDS with a small gap of unknown sequence but not the 3’half. Therefore, we updated the Cre reference with the assembled sequence and repeated these steps of alignment, selection, and de novo assembly until the assembly result reaches the 3’ half of the Ucp1 exon #1, which happened after 4th round. The final assembly results were assessed and annotated using known references and Blast⁶⁷.

Figure design.

Figures were made in Adobe Illustrator. Several figures were created with BioRender.com.

QUANTIFICATION AND STATISTICAL ANALYSIS

Data are presented as mean+s.e.m., unless stated otherwise. Unpaired t-test, analysis of variance (one or two ways) followed by Tukey’s multiple comparisons, Chi-square and Log-rank (Mantel-Cox) as appropriate, were used to determine statistical significance. No pre-test was used to choose sample size. Statistical analysis was done using GraphPad Prism except for global RNA expression (see methods). The number of mice used per experiment is stated in each figure legend. In all panels, *P < 0.05, **P < 0.01, ***P < 0.001.

Highlights:

• Hemizygous Ucp1-Cre^Evdr mice exhibit substantial brown and white fat tissue dysregulation.

• Homozygous Ucp1-Cre^Evdr mice display high mortality, growth defects, and craniofacial abnormalities.

• The Ucp1-Cre^Evdr transgene integration resulted in major genomic disruptions affecting multiple genes.

• The Ucp1-Cre^Evdr transgene retains a possibly functional extra Ucp1 copy.