GAPDH, β-Actin, and β-Tubulin Display Age-Dependent Protein Expression Changes in the Mouse Cortex during Development

Glyceraldehyde 3-phosphate dehydrogenase (GAPDH), β-actin, and β-tubulin are well-recognized housekeeping proteins that play essential roles in basic cellular functions [1]. GAPDH is a key glycolytic enzyme that facilitates the production of cellular energy [2], while β-actin and β-tubulin are structural components of the cytoskeleton [3, 4]. GAPDH, β-actin, and β-tubulin are ubiquitously expressed in all neural cell types and brain regions [5], and their high abundance and presumed stable expression have made them popular reference controls for data normalization in neuroscience protein expression studies [6]. Besides their well-established functions as housekeeping proteins, emerging studies have demonstrated critical roles for these proteins in developmental and pathological processes [7‒11]. However, outside their use as reference controls, few studies have examined how the expression patterns of these proteins change throughout mammalian brain development to adulthood.

GAPDH is an enzyme that catalyzes the conversion of glyceraldehyde-3-phosphate to 1,3-bisphosphoglycerate to facilitate the production of ATP, the fundamental energy source for cellular activity [2]. Other functions of GAPDH have also been reported, including DNA repair, RNA export, membrane fusion and transport, cytoskeletal dynamics, and cell death [12]. Research has implicated the involvement of GAPDH-mediated cell death pathways in neurodegenerative disorders, such as Alzheimer’s disease [13‒16], Parkinson’s disease [17], and Huntington’s disease [18‒20], but the precise mechanisms are not well understood. β-Actin forms the cytoskeletal microfilaments in neurons and glia and serves various functions in cell motility, maintenance of cell shape and polarity, and cell division [3]. β-Tubulin makes up the highly dynamic microtubules that facilitate cell movement and shape and serves as tracks for vesicle and organelle transport [4]. Both β-actin and β-tubulin play critical functions in numerous aspects of brain development, including cell division, neuronal migration, axon specification and outgrowth, dendritogenesis, and spine formation [9, 10, 21‒23]. Mutations in genes encoding β-actin (ACTB) are associated with brain malformations and neuronal migration defects, developmental delay, intellectual disability, seizures, and deafness [24, 25]. Similarly, mutations in genes encoding for various isotypes of β-tubulin (e.g., TUBB2A, TUBB2B, TUBB3) cause a spectrum of complex brain malformations (termed “tubulinopathies”) characterized by abnormal neuronal migration, organization, and axon tract formation, intellectual disability, and seizures [10, 26‒30]. Considering the dynamic structural and functional changes that occur during brain development and the roles of GAPDH, β-actin, and β-tubulin in related biological processes, here, we examined whether the protein levels of GAPDH, β-actin, and β-tubulin change in mouse cortex from embryonic and early postnatal ages to adulthood using western blot analysis with a total protein normalization approach.

All animal procedures were approved by The University of Texas at Dallas Institutional Animal Care and Use Committee (IACUC) and performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals and ARRIVE guidelines. All efforts were made to minimize suffering. Animals were housed in light-, temperature-, and humidity-controlled rooms on a 12:12 light/dark cycle.

Cortical tissue was collected from male and female CD-1 mice (Charles River Laboratories, RRID:IMSR_CRL:022) for two sample sets: embryonic and postnatal. For the embryonic set, tissue was collected on embryonic day (E) 15, E17, and postnatal day (P) 0. For the postnatal sample set, tissue was collected from littermate animals at P0, P5, P10, P15, P20, and 2.5-month-old adult age. The P0 samples for the two sets were from different cohorts of animals. A total of eight different litters from eight time-pregnant dams were used for this study. The numbers of animals for each age-group and the litter they came from are E15: 7 (litter #1), E17: 7 (litter #2), P0: 7 (litter #3), P0–P20: 7–8 (litters #4–6, 2–3 animals from each litter in each age-group), adult: 8 (litters #7–8, 4 animals from each litter).

For embryonic tissue collection, time-pregnant mice were deeply anesthetized with isoflurane and decapitated. Embryos were removed from the abdominal cavity and whole cortical hemispheres were rapidly dissected in ice-cold phosphate-buffered saline (PBS; 137 mm NaCl, 2.7 mm KCl, 10 mm Na2HPO4, 1.8 mm KH2PO4, pH 7.4) under a stereomicroscope (Motic SMZ 171). For postnatal tissue collection, mice were anesthetized with isoflurane, decapitated, and whole cortical hemispheres were dissected as described above. Isolated tissues were immediately flash frozen in dry ice and stored at −80°C until used. Each sample consists of one whole cortical hemisphere from an individual animal.

Whole cortical hemispheres were homogenized by passing the tissue through a 27-gauge needle attached to a 1 mL syringe 8–10 times in ice-cold homogenization buffer (0.32 mm sucrose, 1 mm EDTA, 5 mm HEPES, 100 mm Tris HCl, pH 7.4) containing protease and phosphatase inhibitor cocktail tablets (Roche). Samples were kept on ice throughout the entire procedure. The total protein concentration of the homogenates was determined using Pierce BCA Protein Assay (Thermo Scientific #23227) and NanoDrop One UV-Vis Spectrophotometer (Thermo Scientific). All homogenates were diluted in Laemmli buffer (4X: 0.25 m Tris HCl, pH 6.8, 6% SDS, 40% bromophenol blue, 200 mm dithiothreitol) and 100 mm Tris HCl, pH 7.4, to a final concentration of 1 µg/µL. Diluted samples were stored at −20°C until used.

Equal amounts of total protein from each sample (10 µg, corresponding to 10 µL of sample) were loaded onto 26-well Criterion TGX Stain-Free Precast Gels, 4–15% gradient (Bio-Rad #5678085), and resolved by SDS-PAGE for 35–40 min at 200 V. Following SDS-PAGE, the gels were activated and imaged by UV exposure for 5 min using the Stain-Free gel setting on the ChemiDoc Imaging System (Bio-Rad). Resolved proteins on the gels were then transferred onto low fluorescence PVDF membranes (Bio-Rad #1620264) using a tank transfer system (Bio-Rad) for 1 h at 100 V. The total protein on the membranes was then imaged using the Stain-Free blot setting on the ChemiDoc Imaging System. Following total protein imaging, membranes were incubated in blocking buffer (5% nonfat milk, 1 mm Na₃VO₄ in PBS + 0.1% Tween 20) for 1 h at room temperature. Membranes were then incubated in primary antibodies overnight at 4°C and in secondary antibodies for 1 h at room temperature. Membranes were washed 3 × 10 min in PBS + 0.1% Tween 20 after each antibody incubation step. Immunoreactive protein bands were developed using SuperSignal West Pico PLUS or SuperSignal West Atto chemiluminescence reagents (Thermo Scientific), as appropriate, and imaged on the ChemiDoc Imaging System. Each target protein was probed on separate membranes. An overview of the Stain-Free total protein detection and western blot workflow is summarized in Figure 1.

All primary and secondary antibodies were from Cell Signaling Technology. All primary antibodies were rabbit monoclonal antibodies used at 1:2000 dilution: GAPDH (#5174, RRID:AB_10622025), β-actin (#4970, RRID:AB_2223172), β-tubulin (#2146, RRID:AB_2210545), and Akt (#4685, RRID:AB_2225340). The HRP-linked anti-rabbit IgG secondary antibody (#7074, RRID:AB_2099233) was used at 1:5000 dilution.

Optical densities (ODs) of the total protein and target protein bands were measured using ImageJ (NIH, RRID:SCR_003070). The ODs of the total protein were obtained by measuring a region of interest (ROI) spanning 24–225 kDa for each sample lane (Fig. 2a). The ODs of the target protein bands were obtained by measuring an ROI surrounding the band at the correct molecular weight in each lane and subtracted from the average background signal (Fig. 2a). ROIs were kept identical in size across lanes on the same membranes. Analyses of the coefficient of variation (CV, where CV = standard deviation/mean) were performed using relative OD values obtained by dividing raw values by the P0 mean for each blot. Relative protein levels were calculated by dividing the OD of the target protein bands by the OD of the total protein within the same lane and subsequently dividing by the P0 mean for each blot. Data are shown as a percentage of the P0 mean (% P0).

All statistical analyses were performed using Prism 10 (GraphPad, RRID:SCR_002798). CVs of the relative OD values from all samples in a given total or target protein assessment were calculated using the descriptive statistics function. Differences in relative protein levels between age-groups were assessed using one-way ANOVA with Tukey’s post hoc test. The significance level was set at p < 0.05. Data are presented as mean ± standard error of the mean.

Publicly available RNAs-seq datasets from two mouse development series were downloaded from the ENCODE data portal (https://www.encodeproject.org/) [31‒33]. The datasets are published by the ENCODE Consortium and were produced by the laboratories of Dr. Barbara Wold (Caltech) and Dr. J. Michael Cherry (Stanford). The embryonic development series (accession number: ENCSR443OEA) consists of polyA plus RNA-seq data from E10.5 to P0 B6NCrl mouse forebrain tissue (one animal per timepoint). The postnatal development series (accession number: ENCSR561OFW and ENCSR631HFW) consists of total RNA-seq data from P4 to 20-month-old B6CASTF1/J mouse left cortical tissue (one male and one female per timepoint, respectively). The accession numbers for the specific datasets and timepoints used in this study are ENCSR185LWM (E14.5), ENCSR080EVZ (E16.5), ENCSR362AIZ (P0), ENCSR841KGE (P4, male), ENCSR757VTG (P4, female), ENCSR516DAX (P10, male), ENCSR321WYK (P10, female), ENCSR702MVT (P14, male), ENCSR774DTO (P14, female), ENCSR764GMN (P25, male), ENCSR248KDJ (P25, female), ENCSR496PRU (adult, male), and ENCSR219ZXZ (adult, female). The transcript per million (TPM) values for GAPDH (Gapdh), β-actin (Actb), and eight β-tubulin isotypes (Tubb1, Tubb2a, Tubb2b, Tubb3, Tubb4a, Tubb4b, Tubb5, and Tubb6) were obtained at E14.5, E16.5, and P0 (embryonic) and P4, P10, P14, P25, and 2-month-old adult age (postnatal), and the averaged TPM values from 2 technical replicates per timepoint were calculated. The relative RNA expression changes are shown as log2 (fold change [FC] from P0) for the embryonic dataset and log2 (FC from P4) for the postnatal dataset. The threshold for expression changes based on FC alone was set at log2(FC) = 0.5, equivalent to a 1.4-fold increase or decrease in expression.

To quantify the cortical protein levels of GAPDH, β-actin, and β-tubulin across embryonic and postnatal ages, we performed western blot analysis with total protein normalization. In this quantification approach, the abundance of the target protein is normalized to the total amount of protein within the same western blot lane (i.e., within the same sample), thereby removing the need to rely on conventional housekeeping proteins as loading control [34‒36]. The total protein was detected by Stain-Free technology, wherein the polyacrylamide gel used to resolve proteins contains a trihalo compound that cross-links with tryptophan residues in the proteins upon UV activation to make them fluorescent [36]. The total protein can be visualized both in the gel and the membrane following transfer and prior to immunodetection (Fig. 1). We ran separate gels for the embryonic (E15–P0) and postnatal (P0–adulthood) sample sets and each target protein. Representative western blot images of GAPDH, β-actin, and β-tubulin bands and the corresponding total protein images for each sample set are shown in Figure 2b–d. Full-length versions of the blots are provided in online supplementary Figure 1 (for all online suppl. material, see https://doi.org/10.1159/000544064).

We first examined the expression stabilities of total and target proteins across samples using CV analysis of the protein band OD measurements. A CV of >10% was used as the cutoff criterion for stable expression [37]. The CVs of the total protein measurements were considerably low (<10%) in both the embryonic and postnatal sample sets, verifying consistent sample loading and stable protein expression across samples (Fig. 2e). In comparison, the CVs of GAPDH, β-actin, and β-tubulin measurements were higher (21–50%) in both the embryonic and postnatal sample sets, suggesting a larger expression variability of these proteins across samples (Fig. 2e). To determine whether age was a source of variation, we next normalized GAPDH, β-actin, and β-tubulin OD values to the respective total protein level in each sample and expressed the relative protein levels as % P0. No statistically significant changes in GAPDH, β-actin, and β-tubulin protein levels were observed between E15, E17, and P0 (Fig. 2f–h, GAPDH: F(2, 18) = 1.712, p = 0.2086; β-actin: F(2, 18) = 0.1699, p = 0.8451; β-tubulin: F(2, 18) = 1.249, p = 0.3105). However, there was a significant increase in GAPDH protein levels from P0 to adulthood (Fig. 2f, GAPDH: F(5, 39) = 12.12, p < 0.0001). Specifically, GAPDH levels were unchanged between P0 and P5 and began to increase at P10. By P15, the levels had increased by more than 100% and were statistically different from P0 and P5. GAPDH levels remained stable from P15 to adulthood and were significantly higher at P15–adulthood compared to P0 and P5. The doubling of GAPDH levels with age is consistent with the high CV values observed in Figure 2e. Contrary to GAPDH, β-actin and β-tubulin protein levels were significantly decreased from P0 to adulthood (Fig. 2g, h, β-actin: F(5, 39) = 31.53, p < 0.0001, β-tubulin: F(5, 39) = 10.92, p < 0.0001). Specifically, β-actin levels were unchanged between P0 and P5 and significantly declined by ∼30% at P10. The levels remained stable from P10 to adulthood and were significantly lower at P10–adulthood compared to P0 and P5. β-Tubulin levels were unchanged between P0 and P5 and began to decline at P10. By P15, the levels had decreased by ∼40% and were statistically different from P0 and P5. The levels remained stable from P15 to adulthood and were significantly lower at P15–adulthood compared to P0 and P5. Taken together, these data demonstrate an age-dependent change in mouse cortical GAPDH, β-actin, and β-tubulin protein levels that occur around the second week of life, at P10–15.

To examine whether the observed age-dependent protein level changes correlate with RNA transcript level changes, we screened published mouse RNA-seq datasets by the ENCODE Consortium [31‒33]. We obtained gene expression data for GAPDH (Gapdh), β-actin (Actb), and eight isotype-specific β-tubulin (Tubb1, Tubb2a, Tubb2b, Tubb3, Tubb4a, Tubb4b, Tubb5, Tubb6) from an embryonic development series, which consists of polyA plus RNA-seq data from embryonic forebrain tissue, and a postnatal development series, which consists of total RNA-seq data from postnatal cortical tissue (Fig. 3a). All transcripts, except for Tubb1 (<1 TPM), were expressed in the embryonic forebrain and postnatal cortical tissues (Fig. 3b, c). Of note, Tubb6 was expressed at very low levels and was undetectable in the cortex at P25 and adult age, consistent with previous reports in whole brain tissue [38] (Fig. 3b, c, insets). While statistical analysis was not possible due to sample size limitations, the overall transcript expression pattern of the analyzed genes reflected that of the protein data. In particular, no Gapdh or Actb transcript changes were observed in the embryonic forebrain tissue between E14.5, E16.5, and P0 (Fig. 3d). However, in the postnatal cortical tissue, Gapdh levels increased 1.4-fold [log2(FC) = 0.50] between P4 and P10 and 1.6-fold [log2(FC) = 0.71] between P4 and adulthood, while Actb levels decreased 1.6-fold [log2(FC) = −0.66] between P4 and P10 and 2.7-fold [log2(FC) = −1.44] between P4 and adulthood (Fig. 3e). The embryonic expression profiles of the β-tubulin isotypes were variable. No changes in Tubb3, Tubb4b, and Tubb5 levels were observed in the embryonic tissue. In contrast, Tubb2a, Tubb2b, and Tubb4a were 1.5- to 2.3-fold lower [log2(FC) = −0.62 to −1.19], whereas Tubb6 was 1.8-fold higher [log2(FC) = 0.81], at E14.5 compared to P0 (Fig. 3d). Most of the β-tubulin isotypes decreased with age in the postnatal cortical tissue, with Tubb2a, Tubb2b, Tubb3, Tubb5, and Tubb6 levels being 4- to 27-fold lower [log2(FC) = −2.01 to −4.76] in adults compared to P4 (Fig. 3e). Tubb6, the lowest expressed transcript, decreased to undetectable levels (<1 TPM) by P25 (Fig. 3c, inset). Tubb4a displayed a 1.5-fold [log2(FC) = 0.62] increase whereas Tubb4b levels did not change between P4 and adulthood (Fig. 3e). Overall, these findings support a high correlation between GAPDH, β-actin, and β-tubulin transcript and protein levels in the developing mouse cortex.

As housekeeping proteins, GAPDH, β-actin, and β-tubulin are frequently used as internal loading controls for data normalization in western blot analysis. While these proteins can serve as useful references, numerous studies have emphasized the need to validate the expression stability of these proteins for the experimental conditions and tissues of interest to avoid skewed results [6, 35, 39]. Our findings that GAPDH, β-actin, and β-tubulin display age-dependent expression changes in the cortex between P0 and adulthood suggest that they are not suitable protein loading controls for cortical studies spanning this time frame. To test how their use as loading controls would impact the data, we probed for Akt using the same blots previously used to analyze postnatal β-actin levels (in Fig. 2c, e, g) and normalized the Akt protein bands to either total protein or β-actin within the same samples (Fig. 4a). Akt is a well-described protein kinase with numerous cellular functions and high expression in the brain [40‒42]. Normalizing Akt to total protein levels revealed no changes in Akt levels between P0 and P15 and a significant reduction (∼25%) in Akt levels in the adult group compared to P0–P15 (Fig. 4b, Akt/total protein: F(5, 39) = 8.929, p < 0.0001). When Akt was normalized to β-actin, a different pattern was observed: no changes in Akt levels were found between P0 and P5; however, a significant increase (∼25–40%) in Akt levels was observed at P10–adulthood compared to P0 and P5 (Fig. 4c, Akt/β-actin: F(5, 39) = 13.43, p < 0.0001). These results demonstrate that β-actin is not a reliable western blot loading control for studies comparing cortical protein levels between P0 and adulthood and suggest the use of total protein levels for more accurate quantification.

In this study, we show that the protein levels of GAPDH, β-actin, and β-tubulin change in the mouse cortex from birth to adulthood. In particular, the levels of GAPDH increased whereas the levels of β-actin and β-tubulin decreased with age, notably occurring during the second week of postnatal life. Similar changes were also observed at the RNA transcript level. Overall, these findings reveal age-dependent protein level changes in GAPDH, β-actin, and β-tubulin in the mouse cortex during postnatal development and caution against the use of these proteins as reference controls for cortical protein expression studies spanning this timeframe.

Western blot analysis is the gold-standard approach for quantifying protein levels in cell and tissue samples, and housekeeping proteins are often used as internal controls to account for sample loading variations due to their presumed constant expression across conditions. However, numerous studies have shown that housekeeping proteins may be altered by developmental, pathological, and environmental processes and emphasized the need to validate their expression stability in the tissue and conditions of interest to avoid inaccurate quantification [6, 35, 39]. GAPDH, β-actin, and β-tubulin are among the most commonly used protein loading controls for western blot analysis in the neuroscience field [6]. Our findings show that GAPDH, β-actin, and β-tubulin levels change from P0 to adulthood and thus are not reliable loading controls for assessing cortical protein level changes across this time frame. As a proof of concept, we demonstrated that normalizing Akt to total protein and β-actin levels leads to substantially different results. When Akt was normalized to total protein levels, there was a significant reduction in the Akt levels from P0 to adulthood. However, since β-actin levels were lower in adults than at P0, normalizing Akt to β-actin erroneously resulted in higher Akt levels in adults. Normalizing Akt to β-tubulin would presumably lead to similar results as β-actin while normalizing Akt to GAPDH would exaggerate the decrease in adult Akt levels since GAPDH levels increase with age. Therefore, the use of these proteins as loading controls for comparing cortical protein levels between P0 and adulthood could lead to inaccurate results and is not recommended. Since cortical GAPDH and β-actin protein levels were unchanged between E15 and P0, these may be used as loading controls in studies spanning this period. β-Tubulin protein levels were also unchanged between E15 and P0, but given that the transcript levels of specific β-tubulin isotypes showed unique changes and the affinities for specific isotypes are unknown for the β-tubulin antibody used in this study, we interpret these results with caution. In all experiments, the CV values for total protein were lower, indicating higher expression stability across samples, and we recommend using total protein normalization as a more reliable method for data quantification.

Despite being a long-recognized protein, little is known about GAPDH functions and expression patterns during cortical development. Our findings showing an increase in GAPDH transcript and protein levels beginning around P10–P15 suggest that GAPDH expression is developmentally regulated. Given the central role of GAPDH in glycolysis, the increase in GAPDH may be attributed to a shift in energy requirements during these stages of development. Consistent with this notion, the rodent cortex undergoes large growth spurts during the second week of life, including rapid growth of dendritic arbors, peak oligodendrogenesis, and onset of myelination, all of which require increased metabolic demands [43‒46]. Contrary to GAPDH, the structural proteins β-actin and β-tubulin decreased with age, starting around P10. The transcript levels of β-actin and most β-tubulin isotypes, except for Tubb4a and Tubb4b, which were, respectively, increased and not changed, correlated with the overall protein expression profile in the postnatal period. These findings corroborate with previous northern blot studies showing downregulation of Actb, Tubb2, and Tubb3 mRNAs, and upregulation of Tubb4 mRNA, in the rodent brain from early postnatal development to adulthood [47‒49]. β-Actin and β-tubulin play important roles in the formation of the cortex, and these are likely upregulated early on to support these functions [9, 10]. Although both proteins decrease with age, they remain abundantly expressed, consistent with their housekeeping functions. Interestingly, no changes in cortical GAPDH and β-actin protein levels were found between E15 and P0, suggesting constant expression of these proteins in early development. The overall β-tubulin protein levels were also not changed between E15 and P0, but based on the transcript data, we expect isotype-specific β-tubulin proteins to be uniquely expressed during this period.

A limitation of the study is that we cannot define how the housekeeping protein expression pattern changes within specific cell types since whole cortical homogenates were used in the analysis. At E15, during mid-neurogenesis, the cells in the developing cortex are predominantly neuronal progenitor cells and neurons [50]. At P0 and during the early postnatal periods, the number of glial cells increases manyfold in the cortex, and the types and numbers of cells continue to change throughout postnatal brain development [50, 51]. Variations in cortical cell type composition from early development to adulthood may be a source of the observed protein level changes, and future single-cell proteomic studies could provide valuable insights into age-dependent, cell type-specific protein changes. In summary, our study reveals age-dependent changes in cortical GAPDH, β-actin, and β-tubulin protein levels during mouse postnatal cortical development and describes an unprecedented developmental expression pattern for these proteins that warrants further research into their roles in brain development and disease.

The animal study protocol was reviewed and approved by The University of Texas at Dallas Institutional Animal Care and Use Committee (Protocol #2023-0094).

The authors have no conflicts of interest to declare.

This study was not supported by any external sponsor or funder.

D.R.: investigation, formal analysis, visualization, writing – original draft, and writing – review and editing. M.N.: investigation and formal analysis. T.D. and D.P.: formal analysis. D.T.-F.: formal analysis and writing – review and editing. L.H.N.: conceptualization, investigation, formal analysis, visualization, supervision, writing – original draft, and writing – review and editing.

All data generated or analyzed during this study are included in this published article (and its online suppl. information files). Further inquiries can be directed to the corresponding author.