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

Abstract

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 perform comprehensive analysis of the Ucp1-Cre^Evdr line which is widely used for brown fat research. Hemizygotes exhibit major brown and white fat transcriptomic dysregulation, indicating potential altered tissue function. Ucp1-Cre^Evdr homozygotes also show high mortality, tissue specific growth defects, and craniofacial abnormalities. Mapping the transgene insertion site reveals 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 intended genetic manipulations. Overall, comprehensive validation of transgenic mice is imperative to maximize discovery while mitigating unexpected, off-target effects.

UCP1 is involved in regulating thermogenesis, and the Ucp1-Cre mouse line has been widely used for studying brown fat. Here they show that this mouse line carries an extra copy of the Ucp1 gene and displays unexpected changes in fat tissue function. Researchers should be cautious when interpreting studies using this model.

Introduction

Mouse transgenic models such as overexpressors, reporters, and Cre-recombinases empower spatial and temporal genetic manipulation and have become indispensable for elucidating the biological foundations of adipose tissue in 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 the 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 (Supplementary Fig. S1A, B); (2) insertion can result in large genomic abnormalities that are not routinely inspected^6–8 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 phenotypes^13,14; (4) Cre transgenes are largely used in hemizygosity masking phenotypes that would otherwise be evident^15,16; (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 in 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)^17–19. Cre-recombinase lines utilizing the adiponectin promoter enabled the targeting of all adipocytes^20–22. Promoter elements from UnCoupling Protein 1 (Ucp1) have conferred selective Cre-recombinase expression in brown adipocytes. Although other BAT Cre-targeting tools have existed²³, and other tools to target brown adipocytes are used²⁴, two Ucp1-Cre drivers dominate the literature currently: the constitutive Ucp1-Cre^Evdr line from the Rosen lab²⁵ and the tamoxifen-inducible Ucp1-CreERT2^Biat line from the Wolfrum lab²⁶. Both lines show remarkable specificity, full penetrance, and robust activity on brown adipocytes^27–33. 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.

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 tissue-specific 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 be expressed under high thermogenic burden. In conclusion, our results suggest unintended consequences on mouse physiology induced by Ucp1-Cre^Evdr.

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 of 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 (Fig. 1A). We designated them as controls, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice, respectively. We reasoned that this strategy would enable a comprehensive assessment of potential developmental, physiological, and molecular perturbations arising from the potential genetic disruptions that this widely utilized Cre driver line may generate.

Fig. 1. Ucp1-Cre^Evdr homozygosity induces low survival, tissue selective growth retardation, and craniofacial dysmorphologies.

A Experimental strategy for the generation of control, Ucp1-Cre^Evdr hemizygous (1xUcp1-Cre^Evdr) and Ucp1-Cre^Evdr homozygous (2xUcp1-Cre^Evdr) mice. Created in BioRender. Sanchez Gurmaches, J. (2024) https://BioRender.com/i80u802. 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 the 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 the 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 (ratio 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 tests. *P < 0.05, **P < 0.01, ***P < 0.001. Source data are provided as a Source Data file.

To discriminate transgene copy numbers, we developed a quantitative copy number assay to detect Cre in genomic DNA rather than relying on endpoint PCR genotyping (Supplementary Fig. S1C). At three weeks of age, we find significantly fewer 2xUcp1-Cre^Evdr mice than the expected Mendelian ratio of 25% (Fig. 1B). Specifically, only 14.04% of females and 16.06% of males are homozygous across 251 pups from 46 litters (Fig. 1B). Analysis of both sexes together reveals that 2xUcp1-Cre^Evdr comprises just 15.14% of the offspring, reflecting approximately 60% survival (Supplementary Fig. S1D). The sex distribution is unaffected, indicating no differential penetrance between sexes (Supplementary Fig. 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 (Fig. 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 (Supplementary Fig. 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 (Supplementary Fig. S1F, G). In addition, 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 (Fig. 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 (Fig. 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^34,35. 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 (Fig. 1F), dissection of individual fat and lean tissues shows 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 depot weights, including interscapular (iBAT), subscapular (sBAT), and cervical (cBAT) compared to control littermates (Fig. 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 (Fig. 1H). Beyond WAT, only quadriceps mass differs in 2xUcp1-Cre^Evdr females compared to controls (Fig. 1I–K). Male homozygotes also exhibit a similar decrease in body weight and dramatic WAT depletion along with reductions in liver, quadriceps, and gastrocnemius mass (Supplementary Fig. S1H–M). Thus, tissue-specific growth effects underlie the global growth retardation in 2xUcp1-Cre^Evdr, suggesting that certain tissues response to transgene integration is more severe.

Reduction in WAT mass can happen by distinct mechanisms. Histological analysis of iBAT and psWAT reveals no major changes in adipocyte size between genotypes in either females or males (Fig. 1L and Supplementary Fig. S1P). However, adipocytes in pgWAT of 2xUcp1-Cre^Evdr females and males appear to be smaller in size compared to control littermates (Fig. 1L and Supplementary Fig. S1P). These histological results suggest that the changes in psWAT and pgWAT weights may be due to different mechanisms involving the generation of adipocytes and control of their size, respectively.

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 (Supplementary Fig. 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 by qPCR and unchanged compared to 1xUcp1-Cre^Evdr (Supplementary Fig. S1N). Analysis of male fat depots shows similar results (Supplementary Fig. 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 and disrupts genomic integrity

To date, the genomic integration site and insertion genetic consequences 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 cross-linking. Subsequent deep sequencing of PCR products enables mapping of the transgene insertion locus³⁶.

TLA using Cre-specific primers reveals the Ucp1-Cre^Evdr transgene integrated into chromosome 1, cytoband A5 (Fig. 2A and Supplementary Fig. S2A). As expected, Cre primers also detect homology near the endogenous Ucp1 locus in chromosome 8, indicating the inclusion of surrounding Ucp1 genomic sequences in the transgene (Fig. 2A). Primer pairs surrounding Ucp1 produce high signal levels at the Ucp1 locus (Supplementary Fig. S2A, B). TLA maps the concatemer insertion site of Ucp1-Cre^Evdr to chr1:20,962,125-21,016,858 (Fig. 2B). Integration induces a ~ 54 kb deletion flanking the insertion sites, along with a 3’ ~ 280 kb inversion (Fig. 2B). This directly deletes or inverts the entirety or large portions of 4 genes (Efhc1, Tram2, Tmem14a, Gsta3) (Fig. 2B). In addition, the concatemer localizes in close proximity to other 7 genes (Il17a, Il17f, Mcm3, Paqr8, Khdc1a, Khdc1c, Khdc1b). Several noncoding sequences within or in close proximity to the concatemer may also be affected (Fig. 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 (Fig. 2C). Knockout mouse models have not been generated for each coding gene affected by or surrounding the Ucp1-Cre^Evdr transgene (Fig. 2D). However, out those generated, Mcm3 knockout mice show prenatal lethality with complete penetrance³⁷ (Fig. 2D). Although this is quite distinct from what we observe in 2xUcp1-Cre^Evdr mice (Fig. 1B, C and Supplementary Fig. S1D, E), the genomic disruption induced by the Ucp1-Cre^Evdr concatemer may contribute to survival rate.

Fig. 2. The Ucp1-Cre^Evdr transgene is located in Chromosome 1 and disrupts genomic integrity.

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. Created in BioRender. Sanchez Gurmaches, J. (2024) https://BioRender.com/k25w280. 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 Representative Integrative Genomics Viewer (IGV) tracks for psWAT RNA-seq data of the insertion site of the Ucp1-Cre^Evdr transgene of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr females (n = 2 per genotype). F Gene fusion events detected in psWAT of control, 1xUcp1-Cre^Evdr, and 2xUcp1-Cre^Evdr females between Tram2 and Gsta3 (n = 2 per genotype). Unless otherwise noted, data are mean + SEM. Source data are provided as a Source Data file.

To expand on the genomic impact of the Ucp1-Cre^Evdr transgene insertion, we conducted RNA-sequencing analysis of iBAT and psWAT from six-week-old control, 1xUcp1-Cre^Evdr, and 2xUcp1-Cre^Evdr female littermates. Analysis of psWAT reveals a dose-dependent reduction in expression of the exons affected by the transgene (Tram2 exons 3-10 (out of 10) and Efhc1 exons 5–11 (out of 11), with complete absence of expression in 2xUcp1-Cre^Evdr mice (Fig. 2E). The effects on Efhc1 are less pronounced due to its low basal expression in psWAT. Results in iBAT mimic the same pattern (Supplementary Fig. S2C). These transcriptomic findings corroborate and extend the findings by TLA at genomic DNA level. In support of the genetic rearrangements identified by TLA, we identified a fusion transcript between Tram2 exon 2 and Gsta3 exon 2, with abundance increasing from 1xUcp1-Cre^Evdr to 2xUcp1-Cre^Evdr mice in psWAT and iBAT (Fig. 2F and Supplementary Fig. S2D). These transcriptomic data provide evidence of extensive chromosomal rearrangements at the transgene insertion site, underscoring the complex genomic consequences of the Ucp1-Cre^Evdr transgene.

Overall, our findings at genomic and transcriptomic levels show that the Ucp1-Cre^Evdr transgene produces significant genomic alterations, which may contribute to the phenotypic changes observed in these mice.

The UCP1-Cre^Evdr Transgene harbors an extra Ucp1 gene copy

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 ~ 230 Mb of chromosome 8 surrounding the Ucp1 gene, is incorporated with the transgene in chromosome 1 (Fig. 3A). 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 from 1xUcp1-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 (Fig. 3B). The Cre coding sequence is followed at 3’ by a 3’UTR and a short sequence of unknown function (Fig. 3B). 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 (Fig. 3B). The presence of Ucp1 coding mRNA sequence within the Ucp1-Cre^Evdr transgene transcript suggests that the extra copy of Ucp1 gene may be expressed.

Fig. 3. The Ucp1-Cre^Evdr transgene harbors an extra Ucp1 gene.

A Coverage of BAC 148M1 inserted within the Ucp1-Cre^Evdr transgene as determined by TLA. Created in BioRender. Sanchez Gurmaches, J. (2024) https://BioRender.com/d78y093. B 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. Created in BioRender. Sanchez Gurmaches, J. (2025) https://BioRender.com/z95c201. C Copy number assay of Cre of control, 1xUcp1-Cre^Evdr, and 2xUcp1-Cre^Evdr mice. n = 3 per genotype. D Copy number assay of Ucp1 of control, 1xUcp1-Cre^Evdr, and 2xUcp1-Cre^Evdr mice. n = 8 per genotype. E Absolute copy number by ddPCR of Cre of control, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice. n = 3 per genotype. Unless otherwise noted, data are mean + SEM. Source data are provided as a Source Data file.

An intrinsic limitation of the TLA protocol is that it cannot discern the number of repetitions occurring within the concatemer. To 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 qPCR-based copy number assays against the Ucp1 intron 3 to assess the number of copies of the Ucp1 gene in genomic DNA of controls, 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice (Fig. 3C). Control littermates were used as reference for two copies of Ucp1 gene (Fig. 3D). In contrast, 1xUcp1-Cre^Evdr mice have three copies and 2xUcp1-Cre^Evdr mice have four copies of the Ucp1 gene (Fig. 3D). However, this assay requires calibration with reference samples, limiting its ability to discern Cre copy number within the transgene concatemer. To quantify the absolute copy number of Cre, we used digital droplet PCR (ddPCR)³⁸ 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 (Fig. 3E). 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.

The UCP1-Cre^Evdr Transgene induces profound effects in BAT and WAT biology

The differences in WAT size suggest the possibility that the Ucp1-Cre^Evdr transgene directly impacts fat biology. To test this hypothesis, we used unbiased whole genome expression profiling of iBAT and psWAT from control, 1xUcp1-Cre^Evdr, and 2xUcp1-Cre^Evdr female littermate mice. Unsupervised hierarchical clustering analysis based on differentially expressed genes and multidimensional scaling (MDS) plots revealed distinct global transcriptomic profiles among control, 1xUcp1-Cre^Evdr, and 2xUcp1-Cre^Evdr female mice in both iBAT and psWAT (Supplementary Fig. S3A–D). iBAT and psWAT samples from 1xUcp1-Cre^Evdr mice clustered more closely with those from 2xUcp1-Cre^Evdr mice than with controls (Supplementary Fig. S3B, D). This pattern suggests that the presence of a single copy of the Ucp1-Cre^Evdr transgene substantially alters the transcriptional landscape of both BAT and WAT, potentially impacting their biological functions.

In 1xUcp1-Cre^Evdr iBAT, 569 genes are upregulated and 597 downregulated compared to controls (Fig. 4A). Even more dramatic effects are evident in 1xUcp1-Cre^Evdr psWAT, with 3742 genes upregulated and 3130 downregulated (Fig. 4B). 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 (710) (Fig. 4C, D). A lower number of genes are significantly different in iBAT and psWAT when comparing 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr (Supplementary Fig. S3E, F). In summary, unbiased transcriptional profiling indicates that just one copy of Ucp1-Cre^Evdr transgene may profoundly impact the molecular state of both brown and white adipose tissues, even in tissues where the transgene is not expressed as psWAT suggesting major effects from transgene insertion.

Fig. 4. One copy of Ucp1-Cre^Evdr induces profound dysregulations of BAT and WAT biology.

A RNA-seq comparing female control and 1xUcp1-Cre^Evdr iBAT gene expression (left). Each dot represents one gene. Corresponding GO analysis (middle) and upstream transcriptional regulators by TRRUST (right). Genes, pathways, and transcriptional regulators significantly enriched in controls are labeled in orange and those enriched in 1xUcp1-Cre^Evdr are labeled in red. (n = 2 per genotype). B RNA-seq comparing female control and 1xUcp1-Cre^Evdr psWAT gene expression (left). Each dot represents one gene. Corresponding GO analysis (middle) and upstream transcriptional regulators by TRRUST (right). Genes, pathways, and transcriptional regulators significantly enriched in controls are labeled in orange, and those enriched in 1xUcp1-Cre^Evdr are labeled in red. (n = 2 per genotype). C RNA-seq comparing female control and 2xUcp1-Cre^Evdr iBAT gene expression (left). Each dot represents one gene. Corresponding GO analysis (middle) and upstream transcriptional regulators by TRRUST (right). Genes, pathways, and transcriptional regulators significantly enriched in controls are labeled in orange, and those enriched in 2xUcp1-Cre^Evdr are labeled in brown. (n = 2 per genotype). D RNA-seq comparing female control and 2xUcp1-Cre^Evdr psWAT gene expression (left). Each dot represents one gene. Corresponding GO analysis (middle) and upstream transcriptional regulators by TRRUST (right). Genes, pathways, and transcriptional regulators significantly enriched in controls are labeled in orange, and those enriched in 2xUcp1-Cre^Evdr are labeled in brown. (n = 2 per genotype). For RNA-seq, differential genes were selected by false discovery rate (FDR) < 0.05 with no fold-change cut-off. Source data are provided as a Source Data file.

Gene ontology (GO) by biological process analysis of the significantly altered genes reveals downregulation of mitochondrial activity GO terms (e.g., electron transport chain, respiratory chain, energy generation) in 1xUcp1-Cre^Evdr psWAT compared to control (Fig. 4B). Conversely, mRNA biology pathways are upregulated in 1xUcp1-Cre^Evdr psWAT (Fig. 4B). 1xUcp1-Cre^Evdr iBAT displays downregulation of pathways related to muscle identity compared to control littermates while 1xUcp1-Cre^Evdr iBAT is enriched in immune response GO terms (Fig. 4A). Similar patterns are observed in psWAT of 2xUcp1-Cre^Evdr mice exhibiting suppressed energy generation pathways and elevated mRNA biology (Fig. 4D). In contrast to 1xUcp1-Cre^Evdr iBAT, 2xUcp1-Cre^Evdr iBAT uniquely show upregulation of fatty acid and purine metabolisms while down regulating muscle-related GO terms (Fig. 4C). iBAT of 2xUcp1-Cre^Evdr mice is enriched in genes related to multiple terms of fatty acid metabolism while psWAT of 2xUcp1-Cre^Evdr is depleted of them when compared to 1xUcp1-Cre^Evdr samples (Supplementary Fig. S3E, F). Collectively, these results poised a scenario in which a single copy of the Ucp1-Cre^Evdr transgene may affect energy and lipid metabolism in iBAT and psWAT, and these effects are heightened in 2xUcp1-Cre^Evdr mice.

To elucidate the upstream regulators driving gene expression changes, we employed transcriptional regulatory relationships unraveled by sentence-based text mining (TRRUST)³⁹. Control iBAT compared to both 1xUcp1-Cre^Evdr and 2xUcp1-Cre^Evdr mice showed significant enrichment of muscle-related transcription factors (ie, Myog, Myf5, Mef2c, Myod1) (Fig. 4A, C). This finding aligns with the muscle-related GO terms found in these samples. Conversely, psWAT of 2xUcp1-Cre^Evdr mice exhibited decreased activation of critical adipogenic and lipogenic regulators, including Pparg, Ppard, Ppara, and Srebf1, when compared to psWAT of control littermates (Fig. 4D). This may underlie the selective reduction in WAT mass observed in 2xUcp1-Cre^Evdr mice (Fig. 1H and Supplementary Fig. S1J).

We next used qPCR to verify whether thermogenic gene expression is affected by the Ucp1-Cre^Evdr transgene in a larger cohort. 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 (Supplementary Fig. S4A). 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 (Supplementary Fig. S4B). These suppressions are amplified in 2xUcp1-Cre^Evdr psWAT with reductions of 98% in Ucp1, 69% in Prdm16, and 94% in Cidea (Supplementary Fig. S4B). Similar thermogenic depletion is evident in 2xUcp1-Cre^Evdr pgWAT of females, including 96% lower Ucp1 expression on average (Supplementary Fig. S4C). Male iBAT and psWAT, but not pgWAT, show similar results to females suggesting a gender-specific effect on pgWAT (Supplementary Fig. S4D–F). Western blot analysis of iBAT and psWAT lysates from control, 1xUcp1-Cre^Evdr, and 2xUcp1-Cre^Evdr mice confirmed no effects on UCP1 levels in iBAT but a dramatic decrease in psWAT of 2xUcp1-Cre^Evdr mice (Supplementary Fig. S4G, H). Protein levels of key components of each electron transport chain complex are not different between genotypes in either iBAT or psWAT. This suggests that mitochondrial content and general mitochondrial respiration may be not affected by carrying Ucp1-Cre^Evdr transgene in these two fat depots. Next, we analyzed the potential effects of the Ucp1-Cre^Evdr transgene on other thermogenic cycles^40–43. Key markers of creatine (i.e., Ckb, Alpl, Slc6a8), calcium (i.e., Atp2a2, Gamt, Gatm), and fatty acid -triglyceride (i.e., GK, Hsl, Atgl) cycling are largely unchanged in iBAT across genotypes (Supplementary Fig. S4I). The effects in psWAT and pgWAT are larger with selective markers in each of all three cycles decreased in 2xUcp1-Cre^Evdr mice (Supplementary Fig. S4J, K). Muscles may also be affected by Ucp1-Cre^Evdr transgene or compensate for changes in fat tissues trough inter-tissue communication. However, none of the genes analyzed in the quadriceps of controls, 1xUcp1-Cre^Evdr, and 2xUcp1-Cre^Evdr females are affected by genotype (Supplementary Fig. S4L). Collectively, these results indicate that the presence of the Ucp1-Cre^Evdr transgene preferentially impairs the expression of UCP1-dependent thermogenesis genes in psWAT.

The UCP1-Cre^Evdr Transgene influence the adaptation to cold and thermoneutrality

Given the profound effects of Ucp1-Cre^Evdr on thermogenic gene expression, we hypothesized that the transgene alone, without additional genetic manipulation, would impact adaptation to environmental temperatures. To do this, we generated control and 1xUcp1-Cre^Evdr female littermates by crossing 1xUcp1-Cre^Evdr males to wild-type females. We first investigated the effects of acute cold exposure on control and 1xUcp1-Cre^Evdr females. Room temperature-acclimated control and 1xUcp1-Cre^Evdr female mice were exposed to 4 °C for 6 h. 1xUcp1-Cre^Evdr mice exhibit lower core body temperature, measured by a rectal thermometer, before cold exposure which normalized to control levels during cold (Fig. 5A). Under room temperature conditions, 1xUcp1-Cre^Evdr mice also display lower tail but higher iBAT surface temperatures compared to controls (Fig. 5B, C). This suggests a scenario in which 1xUcp1-Cre^Evdr mice show tail vasoconstriction and elevated iBAT thermogenesis as a physiological mechanism to increase their lower core body temperature. Upon cold exposure, 1xUcp1-Cre^Evdr tail temperature normalizes while iBAT temperature is higher than the control (Fig. 5B, C). However, over time at cold temperatures, 1xUcp1-Cre^Evdr core temperature tends to drop faster than control littermates (Fig. 5A). Together, these data reveal dysfunctional thermogenic regulation and body temperature control in mice harboring just one copy of the Ucp1-Cre^Evdr transgene.

Fig. 5. The UCP1-Cre^Evdr Transgene Influence the Adaptation to Cold.

A Rectal temperature of control and 1xUcp1-Cre^Evdr females undergoing acute cold challenge. n = 3 controls, 4 1xUcp1-Cre^Evdr. B Tail temperature of control and 1xUcp1-Cre^Evdr females undergoing acute cold challenge. n = 3 controls, 4 1xUcp1-Cre^Evdr. C BAT temperature of control and 1xUcp1-Cre^Evdr females undergoing acute cold challenge. n = 3 controls, 4 1xUcp1-Cre^Evdr. D qPCR analysis of iBAT of control and 1xUcp1-Cre^Evdr females after cold challenge or maintained at room temperature. n = 3. E qPCR analysis of psWAT of control and 1xUcp1-Cre^Evdr females after cold challenge or maintained at room temperature. n = 3. F qPCR analysis of pgWAT of control and 1xUcp1-Cre^Evdr females after cold challenge or maintained at room temperature. n = 3. G Representative H&E images of fat depots from control, 1xUcp1-Cre^Evdr females after cold challenge. n = 4. Scale bar, 50 μm. Unless otherwise noted, data are mean + SEM, and statistical significance was calculated using t test. *P < 0.05, **P < 0.01, ***P < 0.001. Source data are provided as a Source Data file.

After 6 h of cold exposure, 1xUcp1-Cre^Evdr mice exhibit slightly greater body weight loss compared to controls (Supplementary Fig. S5A). iBAT weight is unchanged between genotypes; however, psWAT and pgWAT are smaller in 1xUcp1-Cre^Evdr mice (Supplementary Fig. S5A), 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) (Fig. 5D). 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 (Fig. 5E). pgWAT shows no differences after cold treatment between controls and 1xUcp1-Cre^Evdr mice (Fig. 5F). 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 (Fig. 5G). Together, these data suggest a scenario of impaired psWAT thermogenic activation in response to acute cold stress in mice harboring the Ucp1-Cre^Evdr transgene.

Next, we investigated the effects of two-week thermoneutral (TN) adaptation (30 °C) on control and 1xUcp1-Cre^Evdr female mice. While body weight remains unaffected (Supplementary Fig. S5B), 1xUcp1-Cre^Evdr females exhibit lower iBAT weight after TN exposure, with no changes observed in psWAT or pgWAT weights (Supplementary Fig. S5B).

Gene expression analysis reveals minimal differences in iBAT between control and 1xUcp1-Cre^Evdr mice after TN adaptation (Supplementary Fig. S5C). However, psWAT of 1xUcp1-Cre^Evdr females display a general downregulation of thermogenic markers, including a 2.8-fold decrease in Ucp1 expression, although only Prdm16 and Cox7a reach statistical significance (Supplementary Fig. S5D). In pgWAT, 1xUcp1-Cre^Evdr females show a less pronounced but significant decrease in Ppgc1a, Cox7a, and Elovl3 expression levels (Supplementary Fig. S5E). Western blot analysis confirmed a TN-induced reduction in iBAT UCP1 protein levels in both genotypes, with no significant inter-genotype differences at either environmental temperature (Supplementary Fig. S5F). UCP1 protein levels in other fat depots were below the detection threshold under these conditions. Histological examination corroborated the tissue weight data, revealing smaller lipid droplets in iBAT of 1xUcp1-Cre^Evdr mice, but no apparent differences in psWAT and pgWAT adipocyte size after TN exposure (Supplementary Fig. S5G). In conclusion, the presence of a single copy of the Ucp1-Cre^Evdr transgene is sufficient to induce subtle yet biologically relevant changes in adipose tissue mass, thermogenic gene expression, and histological adaptations in response to both cold and thermoneutral conditions.

UCP1-Cre^Evdr transgene may express 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)⁴⁴ harboring LoxP sites flanking exon 2 (Figure S6A, B). By crossing Ucp1-floxed mice with Ucp1-Cre^Evdr animals, we would selectively ablate endogenous Ucp1 while preserving the transgenic variant. We hypothesized that 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 mice homozygous for Ucp1-floxed (Ucp1-fl/fl) and positive for Ucp1-Cre^Evdr (aka. Ucp1-fl/fl^Ucp1-CreEvdr mice)(Fig. 6A). Concurrently, we observed an excess proportion of heterozygous Ucp1-floxed (Ucp1-fl/+) carriers of Ucp1-Cre^Evdr (aka. Ucp1-fl/+^Ucp1-CreEvdr mice) (Fig. 6A). As explained above, embryonic lethality due to Ucp1 deficiency is not expected.

Fig. 6. The UCP1-Cre^Evdr transgene may express high levels of UCP1.

A Expected and observed genotypes obtained from end-point PCR and FRT copy number assay. n = 131 pups. Statistical significance was calculated using the 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/fl^Ucp1-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/fl^Ucp1-CreEvdr. E Representative H&E images from control, Ucp1-fl/+^Ucp1-CreEvdr and Ucp1-fl/fl^Ucp1-CreEvdr males. n = 4. Scale bar, 50 μm. F qPCR analysis of Ucp1 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 of control, Ucp1-fl/+^Ucp1-CreEvdr and Ucp1-fl/fl^Ucp1-CreEvdr males. H Representative whole mount confocal images of iBAT, sBAT, psWAT, and pgWAT of R26R-mTmG (controls), Ucp1-Cre^Evdr;R26R-mTmG and Ucp1-fl/fl^Ucp1-CreEvdr;R26R-mTmG mice at 6 weeks of age. Scale bar, 100 μm. I Quantification of the number of adipocytes in each depot labeled by membrane GFP. n = 3 per genotype. J Body weights of control and Ucp1-fl/fl^{Ucp1-CreERT2Biat} males. n = 7 control, 6 Ucp1-fl/fl^{Ucp1-CreERT2Biat}. K BAT weights of control and Ucp1-fl/fl^{Ucp1-CreERT2Biat} males. n = 7 control, 6 Ucp1-fl/fl^{Ucp1-CreERT2Biat}. L WAT weights of control and Ucp1-fl/fl^{Ucp1-CreERT2Biat} males. n = 7 control, 6 Ucp1-fl/fl^{Ucp1-CreERT2Biat}. M Western blot of iBAT protein lysates of control and Ucp1-fl/fl^{Ucp1-CreERT2Biat} males. N 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 within the time point. O 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 within the time point. P 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 within the time point. Q 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 within the time point. Unless otherwise noted, data are mean + SEM. Statistical significance was calculated using an unpaired t test or one-way ANOVA followed by Tukey’s multiple comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001. Source data are provided as a Source Data file.

Thus, we ponder the hypothesis that the transgenic Ucp1 gene within the Ucp1-Cre^Evdr concatemer (Fig. 3A) 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 the floxed allele (tm1c) of the Ucp1tm1a(EUCOMM)Hmgu mice from INFRAFRONTIER GmbH (Supplementary Fig. S6A, B). Next, we designed a copy number assay specific to detect this FRT sequence (Supplementary Fig. S6A, B). We used wildtype, Ucp1-fl/+, and Ucp1-fl/fl mice to calibrate our assay to zero, one, and two copies of FRT (Fig. 6B). 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 (Fig. 6B). Genotyping with the FRT copy number assay revealed expected Mendelian ratios of control, Ucp1-fl/+^Ucp1-CreEvdr, and Ucp1-fl/fl^Ucp1-CreEvdr progeny (Fig. 6A), 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 (Supplementary Fig. S6C). BAT depot weights are similar between controls and Ucp1-fl/+^Ucp1-CreEvdr (Fig. 6C). However, Ucp1-fl/fl^Ucp1-CreEvdr mice display marked ~ 2-fold increase in weight in all BAT analyzed (Fig. 6C). WAT, liver or muscle tissue weights are unchanged across genotypes (Fig. 6D and Supplementary Fig. S6D–F). Histological examination reveals the iBAT hypertrophy in Ucp1-fl/fl^Ucp1-CreEvdr mice is attributable to uniformly enlarged brown adipocytes engorged with excessive lipid (Fig. 6E). In contrast, WAT depots morphology is largely unaffected by genotype (Fig. 6E). In summary, targeted BAT-specific Ucp1 ablation elicit pronounced BAT growth with increased lipid accumulation while no impacting WAT weights and morphology.

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 (Fig. 6F). Cre mRNA is equal in iBAT of Ucp1-fl/+^Ucp1-CreEvdr and Ucp1-fl/fl^Ucp1-CreEvdr mice (Supplementary Fig. S6G). iBAT also shows a compensatory increased expression of classic thermogenic key markers Ppargc1a, Cox7a, and Elovl3 (Supplementary Fig. S6J). In psWAT, Ucp1 expression is unaltered by genotype; however, psWAT of Ucp1-fl/fl^Ucp1-CreEvdr mice display elevated expression of Cidea, Cox7a, and Elovl3 (Fig. 6F and Supplementary Fig. S6K). pgWAT gene expression is largely unchanged in Ucp1-fl/+^Ucp1-CreEvdr and Ucp1-fl/fl^Ucp1-CreEvdr mice compared to control littermates (Fig. 6F and Supplementary Fig. S6L). 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 (Supplementary Fig. S6H, I). Thus, BAT-targeted Ucp1 ablation induces depot-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. 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 (Fig. 6G). We and others have previously shown that Ucp1-Cre^Evdr drives recombination in essentially all brown adipocytes^29,45. However, the high residual UCP1 protein opens the possibility that Ucp1-Cre^Evdr is not active in a substantial proportion of brown adipocytes in the context of deleting Ucp1 floxed alleles. To solve this question, we crossed Ucp1-fl/fl^Ucp1-CreEvdr and 1xUcp1-Cre^Evdr mice to the dual membrane fluorescent reporter mTmG⁴⁶. In this reporter, Cre-mediated recombination eliminates the membrane-bound Tomato cassette and activates the expression of membrane-bound eGFP. The accumulation of fluorescence reporters in the plasma membrane permits fine resolution to single brown and white adipocytes^{28,29,47–52}. Whole mount confocal analysis of iBAT and sBAT from 1xUcp1-Cre^Evdr mice show near-complete Cre-mediated recombination in brown adipocytes in 6-week-old mice (Fig. 6H, I). Surprisingly, the same results are obtained in Ucp1-fl/fl^Ucp1-CreEvdr BAT depots (Fig. 6H, I). On the other hand, no adipocytes in psWAT and pgWAT show prove of active Cre in either genotype (Fig. 6H, I). Thus, UCP1-Cre^Evdr activity is maintained in all brown adipocytes of Ucp1-fl/fl^Ucp1-CreEvdr mice. The paradoxical preservation of the UCP1 protein is unlikely to originate from the endogenous Ucp1 genes, which should be efficiently deleted in this context. Instead, these findings suggest the hypothesis that, under these conditions, the transgenic Ucp1 gene within the Ucp1-Cre^Evdr concatemer may be expressed.

To control that Ucp1 floxed allele can be efficiently deleted in brown adipocytes, 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 (Fig. 6J–L and Supplementary Fig. S6M–O). Critically, tamoxifen treatment leads to highly efficient ablation of UCP1 protein in iBAT (Fig. 6M). This confirms that the Ucp1-floxed allele can be efficiently deleted. These findings support the hypothesis that the persistent UCP1 expression in Ucp1-fl/fl^Ucp1-CreEvdr mice stems from the ectopic Ucp1 gene within the UCP1-Cre^Evdr transgene concatemer.

To test for functionality, we performed acute cold challenges by exposing mice at 6 °C for 6 h. Intriguingly, Ucp1-fl/fl^Ucp1-CreEvdr mice exhibit rather elevated BAT temperatures before starting cold exposure, compared to littermate controls (Fig. 6N–P). Unexpectedly, Ucp1-fl/fl^Ucp1-CreEvdr mice are proficient in maintaining core body temperature during cold exposure at the same level as littermate controls (Fig. 6P). In stark contrast, Ucp1-fl/fl^{Ucp1-CreERT2Biat} mice rapidly become hypothermic at 6 °C, reflecting the efficient ablation of UCP1 in BAT (Fig. 6Q). 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 of 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^53,54. However, these aP2-Cre models were found to inefficiently target mature adipocytes while exhibiting broad recombination in the brain, endothelial cells in adipose tissues, macrophages, adipocyte progenitors, and elsewhere^{5,17,21,55–58}. This prompted the development of Cre lines driven by the promoter of adiponectin^20,21. However, Ucp1-Cre^Evdr also exhibits widespread brain activity^45,59, including regions controlling feeding and non-shivering thermogenesis^60–62. Importantly, very low endogenous Ucp1 expression partially overlaps with these brain areas^45,59, suggesting Ucp1-Cre^Evdr partially recaptures native Ucp1 regulation. Ucp1-Cre^Evdr has also recently been shown to be expressed in subsets of epithelial cells in a number of organs (ie, brain, kidneys, mammary gland)⁶³. 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^64–66. Random BAC transgene insertion is frequently associated with substantial genomic alterations, often disrupting gene coding sequences and creating small or large rearrangements^6–8. 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, sequencing approaches such as whole genome sequencing and TLA enable fine mapping of insertion locus, disruption effects, and integrated sequences³⁶. For instance, whole genome sequencing and TLA both revealed Adipoq-Cre^Evdr transgene inserted into the Tbx18 gene on chromosome 9^20,67,68, perturbing Tbx18 expression and adding passenger gene copies with possible widespread effects⁶⁸. In addition, BAC transgenes normally integrate as concatemers leading to multiple full or partial copies of the transgene^3,69,70. Using ddPCR, we find that the Cre coding sequence is present in a single copy in Ucp1-Cre^Evdr mice. This was also the case for the Adipoq-Cre^Evdr transgene⁶⁸. 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 ddPCR³⁸. 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 long-range genome sequencing^7,71,72. In addition, transgenic strains are bred for generations, allowing the 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 the 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^6–16. 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 tissue gene expression changes⁶⁸, the Ucp1-Cre^Evdr effects are considerable. The large remodeling of gene expression and other factors of psWAT of 1xUcp1-Cre^Evdr show that even without expression of the transgene, the effects can be notorious. This opens the possibility that the Ucp1-Cre^Evdr transgene may reprogram any no-UCP1 expressing cell due to genomic disruption or others. For example, the expression of the genes directly affected by the deletion caused by the transgene insertion (i.e., Efhc1 and Tram2) shows their highest expression in testis. This suggests possible effects on the testis by the deletion of these genes. Overall, transgene insertion effects are diverse and context-dependent and may have to be researched in each. Thus, thorough characterization of each mouse transgenic 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 underpowered⁶⁷. The unexpectedly 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, maybe a strong paradigm to ensure the 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^6–16. 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.

The effects of the Ucp1-Cre^Evdr transgene on adipose tissue biology may have significant implications for the interpretation of studies utilizing this widely used Cre-driver line. The transcriptomic remodeling in iBAT and psWAT of hemizygous Ucp1-Cre^Evdr mice may confound the interpretation of gene expression analyses in genetically modified mice generated with this Cre driver, particularly when compared to the commonly used Cre-negative littermate controls. Although the physiological impacts of Ucp1-Cre^Evdr hemizygosity on cold and thermoneutrality adaptation appear modest, their potential interaction with additional genetic modifications warrants careful consideration. Our findings underscore the need for further investigation into the broader implications of the Ucp1-Cre^Evdr transgene. This includes long-term effects on BAT activity (e.g., prolonged cold exposure or high-fat diet-induced obesity) and potential influences on non-adipose tissues (e.g., liver and brain) which may affect the interpretation of both previous and future results. While previous research using this model has undoubtedly provided valuable insights, future studies may benefit from incorporating additional controls to eliminate possible Ucp1-Cre^Evdr-specific effects. This could include control groups harboring only the Ucp1-Cre^Evdr transgene and validation of key findings using orthogonal Cre drivers (e.g., inducible Ucp1-CreERT2^Biat). Though challenging, these considerations could isolate transgene-specific effects from those of intended genetic manipulations. This reevaluation will enhance our understanding of adipose tissue biology and metabolic regulation.

The persistence of significant Ucp1 mRNA and UCP1 protein levels, coupled with unexpected cold tolerance in Ucp1-f l/f l^Ucp1-CreEvdr mice despite efficient Cre-mediated recombination (as verified by the R26R-mTmG reporter), underscores the potential for transgene-derived gene products to significantly impact experimental outcomes. The putative production of UCP1 by the Ucp1-Cre^Evdr transgene itself warrants careful consideration. Notably, out of 95 references in the MGI database employing Ucp1-Cre^Evdr mice, only five provide confirmatory results using the inducible Ucp1-CreERT2^Biat model. While such validation enhances confidence, discrepancies between constitutive and inducible Cre drivers may arise due to differences in the temporal activity of Cre recombinase. More broadly, these results highlight the profound and often unanticipated impacts of passenger sequences in transgenic constructs.

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. In addition, the 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.

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 the scientific rigor of BAC transgenics as future tools to catalyze discovery.

Methods

Lead contact and material availability

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

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), and Ucp1-CreER mice were described before²⁶. Ucp1 flox mice were obtained from the EUCOMM program (C57BL/6N-Ucp1^{tm1a(EUCOMM)Hmgu}/Ieg) after the 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 12 h light/dark cycle and kept in ventilated racks fed ad libitum with a standard chow diet (5053 - PicoLab® Rodent Diet 20), with bedding changed every two weeks. See figure legend for specific age and number of mice used. All mice used within experiments were littermates unless otherwise noted in figure legends. 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.

MGI alleles and publications search

The MGI database was searched with the 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 the location indicated as “unknown” were assigned as the location of the transgene unidentified group. The remainder of mouse models were automatically assigned to transgene with a known location group. To analyze Cre-driver mouse models, the symbol of each mouse model was search for containing “Cre”. Out of the 1968 mouse models found, the location was assigned as above. A number of publications assigned to specific transgenic mice were found in MGI in July 2023.

GWAS-associated traits and mouse mortality phenotype search

Mouse mortality phenotype associated with gene knockouts was searched in the MGI database. GWAS-associated traits to specific genes were searched 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 2 mg/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. The adipose tissue notation used here was described previously⁵¹. Mice were dissected in the 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

The 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) 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 h. 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 the 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 Supplementary 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 10 mg 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 Supplementary 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)(Supplementary Table S4). qPCR was performed on a QuantStudio 6Flex real-time PCR system using the following protocol: 95 °C for 10 min 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

The reaction mixture was composed of 10 µL 2x ddPCR Supermix (without dUTPs; Bio-Rad, Hercules CA), 1 µL each of the proves against housekeeping and target gene (Bio-Rad)(Supplementary Table S4), 1 µL of HaeIII (NEB, R0108), 50 ng of DNA template, and adjusted to a final volume of 20 µL. 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 min, 94 °C for 30 s, and 60 °C for 1 min (40 cycles 2 °C/sec ramp rate), a 10 min 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-week-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 reported³⁶, with six independent pairs of primers (Supplementary Table S3) using the mouse mm10 genome as a host reference.

RNA-seq analysis

RNA-seq reads were aligned to UCSC mouse genome 10 mm using STAR aligner⁷³. Only uniquely aligned reads were used for downstream analysis. Raw read counts for each gene were measured using FeatureCounts in the subread package⁷⁴ with an option, “-s 2 -O–fracOverlap 0.8”. Differential gene expression analysis was performed using EdgeR⁷⁵. MDS plot was generated using EdgeR. Genes with FDR < 0.05 were selected as differentially expressed genes. Gene ontology analysis was performed using Enrichr⁷⁶. Data is deposited into the Gene Expression Omnibus database (GSE276292).

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 was aligned to the Cre reference. Then, we pooled the selected read pairs and performed de novo assembly using Trinity⁷⁷. Given the design of Cre transgene, we anticipated that the assembled sequence should fully cover Cre CDS and span from the 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 reached the 3’ half of the Ucp1 exon #1, which happened after the 4th round. The final assembly results were assessed and annotated using known references and Blast⁷⁸.

Fusion gene detection

To examine if there is any fusion gene event around the transgene insertion locus, we performed gene fusion analysis using STAR-Fusion (v1.10.1)⁷⁹. STAR-Fusion reference was built using prep_genome_lib.pl with mm10 mouse genome and Gencode gene annotation vM25. STAR-Fusion command was run with a default option. Gene fusion events detected around the transgene locus consistently in 1xUcp1-Cre^Evdr or 2xUcp1-Cre^Evdr were selected for presentation.

Software used in figures

All graphs were made in GraphPad. Figures were made in Adobe Illustrator. Due to legal requirements, BioRender is acknowledged in the figure legends.

Quantification and statistical analysis

Data are presented as mean + s.e.m. unless stated otherwise. Two-tailed 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 the 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.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conceptualization and study design: J.S.G. Data collection: M.H. performed most experiments; O.I., R.M., A.S., M.G., and J.S.G. contributed to several experiments; C.L.B.P. performed skeletal staining and head measurements; M.D. performed ddPCR. Data analysis: M.H. and J.S.G. RNA-seq data analysis: H.W.L. and C.A. Data interpretation: J.S.G. Supervision: J.S.G.; SAB supervised C.L.B.P. H.W.L. supervised C.A. Manuscript writing – Original draft: J.S.G. Manuscript writing – Review and editing: all authors. All authors approved the final manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

The RNA sequencing data generated in this study have been deposited in the NCBI Gene Expression Omnibus database under accession code GSE276292. The data generated in this study are provided in the Supplementary Information/Source Data file. Source data are provided in this paper.

Competing interests

The authors declare no competing interest.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work the author(s) used Calude.ai in order to improve readability. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-024-54763-4.