The Cre/lox system: Cre-ating unintended damage

INTRODUCTION

One of the pivotal advances in molecular biology has been the ability to target mammalian gene expression in a site-specific manner. This is achieved using the Cre/lox technique in which the enzyme cyclization recombinase (Cre) from the bacteriophage P1 recognizes certain 34-bp loxP sites and removes intervening DNA through recombination (8, 9). Thus, a murine gene (or key exon) of interest can be flanked by loxP sites and removed by Cre recombinase (2). Spacial specificity is obtained by putting the expression of Cre under the control of specific promoters that are only active in the cell type of interest, and inducible promoters allow for temporal control of Cre-mediated recombination. This Cre/lox system has been important for defining the function of proteins in different mammalian tissues (7) and for lineage tracing studies when used with a reporter mouse containing a stop codon flanked by loxP sites.

EXPRESSION OF CRE AND GLOMERULAR PATHOLOGY

The Cre enzyme itself has long been assumed to be inactive in the absence of a gene flanked by loxP sites (i.e., floxed), and most investigators will used the floxed littermates without Cre as controls. In an article published in a recent issue of the American Journal of Physiology-Renal Physiology, Balkawade et al. (1) showed that uninjured mice with podocyte-specific expression of Cre (NPHS2-Cre) have subtle but definite glomerular pathology. There were no gross functional differences between NPHS2-Cre mice or wild-type mice as measured by proteinuria and creatinine, consistent with the lack of pathology previously reported (5). More thorough investigation by the authors revealed that the number of cells staining for Wilms tumor 1 (WT1), a marker of podocytes, was decreased in NPHS2-Cre^+/+ mice at 24 wk of age. Additionally, statistically significant differences in glomerular basement membrane (GBM) thickness, as measured by electron microscopy, were noted between wild-type (186.8 ± 3.6 nm) versus NPHS2-Cre^+/− (236.9 ± 6.5 nm) or NPHS2-Cre^+/+ (256.8 ± 10 nm) mice (1).

To investigate the mechanism whereby NPHS2-Cre increased GBM thickness, the authors examined protein and transcriptional regulation of the individual laminin chains that comprise mature laminin-α₅β₂γ₁ (LAM-521). Both α₅- and β₂-chains were suppressed, whereas γ₁-chains were increased, in NPHS2-Cre^+/+ compared with wild-type mice. This would be consistent with a switch to immature laminin-α₁β₁γ₁ (LAM-111) but does not prove this as α₁- and β₁-chain levels were not assessed. Also, it is curious that NPHS2-Cre^+/− mice actually had less γ₁ protein expression than wild-type mice despite increased GBM thickness, making this explanation less likely. To further probe these NPHS2-Cre-dependent changes in podocytes and GBM thickness, RNA sequencing was performed on isolated glomeruli of NPHS2-Cre^+/+ and wild-type mice. There were 230 genes with statistically significant alterations in expression between the 2 groups, including 15 matrix-associated genes and 17 apoptosis/cell death-related genes. Some of the genes upregulated in NPHS2-Cre mice have antifibrotic roles (e.g., cytoglobin), indicating that many of these transcripts are altered in response to injury rather than as the cause. Although the number of mice used for RNA sequencing was low (2 mice/genotype), this brief report carefully documents structural changes in the glomeruli of NPHS2-Cre mice and associated alterations in genes involved in matrix production and cell survival (1).

It is reasonable to question the significance of these changes in GBM thickness and WT1⁺ cells given that no major differences in proteinuria or renal function were detected. Aside from the issue that creatinine is notoriously insensitive as a measure of renal function, it is important to recognize that mice generated by the Cre/lox technique are often used to assess the cell-specific role of a protein in renal injury. Relatively subtle pathology in uninjured mice can predispose to greater susceptibility to injury, and an observed phenotype may be caused by Cre-dependent effects and not solely by conditional deletion of the targeted gene. This is the first report to document Cre-dependent damage in the kidney, but injury in other organs has been previously published. Using the α-myosin heavy chain (α-MyHC) promoter to drive Cre expression in the heart, investigators have reported a reduction in cardiac function at 6 mo with fibrosis, inflammation, and DNA damage (6). Similarly, targeting lung epithelia using Cre under control of the human surfactant protein C promoter caused cystic lungs with excessive apoptosis (3). The phenotypes of extrarenal Cre activity were more pronounced than that of NPHS2-Cre, but DNA damage and epithelial apoptosis appear to be common Cre-mediated pathologies. Although NPHS2-Cre-mediated apoptosis was not shown directly, the decrease in WT1⁺ cells in Cre-containing mice and the increase in transcripts of DNA damage/apoptosis genes by RNA sequencing is suggestive (1).

The large difference in phenotype severity between NPHS2-Cre and that published in the heart and lungs raises the questions of why does this variability occur? The amount and duration of Cre expression, which varies based on the promoter, may account for these differences. Consistent with this idea, much of the cardiac toxicity induced by α-MyHC-Cre was averted by transient induction of Cre using a tamoxifen-inducible α-MyHC promoter (4). Another question raised in Balkawade et al.’s work is how does Cre cause this unintended injury? Although Cre preferentially mediates recombination between pairs of loxP sites, it can induce recombination in endogenous sites that have up to 10 mismatches to the canonical 34-bp loxP (10). These cryptic or degenerate loxP sites were identified in transcriptionally active cardiac genes, and changes in transcription and protein expression of these genes were noted when α-MyHC-Cre was present (6). Another possible explanation is that transgene insertion site or copy number could mediate both the phenotype variability and Cre-mediated injury. Although not investigated with NPHS2-Cre, this possibility was considered less likely with the Cre-mediated cardiotoxicity based on insertion site mapping (6).

IMPLICATIONS

So, what are the implications of this study for the field of renal injury and the Cre/lox system? The obvious answer is that Cre-containing mice without the floxed alleles should also be used as controls. This approach requires increasing the mouse colony costs as additional breeders would be necessary. It is important to maintain equivalent background strains between Cre-containing control and Cre plus floxed allele mice as there is variability in Cre-dependent effects based on genetic background (4). Adding additional controls to costly mouse experiments, in addition to the National Institutes of Health mandate to study both sexes, can be a hard pill to swallow, particularly when the budgets for studies are becoming more restrictive. Certainly, there is a case to be made for increasing the budgets for models of injury using genetically altered mice. However, ignoring the direct effect of Cre itself has even more costly consequences as the Cre/lox system is used to generate important preclinical data for potential therapeutics.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

L.G. drafted manuscript; L.G. edited and revised manuscript; L.G. approved final version of manuscript.