Dysfunction of grey matter NG2 glial cells affects neuronal plasticity and behavior

NG2 glia represent a distinct type of macroglial cells in the CNS and are unique among glia because they receive synaptic input from neurons. They are abundantly present in white and grey matter. While the majority of white matter NG2 glia differentiates into oligodendrocytes, the physiological impact of grey matter NG2 glia and their synaptic input are ill defined yet. Here we asked whether dysfunctional NG2 glia affect neuronal signaling and behavior. We generated mice with inducible deletion of the K+ channel Kir4.1 in NG2 glia and performed comparative electrophysiological, immunohistochemical, molecular and behavioral analyses. Focussing on the hippocampus, we found that loss of the Kir4.1 potentiated synaptic depolarizations of NG2 glia and enhanced the expression of myelin basic protein. Notably, while mice with targeted deletion of the K+ channel in NG2 glia showed impaired long term potentiation at CA3-CA1 synapses, they demonstrated improved spatial memory as revealed by testing new object location recognition. Our data demonstrate that proper NG2 glia function is critical for normal brain function and behavior.


258
To get further insight as to how targeted deletion of Kir4.1 affects myelination, axons and 259 myelin sheaths were stained with antibodies against SMI312 and MBP and analyzed at higher 260 spatial resolution using expansion microscopy (Chen et al., 2015;Chozinski et al., 2016).

261
Analysis was confined to the stratum lacunosum moleculare because of the high MBP 262 immunoreactivity seen in this layer ( Fig. 5B-D). Based on the immunostainings, 3D 263 isosurfaces of axons (SMI312) and myelin sheaths (MBP) were constructed, further referred 264 to as particles (Fig. 5E) where the ratio of alternations correlates with the strength of working memory (N = 17 and 19 280 for control and Kir4.1 ko mice, respectively). Neither genotype nor sex influenced the 281 behavior of the mice in this paradigm, and no interaction between sex and genotype was 282 found (Fig. 6A).

283
Testing novel object location recognition (NOLR) is a sensitive tool to evaluate spatial 284 memory. Animals, which remember the original location of an object spend more time  To test social memory, we applied a partner recognition test. Similar to the novel object 290 location recognition test, a positive novelty preference value is an indicator of recognition.

291
Using 1 h inter-trial intervals we found a significant sex effect (p = 0.007), because females 292 showed a higher partner recognition ability. However, the genotype of the animals did not 293 influence partner recognition, and sex -genotype interaction was also not present. Using 294 longer intervals (2 h, 4 h, 8 h) neither genotype nor sex influenced social memory (Fig. 6C).

295
Sex -genotype interaction was also not present. Thus, our result suggests that mice with 296 targeted deletion of Kir4.1 in NG2 glia have similar social memory as control mice.

297
Finally, motor and coordination abilities of male and female controls and mice with Kir4.1 298 deletion in NG2 glia were compared using the beam walking test. Here, a lower latency to 299 reach the goal box indicates better coordination ability. On the wide (28 mm diameter) and 300 middle-sized beams (14 mm), Kir4.1 ko mice reached the goal box significantly faster than 301 the controls (14 mm, p = 0.03; 28 mm, p = 0.03) (Fig. 6D middle, right). Sex effects or 302 genotype -sex interactions were not present. Testing the animals on the thinnest rod (8 mm 303 diameter) we found a significant interaction between sex and genotype (p < 0.01) but no 304 effect of genotype or sex alone.

305
Together, our data reveal that NG2 glia-specific deletion of Kir4.1 improves spatial memory 306 and motor abilities. Interestingly, deletion of Kir4.1 did not lead to a generally improved 307 learning and memory, because working memory or social declarative memory remained 308 unchanged in the ko mice.  Our study confirms that Kir4.1 channels mediate the main K + conductance in adult NG2 glia.

331
These channels not only determine passive membrane properties but also regulate the 332 efficiency of NG2 glia depolarization upon synaptic input.

333
In the naive hippocampus, a single NG2 glial cell receives input from less than 20  Notably, short-term synaptic plasticity, which is a measure of vesicle release probability and 339 related to presynaptic Ca 2+ levels (Zucker and Regehr, 2002), was not affected in Kir4.1 340 deficient NG2 glia synapses. factors, such as PDGF, FGF-2 NT-3 and NGF, or its receptors, might account for the 368 subregional differences in proliferation and differentiation (Barres et al., 1994;Cohen et al., 369 1996;Mason and Goldman, 2002;Wilson et al., 2003).

370
However since neither the proliferation rate of NG2 glia nor the density of GSTpi + cells 371 differed between genotypes we conclude that targeted deletion of Kir4.1 did not stimulate 372 NG2 glia differentiation in the hippocampus. This is in line with a study by Larsen et al. 373 (2018) who found no changes in NG2 glia differentiation 2 weeks after tamoxifen-induced  To further evaluate how deletion of Kir4.1 from NG2 glia affected myelination, expansion 388 microscopy was performed allowing for higher (factor 4-5) spatial resolution than   Notably, improved performance of Kir4.1 ko mice was confined to the NOLR test, while no 437 differences were found in working memory (Y-maze) or social memory (partner recognition 438 test). It has to be considered that not only the hippocampus is involved in executing these 439 memory types. The prefrontal cortex contributes to short term memory performance in the Y- with some behavioral tasks being more susceptible to the deletion-induced alterations than 447 others, but hippocampal memory deficits and motor deficits were not observed. Activity-dependent myelination is an important mechanism to ensure proper synaptic function 451 and spiking synchronicity, motor skills and memory performance. Our data imply that these 452 processes are also influenced by NG2 glia and that this is not simply dependent on their 453 differentiation status. Rather, it is likely that NG2 glia exert its effect by signaling on 454 neighboring oligodendrocytes, either directly or indirectly via influencing neuronal activity.

455
Future work has to identify the mechanisms and factors by which NG2 glia accomplish these 456 actions.   Passive membrane properties were determined after establishing the whole-cell configuration.

511
The average of 10 consecutive transients evoked by 50 ms lasting 10 mV voltage step was 512 used to determine series resistance (Rs), membrane resistance (Rm) and membrane 513 capacitance (Cm). As it is currently impossible to properly determine resistances above 3 GΩ 514 due to technical reasons, data above this threshold was excluded from analysis. mPSPs were 515 recorded (8 min periods) in aCSF containing TTX (0.5 µM; Abcam). NBQX (10 μM; Tocris) 516 and/or Picrotoxin (150 µM; Abcam) were applied via the perfusion system to separate 517 GABAergic and glutamatergic inputs. mEPSPs were detected using the template search of 518 pClamp (Molecular devices) and evaluated by visual inspection. Cells with less than 3 events 519 detected were rejected. Amplitudes, kinetics and frequencies of mEPSPS were analyzed by a 520 custom-written macro in IGOR Pro Software (WaveMetrics).  responses recorded before (baseline, 10 min) and after TBS were normalized to the mean 555 fEPSP slope during baseline recording and the ratio of fEPSP slopes was calculated (paired-556 pulse ratio= slope 2 / slope 1). Recordings with more than 10% fEPSP slope variability during 557 baseline recording were excluded.  Invitrogen) were added. Products were identified with gel electrophoresis using a molecular 618 weight marker (Low molecular weight marker, New England Biolabs, Frankfurt, Germany).

619
As a positive control, RT-PCR for total RNA from mouse brain were run in parallel. Negative