Codon Bias Can Determine Sorting of a Potassium Channel Protein

Abstract

Due to the redundancy of the genetic code most amino acids are encoded by multiple synonymous codons. It has been proposed that a biased frequency of synonymous codons can affect the function of proteins by modulating distinct steps in transcription, translation and folding. Here, we use two similar prototype K⁺ channels as model systems to examine whether codon choice has an impact on protein sorting. By monitoring transient expression of GFP-tagged channels in mammalian cells, we find that one of the two channels is sorted in a codon and cell cycle-dependent manner either to mitochondria or the secretory pathway. The data establish that a gene with either rare or frequent codons serves, together with a cell-state-dependent decoding mechanism, as a secondary code for sorting intracellular membrane proteins.

Keywords: codon usage; dual sorting; effect of synonymous codon exchange; membrane protein sorting.

Figure 1

Codon optimization affects the sorting pattern of the Kesv channel. (A,B) Fluorescent images of HEK293 cells transfected with either Kesv_wt (A) or Kesv_op (B). Images show: GFP tagged channels (green, first column), fluorescence from mitochondrial marker (COXVIII::mCherry) (magenta, second column) and merging of magenta and green channels (third column). A magnification from areas marked in overlay images is shown in the fourth column. Letters in the images refer to cytosol (c), nucleus (n) and mitochondria (m). (C) Mean relative distribution (±SD; n ≥ 3) for localization of channels in mitochondria (black), secretory pathway (blue) or unsorted (orange) in HEK293 cells transfected with Kesv_wt (n = 259 cells), Kesv_op (n = 245 cells) or Kesv_ran (n = 120 cells). (D) Mean ratio ± SD of fluorescence intensity in mitochondria versus adjacent cytosol (f_mito/f_cyt) in cells transfected with Kesv_wt or Kesv_op. A Student t-test predicts high statistical significance between the two conditions (*** p < 0.0001). Scale bars 10 µm.

Figure 2

Codon-sensitive sorting is channel specific and not an artifact of the experimental system. (A) Mean relative distribution ± SD for localization of the Kesv channel in mitochondria (black), SP (blue), or unsorted (orange) in HEK293 cells transfected with Kesv_wt. Images as in Figure 1 were analyzed by co-expression of the channel with marker proteins COXVIII::mCherry (n = 4, n = 63 cells) or HDEL::mCherry (n = 3, n = 55 cells) for mitochondria and ER, respectively. In separate experiments mitochondria and ER of Kesv_wt transfected cells were labeled with fluorescent dyes Mito- (n = 6, with ≥ 144 cells per condition) or ER-tracker, respectively (n = 3, ≥116 cells per condition). (B) Mean relative distribution of Kesv_wt and Kesv_op in HEK293 as in (C) from cells transfected transiently with either 0.1 or 1 µg DNA (n = 3 with ≥ 121 cells per treatment). (C) Fluorescent images of HEK293 cells transfected with Kcv_wt or Kcv_op. Images in two top rows show: GFP tagged Kcv channels (green, first column), fluorescence from ER marker HDEL::mCherry (magenta, first and second row) and overlay of magenta and green channels in third column. Images in the third row show overlay of the GFP (green) and COXVIII::mCherry (magenta) channel for HEK293 cells transfected with either Kcv_wt or Kcv_op. Inset: Mean relative distribution (n ≥ 220 cells) for localization of the channel in SP (blue), or unsorted channels (orange) in HEK293 cells transfected with Kcv_wt or Kcv_op. Scale bars 10 µm.

Figure 3

Sensitivity of channel sorting to codon optimization is conserved in mammalian cells. Fluorescent images of HeLa (A) and COS-7 cells (B) transfected with Kesv_wt. In both cell types the channel exhibited either a clear-cut sorting to the mitochondria (top row) or a unsorted phenotype with GFP fluorescence throughout the cell (lower row). Images show: the GFP tagged Kesv channel (green, first column) and fluorescence from mitochondrial marker COXVIII::mCherry (magenta, second column). Merged images are in the third column. (C) Fluorescent images of different mammalian cells transfected with Kesv_op. The images are overlays of GFP fluorescence (green) and fluorescence from mitochondrial marker COXVIII::mCherry (magenta) (HeLa, CHO, COS-7) or from ER marker HDEL::mCherry (HaCaT). Scale bars 10 µm. (D) Mean relative distribution ± SD for localization of the wt channel (Kesv_wt, left) or codon-optimized channel (Kesv_op, right) in mitochondria (black), secretory pathway (blue) or unsorted channels (orange). (Data from n ≥3; n ≥ 56 cells per condition).

Figure 4

Complex sorting of the Kesv channel and its mutant in HEK293 cells transfected with chimera of genes with wt and optimized codons. (A) Schematic domain architecture of the Kesv channel monomer with transmembrane α-helixes including the N-terminal helix (NH), outer (TM1) and inner (TM2) transmembrane domain and pore helix (PH) (central panel) (top) and composition of Chimeras C1 to C8 comprising parts of the Kesv_wt (grey) and Kesv_op genes (red). (B) Mean relative distribution (±SD; n = 3, n ≥ 120 cells per chimera) of channels in mitochondria (black), SP (blue) and unsorted channels (orange) in HEK293 cells transfected with corresponding genes. The green bars represent cells in which the channel was present within the same cell in the mitochondria and in the SP. (C) Fluorescent image of a HEK293 cell transfected with Chimera C4. The images show distribution of GFP tagged chimera (green, left column), mitochondrial marker COXVIII::mCherry (magenta, second column), and an overlay of magenta and green channel (third column). The part indicated in the overlay is magnified in the fourth column with blue arrows and white arrows indicating presence of GFP in SP (white arrow) and mitochondria (blue arrow), respectively. (D) Top: schematic domain architecture as in A indicating position 113 in which the AA motive GML was inserted in TM2. Central: fluorescent images of HEK293 cells transfected with Kesv113GML from wt (113GML_wt) or codon-optimized (113GML_op) gene. Images show: the GFP tagged channel (green, first column) and fluorescence from mitochondrial marker COXVIII::mCherry (magenta, top) or from ER marker, HDEL::mCherry (magenta, down) as well as overlay of magenta and green channels (third column). Bottom: relative distribution of channels in mitochondria (black), SP (blue) and unsorted channels (orange) in HEK293 cells (numbers in brackets) transfected with corresponding genes. Scale bar in (C,D) 10 µm.

Figure 5

Sorting pattern of the Kesv channel as a function of parameters, which can affect protein synthesis. (A) Relative distribution of the Kesv channel in mitochondria from data in 4A as a function of estimated free energy of RNA structures derived for Kesv_wt, Kesv_ran, Kesv_op and chimeras C1-C8. Energies were calculated for channel coding the RNA sequence only. The line shows linear fit with a correlation coefficient of 0.44. (B) Light triggered transcription of Kcv_op channel shifts sorting propensity from SP to mitochondria. Top: confocal images of representative HEK293 cell expressing GFPtagged Kcv_op (green, 1st panel) and stained with MitoTrackerRed (magenta, 2nd panel); merging of green and magenta channels is shown in the 3rd panel with a blow up of the indicated area in the 4th panel. Bottom: relative distribution for localization of the Kcv_op channel in mitochondria, SP, or unsorted channels in HEK293. Protein was expressed in HEK293 cells by conventional transfection (CT) or under control of a light-sensitive EL222 system [25,44]. In the latter case, transcription was induced by a pulsed blue light of 120 µE, which was applied for 16 h prior to imaging. Light pulses were 10 s/20 s on followed by 60s of darkness. Data are from n cells in N independent experiments: CT-Kcv_Op n = 3, n = 237 and EL222-Kcv_Op n = 3, n = 118. Scale bars 10 µm. (C) Sorting of the channel to the three destinations (mitochondria: black, SP: blue and unsorted: orange) was estimated as in Figure 1 in HEK293 cells transfected with Kesv_wt (left panel) and Kesv_op (right panel). Cells were kept either at 37 °C or 25 °C. Lowering the temperature is unfavorable for channel sorting to the mitochondria. The lower temperature favors non-sorting and sorting to the SP. Mean values ± S.D. of n = 3 experiments with ≥ 270 cells per temperature. Color coding is the same as in B.

Figure 6

Sorting pattern of the Kesv channel in HEK293 cells is affected by cell cycle. (A) Analysis of DNA content in HEK293 cells by flow-cytometry in the absence and presence of cell cycle blocker R0-3306. Representative histograms of control cells (blue) and cells pretreated for 48 h with 7 μM RO-3306 (red) as function of Propidium Iodide (PI) intensity. The respective cell cycle phases are indicated by colored bars. (B) Relative distribution (±SD, n = 3, n ≥ 150 cells) of Chimera C1 in mitochondria (black), SP (blue), dual location in mitochondria and SP (green) as well as non-sorted channels (orange) as a function of cells in G2 in control cells (36 ± 3%) and RO-3306 treated cells (74 ± 1.5%). Cells were transfected 24 h after exposure to inhibitor and imaged 24 h later. (C) Mean relative distribution (±SD, n = 3, n ≥ 120 cells) of Kesv channel in mitochondria (black), SP (blue), dual location in mitochondria and SP (green) as well as non-sorted channels (orange) in HEK293 cells transfected with Chimera C1. Cells were supplied in culture medium with the indicated concentrations of glucose.