An Atoh1 CRE Knock-In Mouse Labels Motor Neurons Involved in Fine Motor Control

Abstract Motor neurons (MNs) innervating the digit muscles of the intrinsic hand (IH) and intrinsic foot (IF) control fine motor movements. The ability to reproducibly label specifically IH and IF MNs in mice would be a beneficial tool for studies focused on fine motor control. To this end, we find that a CRE knock-in mouse line of Atoh1, a developmentally expressed basic helix-loop-helix (bHLH) transcription factor, reliably expresses CRE-dependent reporter genes in ∼60% of the IH and IF MNs. We determine that CRE-dependent expression in IH and IF MNs is ectopic because an Atoh1 mouse line driving FLPo recombinase does not label these MNs although other Atoh1-lineage neurons in the intermediate spinal cord are reliably identified. Furthermore, the CRE-dependent reporter expression is enriched in the IH and IF MN pools with much sparser labeling of other limb-innervating MN pools such as the tibialis anterior (TA), gastrocnemius (GS), quadricep (Q), and adductor (Ad). Lastly, we find that ectopic reporter expression begins postnatally and labels a mixture of α and γ-MNs. Altogether, the Atoh1 CRE knock-in mouse strain might be a useful tool to explore the function and connectivity of MNs involved in fine motor control when combined with other genetic or viral strategies that can restrict labeling specifically to the IH and IF MNs. Accordingly, we provide an example of sparse labeling of IH and IF MNs using an intersectional genetic approach.


Introduction
Motor neurons (MNs) innervating the muscles in the digits of the hand allow for exquisite control of fine motor movements required for dexterous skills, such as writing or sewing. Primates are known for their ability to precisely control individual digits of the hand, but skillful movements are not limited to the hand, as people born without arms are able to dexterously manipulate tools with their toes (Shubin et al., 1997;Sustaita et al., 2013;Dempsey-Jones et al., 2019). Precisely how MNs innervating the intrinsic hand (IH) and intrinsic foot (IF) muscles may differ in terms of function and connectivity compared with MNs that innervate limb muscles is unknown. Presumably, IH and IF MNs have unique properties and connectivity that contribute to their role in fine motor control, however, examination of the function of specifically IH and IF MNs is underexplored.
In part, lack of exploration of the IH and IF MNs is because of an inability to genetically distinguish these MNs from other limb MNs in mice. Mice also demonstrate remarkable manipulative dexterity when performing fine motor tasks such as eating dried pasta or grasping a pellet (Tennant et al., 2010;Yoshida and Isa, 2018) and, thus, represent a model organism to study fine motor skills. Developing a genetic tool that labels IH and IF MNs would have significant utility in interrogating the function and circuitry of fine motor control. To this end, we found that digit-innervating MNs were labeled using CRE-loxP recombination driven by the bHLH transcription factor atonal homolog 1, Atoh1, a transiently expressed gene in the dorsal-most part of the developing neural tube that is also known to specify excitatory (Vglut2 1 ) spinal cord neurons that project rostrally to the hindbrain (Bermingham et al., 2001;Gowan et al., 2001;Sakai et al., 2012;Yuengert et al., 2015;Pop et al., 2020).
Here, we explore the features of the digit-innervating MNs labeled by Atoh1 CRE-LoxP recombination. We find that labeling of IH and IF MNs is ectopic because an Atoh1 mouse line expressing FLPo recombinase that we developed in the lab does not label IH and IF MNs, although it does label the expected Atoh1-lineage neurons in the intermediate spinal cord. The Atoh1 CRE knock-in mouse labels a mixture of a-MNs and g -MNs in the IH and IF MN pools postnatally, while sparsely labeling other limb-innervating MNs. We further endeavor to show the utility of the Atoh1 CRE knock-in mouse using an intersectional genetic approach and discuss possible future methodologies to interrogate fine motor function and circuitry using this mouse strain.
The Atoh1 P2A-FLPo/1 mouse was generated using the Easi-CRISPR approach . Briefly, a long single stranded DNA cassette consisting of a viral peptide self-cleaving sequence [porcine teschovirus-1 2A (P2A); Kim et al., 2011] and the codon optimized flippase recombinase sequence (FLPo) were inserted after the last amino acid codon and before the stop codon of Atoh1. C57Bl/6N zygotes were microinjected with ribonucleoprotein complexes of Cas9 protein, tracrRNA, and crRNA (59-TGACTCTGATGAGGCCAGTT-39) along with a ssDNA megamer for homologous recombination (1497 bp containing 60 bp each 59 and 39 homology arms and the P2A-FLPo sequence; reagents were procured from IDT, microinjection service was provided by the UTSW Transgenic Mouse Core Facility). Assembling CRISPR reagents and microinjections were performed as previously described (Jacobi et al., 2017;Miura et al., 2018). The live born mice were first screened for insertion of the P2A-FLPo sequence and of those that were positive, one of the mice contained the full-length cassette. The cassette contained a minor mutation at the end of FLPo (the last isoleucine amino acid was changed to a serine), which could have occurred possibly because of an imprecise DNA repair event. Nevertheless, this amino acid change does not seem to affect the enzymatic function of FLPo. For genotyping, wild-type (WT) 321-bp and mutant 642-bp PCR products were detected using the following primers: WT forward 59-CCCTAACAGCGATGATGGCACAGAAGG-39, WT reverse 59-GGGGATTGGAAGAGCTGCAGCCGTC-39, and MUT reverse 59-CGAACTGCAGCTGCAGGCTGGA CACG-39. Note that because the P2A sequence selfcleaves near its C terminus, 21 amino acids of the P2A sequence are fused to the C terminus of ATOH1.

Tissue processing
Embryos were timed as embryonic day (E)0.5 on the day the vaginal plug was detected and P0 on the day of birth. Pregnant females were euthanized with CO 2 and cervical dislocation, embryos dissected out of the uterus, and spinal cords dissected out. Embryonic spinal cords (E14.5) were fixed in 4% paraformaldehyde (PFA)/PBS for 2-3 h at 4°C. Early postnatal animals (postnatal day 7 (P7) or younger) were cooled on ice, decapitated, their spinal cords dissected out, and fixed in 4% PFA/PBS for 2 h at 4°C. Mice older than P14 were anesthetized with avertin (2,2,2-tribromoethanol; 0.025-0.030 ml of 0.04 M avertin in 2-methyl-2butanol and distilled water/g mouse) and transcardially perfused, first with 0.012% w/v Heparin/PBS and then 4% PFA/PBS. A dorsal or ventral laminectomy exposed the spinal cord to the fixative. The spinal cords were fixed for 2 h and the brains overnight at 4°C. Tissue was washed in PBS for at least 1 d and cryoprotected in 30% sucrose dissolved in deionized water. Tissue was marked with 1% Alcian Blue in 3% acetic acid on one side to keep orientation and were then embedded in OCT (Tissue-Tek Optimal Cutting Temperature compound). Tissue was sectioned using a Leica CM1950 Cryostat.

CTB muscle injections
Mice aged P14 were anesthetized using isoflurane and prepared for injections into muscle. An approximate total of 500-750 nl of cholera toxin subunit B (CTB) Alexa Fluor 488 or 647 conjugate (Invitrogen; Nanoject II, Drummond Scientific) was injected into two to three different locations in the left forepaw (IH MN pool) or hindpaw (IF MN pool) in 50.6 nl increments. For injections into the tibialis anterior (TA), gastrocnemius (GS), quadricep (Q), or adductor (Ad), the area of the skin above the muscle was shaved and 70% ethanol and betadine (Avrio Health L.P.) applied. An incision was made above the muscle and 500-750 nl of the CTB-conjugated Alexa Fluor was injected into three to four different locations directly into the muscle. The incision was closed with surgical glue (Henry Schein Medical). Carprofen (5 mg/kg) was administered daily 3 d after surgery. Spinal cords were harvested 5 d after injection. For injections at P0, mice were anesthetized with isoflurane and injected with ,250 nl CTB-488 or 647 in one or two different locations of the forepaw and hindpaw and harvested 3 d later.

Experimental design and statistical tests
All details for number of sections counted, biological replicates, and male and female tissue analyzed are given in Results. No statistical tests were required as quantitation of the percentage of particular markers in any given MN pool were not directly compared with each other. Mean 6 SEM are reported throughout. For samples with n = 2, the SEM is equal to half of the range between the two data points.

Results
Atoh1 Cre/1 knock-in mice label MN pools involved in fine motor control We observed using CRE-lineage tracing strategies (Atoh1 Cre/1 knock-in mice crossed to tdTomato (TOM) reporter mice (R26 LSL-Tom , Ai14)) (Yang et al., 2010;Madisen et al., 2010) that subsets of MNs expressing ChAT were labeled in the spinal cord (Fig. 1A,B, arrows and arrowheads). Based on the anatomic location of the MN pools along the rostral-caudal axis, we predicted that the Atoh1 Cre/1 line labeled MNs of the IH and IF in thoracic 1 (T1) and lumbar 6 (L6) areas of the spinal cord ( Fig.  1C; Watson et al., 2009). We injected the forepaw and hindpaw with the retrograde tracer CTB conjugated to Alexa Fluor 488 (CTB-488) and verified that the IH and IF MN pools were Atoh1 Cre/1 TOM 1 MNs (Fig. 1A,B,   involved in the proprioceptive system (Yuengert et al., 2015;Pop et al., 2020).
To see whether the labeling of Atoh1 Cre/1 TOM 1 MNs was specific to the IH and IF MN pools, we injected CTB-488 into the TA, GS, Q, and Ad muscles and found that those MN pools had much fewer TOM 1 MNs (Fig. 1D, arrows and arrowheads). In addition, we sampled sections throughout the rostral-caudal axis of the spinal cord in Atoh1 Cre/1 mice and found that few other MN pools had TOM 1 expression (Fig. 2, arrows). Altogether, the TOM 1 MNs labeled ;60% of the IH and IF MN pools at P19 [IH: 59 6 4%, n = 4, 1:3 male (M): female (F), four to eight half sections/n; IF: 64 6 3%, n = 5, 1:4 M:F, two to eight half sections/n; TA: 7 6 2%, n = 4, 2:2 M:F, two half sections/ n; GS: 14 6 2%, n = 4, 2:2 M:F, two to four half sections/n; Q: 13 6 4%, n = 4, 2:2 M:F, one to three half sections/n; Ad: 13 6 3%, n = 4, 2:2 M:F, one to six half sections/n. MN areas located by CTB-488 injection into appropriate muscle group; Fig. 1E]. We estimate that the total number of ChAT 1 MNs in the IH and IF MN pools at P19 on one side is 410 6 72 neurons for the IH MN pool (n = 3, 1:2 M:F, three to four half sections/n) and 337 6 7 SEM neurons for the IF MN pool (n = 3, 0:3 M:F, three half sections/n). Counts of sections represented a tenth of the MN pool, so the estimated total number of ChAT 1 neurons were the final counts multiplied by ten. Because the TOM 1 MNs comprised ;60% of the IH and IF MN pools, we hypothesized that the Atoh1 Cre/1 mice might be labeling a particular functional class of MNs such as fast or slow twitch MNs. However, we found that only a subset of the IH and IF TOM 1 MNs were labeled by the marker for fast MNs, MMP9, indicating that the IH and IF TOM 1 MNs are a mixture of fast and slow MNs (IH: 59 6 8%, n = 4, 3:1 M:F, three half sections/n; IF: 74 6 3%, n = 5, 3:2 M:F, three half sections/n; Fig. 1F,G, arrows; Kaplan et al., 2014).
we were unable to detect any signal at the mRNA level (our unpublished observations). Therefore, to corroborate the labeling of IH and IF TOM 1 MNs in the Atoh1 Cre/1 mouse line, we crossed these mice to an Atoh1 P2A-FLPo/1 mouse and FLPo-dependent GFP reporter mouse (R26 FSF-EGFP/1 ) such that neurons from the Atoh1 Cre/1 mouse line were labeled with tdTomato and those from the Atoh1 P2A-FLPo/1 mouse were labeled with EGFP. Strikingly, while the IH and IF MN pools were TOM 1 , they were clearly GFP - (Fig. 1H, insets) indicating that the TOM 1 IH and IF MNs are ectopically labeled in the Atoh1 Cre/1 mice, either because of differences in CRE or FLPo recombinase expression themselves or differences in recombination efficiency in the CRE and FLPo lines. Notably, Atoh1-lineage neurons in the intermediate spinal cord had substantial overlap of GFP 1 and TOM 1 fluorescence (Fig. 1H, arrowheads) indicating these neurons are reliably from the Atoh1-lineage.

Discussion
To understand how dexterous movements of the hand and foot is achieved, we must have some knowledge about the function and circuitry of MNs involved in fine motor control. Thus, obtaining genetic tools in mice that could reproducibly label the IH and IF MNs could help address questions as to how these neurons differ perhaps in their electrophysiological properties and connectivity compared with MNs innervating limb muscles involved in gross motor control. To this end, we found that MNs that innervate the IH and IF are labeled in Atoh1 Cre/1 mice. We found that labeling of the IH and IF MNs is ectopic and occurs postnatally resulting in ;60% of IH and IF MNs being labeled by approximately three weeks of age with a slight preference for labeling a-MNs over g -MNs. Here, we discuss some possible uses for the Atoh1 Cre/1 mouse line for future studies interrogating fine motor control circuits.
If IH and IF MNs could be isolated specifically without labeling other neurons in the nervous system, then a number of genetic tools could be used to either manipulate the activity of these MNs (i.e., optogenetic and/or chemogenetic approaches) or identify the inputs to these MNs (i.e., transsynaptic rabies virus tracing). One approach to isolate IH and IF MNs would be to use an intersectional genetic strategy crossing the Atoh1 Cre/1 mice to mice expressing FLPo recombinase in the IH and IF MNs, but not in other overlapping Atoh1 Cre/1 domains. We attempted an initial intersectional cross of Atoh1 Cre/1 to Chat IRES-FLPo/1 and found sparse labeling of the IH and IF MNs because of inefficient recombination in these MNs by the Chat IRES-FLPo/1 line. While this particular intersectional cross could be useful for sparse labeling of the IH and IF MNs for anatomic studies, one must keep in mind that Atoh1 Cre/1 and Chat IRES-FLPo/1 intersect in other areas of the central nervous system such as occasionally in other MN pools (Fig. 2) and in Atoh1-lineage cholinergic neurons in the pedunculopontine tegmentum (PPTg) and the lateral dorsal tegmentum (LDTg) (our unpublished observations; Rose et al., 2009). Thus, an alternate cross is required to separate out the IH and IF MNs.
Moving forward, we propose that other FLPo-recombinase lines could be used in conjunction with the Atoh1 Cre/1 mouse line to isolate the IH and IF MNs. For example, FLPo driven by Hb9, Fign, or Cpne4, which we have shown colocalize with the IH and IF MNs labeled in the Atoh1 Cre/1 line, but not in other Atoh1-lineage neurons in the intermediate spinal cord (Fig. 3), would be promising candidates. Hb9, however, has the same caveat as Chat in that other MN pools throughout the spinal cord may intersect with Atoh1 Cre/1 expression. IH and IF MNs were found to contain a repertoire of unique molecular markers compared with neighboring limb-innervating MNs suggesting a unique developmental program (Mendelsohn et al., 2017). Indeed, the transcriptional landscape in IH and IF MNs appears to preferentially activate the CRE recombinase in Atoh1 Cre/1 mice. Of these unique molecular markers, other candidate genes include Ecrg4, Reg3b, Serpinf1, and Pirt (Mendelsohn et al., 2017), although these would need further characterization of spatial and temporal overlap with the Atoh1 Cre/1 line to determine their suitability for future studies. Furthermore, Osmr and Col8a1 could be considered for exploration of specifically IF MNs (Mendelsohn et al., 2017).
An alternate strategy would be to use viruses to restrict labeling to just the IH or IF MN pools. For example, the Atoh1 Cre/1 line could be crossed to an intersectional line expressing a reporter of choice (i.e., fluorescent, optogenetic, or chemogenetic reporter) and an AAV-FLPo injected into either the spinal cord or in the forepaw or hindpaw to be taken up retrogradely to the IH and IF MNs. However, a caveat of this approach is the potential damage to the spinal cord or muscles of the forepaw and hindpaw because of the needle injection. Thus, appropriate injection controls would be needed with this approach.
Altogether, we present here that the Atoh1 Cre/1 mouse consistently labels MNs of the IH and IF and that the Atoh1 Cre/1 mouse could be used to probe the function and connectivity of MNs in fine motor control. Our findings also serve as a cautionary tale of relying on CRErecombinase mouse lines to faithfully report endogenous gene expression and speak to the need for careful follow-up experiments to appropriately interpret reporter expression.