The CaMKK2 KO male mouse brain and spinal cords were provided as dissected snap-frozen tissue by Dr. Uma Sankar, Indiana University School of Medicine, USA. The CaMKK2 KO mouse was generated by targeted deletion of exons 2–4 flanking sequence (Anderson et al., 2008). The CaMKK2 KO mice used in this study were from C57BL/6J background and approximately 3–4 months old. The 6 and 14 months old female 3xTg-AD mice cerebral cortex, hippocampus and serum samples were provided by Dr. Benedict C Albensi, University of Manitoba, Canada. The 3xTg-AD is a triple-transgenic model of AD harboring PS1(M146V), APP(Swe) and tau (P301L) transgenes (Oddo et al., 2003). The postmortem human CSF and serum samples from AD patients and unaffected controls were obtained through National Institute of Health (NIH) NeuroBioBank, USA (Request number #937).

DRGs from adult male Sprague-Dawley rats were dissected and dissociated as described previously (Calcutt et al., 2017; Sabbir et al., 2018) and cultured in defined Hams F12 media containing 10mM D-glucose (N4888, Sigma, St Louis, MO, USA) supplemented with modified Bottenstein's N2 additives without insulin (0.1 mg/ml TF, 20 nM progesterone, 100 μM putrescine, 30 nM sodium selenite, 0.1 mg/ml BSA; all additives were from Sigma, St Louis, MO, USA) (Akude et al., 2011; Roy Chowdhury et al., 2012; Saleh et al., 2013; Calcutt et al., 2017). In all experiments, the media was also supplemented with 0.146 g/L L-glutamine, a low dose cocktail of neurotrophic factors (0.1 ng/ml NGF, 1.0 ng/ml GDNF and 1 ng/ml NT-3 – all from Promega, Madison, WI, USA), 0.1 nM insulin and 1X antibiotic antimycotic solution (A5955, Sigma).

The HEK293 cell line was a gift from the Dr. Asuka Inoue laboratory, University of Tohoku, Japan. The HepG2 cell line was a gift from Dr. Jeffrey Wigle laboratory, University of Manitoba, Canada. The cells were cultured in DMEM supplemented with heat-inactivated 5% FBS, 0.146 g/L L-glutamine and 1X antibiotic antimycotic solution (A5955, Sigma). The CaMKK2 KO cell lines were generated by transfecting HEK293 and HepG2 cells with CaMKK2 CRISPR/Cas9 KO plasmid (Sc-400928) and CaMKK2 homology-directed repair (HDR) plasmid (Sc-400928-HDR) using lipofectamine reagent (Table 1). The cells were selected with 2 μg/ml puromycin following 24 h of transfection and subsequently plated as single cell/well in 96 well plates and allowed to develop colonies which were then screened for the reporter red fluorescent protein (RFP) and CaMKK2 protein expression by confocal microscopic imaging and immunoblotting, respectively. The floxed-RFP reporter cassette in the HDR construct was excised by transiently expressed by Cre-recombinase vector (Sc-418923). The lists of CRISPR/Cas9 constructs are given in Table 1.

Knockdown of CaMKK2 in cultured DRG neurons was achieved by transfecting cells with lipid nanoparticles (LNP) encapsulated cocktail of 3 siRNAs specific to exon 5, 8, and 12 of CaMKK2 gene (Table 1) (Rungta et al., 2013). The siRNA-LNPs were prepared by mixing appropriate volumes of different cationic lipid stock solutions in ethanol with an aqueous phase containing siRNA multiplexes using a microfluidic micromixer by Precision NanoSystems Inc. For encapsulation, the desired amount of siRNAs (0.056 mg siRNA/ micromole of lipid) was dissolved in the formulation buffer (25 mmol/L sodium acetate, pH 4.0). Subsequently, 1 × volume of the lipid mixtures in ethanol and 3 × volumes of the siRNA in formulation buffer were combined in the microfluidic micromixer using a dual syringe pump to generate the LNPs. The LNP particles containing siRNA were added to the DRG culture and neurons were allowed to grow for 48 h, after which the proteins were analyzed. The LNP based delivery of siRNA was not cytotoxic and did not affect the viability of DRG neurons after 48–74 h of exposure as reported previously in the neuronal cells (Rungta et al., 2013).

IEF separates proteins based on the isoelectric point (pI) which depends on the net charge in the protein. During focusing, the proteins will migrate to the point on an immobilized pH gradient where the net charge of the protein is zero (Freeman and Hemby, 2004). The charge-separated proteins were further separated in the 2nd dimension SDS-PAGE based on their molecular weight. Total cellular proteins were precipitated and dissolved in the rehydration buffer containing 8 M Urea, 2% CHAPS, 50 mM dithiothreitol (DTT) and 0.2% Bio-Lyte ampholytes pH3-10 (Bio-Rad, Cat no. 1632094). The dissolved proteins were then incubated in IPG strips (ThermoFisher) for 1 h and focused at 175 volts (V) for 15 min, 175–2,000 V ramp for 45/109 min (depending on pH gradient) and 2,000 V for 30 min. In some experiments, IPG strips from GE Healthcare Life Sciences Immobiline DryStrip (pH 3–10 L) and Bio-Rad ReadyStrip IPG strips (pH 4-7L) were used and the proteins were focused using Bio-Rad Protean®i12 IEF system as per manufacturer's recommendations. After focusing, the proteins in the strips were reduced (by DTT), alkylated (by Iodoacetamide) and resolved by 2D SDS-PAGE. For protein profiling experiments, the gel was stained with colloidal Coomassie, imaged and the protein spots were compared (Dyballa and Metzger, 2009). For immunoblotting experiments, the gels were transferred to nitrocellulose membrane and immunoblotted using different antibodies (Table 1).

In-Gel digestions of the excised gel-spots were performed as follows. The spots were sliced, destained, dehydrated and dried. the dried gel slices were then rehydrated in 20 μl of 12 ng/μl Trypsin Gold (V5280, Promega) in 0.01% ProteaseMAX^TM Surfactant (Trypsin enhancer, V2071, Promega):50 mM NH₄HCO₃ for 10 min and then overlaid with 30 μl of 0.01% ProteaseMAX^TM Surfactant:50mM NH₄HCO₃, gently mixed and incubated overnight at 37°C on a horizontal shaker. The eluted peptides were cleaned by Pierce^TM C-18 tips (ThermoFisher, 87782) and analyzed by tandem mass spectrometry (MS) analysis using AB SCIEX TripleTOF™ 5600 System (Applied Biosystems/MDS Sciex, Foster City, CA) at Manitoba Centre for Proteomics and Systems Biology, University of Manitoba.

The HEK293 cells were transiently transfected with Flag-TF in serum-starved condition for 14 h. The cells were then crosslinked using 2% freshly prepared paraformaldehyde in phosphate-buffered Saline (PBS) for 10 min. The crosslinking was stopped with Tris and washed thrice in PBS for 10 min each. The protein lysates were prepared in PBS containing 1% IGPAL630 (Sigma), sonicated and reduced (DTT). The crosslinked samples were not boiled but the native proteins were boiled for 10 min in a water bath. The Flag-TF was immunoprecipitated by incubating the protein lysates with 8 μg anti-Flag antibody/50 ul Dynabeads protein-G (F3165: Sigma, 10003D: ThermoFisher Scientific) for overnight and washed thrice using PBS with 1% IGPAL630 for 10 min each. The proteins were denatured, reduced and separated by SDS-PAGE. The protein gels were stained with Oriole fluorescent stain (Bio-Rad, Cat no.1610496) or transferred to nitrocellulose membrane and immunoblotted.

In BN-PAGE/SDS-PAGE, the first dimension native page separates multiprotein complexes and the 2nd dimension denatured and reduced SDS-PAGE separates the interacting protein components in the MPC which appears on a vertical line (Sabbir et al., 2016). The first dimension BN-PAGE and the 2nd dimension SDS-PAGE was performed as described previously (Sabbir et al., 2016). Briefly, the cell lysates were prepared in 1X PBS supplemented with 1X Halt protease and phosphatase inhibitor cocktail (1861281, Thermo Scientific) and 1.5% n-Dodecyl β-D-maltoside (D4641, Sigma) and sonicated. The proteins were then separated in 4–15% gradient blue-native polyacrylamide gel. The gel strips (individual lanes) were carefully excised including the 3.2% stacking gel and immersed in the Laemmli sample buffer containing freshly prepared DTT (54 mg/ml). The gel slices were incubated in sample buffer for 30 min at 55°C temperature and then the proteins in the gel slices were separated in the 2nd dimension SDS-PAGE and immunoblotted.

The FITC-conjugated TF uptake assay was performed by overexpressing RFP tagged TFR plasmid (Table 1) in wild-type and CaMKK2 KO HEK293 cells. The cells were grown overnight in serum-free media and then incubated with 25 μM FITC-TF in serum-free media. Live confocal time-lapse images were taken using an environment controlled LSM510 confocal microscope. The cells were also pulse labeled with 25 μM FITC-TF for 1 h and then washed and overnight incubated in serum-free media and imaged. LSM510 confocal images acquisition parameters were the same for all images.

Wild-type and CaMKK2 KO (clone A5) cells were cultured for 48 h in serum-free media. The culture media was then replaced with a salt-glucose solution containing 114 mM NaCl, 0.22% NaHCO3, 5.29 mM KCl, with or without CaCl2.2H2O (with calcium: 2 mM CaCl2+1 mM BaCl2; without calcium: 3 mM BaCl2), 10 mM HEPES, 10 mM Glucose, 1 mM MgCl2 and supplemented with 5 μM Fluo-4AM dye (ThermoFisher, F14210). The cells were loaded with cell-permeant Fluo-4 for 15 min, washed and then time series images were captured at 30 s interval following 10 μM Muscarine (Sigma, M6532) treatment. Zeiss LSM410 confocal microscope with a controlled humidified atmosphere containing 5% CO2 at 37°C was used to capture time lapse images. The mean Fluo-4 intensity was calculated from time series images using ImageJ time-series analyzer plugin.

Relative quantification of proteins was done by SDS-PAGE separation of total proteins followed by transfer to the nitrocellulose membrane and immunoblotting based detection using HRP-conjugated secondary antibodies. The chemiluminescence was detected and imaged using ChemiDoc^TM imaging system and Image Lab software version 5.0 build-18 (BioRad). Table 1 summarizes all the primary antibodies and other reagents used in this study.

Statistical analysis was performed using Prism version 7.00 (GraphPad Software). The mean of more than 2 groups were compared using one-way ANOVA (randomized) followed by multiple comparison tests (Siegel, 1956; Dunn, 1964). The mean of multiple experimental groups were compared with the control group by Dunnett's post-hoc multiple comparison test, whereas, the mean between two experimental groups were compared by Sidak's post-hoc multiple comparison test (Dunn, 1964). Comparisons between two groups were performed using Student's t-test (unpaired). Differences were considered significant with P <0.05.

In order to understand the role of CaMKK2 in neurons, CaMKK2 was knocked down in cultured adult rat primary DRG neurons (Figure 1A) and total cellular proteins were fractionated by IEF/SDS-PAGE to detect and compare differentially charged protein fractions between control and KD samples (Figure 1B). Comparison of the focused proteins revealed the difference in the abundance of multiple ~75 kDa protein spots at pH~3-4 (Figure 1B, blue rectangles), excised and analyzed by in-gel tryptic digestion followed by mass spectrometric identification of the proteins (Figure 1C). Mass spectrometry identified the protein spots as TF and revealed multiple potential phosphorylated Ser-381/389/409/500/511/512, Thr-392/393/586, and Tyr-257/333/336/338 residues in the scrambled control sample which was absent in KD condition (Table 2, Figure 1E, and Supplementary Figure 1). The phosphorylated residues were plotted on TF with respect to the functional domains, TFR binding motif and iron binding sites (Supplementary Figure 1). The phosphorylated TF (P-TF) residues were not detected in the CaMKK2 KD neurons. Crystal structure of TF revealed that Ser381/389 and Thr392/393 residues overlapped with the TFR binding site (Figures 1E–G). In addition, immunoblotting based quantification revealed a significant reduction of total TF in the CaMKK2 KD neurons (Figure 1D). In some mass spectrometry spectrums, the fragmentation pattern could not resolve specific phosphorylated residues due to the presence of closely spaced Ser/Thr/Tyr residues and fragmentation between them did not occur (Table 2). Therefore, potential phosphorylated residues in the respective peptide fragment were separated by slash punctuation mark (Table 2). The mass spectrometry data is accessible in the Global Proteome machine (GPM) database (http://hs2.proteome.ca/tandem/thegpm_tandem.html) using respective GPM accession number (Table 2 and Figure 1C).

The molecular weight and pI of human/rat TF (SwissPort: P02787/P12346) is 77/76 kDa and 6.81/7.14, respectively. The defined neuronal culture media was supplemented with partially saturated (with iron) recombinant human TF. Therefore, cross-species mixture of TF was expected in the protein analysis. IEF/SDS-PAGE followed by immunoblotting using anti-TF antibody revealed that the charged fractions of TF differed between scrambled siRNA control and CaMKK2 KD DRG neurons (Figures 2A,B). TF appeared as 3 major charged fractions at pH~3-4, ~5-6, and ~9-10 (Figure 2B, red rectangles). Previous mass spectrometric analysis revealed that the pH~3-4 fractions corresponded to potential Ser/Tyr/Thr phosphorylated TF and therefore, hereafter designated as P-TF. Phosphorylated TF epitope specific antibody is not available, therefore, in the subsequent studies, the pH~3-4 fraction was quantified as a measure of P-TF. The relative amount of P-TF was significantly decreased in CaMKK2 KD neurons (Figures 2B,C, red arrow). This supported previous protein profiling based observation (Figure 1B, blue rectangles). The CaMKK2 KD DRG neurons exhibited the presence of additional ~130/>180 kDa focused spots in pH ~5-6 and pH~9-10 regions, whereas scrambled control cells exhibited >180 kDa spots in pH ~9-10 region only (Figure 2B, blue rectangles). The high molecular weight (HMW) forms may be due to PTMs that added additional molecular mass, for example, glycosylation. TF is known N-linked glycosylated at multiple residues (Satomi et al., 2004). Treatment with a deglycosylation mix (protein deglycosylation mix, NEB), containing all enzymes to remove N- and O-linked glycans caused the disappearance of the HMW TF indicating potential glycosylation (data not shown).

CRISPR/Cas9 based CaMKK2 KO HEK293 and HepG2 cell lines were generated to study the effect of CaMKK2 on TF phosphorylation. The HepG2 and HEK293 cell lines were derived from human hepatocellular carcinoma (liver in origin) (Aden et al., 1979) and embryonic kidney, respectively (Shaw et al., 2002). Human Protein Atlas (www.proteinatlas.org) based gene expression data revealed that HepG2 cells express TF but HEK293 cells do not express TF (Uhlen et al., 2005). Immunoblotting revealed that HepG2 strongly express TF whereas HEK293 cells do not express TF or expresses below the detection limit (Figures 2D, 3A). HEK293 and HepG2 cells expressed 2 isoforms (isoform 1 and 2) of CaMKK2 at ~60–75 kDa region and the relative expression of CaMKK2 differed between the cell lines (Figure 2D). In CaMKK2 gene ablation strategy, the CRISPR/Cas9 KO plasmids and CaMKK2 HDR plasmids were designed to cause targeted insertion of a reporter/selection cassette between exon 4 and 13 of the CaMKK2 gene which would effectively eliminate expression of all transcriptional isoforms. Immunoblotting revealed loss of expression of the CaMKK2 isoforms in 2 independently selected CaMKK2 KO HepG2 clonal cell lines (Figure 2D). Total TF level was significantly increased in the CaMKK2 KO HepG2 cell lines (Figures 2D,E). IEF/SDS-PAGE revealed 2 major charged fractions of TF in the pH~3-4 and ~5-6 regions in HepG2 cells (Figure 2F, red rectangles). The P-TF level was significantly decreased in the CaMKK2 KO HepG2 cells (Figures 2F,G, red arrow).

Immunoblotting revealed loss of CaMKK2 isoforms in the CaMKK2 KO HEK293 clonal cell lines (Figure 3A). HEK293 cells do not express TF or express it below the detection limit, therefore, Flag-tagged TF was expressed in the wild-type and CaMKK2 KO HEK293 cells to study the effect on TF phosphorylation (Figure 3A). IEF/SDS-PAGE revealed 2 major fractions of TF in the pH~3-4 and ~5-6 regions in HEK293 cells (Figure 3B). The P-TF level was significantly reduced in Flag-TF expressed CaMKK2 KO HEK293 cells compared to the wild-type (Figures 3B,C, Red arrow). In addition, Flag-TF appeared as a HMW fraction in the CaMKK2 KO HEK293 cells compared to the wild-type cells, as observed previously in CaMKK2 KD neurons (Figure 3B, Blue Square). This indicates difference in potential glycosylated form of TF.

In order to see if CaMKK2 is associated with TF, Flag-TF was immunoprecipitated from wild-type and CaMKK2 KO HEK293 cells and immunoblotted using anti-TF, anti-CaMKK2 and anti-pan-P-Ser specific antibodies (Figure 3D). The pan-P-Ser specific antibody was used because P-Ser is the abundant phosphorylation detected in the pH~3-4 fraction of TF. Protein gel staining, as well as immunoblotting, confirmed pull-down of Flag-TF (Figure 3D, left panel). However, CaMKK2 was not detected in the immunoprecipitated fraction of TF, neither any bands equivalent to P-TF(Ser) (Figure 3D, right panel).

The physical interaction between CaMKK2 and TF may be “hit and run” type with short residence time. Therefore, formaldehyde crosslinking followed by gel mobility shift assay (Klockenbusch and Kast, 2010) was used to study potential association between CaMKK2 and TF. Flag-TF was transiently expressed in the wild-type and CaMKK2 KO HEK293 cells and crosslinked using formaldehyde. Formaldehyde is the shortest cross-linker with a crosslinking distance of 2.3–2.7Å (Klockenbusch and Kast, 2010). The non-crosslinked and crosslinked proteins in Flag-TF expressed wild-type and CaMKK2 KO cells were resolved by SDS-PAGE and immunoblotted (Figure 3E). Native non-crosslinked CaMKK2 isoforms appeared within 63–75 kDa (black square) whereas crosslinked CaMKK2 appeared as 3 major bands at ~75, ~80–90, and ~135–180 kDa (red squares, numbered 1–3) (Figure 3E, left panel). The corresponding CaMKK2 bands were absent in the KO cells as expected (Figure 3E). Non-crosslinked Flag-TF appeared at 75 kDa (black square) whereas crosslinked Flag-TF appeared as 2 major shifted bands at ~180–240 and >240 kDa (red squares, numbered 4–5) (Figure 3E). The formaldehyde crosslinked TF associated bands displayed a comparative difference between the wild-type and CaMKK2 KO cells (Figure 3D, compex-5) indicating a difference in the association with interacting proteins. The relative position of the native CaMKK2 and Flag-TF shifted to multiple HMW bands in the crosslinked samples which suggested a shift in mass due to crosslinking with interacting proteins, however, the absence of any overlap between CaMKK2 and TF associated bands indicates the absence of any direct interaction.

BN-PAGE/SDS-PAGE was used to study the dynamics of TF and CaMKK2 associated protein complexes (Figure 3F). In BN-PAGE/SDS-PAGE, the first dimension native BN-PAGE separates the multiprotein complexes (MPCs) and the 2nd dimension denatured SDS-PAGE separates interacting protein components in the MPC which appears on a vertical line (Sabbir et al., 2016). BN-PAGE/SDS-PAGE revealed that TF formed 4 major MPCs at ~720, ~480, ~242, and ~100 kDa in the wild-type DRG tissues (Figure 3F; blue, green, and red circles). CaMKK2 isoforms formed MPCs at >1,200 and ~66 kDa (Figure 3F, orange and pink circles). The CaMKK2 MPCs were absent in the CaMKK2 KO mouse DRG tissues (Figure 3F). The absence of any vertical alignment of the native CaMKK2 and TF associated MPCs indicates that these proteins do not associate directly in a stable protein complex. The relative position of the ~720 and ~480 kDa TF MPCs shifted (blue and green circle) and the ~242 and ~100 kDa TF MPCs (red circle) disappeared in the CaMKK2 KO DRG tissues (Figure 3F). This indicates that the dynamic appearance of TF associated MPCs depends on CaMKK2.

The dynamics of TF and CaMKK2 associated MPCs were also studied in Flag-TF overexpressing wild-type and CaMKK2 KO HEK293 cells (Figure 3G). In wild-type HEK293 cells, TF appeared as multiple MPCs at >1,200, ~1,200–480, and 242 kDa (blue and red rectangles) and CaMKK2 isoforms appeared at 242 kDa (red rectangle), respectively (Figure 3G). This difference indicates potential cell type-specific difference in the interacting proteins. In CaMKK2 KO HEK293 cells, TF associated >1,200 and ~1,200–480 kDa MPCs were comparatively decreased and CaMKK2 isoform associated MPC disappeared (Figure 3G). In wild-type HEK293 cells, TF and CaMKK2 isoforms both vertically aligned at 242 kDa, but the 242kDa TF MPC was not shifted in its relative position in CaMKK2 KO condition which indicates that these proteins are not associated in the same complex though they appeared vertically aligned in the BN-PAGE.

RFP tagged TFR was expressed in the wild-type and CaMKK2 KO HEK293 cells and FITC-conjugated TF (FITC-TF) uptake and internalization were studied to see the effect of CaMKK2 loss (Figure 4). RFP-TFR was localized in the cell membranes (Figure 4A, red arrow). Internalized FITC-TF associated vesicular structures were significantly reduced in the CaMKK2 KO cells compared to the wild-type cells following 4 h of 25 μg/ml FITC-TF treatment indicating reduced internalization (Figures 4A,B, red square). In addition, pulse treatment with 25 μg/ml FITC-TF for 4 h followed by replacement of the extracellular fluid with serum-free medium and subsequent 24 h of incubation revealed significant retention of membrane-bound FITC-TF in CaMKK2 KO cells compared to the wild-type cells (Figures 4C,D, red arrows, and squares). Overall, this indicates that loss of CaMKK2 significantly altered TF uptake and internalization which associated with reduced P-TF in CaMKK2 KO HEK293 cells.

In order to see if a loss of CaMKK2 leads to the abnormal intracellular Ca²⁺ response, we treated wild-type and CaMKK2 KO HEK293 cells with 10 μM muscarine and measured the temporal dynamics of intracellular Ca²⁺ using cell-permeant fluorescent calcium indicator (Fluo-4). The HEK293 cells endogenously express muscarinic receptors and exhibit G-protein coupled intracellular Ca²⁺ release response upon muscarine treatment (Hussmann et al., 2011). Muscarine treatment elicited an immediate spike (within seconds) followed by a sustained elevation in intracellular Ca²⁺ in the wild-type cells in presence of Ca²⁺ in the extracellular environment (Supplementary Figure 2A, blue line). In CaMKK2 KO HEK293 cells, muscarine treatment caused a similar spike followed by a drop and then an exponential increase in intracellular Ca²⁺ (Supplementary Figure 2A, red line). At 3.5 min of muscarine treatment, the intracellular Ca²⁺ release was significantly lower in CaMKK2 KO cells compared to the wild-type cells which was reversed at 10 min of treatment (Supplementary Figures 2B,C). Thus, the dynamics of muscarinic receptor ligand-induced Ca²⁺ response was altered by loss of CaMKK2. Interestingly, in absence of Ca²⁺ in the extracellular media, CaMKK2 KO cells exhibited a significantly reduced amount of Ca²⁺ release compared to the wild-type cells (Supplementary Figures 2E,F). These indicate that CaMKK2 loss may affect plasma membrane/ endoplasmic reticulum-bound ion channels as well as Ca²⁺ buffer proteins.

TF promoter-trapped GFP reporter expression in the mouse central and peripheral nervous system (CNS and PNS) (GENSAT: Gene Expression Nervous System Atlas project, Rockefeller University, New York, USA) revealed that TF is expressed in the neurons of the spinal cord, olfactory bulb, cortex and cerebellum (Figure 5A). Immunoblotting revealed expression of CaMKK2 isoforms equivalent to 75 and 70 kDa proteins in the DRG (b), cerebellum (c), olfactory bulb (d), cortex (e) and liver (f) tissues (Figure 5). In CaMKK2 KO mice, protein bands corresponding to CaMKK2 isoforms were absent as expected (Figures 5B–F). The relative amount of TF was significantly decreased in CaMKK2 KO DRGs (g) and liver (k), significantly increased in the olfactory bulb (i) and cerebellum (j), and remained unaltered in cortex (k) tissues compared to the wild-type mice (Figure 5). This indicates that the relative abundance of TF was regulated by CaMKK2 in a tissue-specific manner.

IEF/SDS-PAGE revealed that TF appeared as 2 charged fractions in pH~3-4 and ~5-6 regions in the DRG tissues (a) and 3 charged fractions in pH~3-4, ~5-6, and ~9-10 regions in the olfactory bulb (d), cerebellum (e), cortex (f), and liver (g) tissues (Figure 6). In DRG tissues, the P-TF focused spot in pH~3-4 fraction (peak profile numbered 2 in Figure 6B, black arrow), was quantified and found significantly reduced in the CaMKK2 KO DRGs compared to the wild-type (Figure 6C). The relative amount of P-TF was significantly reduced in the CaMKK2 KO mice olfactory bulb (h), cerebellum (i), cortex (j), and liver (k) tissues (Figure 6, red rectangles). This indicates that loss of CaMKK2 ubiquitously reduced TF phosphorylation.

The triple transgenic 3xTg-AD mice were used to study the charged fractions of CaMKK2 and TF during the progression of AD. Only female 3xTg-AD mouse mice were used to avoid gender reported differences in neuropathology and behavior (Hirata-Fukae et al., 2008; Gimenez-Llort et al., 2010; Garcia-Mesa et al., 2011; Hebda-Bauer et al., 2013). A progressive increase in the Aβ peptide deposition was detected in some brain regions of 3xTg-AD mice as early as 3–4 months (Oddo et al., 2003). Synaptic transmission and long-term potentiation were impaired at 6 months in 3xTg AD mice (Oddo et al., 2003). Conformationally altered and hyperphosphorylated tau was detected in the hippocampus of 3xTg-AD mice at 12–15 months (Oddo et al., 2003). Therefore, 6 months and 14 months were considered as an early and late stage of AD in the 3xTg-AD mouse model and studied.

The molecular weight and pI of the mouse CaMKK2 isoforms (transcript ID: ENSMUST00000111668.7 and ENSMUST00000200109.4, respectively) were theoretically predicted as 73/59kDa and 5.27/5.31, respectively by ExPASy-Compute pI/MW tool (Gasteiger et al., 2003). IEF/SDS-PAGE of age and sex-matched early 3xTg-AD and wild-type cortex tissues revealed the presence of ~73 and ~59 kDa CaMKK2 proteins corresponding to isoform-1 and -2, respectively (Figure 7A). Plot profile of the focused CaMKK2 isoform 1 spots (Figure 7B) revealed a comparative difference between the wild-type and 3xTg AD mice in the intensity of the red arrow marked peak which is more negative shifted. IEF/SDS-PAGE study of the immunoprecipitated CaMKK2 from mammalian cells by Green et al. showed that the multiple CaMKK2 focused spots as observed in this study were positive for anti-P-Ser antibody and mutation of S129A, S133A, and S137A lead to the disappearance of the majority of the spots (Green et al., 2011). Validated P-CaMKK2 antibodies are not available. Comparatively more negative pI/pH shifted fraction of the CaMKK2 was considered as potentially phosphorylated and relative quantification revealed significant reduction of the CaMKK2 negative charged fraction (red arrow marked fraction) in the early 3xTg-AD cortex tissues compared to the wild-type mice (Figures 7A–C). TF in the early and late 3xTg-AD cortex tissues appeared as 4 major charged fractions in pH~10, ~7-8, ~5-6, and ~3-4 regions (Figures 7D,E, colored rectangles). P-TF was significantly decreased in both early and late 3xTg-AD cortex tissues compared to the wild-type mice (Figures 7D–G, red rectangles).

TF appeared as 2 MPCs at ~1,000 and ~720 kDa in 14 months old mice cortex and hippocampus tissues (Figure 8A). The relative abundance of ~1,000 kDa TF MPC was significantly decreased in both the hippocampus and cortex of the late-stage 3xTg-AD mice (Figures 8A,B, red rectangles). This indicates that the altered negative charged fraction of CaMKK2 in 3xTg-AD mice brain tissues is associated with decreased P-TF which affected the dynamics of TF associated protein complexes.

The relative abundance of TF remained comparatively unaltered in CaMKK2 KO serum (Figure 8C). However, the relative level of P-TF was significantly reduced in CaMKK2 KO serum (Figures 8D,E, red rectangles). This indicates that loss of CaMKK2 decreased serum P-TF level which may serve as a potential biomarker for AD. TF in 3xTg-AD serum appeared as 75 and 50 kDa proteins (p75-TF and p50-TF) (Figure 9A). The relative abundance of TF in early-stage 3xTg-AD mice was not significantly altered (Figures 9A,C), but in the late-stage 3xTg-AD mice, serum TF level was significantly decreased compared to the wild-type (Figures 9B,D). IEF/SDS-PAGE revealed comparatively decreased P-TF in the serum of both early and late 3xTg AD compared to the wild-type (Figures 9E,F, red rectangles). This indicates that the P-TF profile, not the total TF level in the serum reflect the physiological condition of AD in 3xTg-AD mice.

Total TF and the P-TF level were analyzed in the postmortem human CSF and serum samples from EOAD patients (Table 3). Matched Serum and CSF from the same individual was available for 2 age-matched EOAD and 2 unaffected control patients (Table 3). Matched serum and CSF samples were available from 3 LOAD and 1 unaffected control patients (Table 3). Total TF was significantly reduced in the CSF from EOAD patients compared to the age-matched unaffected controls (Supplementary Figures 3A,D). IEF/SDS-PAGE of the CSF samples from the same group of EOAD patients revealed comparative reduction/loss of P-TF compared to the unaffected control, except one patient (patient ID: 12772, Table 3) who was diagnosed with both AD and schizophrenia (Figures 10A,B, red rectangles). Matched serum from 2 EOAD patients (ID: 12772 and 13373) exhibited unaltered total TF level (Supplementary Figures 3C,F) but the P-TF level was reduced compared to the age-matched unaffected controls (Figures 10C,D, red rectangles). In addition, serum TF profile in these 2 EOAD patients resembled the CSF TF profile in terms of reduced P-TF level. This indicates that serum total TF level was not predictive of AD but the P-TF profile predicted diseased state. This supports P-TF (pH~3-4 fractions) profile as a potential biomarker for early-onset AD.

Total TF was comparatively reduced in the CSF of LOAD patients compared to the unaffected control but the serum TF level remained unaltered (Supplementary Figures 3G,H). IEF/SDS-PAGE revealed a comparative reduction of P-TF in both serum and CSF from the LOAD patients compared to the age-matched unaffected control (Figures 10E,F). This indicates that the P-TF profile in both serum and CSF may serve as a potential biomarker for the late-onset AD.

MS filed a US provisional patent on 4th October 2018 (Serial number: 62/741,148) entitled “Novel biomarker for Alzheimer's disease in human.”