Context-dependent requirement of G protein coupling for Latrophilin-2 in target selection of hippocampal axons

The formation of neural circuits requires extensive interactions of cell-surface proteins to guide axons to their correct target neurons. Trans-cellular interactions of the adhesion G protein-coupled receptor latrophilin-2 (Lphn2) with its partner teneurin-3 instruct the precise assembly of hippocampal networks by reciprocal repulsion. Lphn2 acts as a repulsive receptor in distal CA1 neurons to direct their axons to the proximal subiculum, and as a repulsive ligand in the proximal subiculum to direct proximal CA1 axons to the distal subiculum. It remains unclear if Lphn2-mediated intracellular signaling is required for its role in either context. Here, we show that Lphn2 couples to Gα12/13 in heterologous cells; this coupling is increased by constitutive exposure of the tethered agonist. Specific mutations of Lphn2’s tethered agonist region disrupt its G protein coupling and autoproteolytic cleavage, whereas mutating the autoproteolytic cleavage site alone prevents cleavage but preserves a functional tethered agonist. Using an in vivo misexpression assay, we demonstrate that wild-type Lphn2 misdirects proximal CA1 axons to the proximal subiculum and that Lphn2 tethered agonist activity is required for its role as a repulsive receptor in axons. By contrast, neither tethered agonist activity nor autoproteolysis were necessary for Lphn2’s role as a repulsive ligand in the subiculum target neurons. Thus, tethered agonist activity is required for Lphn2-mediated neural circuit assembly in a context-dependent manner.


18
Latrophilins (Lphn1-3) are highly expressed in the brain and were originally identified as responders 19 to ɑ-latrotoxin, a neurotoxin from black widow spider venom that causes profound release of 20 neurotransmitters from nerve terminals (Davletov et al., 1996). They belong to the family of adhesion 21 G protein-coupled receptors (aGPCRs), capable of eliciting intracellular effects through coupling with 22 heterotrimeric G proteins (Lelianova et al., 1997). Additionally, as cell adhesion molecules, latrophilins residues immediately C-terminal to the GPS constitute the tethered agonist peptide (TA) (also known 53 as the Stachel or stalk peptide), which upon exposure binds within the transmembrane domain to 54 activate heterotrimeric G proteins . 55 While our previous in vivo work established that interaction between Ten3 and Lphn2 was 56 required for precise circuit assembly (Pederick et al., 2021), it did not examine how this might depend 57 4 on Lphn2-mediated signaling mechanisms. Here, we modified our previous hippocampal model to 58 develop a Lphn2 misexpression assay ( Figure 1D). We misexpressed Lphn2 in either CA1 axons or 59 the subiculum target and assessed the impact on normal proximal CA1→distal subiculum axon 60 targeting. We found that ectopically expressing wild-type (WT) Lphn2 in proximal CA1 axons causes 61 their mistargeting to the proximal subiculum. This provided us with a robust platform to interrogate 62 whether TA activity or autoproteolytic cleavage is required for axon targeting in the Lphn2 63 misexpression system. When misexpressed in CA1, Lphn2 TA activity was required for Lphn2-64 mediated axon targeting. By contrast, when we misexpressed Lphn2 in subiculum target neurons, 65 both TA activity and autoproteolysis were dispensable for Lphn2-mediated axon repulsion. Thus, our 66 data support that Lphn2 G protein coupling is required in axons but not target neurons during precise 67 circuit assembly.

70
Misexpression of WT Lphn2 in proximal CA1 leads to axon mistargeting in the subiculum 71 To investigate the role of Lphn2-mediated G protein activity in hippocampal axon targeting, we first 72 designed a gain-of-function assay in which we misexpressed Lphn2 in proximal CA1 neurons. We 73 hypothesized that this ectopic expression would cause proximal CA1 axons to avoid Ten3-positive 74 distal subiculum and incorrectly target the proximal subiculum, and that this platform could provide us 75 an assay to test Lphn2 mutants with defects in various functions to determine whether WT Lphn2 76 mistargeting is compromised. 77 To test our hypothesis, we used a dual injection strategy to ectopically express Lphn2 in 78 proximal CA1 and trace its axons into the subiculum (Figure 1-figure supplement 1). At postnatal 79 day 0 (P0), lentivirus expressing Cre (LV-Cre) (control) or Cre-Lphn2 (LV-Cre-Lphn2) was injected 80 into proximal CA1, followed by injection of a Cre-dependent membrane-bound mCherry (AAV-DIO-81 mCherry) into proximal CA1 in the same mice at approximately P42 ( Figure 1D). As expected, in 82 5 control animals (LV-Cre), Cre+ proximal CA1 axons targeted the most distal parts of the subiculum 83 ( Figure 1E). By contrast, when Lphn2 was misexpressed in proximal CA1 (LV-Cre-Lphn2), Cre+ 84 proximal CA1 axons targeted the most proximal parts of the subiculum (Figure 1F). To analyze the 85 location of proximal CA1 axons in the subiculum, we calculated the fraction of axon intensity within 86 thirds of the subiculum across the proximal/distal axis ( Figure 1G). Proximal CA1 axons 87 misexpressing Lphn2 are located significantly more in the proximal third of the subiculum and 88 significantly less in the distal third of the subiculum when compared to control axons ( Figure 1H). 89 These data supported our hypothesis that ectopic expression of Lphn2 in proximal CA1 axons 90 causes mistargeting to the proximal subiculum. Having established the effect of wild-type (WT) Lphn2 91 misexpression in proximal CA1 axons, we next sought to characterize G protein coupling of WT showed that Lphn3, another member of the latrophilin family of aGPCRs, couples principally to 98 Gα12/13, and also more weakly to Gαq, using a combination of gene expression assays and an 99 activation strategy that permitted acute exposure of the TA in a live-cell system (Mathiasen et al., 100 2020). Thus, we began our signaling characterization of Lphn2 similarly using a WT full-length (FL)

101
Lphn2 construct, and a constitutively active construct termed Lphn2-CTF (Figure 2A) up to the GPS and instead has only a methionine residue before the TA. We tested expression of 105 these constructs in mammalian cells using immunoblotting and showed that Lphn2-CTF ran at the 106 expected truncated position (~72 kDa) and that WT Lphn2 ran at both molecular weights 107 6 corresponding to uncleaved (~164 kDa) and cleaved (~72 kDa) positions ( Figure 2B). This result for 108 FL Lphn2 is similar to previous work that characterized autoproteolysis of Lphn1 (Araç et al., 2012) as 109 well as to our recent findings in Lphn3 (N.A. Perry-Hauser, unpublished data).

110
To infer activity of these constructs in G protein signaling pathways, we used the 111 luminescence-based gene expression assay for serum response element (SRE), which produces a 112 robust response in our previous studies of Lphn3 (Mathiasen et. al., 2020). In our assay design, SRE 113 action is coupled to the transcription and translation of firefly luciferase; this readout is then 114 normalized to the control reporter, Renilla luciferase, expressed from the same plasmid under a 115 constitutive promoter. We found that Lphn2-CTF significantly enhanced signaling over WT Lphn2 for 116 SRE gene expression at varying levels of cDNA transfection ( Figure 2C). Since the SRE assay 117 reports on signaling by G12/G13 as well as Gq, we tested whether Gα12/13 or Gαq was the primary 118 contributor to this response using a selective Gαq inhibitor, YM-254890 (Figure 2-figure   119 supplement 1). We did not observe a significant effect upon addition of the inhibitor, suggesting that 120 Lphn2 signals through Gα12/13.

121
To verify our result in the context of acute G protein activation, we next tested how TA 122 exposure affects G protein activation in a bioluminescence resonance energy transfer (BRET) assay 123 ( Figure 2D). We designed a synthetically-activatable Lphn2 construct based on a recent publication 124 that took advantage of the protease enterokinase . Enterokinase selectively 125 recognizes the trypsinogen substrate sequence DDDDK and cleaves after the lysine residue, thereby 126 exposing the native TA. Thus, we cloned a Lphn2 construct that included a modified hemagglutinin 127 signal peptide, the P2Y12 N-terminal extracellular sequence (amino acids 1-24), a flexible linker 128 (GGSGGSGGS), the enterokinase recognition site (DYKDDDDK), and the truncated Lphn2-CTF 129 sequence. We tested this construct in a Gβγ-release assay where energy transfer was monitored 130 between the membrane-anchored luminescent donor, GRK3-ct-Rluc8, and the fluorescent acceptor, Gγ-Venus ). This assay was performed in a HEKΔ7 cell line with targeted deletion 132 7 of Gα12 and Gα13, as well as Gαs/olf, Gαq/11, and Gαz (Alvarez-Curto et al., 2016) to enable systematic 133 re-introduction of the Gα subunits. As expected, in the absence of Gα subunits no BRET signal was 134 observed; however, when Gα12 or Gα13 was re-introduced to cells expressing the Lphn2 construct 135 there was a significant increase in the BRET signal upon treatment with enterokinase ( Figure 2E). 136 This increase was not observed upon co-expression of the receptor with Gαs, Gαi1, or Gαq ( Figure   137 2-figure supplement 2). This suggests that the increase in cAMP reported previously for the Lphn2   Mutating conserved residues F831A and M835A in the TA impairs G protein-coupling activity 146 Previous studies suggest that the third and seventh residues of aGPCRs are required for TA-147 mediated G protein activation (Stoveken et al., 2015). We hypothesized that mutating these residues We then proceeded to test these constructs in our SRE gene expression system ( Figure 3A). As 158 hypothesized, both the FL and truncated Lphn2 had dramatically impaired responses to SRE across 159 varying levels of cDNA transfection. 160 To confirm that the reduced SRE response was due to impaired G protein coupling and not 161 simply to impaired proteolysis, we cloned the CTF of our Lphn2-F831A/M835A mutant into our 162 enterokinase-activatable construct. We then tested our construct in the Gβγ-release assay and   (Figure 3-figure supplement 2), although the total 179 fraction of axon intensity was significantly lower in the distal subiculum ( Figure 3E). 180 Collectively, these findings suggest that TA activity or autoproteolysis is required for Lphn2-181 mediated miswiring of proximal CA1 axons. Mutating residue T829G in the TA renders Lphn2 cleavage deficient but preserves the ability 184 of the TA to activate G protein 185 While misexpressing Lphn2-F831A/M835A failed to cause proximal CA1 axons to mistarget to the 186 proximal subiculum, we could not definitively link this result to impaired TA activity since the Lphn2-187 F831A/M835A mutant was also resistant to autoproteolytic cleavage ( Figure 3B). Since our initial 188 efforts to find a TA mutant with impaired G protein signaling that retained normal autoproteolytic 189 cleavage were unsuccessful, we designed a construct that rendered Lphn2 resistant to  196 We next assessed G protein signaling for Lphn2-T829G using our SRE gene expression comparable to Lphn2-CTF suggesting that the mutated TA is fully active if exposed. We therefore 202 cloned the CTF of the T829G mutant into our enterokinase-activatable construct and tested BRET 203 signaling following re-introduction of Gα proteins ( Figure 4C). The T829G-CTF retained BRET 204 signaling comparable to WT Lphn2-CTF for Gα12 and Gα13, with no discernable BRET response for 205 Gαs, Gαi1, or Gαq (Figure 2-figure supplement 2). Taken together, these data supported that 206 Lphn2-T829G is resistant to autoproteolytic cleavage but maintains a functional TA.  Neither TA activity nor autoproteolysis is required for Lphn2's action as a repulsive ligand 228 We previously showed that misexpression of Lphn2 in distal subiculum target neurons causes 229 proximal CA1 axons to avoid this area, suggesting that Lphn2 acts cell non-autonomously as a 230 repulsive ligand in directing target selection of proximal CA1 axons (Pederick et al., 2021). In the 231 context of repulsive axon guidance, proteolysis has been proposed as a mechanism to disassemble 232 the extracellular binding complex after repulsive signaling, which is necessary for repulsion (Hattori et   In this study, we utilized a combination of in vivo axon target selection and in vitro cell signaling 256 assays to determine if Lphn2 G protein signaling is required for its role as a neural wiring molecule.

257
First, we showed that Lphn2 misexpression can cell-autonomously misdirect proximal CA1 axons to 258 proximal subiculum, establishing an assay to test the requirements of Lphn2 G protein signaling when 259 it acts as a receptor (Figure 1). Second, we identified the G protein interaction partners of Lphn2 260 (Figure 2) and validated point mutations that disrupt TA activity and/or autoproteolysis of the GPS 261 region (Figures 3, 4). Third, we showed that when Lphn2 is misexpressed in CA1 axons TA activity is 262 required for Lphn2's ability to misdirect axon targeting (Figures 3, 4). Finally, when Lphn2 acts as a 263 repulsive ligand in subiculum target neurons, we demonstrated that neither TA activity nor 264 autoproteolytic cleavage is required for the receptor's ability to repel Ten3+ proximal CA1 axons 265 ( Figure 5). Taken together, these findings highlight the importance of Lphn2 G protein signaling 266 during precise circuit assembly in a context-specific manner. Our results also support that while 267 aGPCR GPS cleavage is dispensable Lphn2's role as a receptor to direct axon targeting, an intact TA 268 is essential.  They argue that TA exposure likely occurs in a stepwise mechanism where the TA is uncovered 303 along its N→C axis. Thus, it is possible that an intact complex of aGPCR's NTF and CTF could 304 unmask the TA sufficiently for interaction with the 7TM, resulting in receptor activation.

305
The ability of TA exposure to occur in intact aGPCRs could provide an explanation for why our 306 Lphn2-T829G mutant displays a partial axon mistargeting phenotype (Figure 4). Even though the 307 14 Lphn2-T829G mutant cannot undergo autoproteolytic cleavage, it still retains a functional TA that is 308 able to initiate G protein signaling if transiently unmasked. This is likely why activation of FL Lphn2-309 T829G is less robust than the WT Lphn2, which can more readily unmask the TA.

311
However, this mutant also has impaired G protein coupling activity even with full exposure of the TA.

312
This explains why we observed essentially normal axon targeting to the distal subiculum when we 313 overexpressed Lphn2-F831A/M835A in proximal CA1 (Figure 3). Even if the TA of Lphn2-314 F831A/M835A becomes unmasked it still cannot initiate TA-mediated receptor activity.  Interestingly, neither autoproteolytic cleavage nor TA activity is required for Lphn2 to act cell 325 non-autonomously as a repulsive ligand in target neurons (Figure 5). This suggests that cleavage of 326 Lphn2 is not required for repulsion in this context and implies that another mechanism mediates 327 disassembly of the extracellular binding complex, which is required for retracting axons to pull away 328 from the targets. Other potential mechanisms to disassemble the extracellular binding complex            axons must overlap with lentivirus injection site in subiculum ( Figure 5). All mice that fulfilled these 508 criteria are reported in Fig. 1, 3, 4 and 5 and were included in quantifications. Images of injections 509 sites (5× magnification) and projections (10× magnification) were acquired of every other 60-μm 510 sagittal section using a Zeiss epifluorescence scope. Due to variation in injection sites within each 511 mouse, exposure was adjusted for each mouse to avoid saturation. Fluorescence intensity 512 measurements on unprocessed images were taken using FIJI and data processing was performed 513 using MATLAB.

514
For injection site quantification, a 30-pixel-wide segmented line was drawn from proximal CA1 515 to distal CA1 using DAPI signal as a guide. For projection quantification in subiculum, a 200-pixel-516 wide segmented line was drawn from proximal subiculum to distal subiculum through the cell body 517 layer using only DAPI as a guide. From this point, injection site and projection images were 518 processed the same. Segmented lines were straightened using the "Straighten" function, background 519 subtraction was performed using the "Subtract" function and intensity values were measured using  Representative images (Figures 1, 3 and 4) were taken using a Zeiss LSM 780 confocal microscope 534 (20× magnification, tile scan, max projection).

535
In Figure 5, the experimental and data analysis procedures were identical to Pederick et al.,  To quantify average axon intensity in GFP+ and adjacent GFP-regions in subiculum targets 540 ( Figure 5F), we restricted analysis to the most distal 20% of the subiculum. To determine the GFP+ 541 region we identified the intensity-weighted central row using the summed fluorescence of each row 542 and determined the minimal symmetric window of rows around the central row that encompassed at 543 least 50% of the total intensity in the restricted GFP image. This defined a rectangle in the original 544 image that we designated as the GFP+ region. We then computed the mean fluorescence intensity in 545 this region for the mCh channel. We used the two rows above and below (lateral and medial) the 546 designated GFP+ region as the adjacent GFP-region and computed the mean mCh fluorescence 547 across these four rows. To determine mCh fluorescence differences in GFP+ versus GFP-regions, 548 we divided the mCh intensity in the GFP+ region by the mCh intensity in the GFP-region for each 549 mouse (i.e., GFP+/GFP-). mCh fluorescence intensity GFP+/GFP-was compared across groups 550 using a one-way ANOVA with Tukey's multiple comparisons test using Prism 9 (GraphPad). Three-551 dimensional mountain plots were generated using the 'surf' function.