Challenges of in vivo protein localization in plants seen through the DEK1 protein lens

ABSTRACT

During the last 25 y, fluorescent protein tagging has become a tool of choice to investigate protein function in a cellular context. The information gathered with this approach is not only providing insights into protein subcellular localization but also allows contextualizing protein function in multicellular settings. Here we illustrate the power of this method by commenting on the recent successful localization of the large membrane DEK1 protein during three-dimensional body formation in the moss Physcomitrella patens. But as many approaches, protein tagging is not exempt of caveats. The multiple infructuous (failed) attempts to detect DEK1 using a fluorescent protein tag present a good overview of such potential problems. Here we discuss the insertion of different fluorescent proteins at different positions in the PpDEK1 protein and the resulting unintended range of mutant phenotypes. Albeit none of these mutants generated a detectable fluorescent signal they can still provide interesting biological information about DEK1 function.

The use of a fluorescent protein tag to detect in vivo protein accumulation has revolutionized cell biology.¹In vivo detection of a tagged protein at its genomic native locus under its native promoter currently represents the optimal tool to understand protein function in its cellular and multicellular context. One obvious advantage is elimination of the potential problem of protein over-accumulation encountered using an exogenous promoter to drive protein expression.² Although not widely reported in the literature, gene tagging at the native locus may fail to yield a detectable signal. Factors causing this may include the amount of protein,³ the pH associated with a protein-specific localization^4,5 or interactions with other cellular components that destabilizes the fluorophore.⁶ In addition, adding an endogenous protein tag of 25 to 80 KDa to a protein may generate dominant interactive effects that can interfere with protein activity and potentially affect the interpretation of the observed localization.^7,8

Here we use the recent localization of DEK1 (Defective kernel 1) in the moss Physcomitrella patens⁹ to illustrate some of the pitfalls described above. DEK1 is a large modular transmembrane protein (see Figure 1a for a schematic representation) indispensable for three-dimensional growth (3D) in land plants.^12–14 Although its precise molecular mechanism remains to be elucidated, DEK1 appears to act as a cell surface sensor,¹⁵ potentially through a mechanosensitive mechanism¹⁶ that is transmitted through the activity of its calpain domain.¹³

Figure 1.

DEK1-TOMATO^int displays two distinct subcellular localizations pattern in Physcomitrella patens. (a) Schematic representation of the Physcomitrella patens DEK1 protein showing insertion of the successful fluorescent tag (dek1-tomato^int,⁹) in red and the unsuccessful tag positions in gray (See Table 1 for detailed description of the tagged DEK1-strains.). DEK1 consists of a transmembrane domain with 23 membrane spans (MEM) and a cytoplasmic domain (Linker + Calpain). Domains and subdomains were previously described in detail by Johansen and collaborators.¹⁰ The numbered vertical green boxes represent predicted transmembrane segments. The protein model was generated using IBS.¹¹(b–f) Confocal microscopy average stack projection of dek1-tomato^int gametophytic tissues. Left, right and center panels show DEK1-TOMATO^INT specific signal, chlorophyll auto-fluorescence specific signal and their overlay, respectively. (b–d) Immature phyllid (partial, right) and an immature archegonium (left) displaying the typical signal of growing 3D tissue: a strong subcellular polarized DEK1-TOMATO^INT fluorescent signal at the interface of recently divided cells. Bar: 50 μm. (e–g) dek1-tomato^int developing antheridium (*) showing the similar typical signal of 3D tissue as in (b) and mature antheridium (**) displaying a strong spermatozoid-localized signal whilst the signal in of the antheridial jacket cells disappeared. Bar: 25 μm.

The dek1-tomato^int transgenic strain (Figure 1) containing the tdTomato sequence inserted inside the PpDEK1 genomic locus allowed DEK1 subcellular detection.⁹ Initially, a fluorescent signal was undetectable using standard fluorescent confocal microscopy, leading us to apply the recently developed time-gating confocal fluorescent microscopy.¹⁷ Here the protein DEK1-TOMATO^int signal is detectable from the onset of 3D growth of the gametophore bud and during phyllid development (Figure 1b–d). Notably, the signal is present at the plasma membrane interface of developing cells and absent from the rest of the plasma membrane of the same cell. This strong polarized subcellular signal is also recapitulated in other 3D developing tissues including the developing archegonium (Figure 1b–d) and the developing antheridium (Figure 1e–g). Surprisingly, in view of the established DEK1 function in 3D cell development, DEK1-TOMATO^int is also detected in the single cell maturing spermatozoid (Figure 1e–g). Potentially, the presence of DEK1-TOMATO^int during spermatozoid maturation may indicate a new function of DEK1. Although dek1-tomato^int gametophyte development is morphologically wild-type, these plants are sterile.⁹ At the functional level, the impact of the tag is cell-type specific, indicating that the tag could affect DEK1 function (either structurally or though a modulation of its calpain activity) through the interactions with other protein(s) only present in gametangia. Alternatively, the tag could affect the specific DEK1 activity at a level only necessary for gametangia function. This represents a good example of the non-intended phenotypic impact a tag can have on specific stage of plant development.

However, dek1-tomato^int was only the last of nine versions of tagged PpDEK1 strains we established (Figure 1a and Table 1). The remaining eight not only did not displayed detectable fluorescent signal but also led a variety of phenotypes that illustrate potential impacts of a tag on a protein function and thereby plant development. All PpDEK1 intragenic taggings were performed using the same two-steps transformation approach as for dek1-tomato^int to generate in-frame translational fusions (for methodological details, see Perroud and collaborators, 2020⁹). The C-terminal tagging trials (also referred as C-terminal knock-ins) were performed following standard homologous recombination in locus C-terminal fusion technique¹⁸ used successfully numerous times by different laboratories that work with P. patens, e.g. for KINID1a and b¹⁹ or for Arginyl-tRNA protein transferase.²⁰ Dek1-tomato^int set aside, none of these PpDEK1 tagged transgenic strains displayed any fluorescent signal when observed with standard fluorescence microscopy, standard confocal microscopy, or time-gated fluorescence confocal microscopy (Figure 1a, Table 1).

Table 1.

DEK1 tagging trials.

Name (by domain insertion)	Insertion site (aa of PpDEK1)	Fluorophore type	Gametophore phenotype	Fertility
dek1-tomatoint*	1622	tdTomato	WT	no
mem-1	97	eGFP	Δdek1	no
mem-2	769	mCherry	WT	WT
mem-3	769	tdTomato	WT	WT
linker-4	1119	eGFP	Δdek1	no
linker-5	1622	mCherry	WT	no
calpain-Cter-6	2173	eGFP	narrow phyllid	no
calpain-Cter-7	2173	eGFP-15 aa spacer	narrow phyllid	no
calpain-Cter-8	2173	mCherry	narrow phyllid	no
calpain-Cter-9	2173	3XeGFP	narrow phyllid	no

* For the detailed description of dek1-tomato^int, see.⁹ The gray shading indicates the detectable fluorescent signal unique to this transgenic.

Mem-2 and mem-3 represent probably the most frustrating failure to generate a localization strain. These strains, tagged at the same position in PpDEK1 with either mCherry or tdTomato sequence did not show any developmental stage defects and were furthermore fully fertile. Additionally, transcript amplification showed the proper splicing of the gene and the presence of the tag sequences. This indicates that even in the best-case scenario from a developmental perspective, a fluorescent signal may not be always observed. The fluorescent protein tag could be affected during the DEK1 initial protein folding, the DEK1 structure, or by other interacting protein(s) that disturb enough the fluorophore structural integrity to abolish its fluorescent property. Mem-1 and linker-4 illustrate the case in which adding a tag lead to the unfortunate result of generating a null mutant phenotype of the tagged protein. Mem-1 and linker-4 did not recover any developmental steps from the Δdek1 phenotype¹⁴ upon Cre recombinase-mediated removal of the resistance cassette from primary transformants. In these two cases, the addition of the fluorescent protein DNA sequence at these specific loci could yield improper gene splicing²¹ and thus to the absence of functional DEK1. Alternatively, the presence of a tag at these specific positions could impede totally DEK1 protein synthesis or its further function and lead to the Δdek1 phenotype. The case of Linker-5 points to the importance of the fluorescent tag type to detect a signal. The linker-5 fluorescent protein insertion site is identical to the one used in dek1-tomato^int and similarly to it,⁹ proper transcript splicing and tag sequence presence could be confirmed. But, no above background fluorescent signal was detected with a mCherry monomer in contrast to dek1-tomato^int (Figure 1b–g) that contains a dimer of the same fluorescent protein. Note here that linker-5, like dek1-tomato^int was sterile, confirming the importance of this amino acid site for fertility regardless of the tag type. Finally, general organ morphology can also be impacted by the presence of a tag. The PpDEK1 C-terminal tagging trials, regardless of the fluorophore types (monomeric GFP with different spacer length, monomeric mCherry, trimeric GFP) led uniformly yet to another phenotype. Not only these transformants were all sterile, but also the phyllid shape was altered to displays narrow phyllid, a similar phenotype observed previously with the deletion of the PpDEK1 LG3 domain.¹⁰ At this position, the tag may not necessarily affect DEK1 synthesis or structure but could impact the calpain domain catalytic activity as similarly suggested for Δlg3 by Johansen and collaborators.¹⁰

Taken together these results show the importance of differential in vivo tagging experiments. Successful results as DEK1-TOMATO^int polar localization in cells of 3D organs can comfort and improve the understanding of specific mechanism such as here the multicellular 3D development in land plant. Additionally, DEK1 localization during spermatogenesis shows that this approach can point to uncharted function and open new research avenue. Nevertheless, it is clear that negative results can occur with this approach due potentially to the choice of the fluorescent protein type or its position in any given protein. Moreover, the impact of the tag on the normal plant development can vary with its insertion position. So frustrating these later results are in term of protein detection, they remain useful as they can give functional information about the tagged site or domain.

Funding Statement

This work was supported by the Slovak Research and Development Agency [APVV-17-0570].

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.