Aberrant glial activation and synaptic defects in CaMKIIα-iCre and nestin-Cre transgenic mouse models

Current scientific research is driven by the ability to manipulate gene expression by utilizing the Cre/loxP system in transgenic mouse models. However, artifacts in Cre-driver mouse lines that introduce undesired effects and confound results are increasingly being reported. Here, we show aberrant neuroinflammation and synaptic changes in two widely used Cre-driver mouse models. Neuroinflammation in CaMKIIα-iCre mice was characterized by the activation and proliferation of microglia and astrocytes in synaptic layers of the hippocampus. Increased GFAP and Iba1 levels were observed in hippocampal brain regions of 4-, 8- and 22-month-old CaMKIIα-iCre mice compared to WT littermates. Synaptic changes in NMDAR, AMPAR, PSD95 and phosphorylated CaMKIIα became apparent in 8-month-old CaMKIIα-iCre mice but were not observed in 4-month-old CaMKIIα-iCre mice. Synaptophysin and synaptoporin were unchanged in CaMKIIα-iCre compared to WT mice, suggesting that synaptic alterations may occur in excitatory postsynaptic regions in which iCre is predominantly expressed. Finally, hippocampal volume was reduced in 22-month-old CaMKIIα-iCre mice compared to WT mice. We tested the brains of mice of additional common Cre-driver mouse models for neuroinflammation; the nestin-Cre mouse model showed synaptic changes and astrocytosis marked by increased GFAP+ astrocytes in cortical and hippocampal regions, while the original CaMKIIα-Cre T29-1 strain was comparable to WT mice. The mechanisms underlying abnormal neuroinflammation in nestin-Cre and CaMKIIα-iCre are unknown but may be associated with high levels of Cre expression. Our findings are critical to the scientific community and demonstrate that the correct Cre-driver controls must be included in all studies using these mice.

In the field of biomedical research, genetic engineering is an essential technology to investigate protein function in specific tissues and cell types. The Cre/loxP system is an advanced method of gene targeting that allows for precise control over tissue specificity and timing of gene expression in mice, while avoiding complications that arise during constitutive gene knockout (KO) such as early embryonic lethality. In the Cre/loxP system, Cre recombinase, a prokaryotic enzyme isolated from the P1-bacteriophage, is expressed in genetically engineered mice under the control of a tissue-specific promoter and catalyzes the recombination between two loxP sites flanking an exon of a gene of interest [1][2][3] . Mice with LoxP sites flanking a gene of interest (floxed) are bred to transgenic mice expressing Cre to create gene KO and knock-in (KI) mice 4,5 . Many Cre-driver mice are available that express Cre in diverse patterns within the central nervous system (CNS) to investigate genes of interest within specific regions and cell types of the brain [6][7][8] . The selection of the promoter driving Cre expression is critical, as this controls the timing and location of Cre activity and thus of gene KO.
The calcium/calmodulin-dependent protein kinase II alpha subunit (CaMKIIα) promoter is routinely used to drive Cre expression in excitatory forebrain neurons of the CNS. Numerous strains of CaMKIIα Cre-driver mice have been developed 9 , with the most widely used strains being the original CaMKIIα Cre-driver mouse, the CaMKIIα Cre T29-1 strain 10 and the CaMKIIα-iCre strain 11 . CaMKIIα is a member of the CaMKII family of serine/threonine protein kinases activated by Ca 2+ and is endogenously expressed within pyramidal cells of the hippocampus and the cerebral cortex 12,13 . Within the hippocampus, CaMKIIα is strongly expressed in the

Materials and methods
Study design. Cre-expressing mice are typically bred in a breeding scheme wherein Cre-driver mice are crossed to floxed allele mice to create gene KO, KI or reporter mice. In this study, we find that in the absence of floxed alleles, activated microglia and astrocytes proliferated in a distinct pattern throughout the SR and SO hippocampal layers of CaMKIIα-iCre mice compared to WT littermates. CaMKIIα-iCre mice were analyzed at 4, 8 and 22-months of age and an equivalent number of male and female mice were analyzed for all parameters. The figure legends have specific n values and females are represented as circles and males as triangles in all graphs. Synaptic changes and hippocampal volume loss were analyzed in aged CaMKIIα-iCre mice. We initially observed an aberrant neuroinflammation in the hippocampus of CaMKIIα-iCre mice 11 , and we determined the presence of Cre-associated neuroinflammation in two additional strains of Cre recombinase expressing mice, the CaMKIIα-Cre T29-1 strain 10 and the nestin-Cre mouse model 21 . These mice were analyzed at 4-monthsold. By analyzing these additional mice and WT controls, we established that neuroinflammation and synapse changes revealed in CaMKIIα-iCre were likely caused by Cre recombinase expression. We investigated these effects by utilizing immunofluorescence markers that demonstrate various states of activity, reactivity, and formation of glial cells within areas of the hippocampus that govern synaptic plasticity and function in learning and memory, as well as the quantification of these cell markers. The analysis of these neuronal and synaptic markers highlighted the phenotypic neuroinflammatory effect of Cre on presynaptic and postsynaptic regions within specific areas of the brain. We further confirmed the results of immunofluorescence through western blot analysis by means of probing for neuronal inflammatory and synaptic markers and the quantification of those results.
Ethics statement. All experimental protocols were approved by the Northwestern University Institutional Animal Care and Use Committee (IACUC). All methods were carried out in accordance with relevant guidelines and regulations. All methods reported are in accordance with ARRIVE guidelines.
Animals. Animals were housed and monitored in accordance with IACUC. Mice were maintained on a 12:12 light: dark cycle and consumed a standard rodent diet and water ad libitum. CaMKIIα-iCre mice were a generous gift of Dr. Warren Tourtellotte (Cedars-Sinai) and bred to C57BL/6J mice and maintained as heterozygotes. Homozygous CaMKIIα-Cre T29-1 mice were purchased from The Jackson Laboratory (strain #005359) and crossed with C57BL/6J mice to produce heterozygous mice for analysis. Homozygous Nestin-Cre mice were purchased from The Jackson Laboratory (strain #003771) and crossed with C57BL/6J mice to produce heterozygous mice. Wild-type C57BL/6J littermates were used as controls for all groups in this study. All animals analyzed were included in statistical analysis. To confirm genotypes, all mice used in the study were PCR-genotyped in house for Cre via primers purchased from Integrated DNA Technologies (IDT) and used in accordance with Jackson Laboratory standard genotyping protocols. The CaMKIIα-iCre genotyping protocol involved a two primer reaction, the forward DNA oligo consisted of nineteen bases 5ʹ-AGA AGC CCC AAG CTC GTC A-3ʹ, and the reverse  DNA oligo consisted of eighteen bases 5ʹ-CAG CAG GGA ACC ATT TCC -3ʹ. The Nestin-Cre genotyping protocol  involved a three primer reaction, the forward DNA oligo consisted of twenty-one bases 5ʹ-CCT TCC TGA AGC  AGT AGA GCA-3ʹ, the reverse DNA oligo consisted of twenty bases 5ʹ-GCC TTA TTG TGG AAG GAC TG-3ʹ, and the wildtype forward consisted of twenty-one bases 5ʹ-TTG CTA AAG CGC TAC ATA GGA-3ʹ. The CaMKIIα-Cre T29-1 strain genotyping protocol involved a four primer reaction, the forward DNA oligo consisted of twenty bases 5ʹ-CGT CCA TCT GGT CAG AAA AG-3ʹ, the reverse DNA oligo consisted of twenty bases 5ʹ-TCT TCT TCT  TGG GCA TGG TC-3ʹ, the internal positive control forward consisted of twenty bases 5ʹ-AGT GGC CTC TTC  CAG AAA TG-3ʹ, and the internal positive control reverse DNA oligo consisted of twenty bases 5ʹ-TGC GAC  TGT GTC TGA TTT CC-3ʹ. Tissue extraction and preparation. Animals were euthanized by intraperitoneal injection of xylazine (15 mg/kg) and ketamine (100 mg/kg) followed by transcardial perfusion with 10 ml 1× cold phosphatebuffer saline (PBS) containing phenylmethylsulfonyl fluoride (20 mg/ml in EtOH) 1:1000, dithiothreitol (1 M) 1:10,000, leupeptin (5 mg/ml) 1:10,000 and sodium orthovanadate (200 mM) 1:10,000. Mouse brains were extracted immediately following perfusion and right hemibrains were dissected on ice into two regions: cortex and hippocampus. The tissue was flash frozen in liquid nitrogen and stored at − 80 °C until biochemical analysis.
All tissue lysis buffers contained Halt phosphatase inhibitor (#78420, Thermo Fisher Scientific) and protease inhibitor cocktail III (#535140, Millipore). Hippocampus and cortex tissues were weighed and homogenized with a dounce homogenizer in a 1:10 (w/v) 1× PBS, then centrifuged at 4 °C at 14,000 RPM for thirty minutes. The supernatant was removed as the soluble fraction and stored at − 80 °C. The remaining pellet was extracted by the addition of radioimmuprecipitation assay buffer (RIPA) (50 mM tris, 0.15 M NaCl, 1% octylphenoxypolyethoxyethanol (IGEPAL), 1 mM EDTA, 1 mM EGTA, 0.1% SDS, 0.5% sodium deoxylate at pH 8) at 1:10 (w/v). The samples were sonicated for twenty seconds on ice, followed by centrifugation at 14,000 RPM for 30 min at 4 °C. The supernatant was removed and a bicinchoninic assay (BCA) (#23225, Thermo Fisher Scientific) was performed to measure protein concentration. After determining protein concentration, the appropriate volumes of samples and buffers were calculated to reach the same final protein concentration (1 mg/ml). NuPAGE LDS Sample Buffer (4×) (#NP0008, Thermo Fisher Scientific) with 4% β-Mercaptoethanol (#M6250, Millipore Sigma) was added to all samples followed by heating at 95 °C for 10 min on a heat block.
Immunoblotting. Equal amounts of protein homogenates (20ug) were loaded into NuPAGE midi Bis-tris gels (#WG1403BOX, Thermo Fisher Scientific) set up in Criterion Cell electrophoresis chambers (Bio-Rad) containing 1× MOPS running buffer using 50 mM MOPS (#PHG0007, Millipore Sigma) prepared in 50 mM Tris Base (#DST60040-10000, Dot Scientific), 1 mM EDTA (#50-841-667, Teknova) and 0.1% SDS (#50-751-6948, Quality Scientific). Gels ran at approximately 140-160 V for 1-1.5 h, or until the samples reached the bottom of the gel. Gels were transferred to nitrocellulose membranes in 5× transfer buffer (Bio-rad) and ran for 45 min at a current of 1.1A using the Trans-Blot Turbo Transfer system (Bio-rad). The membrane was washed 3 times in washing buffer (1× PBS, 0.1% TWEEN-20) for 5 min, then blocked in SuperBlock blocking buffer blotting in PBS (#37517, Thermo Fisher Scientific) at room temperature for 1 h. Primary antibodies were prepared in washing buffer with 10% SuperBlock blocking buffer then added to the membranes and allowed to incubate overnight in 4 °C. Primary antibody was then removed, and the membranes were washed in washing buffer 3 times for 5 min at room temperature. Secondary antibodies (horse anti-mouse, #P1-2000 Vector Labs, goat anti-rabbit, #P1-1000 Vector Labs) were prepared in wash buffer with 10% SuperBlock and added to the membranes at 1:10,000. The membranes were incubated in secondary antibodies for 60 min and then washed 3 times for 5 min in washing buffer. The blots were developed using SuperSignal West Femto Maximum Sensitivity Substrate (#34096, Thermo Fisher Scientific) and SuperSignal West Pico PLUS Chemiluminecsent Substrate (#34580, Thermo Fisher Scientific) and imaged on a ProteinSimple FCR imager. Chemiluminescent signals were quantified using AlphaView software (ProteinSimple).
Immunohistochemistry. Left hemibrains were immersion-fixed in 10% formalin and kept in 30% sucrose 1× PBS solution at 4 °C for long-term storage. Hemibrains were mounted coronally and serially sectioned into 12-well plates at 30 microns thick using a freezing-sliding microtome and preserved in cryoprotective solution (1× PBS, 30% sucrose and 30% ethylene glycol). Brain sections were washed in 1× tris-buffer saline (TBS) 3 times for 5 min on an orbital shaker at 180 RPM. Sections were transferred to a glycine solution (1XTBS, 0.25% triton X-100, 16 mM glycine) and incubated on an orbital shaker at 180 RPM for 1 h at room temperature. Brain sections were washed again 3 times in 1× TBS for 5 min. Sections were then transferred to blocking solution (1× TBS, 0.25% triton X-100, 5% donkey serum) and allowed to incubate for a duration of 2 h on an orbital shaker at 180RPM. Sections were then washed in 1× TBS, 0.25% triton X-100 and 1% BSA 2 times for 10 min. Primary antibodies were prepared in 1× TBS, 0.25% triton X-100 and 1% BSA, and incubated overnight at 4 °C on an orbital shaker at 110 RPM. The sections were then washed again in 1× TBS, 0.25% triton X-100 and 1% BSA 3 times for 10 min on an orbital shaker at 180 RPM until they were transferred to Alexa fluor-labeled secondary antibodies (Invitrogen) at a dilution of 1:1000 prepared in 1XTBS, 0.25% triton X-100 and 1% BSA and allowed to shake for 2 h at 180 RPM. All sections were washed 3 times for 15 min in 1XTBS and then mounted onto slides using ProLong Gold (#P36934, Thermo Fisher Scientific) and imaged on a Nikon A1 laser scan-

Image quantification.
To measure the volume of the hippocampus, brains were serially sectioned into 30 μm thick sections in a 12-well plate. Sequential sections chosen for analysis were from Bregma coordinates of approximately − 0.94 to − 3.40 mm through the hippocampus (eight sections per brain, 360 μm apart) and areas of interest were traced manually and measured using ImageJ software. Sections were immunostained as described above using anti-NeuN primary antibody and DAPI (4′,6-diamidino-2-phenylindole), and then imaged with a 4× objective using a Ti2 wide-field microscope. Volume was calculated using the following formula: volume = (sum of area) × 0.36 mm. For immunofluorescence quantification of Iba1, GFAP, synaptoporin and NeuN-covered area, three to four coronal sections from Bregma coordinates of about − 1.70 to − 3.52 mm were obtained. Brain regions were immunostained using anti-Iba1, anti-GFAP, anti-synaptoporin and anti-NeuN primary antibodies as described above, and images were captured using a 10× objective and a Ti2 widefield microscope. Nikon NIS-Elements Software (Northwestern University Nikon Imaging Centre) was used to set intensity and size thresholds to eliminate background staining. To calculate Iba1, GFAP, synaptoporin and NeuN-covered area within the hippocampus or cortex, SR, SO or pyramidal layers, each area was traced manually using NIS-Elements, and a binary channel was created for each region of interest. The average of three sections was obtained to calculate the covered area in the SR, SO, pyramidal layers, hippocampus and cortex for each signal in each mouse. All imaging, section selection, tracing, and volume analysis were performed by someone blind to the genotypes of the animals.

Statistical analyses.
Statistical tests were performed using GraphPad Prism software v7.05 (http:// graph pad. com/ scien tific-softw are/ prism/). Detailed test information including the specific statistical tests, number of replicates, and P values are stated in the figure or the figure legend. For all quantifications, the data are plotted as the mean ± SEM. Significance was concluded when P < 0.05, indicated by *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Results
Activation and proliferation of microglia and astrocytes in the stratum oriens and radiatum in the hippocampus of CaMKIIα-iCre mice. While investigating the effect of gene knockout on glial cell activity in floxed allele CaMKIIα-iCre mice, we incidentally observed abnormal neuroinflammation in hippocampal brain regions of 8-month-old CaMKIIα-iCre mice that do not express floxed alleles. Therefore, we sought to investigate and characterize this abnormal phenotype in CaMKIIα-iCre mice in this study. Significant increases in microglia marker Iba1 and astrocytic protein GFAP in the hippocampus were observed in 8-month-old CaMKIIα-iCre mice compared to wild-type (WT) littermates of the same genetic background (C57/BL6) (Fig. 1A,D,E). We also detected increased expression of CD68, which is highly expressed during microglia activation 26 and C3, which is expressed by reactive astrocytes 27 (Fig. 1B,C). We assessed CaMKIIα-iCre and WT mice at 4-months of age and found trends toward an increase in Iba1 and GFAP to levels comparable to 8-months in CaMKIIα-iCre mice (Fig. 1D,E). The activation and proliferation of microglia and astrocytes observed by immunofluorescence microscopy was confirmed by performing immunoblot analysis for GFAP and Iba1 in hippocampal homogenates of 4-and 8-months old WT and CaMKIIα-iCre mice (Fig. 1F,G). Consistently, at 4 and 8 months, Iba1 and GFAP were increased to comparable levels in CaMKIIα-iCre mice (Fig. 1H,I). To confirm our findings, we also assessed Iba1 and GFAP levels in CaMKIIα-iCre mice that were bred and housed in a separate facility. Here we found increased levels of Iba1 and GFAP in hippocampal homogenates of CaMKIIα-iCre mice compared to WT mice at 2 and 5 months of age (Fig. S1A).
In the hippocampus of CaMKIIα-iCre mice, activated microglia and astrocytes appeared to proliferate and migrate into the pyramidal, SO and SR layers that surround CA1 pyramidal neurons, compared to WT mice ( Fig. 1A-C). Iba1 and GFAP covered area in hippocampal brain regions were significantly increased in 8-monthold mice (Fig. 1D-I), so we next analyzed the distribution of Iba1 and GFAP-positive glial cells across the www.nature.com/scientificreports/ CA1 pyramidal neurons, SR and SO layers of the hippocampus in 8-month-old WT and CaMKIIα-iCre mice. The most significant increase in GFAP and Iba1 was in the SR layers surrounding CA1 pyramidal neurons in CaMKIIα-iCre mice (Fig. S1B,C). Significant increases were also observed in the SO (Fig. S1D,E) and pyramidal layers (Fig. S1F,G). Together, elevated levels of Iba1, CD68, GFAP and C3 indicated that microglia and astrocytes were in a persistent reactive, activated state in CaMKIIα-iCre mice. These results reveal increased expression of CD68 that is highly linked to microglial activation and increased expression of C3, suggesting an upregulation of reactive astrocytes. These changes occurred in the pyramidal, SR and SO layers of the hippocampus and could subsequently result in the adaptation of a neurotoxic hippocampal phenotype that disrupts synaptic processes within the brain. www.nature.com/scientificreports/ Activation and proliferation of microglia and astrocytes in the cortex of female CaMKIIα-iCre mice. The CaMKIIα promoter is expressed at high levels by pyramidal neurons of both the hippocampus and the cerebral cortex 11 , so we next analyzed whether the expression of iCre in pyramidal neurons of the cortex was associated with increased Iba1 and GFAP. Immunofluorescence microscopy revealed no differences in Iba1 and GFAP covered area in cortical brain regions of 8-month-old CaMKIIα-iCre mice compared to WT mice when male and female mice were analyzed together ( Fig. 2A-D). However, when quantifications were performed separately, there was a trend toward an increase in Iba1 (Fig. 2E) and GFAP (Fig. 2F) covered area in female CaMKIIα-iCre mice and a decrease in Iba1 (Fig. 2G) and GFAP (Fig. 2H) covered area in male CaMKIIα-iCre mice compared to WT control mice. Immunoblot analyses revealed similar results; there were no changes in Iba1 and GFAP in cortical brain homogenates when male and female CaMKIIα-iCre and WT mice were analyzed together (Fig. 2I-K), yet significantly increased Iba1 and GFAP was observed in cortical homogenates of female (Fig. 2L,M), but not male (Fig. 2N,O), CaMKIIα-iCre mice compared to WT mice.
To determine sex differences in Iba1 and GFAP in the hippocampus, we analyzed hippocampal GFAP (Fig. 1D,H) and Iba1 (Fig. 1E,H) separately for male and female 8-month-old CaMKIIα-iCre mice and WT mice.
Here, there was a trend toward an increase in GFAP (Fig. S2A,C) and a significant increase in Iba1 (Fig. S2B,D) in male CaMKIIα-iCre mice compared to WT mice. Female CaMKIIα-iCre mice showed significant increases in GFAP (Fig. S2E,G) and Iba1 (Fig. S2F,H) in CaMKIIα-iCre mice in hippocampal brain regions compared to WT control mice. Our results indicate that physiological sex differences in rodent microglia and astrocytes [28][29][30] residing in cortical brain regions caused the differential results in male and female CaMKIIα-iCre.
CaMKIIα-iCre mice show synaptic defects in the SO and SR layers of the hippocampus. Microglia and astrocytes regulate the excitatory activity of neural circuits in the hippocampus, during both development and in the adult brain, by performing activity-dependent synaptic pruning [31][32][33] . In CaMKIIα-iCre mice, microglia and astrocytes migrated to the pyramidal, SR and SO layers (Fig. 1A, Fig. S1B-G) that are enriched with synaptic connections of the hippocampus. Therefore, we hypothesized that activated microglia and astrocytes may engulf excitatory synapses and alter synaptic function in the CaMKIIα-iCre hippocampus.
To investigate this hypothesis, we first tested expression levels of post-synaptic proteins enriched in the SR and SO layers in hippocampal brain homogenates of 4 and 8-month-old CaMKIIα-iCre and WT mice. Ionotropic glutamate receptors AMPAR and NMDAR, postsynaptic density protein 95 (PSD95), CaMKIIα and phosphorylated CaMKIIα (Thr286) were measured by immunoblot analyses (Fig. 3A). No significant differences in PSD95, AMPAR or NMDAR were detected between 4-month-old WT and CaMKIIα-iCre mice (Fig. 3B-D), whereas in 8-month-old CaMKIIα-iCre mice, the expression levels of PSD95 and AMPAR were significantly decreased, while NMDAR was increased (Fig. 3B-D). The total level of CaMKIIα in hippocampal brain homogenates of 4 or 8 month-old CaMKIIα-iCre mice compared to WT mice was unchanged (Fig. 3E), yet the phosphorylation of CaMKIIα at Thr286 was markedly increased in CaMKIIα-iCre mice compared to WT mice (Fig. 3F). To determine whether the changes in synaptic proteins were associated with activated microglia and astrocytes in CaMKIIα-iCre mice, immunofluorescence microscopy was performed for Iba1, GFAP and PSD95 in 8-month-old mice (Fig. 3G). PSD95 staining in dendrites of CA1 neurons projecting into SR hippocampal layers appeared to be reduced and disorganized in CaMKIIα-iCre mice, and an increase in the co-localization of PSD95 and Iba1 was observed, suggesting the possibility of enhanced engulfment of synapses by microglia in CaMKIIα-iCre mice (Fig. 3G).
Presynaptic terminal proteins synaptophysin and synaptoporin are concentrated in the mossy fiber pathway of the hippocampus and were analyzed by immunoblot in 4 and 8-month-old hippocampal homogenates of CaMKIIα-iCre and WT mice (Fig. 4A). Here, no significant differences were observed (Fig. 4B,C), although 8-month-old male CaMKIIα-iCre mice showed a trend towards an increase in synaptoporin levels, and a decrease in synaptophysin levels, compared to WT mice ( Fig. S2I-L). Immunofluorescence imaging for synaptoporin and NeuN confirmed the trending increase in synaptoporin in 8-month-old CaMKIIα-iCre mice (Fig. 4D,E). These results suggest synaptic alterations in the hippocampal regions where activated microglia and astrocytes are found to associate with post synaptic regions that express iCre, such as in the SR and SO.
Hippocampal volume is reduced in aged CaMKIIα-iCre mice. The levels of the post-synaptic proteins in hippocampal homogenates of WT and CaMKIIα-iCre mice were comparable at 4-months, but altered at 8-months, suggesting that synaptic defects in CaMKIIα-iCre mice were caused by the chronic activation of microglia and astrocytes (Fig. 3B-D). To investigate whether hippocampal atrophy followed synaptic defects in CaMKIIα-iCre mice, hippocampus volume was first assessed in 8 and 22-month-old WT and CaMKIIα-iCre mice. The length of the intra pyramidal bundle (IPB) and NeuN immunostaining were also quantified to assess mossy fiber axon organization and neurodegeneration, respectively. IPB length, NeuN covered area and hippocampus volume were unchanged between 8-month-old CaMKIIα-iCre mice and WT mice (Fig. 5A-C), indicating that glial activation and synaptic alterations did not cause hippocampal atrophy in the CaMKIIα-iCre hippocampus at 8-months.
However, when hippocampus volume was measured in CaMKIIα-iCre mice at 22-months, a significant reduction was observed compared to WT mice (Fig. 5D,E). NeuN immunostaining was also significantly reduced in 22-month-old CaMKIIα-iCre mice (Fig. 5F). To determine whether elevated Iba1 and GFAP persisted in 22-month-old CaMKIIα-iCre mice, we measured Iba1 and GFAP covered area in hippocampal brain regions by immunostaining. A significant increase in Iba1 was observed (Fig. 5G), although GFAP levels were similar between WT and CaMKIIα-iCre mice (Fig. 5H). GFAP covered area increased with age in WT mice (13.75% GFAP covered area at 8 months (Fig. 1B) versus 16.01% GFAP covered area at 22 months (Fig. 5H)), whereas GFAP covered area was unchanged with age in CaMKIIα-iCre mice (Figs. 1B, 5H), which may explain the www.nature.com/scientificreports/ www.nature.com/scientificreports/ www.nature.com/scientificreports/ equivalent GFAP levels in WT and CaMKIIα-iCre mice at 22 months. Finally, presynaptic proteins synaptoporin and synaptophysin were unchanged (Fig. 5I-K), while PSD-95 and AMPAR decreased in 22-month-old mice (Fig. 5I,L,M). Altogether, our data suggest that reactive microglia and astrocytes proliferate to the SR and SO hippocampal layers and lead to abnormal expression of postsynaptic proteins, resulting in hippocampal atrophy in aged CaMKIIα-iCre mice.

Glial activation in hippocampal and cortical brain regions of nestin-Cre mice.
To determine whether the increase in activated microglia and astrocytes and synaptic defects observed in the CaMKIIα-iCre mouse line were induced by Cre recombinase itself, and thus common to various Cre mouse strains, we evaluated microglia, astrocytes, and synaptic proteins in the brains of additional Cre-driver mice that express Cre in the hippocampus and cortex. The original CaMKIIα-Cre T29-1 strain 10 that expresses prokaryotic Cre recombinase in the forebrain at a lower level than in the CaMKIIα-iCre mouse 16 and nestin-Cre mice 21 that express high levels of Cre in CNS neuronal precursors were chosen for analysis. 4-month-old CaMKIIα-Cre T29-1 and nestin-Cre mice were first analyzed by immunofluorescence imaging and immunoblotting for Iba1 and GFAP in hippocampal and cortical brain regions. Notably, the distinct proliferation of GFAP and Iba1 to the SR and SO regions of the hippocampus in CaMKIIα-iCre mice (Fig. 1A) was absent in the CaMKIIα-Cre T29-1 strain (Fig. 6A). The localization of Iba1-positive microglia and GFAP-positive astrocytes in hippocampal brain regions also appeared to be similar between nestin-Cre and WT control mice (Fig. 6A). Immunoblot analyses confirmed no differences in the levels of GFAP or Iba1 in hippocampal brain regions of CaMKIIα-Cre T29-1 mice compared to WT mice (Fig. 6B-D). A significant increase in GFAP in the hippocampus was found in nestin-Cre mice (Fig. 6C). www.nature.com/scientificreports/ We next analyzed Iba1 and GFAP expression in cortical brain regions of CaMKIIα-Cre T29-1 and nestin-Cre mice. No changes in Iba1 or GFAP distribution or total covered area in CaMKIIα-Cre T29-1 mice compared to WT controls were observed by immunofluorescence microscopy (Fig. 6E). However, GFAP expression was markedly increased in cortical brain regions in nestin-Cre mice compared to WT or CaMKIIα-Cre T29-1 mice (Fig. 6E). There was also a trend towards an increase in Iba1 in cortical brain regions of nestin-Cre mice, compared to WT or CaMKIIα-Cre T29-1 mice (Fig. 6E,H). The elevations of GFAP and Iba1in the cortex of nestin-Cre mice were confirmed by immunoblot analyses (Fig. 6F-H). www.nature.com/scientificreports/ Synaptic protein changes in the hippocampus and cortex of nestin-Cre mice. We found changes in postsynaptic proteins accompanying microgliosis and astrocytosis in the hippocampus in CaMKIIα-iCre mice. Therefore, we next checked the cortex and hippocampus of nestin-Cre and CaMKIIα-Cre T29-1 mice for expression levels of postsynaptic markers. PSD-95, AMPAR, NMDAR and phosphorylated CaMKII were affected in CaMKIIα-iCre mice (Fig. 3) although unaffected in WT or CaMKIIα-Cre T29-1 mice in the hippocampus (Fig. 7A-E), which is consistent with the comparable expression of GFAP and Iba1 in hippocampal brain regions of CaMKIIα-Cre T29-1 mice and WT mice (Fig. 6). By contrast, Nestin-Cre mice presented with dramatic increases in AMPAR (Fig. 7B), NMDAR (Fig. 7C), and trends toward an increase phosphorylated CaMKII (Fig. 7D) and PSD95 (Fig. 7E). In cortical brain regions, immunoblot analyses showed trends toward a decline in AMPAR and NMDAR expression in CaMKIIα-Cre T29-1 mice compared to WT mice (Fig. 7F-H), while PSD95 and phosphorylated CaMKII levels remained stable (Fig. 7F,I,J). No significant differences in NMDAR, phosphorylated CaMKII and PSD95 were observed in cortical brain regions of nestin-Cre mice, while a trend toward an increase in AMPAR was observed. These results suggest the possibilities that activation of astrocytes had a more robust impact on hippocampal synaptic proteins compared to cortical proteins. It is also possible that changes in synaptic proteins may occur in a manner independent of glial activation in the nestin-Cre mouse.

Discussion
The results in our study establish additional undesirable glial and synaptic changes in mice expressing Cre under control of the CNS promoters CaMKIIα and nestin. In CaMKIIα-iCre mice increased microglial activation as well as heightened astrocytic reactivity occurred in pyramidal, SR and SO hippocampal regions of male and female mice, yet only presented in the cortex of female CaMKIIα-iCre mice. The highly activated state of microglia and astrocytes appeared to reside within layers of the hippocampus that comprise significant populations of interneurons that control excitatory synaptic activity of neural circuits 33 . We speculate the homeostatic state within neural circuits of the hippocampus is disrupted within CaMKIIα-iCre mouse brains. While no changes in presynaptic protein expression of synaptophysin and synaptoporin were observed, alterations in post-synaptic proteins NMDAR, phosphorylated CaMKIIα, PSD95 and AMPAR were likely induced by the long-term state of neuroinflammation characterized by active and reactive microglia and astrocytes. We hypothesize that this chronic inflammation in the hippocampus eventually led to hippocampal atrophy in aged CaMKIIα-iCre mice. We sought to identify whether neuroinflammation and synaptic changes were limited to CaMKIIα-iCre mice by evaluating the brains of additional common Cre-driver mouse models. We assessed the original CaMKIIα-Cre T29-1 strain 10 and the nestin-Cre mouse model 21 . We chose to assess these two strains because they are widely used and express the prokaryotic version of Cre, thus allowing us to investigate whether neuroinflammation and synaptic changes in the CaMKIIα-iCre mice were caused by features unique to iCre. The nestin-Cre model expresses Cre within neural progenitors, while the CaMKIIα-Cre T29-1 strain expresses Cre in forebrain excitatory neurons. Unexpectedly, nestin-Cre mice showed enhanced astrocytic infiltration into cortical and hippocampal brain regions compared to WT controls and we found significant changes in hippocampal AMPARs and NMDARs. In contrast, the levels of GFAP and Iba1 were similar in hippocampal and cortical brain regions of CaMKIIα-Cre T29-1 and WT mice. The mechanism underlying glial activation in the nestin-Cre and CaMKIIα-iCre mice is unknown but may possibly cause downstream synaptic changes and neurodegeneration. Together, our results have broad implications on the interpretation of results derived from studies using CNS Cre-driver mice for gene manipulation, especially those focused on glial or synaptic processes in nestin-Cre and CaMKIIα-iCre mice.
Interestingly, we noted the location of microglia and astrocytes surrounding the synapses of neural circuits in the hippocampus of CaMKIIα-iCre mice, as well as changes in synaptic proteins that are involved in glutamatergic excitatory neurotransmission. Considering the co-localization of postsynaptic protein PSD95 within microglial soma, we hypothesize that activated microglia alter excitatory synapses in the hippocampus, ultimately impeding signal transduction within synapses. This process is mediated by NMDARs that control calcium influx into the post synapse and regulate the postsynaptic targeting of AMPARs. CaMKIIα is regulated by calcium levels and is highly enriched in postsynaptic densities of neuronal synapses and is a key protein involved in regulating synaptic physiology of the hippocampus 34 . The activity of CaMKIIα is controlled by its autophosphorylation at Thr286 in response to prolonged elevation of intracellular calcium levels, which is critical for LTP and memory in mice 35,36 . We observed concomitant changes in AMPARs and NMDARs, and an increase in CaMKIIα phosphorylation in CaMKIIα-iCre mice, so it is possible that constitutive activity of CaMKIIα due to increased autophosphorylation could be a critical event in driving the association of microglia and astrocytes with postsynaptic regions. Signal transduction within synapses of the hippocampus is critical for LTP and LTD and is required for synaptic plasticity and memory storage in the brain, so it is possible that these changes led to hippocampal-dependent behavioral and electrophysiological alterations in CaMKIIα-iCre mice.
We observed neuroinflammatory and neurodegenerative changes in CaMKIIα-iCre mice that express iCre and not in CaMKIIα-Cre T29-1 mice that express prokaryotic Cre. iCre expresses at a higher level than prokaryotic Cre when equal amounts of expression vectors are used 16 , and in nestin-Cre mice that express Cre in all neural progenitors, GFAP and Iba1 increased in Cre expressing brain areas. These observations support the hypothesis that high Cre expression causes neuroinflammation in CaMKIIα-iCre mice and in nestin-Cre mice. It is possible that neuroinflammation may be caused by apoptosis of Cre-expressing neurons, or by altered synaptic activity of Cre neurons leading to the activation of synaptic pruning by microglia. Cre has been previously known to have pathological effects; Cre decreases cell proliferation and increases cell apoptosis in vitro in cultured cell lines 24,37,38 and Cre-mediated damage to nestin-expressing neural progenitors results in microencephaly and hydrocephalus in vivo in nestin-Cre mice 20 www.nature.com/scientificreports/ www.nature.com/scientificreports/ activated microglia and astrocytes lead to synaptic alterations resulting in synapse loss and brain atrophy. The correlation between Cre levels, neuroinflammation and neurodegeneration is not fully understood and should be investigated in further studies. If Cre expression level led to neuroinflammation in nestin-Cre and CaMKIIα-iCre mice, it is possible that our findings could extend to other Cre-driver mice that express high levels of Cre driven by neuronal promoters. Furthermore, Cre expression driven by glial cell promoters may directly affect the activity of microglia and astrocytes. A limitation of this study is that we did not manipulate Cre expression in nestin-Cre and CaMKIIα-iCre mice, as this would likely clarify whether Cre-expression caused the phenotypes observed in our study. Future studies to address this could be performed by pharmacologically inhibiting Cre, comparing homozygous and heterozygous genotypes, or using Cre inducible tamoxifen-treated mice. Moreover, we evaluated CaMKIIα-iCre mice 4, 8 and 22-months-old, while nestin-Cre and CaMKIIα-Cre T29-1 mice were only examined at 4-monthsold. Neuroinflammation in CaMKIIα-iCre mice occurred at 4-months-old, although synaptic changes and hippocampal atrophy did not present until mice were 8 months old and 22-months-old, respectively. It is possible that aging may increase the level of Cre and lead to the more severe phenotypes observed in 8 and 22-month-old CaMKIIα-iCre mice, or that chronic neuroinflammation with age leads to downstream synaptic changes and neuronal loss. Future studies utilizing various strains of heterozygous and homozygous CNS Cre-driver mice should be performed in young and old mice to address these questions.
Other indirect mechanisms that are specific to nestin-Cre and CaMKIIα-iCre mice, and not associated directly with Cre expression, may possibly explain our findings. For instance, Cre transgenic mouse models are typically created by pronuclear microinjection of Cre transgenes, wherein the genomic insertion site of the Cre transgene is random and could interfere with endogenous gene expression or lead to multiple copies of Cre within the same chromosome 40 . In addition, Cre expression has caused unexpected results in Cre-driver mice by performing illegitimate recombination of genes through pseudo lox-P sites 41 . In a recent study examining the overexpression of synaptotagmin-2 in CaMKIIα-iCre, the BAC transgene insertion was hypothesized to disrupt the suppression machinery of the syt2 locus, leading to ectopic overexpression 25 . Therefore, it is conceivable that the expression of endogenous genes was disrupted, either by illegitimate recombination or the random insertion of the Cre transgene. Disruption of loci involved in glial or synaptic function by Cre transgene insertion could potentially lead to the neuroinflammation seen in nestin-Cre and CaMKIIα-iCre mice.
The methods used to create the nestin-Cre and CaMKIIα-iCre mouse models could have also caused the unexpected neuroinflammatory and synaptic alterations in these mice. One key difference between the CaMKIIα-iCre mouse model and other CaMKIIα Cre mice, including the CaMKIIα-Cre T29-1 strain, is the construction and expression of the Cre transgene in a large 170 kb BAC as opposed to use of a traditional small promoter-based strategy. The BAC is superior in that it supports desired endogenous gene expression by containing all required regulatory elements 11 , though the unintentional overexpression of additional genes located in the arms of the BAC may warrant unwanted phenotypes in CaMKIIα-iCre mice. The CaMKIIα-iCre BAC contains genomic sequence 50 kb upstream and 100 kb downstream of the CaMKIIα gene 11 and based on the chromosomal regions surrounding the CaMKIIα locus (https:// www. ncbi. nlm. nih. gov/ gene/ 12322), the genes SLC6A7, ARSI and CDX1 were likely to be incorporated into the BAC and are overexpressed in CaMKIIα-iCre mice 25 . Although these are genes not directly involved with astrocytes or microglia, our results may be linked to the overexpression of SLC6A7 which is a member of the gamma-aminobutyric acid neurotransmitter gene family and encodes a highaffinity mammalian brain L-proline transporter protein 42 . Although reports regarding the function of SLC6A7 are limited, it is known to be expressed in forebrain synaptic terminals involved in glutamatergic pathways 43 , which could lead to inappropriate synaptic activity and activated glia-mediated synaptic pruning.
The nestin-Cre mouse has long been established as having serious adverse physiological problems, the most prominent being microencephaly and hydrocephalus, which have been attributed to Cre toxicity in neural progenitors 20,23,39 . Other physiological issues in nestin-Cre mice include behavioral abnormalities, hypopituitarism, decreased levels of growth hormone and reduced body weight 44,45 . These artifacts may be associated with endogenous gene disruption by random integration of the Cre transgene 23 . However, the construct used to generate the nestin-Cre mouse model has also been implicated 46 . The transgene carries the human growth hormone (hGH) minigene, as it was identified that the sequences incorporated were critical to efficiently express the Cre transgene under the nestin promoter 47 . The expression of hGH on the nestin-Cre transgene results in aberrant hGH expression in the hypothalamus, and subsequently may cause hypopituitarism, metabolic and behavioral phenotypes in nestin-Cre mice 46 . Importantly, hGH is known to regulate glial and neuronal cell differentiation in the CNS 48 . Glial cells express the hGH receptor (GHR) and in GHR deficient mice, astrocytes are smaller and fewer in number 49 . It is unknown whether hGH is overexpressed in the cortex or hippocampus of nestin-Cre mice, but it is possible that overexpression of hGH in nestin-Cre mice could have caused enhanced astrocytic activity. Therefore, while we hypothesize that high Cre levels induce neuroinflammation, additional genes contained on the transgenes in the nestin-Cre mouse and in the CaMKIIα-iCre mouse could be responsible for the results observed in our study.
The use of tissue and cell specific promoters to express Cre in transgenic mice has become a staple for studying the genetics of diseases in translational research. The knowledge gathered thus far would not have been conceivable without the availability of tools such as Cre recombinase to manipulate genetic expression in mice. As the use of Cre-driver mice continues, the possibility of arising phenotypes because of inappropriate genetic recombination, Cre toxicity, endogenous gene disruption and germline recombination must be taken into consideration. We identify and report inflammation and synaptic changes present within two widely used Cre models: nestin-Cre and CaMKIIα-iCre mice. The activation of microglia and astrocytes led to further detrimental effects on synapses and neurodegeneration in CaMKIIα-iCre mice. Although we did not analyze nestin-Cre brains at advanced ages for neurodegeneration, a similar cascade of events would likely follow enhanced glial activity in these mice. The CaMKIIα-Cre T29-1 strain did not exhibit any differences compared to WT mice in our study www.nature.com/scientificreports/ and may be a superior model for genetic manipulation in excitatory forebrain neurons than CaMKIIα-iCre mice, although a lower level of Cre expression may lead to incomplete gene knockout in these mice. Our study is limited by a low number CaMKIIα-Cre T29-1 mice and we only analyzed two additional Cre driver mouse models. Based on our findings, future analyses should be performed to determine whether neuroinflammation is common across Cre-driver mice expressing high levels of Cre in the CNS. Our results demonstrate aberrant neuroinflammation in two widely used Cre-drivers and have broad implications on studies employing Cre-drivers to investigate genes involved in disorders in which the immune system is implicated, such as Alzheimer's disease. Moreover, the alterations in PSDs, NMDARs and AMPARs that are fundamental to processes involving synaptic plasticity, learning and memory, indicate dysfunction at synapses and alterations in these processes in nestin-Cre and CaMKIIα-iCre mice. The results presented in this study raise uncertainties to other Cre expressing mouse lines, as it is likely that other CNS phenotypes could exist. For the past decade, pathological reports of Cre expression have called for the use of Cre-driver lines as controls, yet these mice are typically not included when designing breeding strategies. To clearly interpret results, future Cre-loxP studies must employ appropriate Cre expressing controls to account for any confounding phenotypes that could present because of transgenic Cre recombinase expression. Furthermore, a comprehensive characterization of any abnormal results in Cre-driver mice must be documented, and studies should include all breeding strategies to obtain genetically modified mice.

Data availability
The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.