In this study we compared the behavior of aged B6(SJL)-Apoe^{tm1.1(APOE*4)Adiuj} Trem2^em1Adiuj/J (n = 23), Strain #028709, with WT mice (n = 27), Strain #000664 from the Jackson Laboratory.¹ The mice were divided into three age groups and run in 4 batches (Table 1). Table 1 lists the number of APOE4.TREM2 and WT mice used in these behavioral studies by genotype, gender, and age. Table 2 shows the number of mice by sex, age and genotype for each behavioral test. Please note that batches of the same age were not tested in parallel due to the different times of availability of offspring of each group. To avoid potential confounding effects thereof, we used tightly defined SOPs and automated behavioral scoring to ensure testing was kept as constant as possible between batches. While we did not keep a strict order of the tests’ administration within each group, OFT and HaXha were mostly performed early on (on average the 2.2th, out of 7 tests across the 6 groups), and SPT, NB and MWM later (5.0th, 5.0^th and 5.5th, resp., out of 7 tests). The test battery took 8 weeks to complete per group. Mice had a rest period of at least 24 h between the tests and were acclimated for at least 10 min to the room prior to starting each test. The testing apparatus was cleaned with 70% ethanol after each subject was done.

Missed trials (boldfaced in Table 2) were due to one of three reasons. Six mice were deemed unfit for the MWM upon their initial few trials based on observed distress (prolonged lack of swimming and being fully submerged for several seconds). These trials were performed near the end of the overall series of tests, but the affected mice were not excluded from the few subsequent tests. In several cases tests were skipped in error: for three mice NB was not assessed, and for one mouse SW and for one SPT were also not assessed. For one mouse HC was missed and for two HaxHa was missed, due to PC crashes with incomplete data.

A 12-h/12-h inverted light cycle with lights out at 9:00 a.m. was used in the vivarium. The mice were fed ad libitum chow (Harlan 18% protein rodent diet) and were housed separately in polycarbonate cages (12 × 12 × 25 cm) with controlled humidity (40%) and temperature (22 °C). As we were unable to group house some of the (male) mice, we decided to isolate all of them for consistency. Guidelines established by the National Institute of Health were followed for the care of all animals (2011). The experimental protocols were approved by the Institutional Animal Care and the John B. Pierce Laboratory (JV1-21). The John B. Pierce Laboratory is AAALAC accredited.

Statistical analyses were performed in Prism version 10.4.2 for Mac OS (GraphPad Software, Boston, Massachusetts USA).² Means are reported as means±SEM. For most behavioral assessments, where appropriate, 2-way ANOVA was performed with factors age and genotype and their interaction. Details are given in each tests section. Post-hoc tests were Bonferroni corrected for multiple comparisons. Alpha level was set at 0.05.

All behavioral tests tracking movement (Habituation/cross-habituation, Open Field Test, Hidden cookie test, Morris Water Maze) were assessed by Noldus Ethovision (EthoVision XT, version 10.1, Noldus Information Technology B. V., Wageningen, The Netherlands). A video camera (Basler Gen1cam, Resolution 1280*1024) connected to a Noldus computer system was placed above the test arena. To optimize tracking contrast we used 2 lamps with 5 15 W 5000 K LED bulbs each, aimed at the white ceiling and providing diffuse setup illumination.

This olfactory-dependent behavioral task was performed in a sealed semi-transparent white acrylic box (26x38x16cm) in which a cotton-tipped wood applicator (8 cm long, Dynarex MS-50330) was placed in the center of the box and presented to the mouse, while retained by a 54x54x4mm applicator holder. We added one of 3 odorants to the applicator: mineral oil (MO, control), amyl acetate (AA) (diluted 1% in MO), phenyl ethanol (PE) (1% in MO), and social odor (S, SO) (obtained by swabbing the cage of non-experimental CRE-OMP female mice). A total of 12 trials were done per mouse, where each odorant was presented three times in succession per session to yield the following order: MO1-3, AA1-3, PE1-3 and SO1-3. Each trial consisted of a 2 min odorant exposure and an interval of 1 min between stimuli. The time spent smelling the cotton tip was registered using a camera mounted at the ceiling of the box, and Ethovision identified the orientation and scored the behavior of the animal. Smelling was defined as being oriented toward the applicator tip while the nose is within 2 cm of it, which was validated by us in Coronas-Samano et al. (2016). This test evaluates if mice can spontaneously recognize a novel odorant stimulus by spending more time smelling the applicator (cross-habituation phase, trial 3 vs. new odor 1), as opposed to the time that the mice spent on each stimulus (habituation phase, trial 1 vs. 3 of same odor).

SPT was divided into Day 1 and 2 measuring the volume of water and sucrose solution consumption. Our sucrose preference test was based on Romano et al. (2015). The mice were placed in regular home cages (22.5 × 16.7 × 14 cm) and had access to two drinking tubes, one filled with 20 mL tap water, and the other with 20 mL 10% sucrose (Sucrose, Lot 18,005,800, American bio) diluted in deionized water. The location of the tubes was randomized for two consecutive days to avoid spatial bias. During 6 h the free consumption of water and 10% sucrose solution took place in the presence of ad libitum food during the dark phase. Sucrose preference was the percentage of the total amount of liquid drunk in 6 h that was sucrose, reported per day.

The Open Field was based on Van Dam et al. (2003), Chen et al. (2013), and Galeano et al. (2014) to assess spontaneous exploratory activity and anxiety-related behaviors. Anxiety and exploratory activities were evaluated by allowing mice to freely explore an open field arena for 15 min. The testing apparatus was a classic open field (i.e., a white PVC square arena, 50 × 50 cm, with walls 40 cm high). An Ethovision-connected video camera was placed above the box. Each mouse was placed individually in the center of the arena and the behavior was monitored by the video tracking system (Noldus Ethovision). The central area was arbitrarily defined as a square of 35 × 35 cm (half the total area). The ethological measures included the frequency and duration spent at each area (center and periphery) and locomotor activity.

This test assesses deficits in olfactory guided foraging skills by measuring the latency of the mice to find a buried cookie in one of the corners of their cage. Mice were familiarized with the chocolate cookie (Chips Ahoy!) for a week before the test. A quarter of chocolate cookie (~2.5 g) was placed every 48 h for three times in the home cage on the top of the bedding. Mice received food and water ad libitum. After a week of the familiarization phase, mice were deprived of the cookie for 3 days before the test. On the first day of the test, mice were food restricted (no food present) for 6 h and individually caged with 3 cm of clean bedding. On the first day of the test, a piece of cookie was buried 2 cm deep in one corner of a rat cage (45.6 cm × 23.6 cm) and the latency to finding the cookie was recorded. Then the moused was placed in the opposite corner, and the latency to finding the cookie was recorded using Noldus Ethovision and the mouse removed. An hour and a half later, this was repeated. At the end of the second trial, the mice were taken to the animal room. The next day, mice were placed in the experimental cages and the piece of cookie was buried in a corner different from the last day, and this was tested twice following the same protocol as the first experimental day. Maximum trial time was 15 min.

Spatial reference learning and memory were evaluated in a water maze adapted from that previously described by Morris (1984) and was based on Van Dam et al. (2003), Chen et al. (2013), Galeano et al. (2014), and Javed et al. (2011). The test was performed in a circular stainless steel pool of 90 cm in diameter and 40 cm height, filled with 20 cm of water tainted with a non-toxic white paint (acrylic paint, 20,503 white, Apple Barrel). As it is known that the standard Morris water maze can interfere with physiology, likely due to the stress of heat loss, we opted to refine this method to be less stressful by not maintaining it at room temperature (21 ± 2 °C), but instead between room and core body temperature (37 °C), i.e., at 29 ± 2 °C. The pool was virtually divided into four equal quadrants, labeled north–south–east–west. The Ethovision camera was mounted above the maze. Several visual cues were available in the small asymmetric room including two lamps, wall mounted shelves, a large safety cabinet, a white door and a blacked out window. Mice were not given a habituation session or visual acuity test.

During training, a platform (10 cm in diameter and made of transparent acrylic plastic) was submerged 1 cm below water surface and was placed at a fixed location in one of the quadrants. If the mice were not able to reach the platform within 120 s, they were guided to the platform where they had to stay for 30 s before being returned to their home cage for 30 s on a dry towel. All mice were given four trials per day, being released into each quadrant once, for four consecutive days. The start position was randomized among four quadrants of the pool. During training trials, the latency to reach the escape platform and the path length were measured.

A probe trial was performed 24 h after the last day of training. During the probe trial, mice were allowed to swim in the pool without the escape platform for 120 s. The latency to reach the platform (s), swim distance (cm), and swim speed (cm/s) were recorded using Ethovision. During the probe trial, performance was expressed as the time spent in each quadrant of the MWM, and the number of crossings through the position where the platform used to be during acquisition (using 15 cm diameter target area).

After at least a 24 h rest period, mice were housed individually in their regular home cage in the vivarium under regular conditions and tested for nest building [adapted from Deacon (2006) and Wesson and Wilson (2011)]. Two hours prior to the onset of the dark phase of the light cycle, individual cages were supplied a nestlet made of pressed cotton square (Ancare, UK agent, Lillico). The next morning (∼24 h later) cages were inspected for nest construction. Pictures were taken prior to evaluation for documentation. Nestlet nest construction was scored by 3 investigators blind to genotype using the system of Deacon [please see Deacon (2006) for detailed scoring standard]. Briefly, in this 5-point scale, 1 indicates a > 90% intact nestlet, whereas a 5 indicates a nestlet torn >90% and a clear nest crater.

Piezo foil (PVDF)-based circadian rhythm scoring of sleep/wake rhythm was used as established in Donohue et al. (2008) and used in several AD studies (Duncan et al., 2012; Mang et al., 2014; Sethi et al., 2015). Mice were individually housed for 48 h in clean plastic cages (15x15cm, Signal Solutions LLC, www.sigsoln.com, Lexington, KY) with approximately 5 mm of corn cob bedding lining the floor on top of the PVDF foil and left with an inverted light cycle (as in the colony) in a separate dedicated room using a single 15 W 500 K LED lamp 2 m from the cages aimed at the white ceiling for diffuse light. Regular chow and water was available ad lib. Activity was continuously recorded without disturbance for the 48 h and was analyzed for the circadian rhythm of wakefulness and sleep using the Signal Solutions software.

This portion of the study uses the same strains as the behavioral study and were housed under the same conditions. Here we evaluate the difference in blood volume (BV) between aging B6(SJL)-Apoe^{tm1.1(Apoe*4)Adiuj} Trem2^em1Adiuj/J (n = 17) versus WT (n = 21). Table 3 shows the number of mice used for this study by age, gender, and genotype. Guidelines established by the National Institute of Health were followed for the care of all the animals (2011). The experimental protocols were approved by the Institutional Animal Care and the John B Pierce Laboratory. The John B Pierce laboratory is AAALAC accredited.

Mice were anesthetized with isoflurane (4% for induction, 1.5–3% for maintenance) followed by injections of Atropine (0.04 mg/kg, IP) and Ethiqa XR (3.25 mg/kg, SC). Body temperature was maintained at 37 °C. An incision was made down the midline of the skull between the eyes exposing the dorsal surface of the bulbs, followed by the administration of topical bupivacaine. Custom headplates were placed and secured with MetaBond.

Once the animal was placed in the stereotax a dental drill was used to thin the bone and expose the dorsal side of the olfactory bulbs (dOB). Sterile saline was repeatedly used to saturate the skull to assess the optical clarity. The skull over the bulbs, as well as additional exposed skull and skin edges, were sealed with Krazy Glue for transparency. Next, Carprofen (5 mg/kg, SC) was injected. Mice are kept isolated in recovery cages while body temperature was supported until awake and then given 4–8 days to recover prior to imaging.

Mice were anesthetized with isoflurane (4%) followed by injections of Dexdomitor (0.5 mg/kg, IP), Ketamine (75 mg/kg, IP), and Atropine (0.04 mg/kg, IP) with body temperature maintained at 37 °C. Mineral oil was put on the bulbar window to enhance optical clarity. Intrinsic optical signals from the dOB were recorded using a CCD camera (Redshirt Imaging LLC, Decatur, GA, USA) with 256 × 256 pixel resolution, at a frame rate of 7 Hz. The epifluorescence macroscope used was a custom-made tandem-lens type (Ratzlaff and Grinvald, 1991). Imaging lenses were prime Nikon F-mount (ccd lens: 135 mm f/3.5, used at f/8; object lens: 135 mm f/3.5, used at f/3.5 or 80 mm at f/5.6). A high-power LED with 590 nm spectral peak (M590L4, Thorlabs, Newark, NJ), chosen for its isosbestic wavelength of oxy- and deoxy-hemoglobin to specifically measure blood volume (BV), was aimed via an optical fiber at the dOB, driven by a T-Cube LED Driver (LEDD1B, Thorlabs, Newark, NJ).

The flow dilution olfactometer output was presented through Teflon tubes that converged to a mask positioned to isolate the nose of the mouse. Nine trials using ethyl butyrate (EB) at three increasing concentrations (0.3, 1.0, and 3.0% of saturated vapor pressure, respectively) were carried out after the initial three control trials with no odor presented. Odor was presented for 4 s, 8 s into the start of the trial. A 180 s interval was used between each trial allowing OB activity to ensure return to baseline. Once the EB trials were completed, we used a 5-min time interval to clear out all remaining odor. The same process was repeated for additional controls and methyl valerate (MV) at the same dilutions. After all trials were completed, mice were administered Antisedan (1 mg/kg, SC) and were allowed to recover while supporting body temperature.

MATLAB (The Mathworks Inc., V 2023) was used for dOB arterial region selection, by selecting pixels based on the consistency of blood volume (BV) increase out of the six trials from each concentration. Three regions were evaluated, the first ROI selected pixels with BV increase one SD above baseline for 4/6 trials (Figure 1A), the second selected pixels with BV increase three SD above baseline for 3/6 trials (Figure 1B), and the third ROI being a rectangular selection with no thresholds or assumptions, thereby evaluating BV changes across a large section of each dOB (Figure 1C). For further validation of these regions, mice were imaged during an intravenous tail injection of Fluorescein isothiocyanate (FITC, 250 mg/kg) using a high-power LED of 488 nm (Thorlabs, Newark, NJ) aimed via optical fiber at the dOB, driven by a T-Cube LED Driver (LEDD1B, Thorlabs, Newark, NJ). Imaging was captured using the same imaging setup, at 25 frames/s.

Unless indicated, all optical imaging data analysis was done using custom-written MATLAB (The Mathworks Inc., V 2023).

The Habituation-Cross-habituation test aimed to determine how APOE4.TREM2 mice might display olfactory dysfunction in comparison to WT animals throughout their lifetime (Tables 1, 2; see Methods for full details). This assessment involved evaluating their capacity to distinguish between a sequentially presented set of odors. Mineral oil (MO) functioned as the odorless diluent used as the control stimulus to measure baseline exploration and habituation. Phenyl ethanol (PE) served as a non-trigeminal rose-like odorant, allowing for the selective probing of the olfactory system. Amyl acetate (AA) was selected as a commonly used banana-like food odorant. Lastly, social odor (SO, “S” in figures) was included, as previous research indicated that this stimulus elicited longer exploration times (Crawley and Goodwin, 1980), thereby enhancing the sensitivity of the test. Post-hoc we performed seven Bonferroni-corrected comparisons between the first and third of each stimulus (habituation), and the third and following first presentation (cross-habituation, or “dis-habituation”).

At 9 months, the WT mice showed an overall effect of exploration time across the odors [F_{(1.35, 9.42)} = 8.32, p = 0.02; one-way RM ANOVA, Figure 3a]. Please note that in figures we refer to WT as C57BL6/J. WT mice displayed habituation (•) between the repeated presentation of the same social odors; SO1 (“S1,” 28.7 ± 6.9 s) and SO3 (“S3,” 11.3 ± 6.2 s; ••p = 0.003). They also showed cross-habituation (♦) between the presentation of PE3 (1.8 ± 0.5) and SO1 (28.7 ± 6.9 s; ♦p = 0.04). Interestingly, the 9 months APOE4.TREM2 mice did not show an overall effect on exploration time across the odors [F_{(2.88, 14.4)} = 2.67, p = 0.09; one-way RM ANOVA, Figure 3a]. Therefore, habituation and cross-habituation were not significant. Planned post-hoc comparisons were also not significant. Additionally, we found statistical difference in the exploration time of SO1 between 9 months WT (28.7 ± 6.9 s) and the APOE4.TREM2 group (16.4 ± 5.0 s; * p = 0.03), however there was no overall difference between the groups [F_{(1, 143)} = 0.84, p = 0.36, two-way ANOVA, Figure 3a].

The 13 months, WT mice showed an effect on exploration time across the odors [F_{(3.77, 33.8)} = 13.9, p < 0.0001; one-way ANOVA, Figure 3b, note smaller y-axis range used to improve resolution]. This group showed cross-habituation between PE3 (5.2 ± 1.8) and SO1 (28.6 ± 3.2, ♦♦p < 0.001) and habituation between SO1 and SO3 (10.5 ± 2.9, •p = 0.02). Meanwhile, the APOE4.TREM2 group showed a stimulus effect on exploration duration (F_{2.52, 22.67} = 7.7, p = 0.002, one-way ANOVA, Figure 3b) with cross-habituation between PE3 (5.1 ± 1.6 s) and SO1 (25.6 ± 4.8 s; ♦♦p = 0.007), and habitation between SO1 and SO3 (8.0 ± 2.3 s, •p = 0.04). There were no overall differences between the 13 months groups [F_{(1, 216)} = 0.03, p = 0.86], nor planned post-hoc differences.

At 16 months, WT mice showed no effect of stimulus on exploration time [F_{(1.99, 13.9)} = 2.2, p = 0.15; one-way ANOVA, Figure 3c]. No habituation or cross-habituation was observed. In contrast, the 16 months APOE4.TREM2 mice showed an effect on exploration time across the odors [F_{(2.35, 23.50)} = 11.90, p = 0.0002, one-way ANOVA, Figure 3c]. This group showed habituation between SO1 (40.0 ± 6.9 s) and SO3 (13.7 ± 3.1 s; ••p = 0.004). The cross-habituation was significant between PE3 (6.3 ± 2.0 s) and SO1 (40.0 ± 6.9 s; ♦♦p = 0.009). Interestingly, there was a statistical difference for SO1 between the 16 months WT (13.3 ± 4.8 s) and the APOE4.TREM2 (40.0 ± 6.9 s; **p < 0.0001) [F_{(1, 204)} = 22.71, p < 0.0001, two-way ANOVA, Figure 3c].

These results do not suggest age-related olfactory dysfunction in the APOE4.TREM2 mice, although they did exhibit an unexpected increased novel social odor exploration with age.

Consumption was measured over 2 days in the sucrose preference task, testing the mice’s reward sensitivity for indications of anhedonia. Six hours of free consumption of water and 10% sucrose solution took place in the presence of ad libitum food. Sucrose preference was calculated from the amount of sucrose solution consumed, expressed as a percentage of the total amount of liquid ingested in 6 h. All groups on all days showed highly significant sucrose preference (1-tailed unpaired t-test, p < 0.001).

At 9 months both groups obtained high preference on the first day of the test, WT 87.5 ± 2.5% and APOE4.TREM2 85.7 ± 5.3% consumption. Preference was maintained on the second day, WT 92.5 ± 2.5% and APOE4.TREM2 92.9 ± 1.8%, with no statistical differences between groups [F_(1,26) = 0.05, p = 0.83; two-way ANOVA, Figure 4a].

Similarly, at 13 months, both groups showed high preference on the first day, WT 88.2 ± 3.0% and APOE4.TREM2 85.0 ± 2.7%. Preference was maintained on the second day for both groups WT 95.5 ± 1.6% and APOE4.TREM2 93 ± 2.6% with no statistical differences between groups [F_(1,38) = 1.27, p = 0.27; two-way ANOVA, Figure 4b].

At 16 months both groups again showed similar preference with no statistical difference with WT ingesting 90.0 ± 1.9% and APOE4.TREM2 82.0 ± 5.1% on the first day. Both groups maintained preference on the second day as WT consumed 95.0 ± 1.9% and APOE4.TREM2 91.1 ± 3.5% with no statistical difference [F_(1,31) = 2.64, p = 0.12; two-way ANOVA, Figure 4c].

In addition to preference, the total amount of volume consumed was measured every day. Over the test days, at 9 months the groups showed a time effect [F_(1,26) = 17.8, p < 0.001; two-way ANOVA, Figure 4d], and the WT mice at 9 months increased their consumption from the first day: 6.1 ± 0.4 mL to the second day at 9.8 ± 0.6 mL (p = 0.012). Meanwhile, the APOE4.TREM2 consumed 6.4 ± 0.7 mL on the first day and 8.2 ± 0.8 mL on the second day with no statistical differences (p = 0.39).

The WT at 13 months also showed an effect over days [F_{(1, 38)} = 8.97, p = 0.005; two-way ANOVA, Figure 4e], with an increase in consumption between the first day 6.4 ± 0.7 mL and the second day of test 9.7 ± 1.0 mL (p = 0.031). APOE4.TREM2 mice did not show statistical differences between first day (7.3 ± 0.8) and the second day (8.8 ± 0.7).

Interestingly, at 16 months there was only a marginal effect across days [F_(1,31) = 3.87, p = 0.059], while there was a strong effect between groups [F_(1,31) = 44.63, p < 0.0001; two-way ANOVA, Figure 4f]. WT group consumed more on the first day (7.5 ± 0.6 mL) than the APOE4.TREM2 (3.9 ± 0.6 mL, p < 0.001). This behavior was observed on the second testing day as the WT consumed 9.1 ± 0.5 mL while the APOE4.TREM2 consumed 4.7 ± 0.6 mL (p < 0.0001). Thus, APOE4.TREM2 mice at 16 months consumed significantly less compared to control. Overall, however, we did not find anhedonia in APOE4.TREM2 mice.

Anxiety and exploratory activities were evaluated by allowing mice to freely explore an open field arena for 15 min in a classic open field-testing apparatus. The mice showed an aging effect [F_(2,50) = 8.62, p = 0.0006; two-way ANOVA, Figure 5a]. Specifically, time spent in the central 50% of the total area was lower in APOE4.TREM2 mice at 16 months (124.9 ± 12.0 s) compared to the 9 months WT (216.1 ± 23.4 s, p = 0.0096), 9 months APOE4.TREM2 (213.5 ± 16.7 s, p = 0.014), and 13 months WT (225.1 ± 16.0 s, p = 0.001) (Figure 5a). Genotype was marginally significant [F_(2,50) = 3.49, p = = 0.068].

During these trials the total distance traveled (cm), and velocity (cm/s) were also measured (Figures 5b,c), with a significant effect of age [F_(2,50) = 7.51, p = 0.002] and an age:genotype interaction [F_(2,50) = 5.84, p = 0.006; two-way ANOVA; Figure 5b] for distance, and hence similarly for velocity [age (F_(2,50) = 5.51, p = 0.007] and age:genotype interaction [F_(2,50) = 6.51, p = 0.004; two-way ANOVA), Figure 5c]. The APOE4.TREM2 mice at 13 months traveled (4,138 ± 454 cm), less than the APOE4.TREM2 mice at 9 months (6,578 ± 409 cm; p = 0.005) and to WT at 16 months (6,576 ± 742 cm; p = 0.005). The 13 months APOE4.TREM2 showed lower velocity (4.8 ± 0.5 cm/s) compared to APOE. TREM2 at 9 months (7.4 ± 0.5 cm/s, p = 0.006) and WT 16 months (7.3 ± 0.8 cm/s, p = 0.008).

Additionally, the count of fecal boli was obtained at the end of the test and showed main effect of genotype [F_(2,50) = 3.41, p = 0.042] on the number of fecal boli and an age:genotype interaction [F_(2,50) = 4.98, p = 0.011; two-way ANOVA, Figure 5d]. At 13 months, APOE4.TREM2 mice had a higher count of fecal boli (4.0 ± 0.7) compared to 9 months (0.9 ± 0.4, p = 0.018) and compared to 16 months (1.1 ± 0.6, p = 0.016). These findings demonstrate age dependent increase in anxiety in 16 months APOE4.TREM2 mice.

On the first day of the test, a piece of cookie was buried in one corner and the latency to finding the cookie was recorded. An hour and a half later, another piece of cookie was buried in the same corner as the first trial. The next day, mice were placed in the experimental cages and the piece of cookie was buried in a corner different from the prior day. This was tested twice following the same protocol as the first experimental day. Maximum trial time was 15 min.

Mice did not show statistical differences between genotypes [F_(1,48) = 0.17, p = 0.68; two-way ANOVA, Figure 6a] at 9 months or at 16 months [F_(1,64) = 0.45, p = 0.50; two-way ANOVA, Figure 6c]. Statistical significance was seen in 13 months APOE4.TREM2 mice, as they showed greater latency in finding the cookie compared to WT, specifically on the last trials of each day [F_(1,76) = 21.92, p < 0.0001; two-way ANOVA, Figure 6b, please note larger y-scale range].

Figure 6d compares the latencies averaged over the first experimental day, showing significant effect of age [F_(2,100) = 4.02, p = 0.011], genotype [F_(1,100) = 5.52, p = 0.021] and age:genotype interaction [F_{(2, 100)} = 3.37, p = 0.04; two-way ANOVA]. The APOE4.TREM2 at 13 months showed higher latency (717.5 ± 55 s) compared to 13 months WT (412.5 ± 69 s; p = 0.023). Additionally, 13 months APOE4.TREM2 mice showed differences compared to WT at 9 months (373.6 ± 69.7 s, p = 0.016), to 9 months APOE4.TREM2 (295.4 ± 70.9 s, p = 0.004), and 16 months WT (359.5 ± 95.8 s, p = 0.016). The increased latency in buried cookie detection, especially in 13 months APOE4.TREM2 mice, indicates genotype related dysfunction of olfactory guided foraging tasks.

The MWM assesses spatial memory and learning deficits. Mice swam freely in a pool to allow them to learn the fixed location of the escape platform. On the test day, the platform was removed, and each mouse was allocated 120 s to swim freely. During this time, the latency to the platform area, cumulative time at the platform area, and cumulative time in the quadrant of the platform area were measured.

Latency was only affected by genotype [F_(1,42) = 7.16, p < 0.011; two-way ANOVA; Figure 7a]. At 9 and 16 months there were no differences. Mice at 13 months showed a higher latency (49.4 ± 17.6 s) to find the platform area compared to WT mice (14.5 ± 3.2 s; p = 0.011).

Time at the platform was affected by both genotype [F_(1,44) = 11.61, p < 0.002] and age:genotype interaction [F_(2,44) = 18.81, p < 0.0001; two-way ANOVA; Figure 7b]. The 13 months APOE4.TREM2 mice spent significantly less time at the platform area (0.7 ± 0.2 s) compared to the WT (6.6 ± 0.8 s; p < 0.0001).

There were no statistical differences for the cumulative time in the quadrant (Figure 7c). We can conclude from these results that the APOE4.TREM2 mice display genotype-dependent deficits in spatial memory and learning.

The nestlet building test measures the capacity of mice to build a nest to evaluate innate behavior, motor deficits, as well as overall welfare. Mice are scored out of 5 points, 1 indicates a > 90% intact nestlet, whereas a 5 indicates a nestlet torn >90% and a clear nest crater (Deacon, 2006). Scores were averaged across three raters using anonymized nest pictures.

Two-way ANOVA showed only an interaction effect, of age and genotype [F_(2,46) = 4.51, p = 0.017; Figure 8]. At 9 months both WT and APOE4.TREM2 mice scored similarly (3.4 ± 0.1 vs. 3.5 ± 0.3) without statistical difference. At 13 months, the WT scored 4.3 ± 0.3, higher than APOE4.TREM2 at 3.2 ± 0.3 (p = 0.005). At 16 months, the control group scored 3.1 ± 0.4, similar to the APOE4.TREM2 3.5 ± 0.1 (but note n = 4 for those mice). WT mice scored higher at 13 months than 16 months (p = 0.006). Overall, no impairment in nestlet building in APOE4.TREM2 mice is evident.

The sleep/wake cycle was measured for 2 days to know if the APOE4.TREM2 mice showed sleep alterations. At 9 months, both groups were active when they were in the dark phase of the experiment (Figure 9a, WT Day 1: 18.5 ± 3.3% of dark time was slept and Day 2: 29.9 ± 4.4%; APOE4.TREM2 Day1: 23.3 ± 4.9% and Day 2: 32.4 ± 2.7%). Meanwhile, the percentage of sleep was higher during the light phase than dark phase for both groups (Figure 9a, WT Day 1: 56.8 ± 1.8% and Day 2: 54.1 ± 1.4%; APOE4.TREM2 Day 1: 57.6 ± 1.5% and Day 2: 54.6 ± 1.0%; both post-hoc p < 0.001) [phase effect F_(3,52) = 70.70, p < 0.0001; two-way ANOVA]. Differences were also not found between groups when both days were averaged (Figure 9d, dark phase WT 24.2 ± 3.0% vs. APOE4.TREM2 27.9 ± 3.0%; light phase WT 55.4 ± 1.2% vs. APOE4.TREM2 56.1 ± 1.0%) [F_(1,56) = 0.90, p = 0.35; two-way ANOVA].

At 13 months, the WT and APOE mice increased their sleep time in the dark phase of the experiment across days (Figure 9b, WT Day 1: 15.8 ± 2.1% vs. Day 2: 34.6 ± 1.6%; APOE4.TREM2 Day 1: 17.2 ± 2.7% vs. Day 2: 36.2 ± 1.8%, both post-hoc p < 0.0001), but their higher (both p < 0.0001) % sleep in the light phase did not differ across days (Figure 9b, WT Day 1: 58.4 ± 2.3% vs. Day 2: 53.8 ± 2.2%; APOE4.TREM2 Day 1: 63.2 ± 3.4% vs. Day 2: 57.5 ± 2.3%) [phase effect F_(3,76) = 153.1, p < 0.0001; two-way ANOVA]. Data did not show statistical difference for the total sleep between groups when both days were averaged (Figure 9e, dark phase WT 25.2 ± 2.4% vs. APOE4.TREM2 26.7 ± 2.7%; light phase WT 56.1 ± 1.6% vs. APOE4.TREM2 60.4 ± 2.1%) [F_(1,80) = 0.1.69, p = 0.20; two-way ANOVA].

At 16 months, both groups were also most active in the dark phase (Figure 9c, WT Day 1: 14.6 ± 4.8% vs. Day 2: 25.8 ± 3.5%; APOE4.TREM2 Day 1: 13.8 ± 3.2% vs. Day 2: 23.0 ± 2.3%). Both groups increased their sleep percentage in the light phase (Figure 9c, WT Day1: 65.8 ± 2.1% and Day 2: 57.0 ± 1.3%; APOE4.TREM2 Day1: 61.1 ± 1.5% and Day 2: 56.3 ± 1.9%; both p < 0.0001) [phase effect F_(3,64) = 155.1, p < 0.0001; two-way ANOVA]. Again, no statistical difference between groups was observed for the total percentage of sleep in the two-day average of the dark/light cycle (Figure 9f, dark phase WT 20.2 ± 3.2% vs. APOE4.TREM2 18.4 ± 2.2%; light phase WT 61.4 ± 1.6% vs. APOE4.TREM2 58.7 ± 1.3%) [F_(1,68) = 1.08, p = 0.31; two-way ANOVA]. We conclude that we did not find age or genotype related differences in sleep time across the sleep/wake cycle.

Odor-evoked intrinsic signals were recorded and measured in anesthetized ~0.7 to ~2 years old mice (Table 3) to evaluate whether the APOE4.TREM2 (A4T2) genotype produces vascular hyperresponsiveness, particularly with increasing age. The responses were calculated for 3 types of region of interest (ROI): 1SD above baseline, 3SD above baseline and un-thresholded (Figures 1A–C and Methods). The 3SD ROI type reflects dOB arteries, as verified by FITC injection (Figure 1E, Methods). Arteriolar vessels cannot be selected based on resting light image vessels alone (Figure 1D), which can include non-arterial vessels.

The dOB was imaged using a 590 nm LED, the power spectrum of which matches hemoglobin’s isosbestic point and therefore reports changes in blood volume (BV). An increase in BV increases absorption and hence decreases reflected signal. The average hemodynamic response function shown in Figure 10 represents BV dOB responses to odor presentations (flow dilutions at 0, 0.3, 1, and 3% of saturated vapor pressure) of ethyl butyrate (EB) for 4 s and was calculated based on the 3SD-based ROI across all mice (n = 38, Table 3) and trials (n = 12/mouse for EB). These responses have a rapid (~1 s) onset and a slow offset lasting over 10 s and have larger amplitude with increasing odor concentrations. The data demonstrate that the APOE4.TREM2 (A4T2) mice show increased intrinsic BV responses (i.e., hyperresponsiveness) at each concentration of EB, being especially prominent at 1 and 3% EB, relative to WT (C57).

We calculated the Area Under the Curve (AUC) to assess the overall integrated BV response magnitude over the first 10s post odor onset (sum of % dF/F across 70 frames, i.e., %dF/F * t (10 s) * frame rate (7 frames per sec)). A more negative AUC value signifies greater BV response. Evaluating the AUC ignores the response dynamics to focus on the overall magnitude. When AUC is averaged across mice combining both odorants in each ROI, APOE4.TREM2 mice display vascular hyperresponsiveness to odor stimuli at each concentration. Statistically highly significant odor-induced BV hyperresponsiveness by APOE4.TREM2 mice is evident in the 1SD-based ROI responses (Figure 11A), artery-like 3SD ROI responses (Figure 11B) and even dOB-wide rectangular ROI responses (Figure 11C). Tables 4, 5 demonstrate, using 2-way (age, genotype) and 3-way (age, genotype, sex) ANOVA across all mice and all 24 trials per mouse, that age and genotype separately and in interaction significantly affect intrinsic responses. Sex was not a significant factor, although we note uneven distribution of males vs. females (Table 3). These data demonstrate that the intrinsic odor responses are larger in the APOE4.TREM2 (A4T2) than WT (C57) mice, and more so with age.

As LOAD is strongly dependent on age, we further assessed if the odor-evoked intrinsic responses revealed an aging effect. APOE4.TREM2 mice show an increasing trend of vascular hyperresponsiveness to odor stimuli with age, while the WT mice show no significant BV changes with increasing age in the 1SD-based ROI responses (Figure 12A, APOE4.TREM2: −5.5%_AUC/months, with an r² = 0.29; WT: +2.3%_AUC/months, and r² = 0.09). This is also evident in the artery-like 3SD ROI responses (Figure 12B) and the dOB-wide rectangular ROI responses (Figure 12C). These data emphasize not only the vascular dysfunction seen in the APOE4.TREM2 mice, but further demonstrate that the dysfunction is age dependent. These data validate this model’s translatability in studying the progression of LOAD.

Here we report the neurovascular coupling in the OB of aged A4T2 mice. We hypothesized that in early stages of LOAD, neurovascular hyper-responsiveness will be detected due to compensatory efforts sustaining the degenerating tissues. We hence expected in increase in the ratio of BV to Ca responses.

We confirm findings from our A4T2 cohort in that BV AUC (Figure 13C) in the A4T2 x Thy1-GCamP6f mice also increased with age for 0.3% (p < 0.002) and 1% (p < 0.04) odor, though not for 3%. However, neural responses (AUC of estimated firing rate, Figure 13A) increased with age at all odor concentrations (p < 0.003). The neurovascular coupling, defined as ratio of BV and Ca, did not depend on age at any concentration (Figure 13B).

We explored this further by estimating the temporal hemodynamic transfer function at each concentration of each odor for each mouse. We estimated the odor evoked firing rate (Figure 2A) by deconvolving the GCaMP6f off response time constant. We then derived the modified gamma function (Figure 2E) best fitting the BV response (Figure 2C) after convolution with the neural response (Figure 2D). Goodness of fit (maximum of the cross correlation) between BV and the calculated BV (based on convolving the transfer function with neural activity response) was 0.74 ± 0.28, 0.94 ± 0.06 and 0.99 ± 0.01 (mean ± sd, across all 8 mice and 6 trials per mouse per concentration) for 0.3, 1 and 3% of the odorants, respectively, with a consistent lag of 1 frame. We found no effect of age on the hemodynamic transfer function amplitude (Figure 14A), nor on FWHM (Figure 14B) or time to peak (Figure 14C).

Additionally, we found that the neurovascular coupling increased as function of odor concentration in these LOAD mice, being evident from the increasing amplitude for both odors (Figure 14B). Regression slopes of amplitude (AU) versus concentration were consistently positive (0.0034 ± 0.0023 and 0.0055 ± 0.0045 for EB and MV, respectively) and higher than 0 (p < 0.002 and p < 0.006, 1-sided t-test, respectively). Amplitudes increased 1.76 ± 0.48 and 2.22 ± 0.75 times with a 10-fold (0.3 to 3%) concentration increase for EB and MV, respectively (p < 0.006 and p < 0.005, 1-sided t-test, respectively).